Physical and Chemical

Hydrogeology

Second Edition

P a t r i c k R Domenico
David R Harrls Professor of Geology
Texas ACM University

Franklin 'vis: Scbwartz

Ohio Eminent Scholar in Hydrogeology
The Obi0 State University

John Wiby G Sons, Inc.

New York C bfcbester Weinbeim Brisbane Toronto Singapore
To Diane and Cynthia and
the Memory of Lucy, Phil, and Daniel

ACQUISITIONS EDITOR Cliff Mills

MARKETING MANAGER Kimberly Manzi
PRODUCTION EDITOR Deborah Herbert
DESIGNER AM Mane Renzi
ILLUSTRATION EDITOR Edward Starr
COVER Designed by Carolyn Joseph Photo: Tony Stone Worldwide/David Carriere

This book is printed on acid-free paper. @

The paper in this book was manufactured by a mill whose forest management programs include sustained
yield harvesting of its timberlands. Sustained yield harvesting principles ensure that the numbers of trees
cut each year does not exceed the amount of new growth.

Copyright 0 1990, 1998John Wdey & Sons, lnc. All rights reserved.
No part of this publication may be reproduced, stored in a retrieval system or transmitted
in any form or by any means, electronic, mechanical, photocopying, recording, scanning
or otherwise, except as permitted under Sections 107 or 108 of the 1976 United States
Copyright Act, without either the prior written permission of the Publisher, or
authorization through payment of the appropriate percopy fee to the Copyright
Clearance Center, 222 Rosewood Drive, Danvers, MA 01923, (508) 750-8400, fax
(508) 750-4470. Requests to the Publisher for permission should be addressed to the
Permissions Depamnent, John Wifey & Sons, Inc.. 605 Third Avenue, New York, NY
101580012, (212) 8506011, fax (212) 8506008, E-Mail: [email protected].

Library of Congress CutaJoging in Publication Data

Domenico, P. A. (Patrick A.)

Physical and chemical hydrogeology / Patrick A. Domenico, Franklin
W. Schwaru.-2nd ed.
p. cm.
Includes bibliographical references and index.
ISBN 0471-59762-7 (cloth : alk. paper)
I . Hydmgeology. 1. Schwartz, F. W. (Franklin W.) 11. Title.
GB1003.2.D66 1997
551.49-dc21 97-21776
10 9 8 7 6 5 4 CIP
To Diane and Cynthia and
the Memory of Lucy, Phil, and Daniel

ACQUISITIONS EDITOR Cliff Mills

MARKETING MANAGER Kimberly Manzi
PRODUCTION EDITOR Deborah Herbert
DESIGNER AM Mane Renzi
ILLUSTRATION EDITOR Edward Starr
COVER Designed by Carolyn Joseph Photo: Tony Stone Worldwide/David Carriere

This book is printed on acid-free paper. @

The paper in this book was manufactured by a mill whose forest management programs include sustained
yield harvesting of its timberlands. Sustained yield harvesting principles ensure that the numbers of trees
cut each year does not exceed the amount of new growth.

Copyright 0 1990, 1998John Wdey & Sons, lnc. All rights reserved.
No part of this publication may be reproduced, stored in a retrieval system or transmitted
in any form or by any means, electronic, mechanical, photocopying, recording, scanning
or otherwise, except as permitted under Sections 107 or 108 of the 1976 United States
Copyright Act, without either the prior written permission of the Publisher, or
authorization through payment of the appropriate percopy fee to the Copyright
Clearance Center, 222 Rosewood Drive, Danvers, MA 01923, (508) 750-8400, fax
(508) 750-4470. Requests to the Publisher for permission should be addressed to the
Permissions Depamnent, John Wifey & Sons, Inc.. 605 Third Avenue, New York, NY
101580012, (212) 8506011, fax (212) 8506008, E-Mail: [email protected].

Library of Congress CutaJoging in Publication Data

Domenico, P. A. (Patrick A.)

Physical and chemical hydrogeology / Patrick A. Domenico, Franklin
W. Schwaru.-2nd ed.
p. cm.
Includes bibliographical references and index.
ISBN 0471-59762-7 (cloth : alk. paper)
I . Hydmgeology. 1. Schwartz, F. W. (Franklin W.) 11. Title.
GB1003.2.D66 1997
551.49-dc21 97-21776
10 9 8 7 6 5 4 CIP
 Preface

One main objective in producing this Second Edition of make each chapter as self sufficient as possible. This
Physical and Chemical Hydrogeology was to incorpo- Edition, like the previous one, deliberately contains more
rate the new, broadly-based scientific advances that have material than can be covered in a single course so that
been evident in the field. Most noteworthy in this respect choices have to be made by the instructor. We hope that
is new knowledge on ground-water microbiology, theo- this reorganization makes the task of choice easier.
retical and practical knowledge related to contamination We acknowledge the special contribution of Dr. Ste-
by NAPLs and DNAPLs and multiphase fluids in general, phen Worthington of McMaster University, and the late
new strategies for site clean ups, and risk assessment as Jim Quinlan. They helped by writing sections dealing
a tool for making decisions about contaminated sites. with karst in Chapters 2 and 16. Like all professors, we
However, we have not introduced these new concepts at benefit from the continuing help and assistance of pres-
the expense of other essential material- both traditional ent and former students. Dr. Hubao Zhang worked to
and modern-that makes up the essence of hydrogeol- expand our treatment of well hydraulics through the
ogy. As with the previous edition, the transport of fluid, addition of the code, WELLz, which is included on disk
enerky, and mass in porous media remains the guiding at the back of the book. Dr. Dea-Ha Lee and Dr. Abe
theme throughout the book. Additionally, the Second Springer commented on Chapter 7, with Dr. Springer
Edition preserves the process oriented focus of the origi- providing MODFLOW data for his study site in Ohio.
nal book and continues to emphasize the relationship Dr. Rob Schincariol kindly provided computer generated
between theory and practice. We have again attempted to hydraulic conductivity fields in Chapter 10. Dr. Alan Fryar
retain an understandable style while explaining complex contributed section 16.4 that deals with self-organizing
hydrogeological matters. To this end, as in the previous systems. Gordon McClymont inspired the sections on risk
edition, we include a significant number of worked exam- assessment, and he and Hubao Zhang helped in reviewing
ples and a problem set for most of the chapters. The this material. Our editors at John Wiley 81 Sons have
hook is still intended for students at the advanced under- been extremely supportive and worked to modernize the
graduate or beginning graduate level. layout of the book.
The Second Edition also reflects thoughtful suggestions
and criticisms of colleagues who made clear what ele- Patrick A. Domenico
ments of the original book worked or didn’t work. Clearly Franklin W. Schwartz
some reorganization was called for with an attempt to
 Contents

Chapter 1 Chemical Rock- Water Interactions: Secondary

Introduction 1 Porosity in Sandstones 26
1.1 What Is Hydrogeology? 1 2.5 Uplift, Diagenesis, and Erosion 27
Physical Hydrogeology Before the Early 1940s 2 The Style of Formations Associated with
Chemical Hydrogeology Before the Early 1960s 3 Uplift 27
Post-1960 Hydrogeology 4 Secondary Porosity Enhancement in Carbonate
1.2 The Relationship Between Hydrogeology and Rocks 29
Other Fields of Geology 4 2.6 Tectonism and the Formation of
1.3 Hydrologic Cycle 5 Fractures 29
Components of the Hydrologic Cycle 5 Style of Fracturing 30
Evapotranspiration and Potential Fluid Pressure and Porosity 31
Evapotranspiration 7 Connectivity 31
Infiltration and Recharge 8
Base Flow 8
Hydrologic Equation 10 Cbapter 3
Ground-Water Movement 33
Cbapter 2 3.1 Darcy’s Experimental Law and Field
Tbe W g i n of Porosity and Permeability 13 Extensions 33
2.1 Porosity and Permeability 13 The Nature of Darcy’s Velocity 34
Porosity and Effective Porosity 13 Hydraulic Head: Hubbert’s Force Potential 34
Permeability 15 The Gradient and Ground-Water Flow 36
2.2 Continental Environments 16 Physical Interpretation of Darcy‘s Proportionality
Weathering 16 Constant 36
Erosion, Transportation, and Deposition 17 Units and Dimensions 37
Fluvial Deposits 17 3.2 Hydraulic Conductivity and Permeability of
Eolian Deposits 20 Geologic Materials 37
Lacustrine Deposits 20 Observed Range in Hydraulic Conductivity
Glacial Deposits 20 Values 37
2.3 The Boundary Between Continental and Character of Hydraulic Conductivity
Marine Environments 20 Distribution 38
2.4 Marine Environments 21 Anisotropicity and HeterogeneiQ Within
Lateral and Vertical Succession of Strata 21 Units 39
Ancestral Seas and Their Deposits 22 Heterogeneity Among Units and the C1;issification
The Paleozoic Rock Group 23 of Aquifers 41
The Mesozoic Rock Group 24 Creating Hydraulic Conductivity Averages 4 2
The Cenozoic Rock Group 24 Darcy’s Law for Anisotropic Material 43
Diagenesis in Marine Environments 24 Measurement of Hydraulic Conductivity 44
Porosity Reduction: Compaction and Presslire Laboratory Testing 44
Solution 24 The Search for Empirical Correlations 44
-
ViU Contents

3.3 Mapping Flow in Geologkal Systems 45 5.2 Surface Features of Ground-Water Flow 91
Hydrogeological Cross Sections 46 Recharge-Discharge Relations 91
Potentiometric Surface and Water-Table Maps 47 Ground Water-Lake Interactions 93
Closing Statements 48 Ground Water-Surface Water Interactions 95
3.4 Flow in Fractured Rocks 48 5.3 Some Engineering and Geologic Implications
Continuum Approach to Fluid Flow 48 of Topographic Drive Systems 97
Intergranular Porous Rocks 49 Large Reservoir Impoundments 97
Fractured Rocks 49 Excavations: Inflows and Stability 98
3.5 Flow in the Unsaturated Zone 51 The Sea-Level Canal 98
Hydraulic and Pressure Heads 51 Ground-Water Inflows into Excavations 99
Water Retention Curves 53 The Stability of Excavations in Ground-Water
Discharge Areas 99
Darcy's Law for Variably Saturated Flow 54
Unsaturated Flow in Fractured Rocks 54 Landslides and Slope Stability 101

Chapter 4 Chapter 6
Main Equations of Flow, Boundary Hydraulic Testing: Models, Methods, and
Conditions, and Flow Nets 50 Applications 103
4.1 Organizing the Study of Ground-Water Flow 6.1 Prototype Geologic Models in Hydraulic
Equations 58 Testing 103
4.2 Conservation of Fluid Mass 59 6.2 Conventional Hydraulic Test Procedures and
Main Equations of Flow 60 Analysis 105
4.3 The Storage Properties of Porous Media 62 The Theis Nonequilibrium Pumping Test
Compressibility of Water and Its Relation to Method 105
Specific Storage for Confined Aquifers 63 The Curve-Matching Procedure 107
Compressibility of the Rock Matrix: Effective Assumptions and Interpretations 107
Stress Concept 64 Modifications of the Nonequilibrium
Matrix Compressibility and Its Relation to Equation 108
Specific Storage of Confined Aquifers 65 Time-Drawdown Method 108
Equation for Confined Flow in an Aquifer 67 Distance-Drawndown Method 109
Specific Yield of Aquifers 68 Steady-State Behavior as a Terminal Case of the
4.4 Boundary Conditions and Flow Nets 68 Transient Case 109
Graphic Flow Net Construction 71 The Hantush-Jacob Leaky Aquifer Method 110
Water Table Aquifers 1 12
4.5 Dimensional Analysis 72
6.3 Single-Borehole Tests 114
Recovery in a Pumped Well 114
Cbapter 5 The Drill Stem Test 114
Ground Water in the Bash Hydrologic Slug Injection or Withdrawal Tests 115
cycle 7 5 Response at the Pumped Well: Specific Capacity
5.1 Topographic Driving Forces 75 and Well Efficiency 116
The Early Field Studies 75 6.4 Partial Penetration, Superposition, and
Conceptual, Graphical, and Mathematical Models Bounded Aquuers 118
of Unconfined Flow 76 Partial Penetration 118
Effects of Basin Geometry on Ground-Water Principle of Superposition 118
Flow 78
Bounded Aquifers 120
Effects of Basin Geology on Ground-Water Flow 80
Ground Water in Mountainous Terrain 83 6.5 Hydraulic Testing in Fractured or Low-
Ground Water in Carbonate Terrain 87 Permeability Rocks 122
Ground Water in Coastal Regions 88 Single-Borehole Tests 123
The Fresh Water-Salt Water Interface in Coastal Multiple-Borehole Tests 123
Regions 89 6.6 Some Applications to Hydraulic
The Ghyben-Herzberg Relation 89 Problems 124
The Shape of the Interface with a Submerged Screen Diameter and Pumping Rates 125
Seepage Surface 90 Well Yield: The Step-Drawdown Test 125
Upconing of the Interface Caused by Pumping A Problem in Dewatering 125
Wells 91 A Problem in Water Supply 127
 Contents &
l

6.7 Computer-Based Calculations 128 8.2 Abnormal Fluid Pressures in Active

Code Demonstration 131 Depositional Environments 172
Bounded Aquifers Revisited 131 Origin and Distribution 172
Mathematical Formulation of the Problem I74
Cbapter 7 lsothermal Basin Loading and Tectonic
Ground Water as a Resource 136 Strain 175
7.1 Development of Ground-Water One-Dimensional Basin Loading 176
Resources 136 Extensions of the One-Dimensional Loading
The Response of Aquifers to Pumping 136 Model 177
Yield Analysis 137 Thermal Expansion of Fluids 179
Case Study: The Upper Los Angeles River Area 137 Fluid Pressures and Rock Fracture 182
Management Strategies 139 Phase Transformations 183
Artificial Rechwge 139 Subnormal Pressure 184
Conjunctive Use 141 Irreversible Processes 185
7.2 Introduction to Ground-Water Flow 8.3 Pore Fluids in Tectonic Processes 185
Simulation 142 Fluid Pressures and Thrust Faulting 185
Generalized Modeling Approach 142 Seismicity Induced by Fluid Injection 186
Conceptual Model 142 seismicity Induced in the Vicinity of
Ground-Water Flow Simulation 143 Reservoirs 187
Evaluation of Model Results 144 Seismicity and Pore Fluids at Midcrustal
Model Verification 144 Depths 188
A Note of Caution 144 The Phreatic Seismograph: Earthquakes and
7.3 Formulating a Finite-DifferenceEquation for Dilatancy Models 188
Flow 145
Description of the Finite-Difference Grid 145 Chapter 9
Derivation of the Finite-Difference Equation 146 Heat Transport in Ground-Water Flow 191
7.4 The MODFLOW Family of Codes 147 9. I Conduction, Convection, and Equations of
Solving Systems of Finite-Difference Heat Transport 191
Equations 148 Fourier's Law 192
Modular Program Structure 148 Convective Transport 193
Illustrative Example 148 Equations of Energy Transport 194
Operational Issues 152 The Heat Conduction Equation 195
Time-Step Size 152 The Conductive-Convective Equation 195
Drawdowns at "Pumping" Nodes 152 Dimensionless Groups 196
Water-Fable Conditions 153 9.2 Forced Convection 197
Boundary Conditions 153 Temperature Profiles and Ground-Water
7.5 Case Study in the Application of Velocity 197
MODFLOW 154 Heat Transport in Regional Ground-Water
Model Development 154 Flow 199
Data Preparation and Model Calibration 156 Heat Transport in Active Depositional
Environments 203
Chapter 8 Heat Transport in Mountainous Terrain 205
Stress, Strain, and Pore Fluids 159 9.3 Free Convection 207
8.1 Deformable Porous Media 159 The Onset of Free Convection 207
One-Dimensional Consolidation 159 Sloping Layers 208
Development of the Flow Equation 159 Geological Implications 208
The Undrained Response of Water Levels to Natural
Loading Events 160 9.4 Energy Resources 209
The Drained Response of Water Levels to Natural Geothermal Energy 209
Loading Events 163 Energy Storage in Aquifers 209
Land Subsidence as a One-Dimensional Drained 9.5 Heat Transport and Geologic Repositories
Response 163 for Nuclear Waste Storage 210
Mathematical Treatment of Land Subsidence 165 The Nuclear Waste Program 210
Three-Dimensional Consolidation 169 The Rock Types 2 10
Elastic Properties in Deformational Problems 160 Thermohydrochemical Effects 2 12
Flow Equations for Deformable Media 171 Thermomechanical Effects 2 13
X_ Contents

Cbapter 10 Chapter 12
Solute Transport 215 Chemical Reactions 255
10.1 Advection 215 12.1 Acid-Base Reactions 255
10.2 Basic Concepts of Dispersion 216 Natural Weak Acid-Base Systems 256
Diffusion 218 C0,-Water System 256
Mechanical Dispersion 219 Alkalinity 257
10.3 Character of the Dispersion 12.2 Solution, Exsolution, Volatilization, and
Coefficient 220 Precipitation 258
Studies at the Microscopic Scale 220 Gas Solution and Exsolution 258
Dispersivity as a Medium Property 221 Solution of Organic Solutes in Water 258
Studies at Macroscopic and Larger Scales 221 Volatilization 2 59
10.4 A Fickian Model of Dispersion 223 Dissolution and Precipitation of Solids 262
Solid Solubility 262
10.5 Mixing in Fractured Media 226
12.3 Complexation Reactions 263
10.6 A Geostatistical Model of Dispersion 228
Srdbility of Complexes and Speciation
Mean and Variance 228
Modeling 263
Autocovariance and Autocorrelation
Major Ion Complexation and Equilibrium
Functions 229
Calculations 264
Generation of Correlated Random Fields 230
Enhancing the Mobility of Metals 265
Estimation of Dispersivity 230
Organic Complexation 265
10.7 Tracers and Tracer Tests 231
12.4 Reactions on Surfaces 266
Field Tracer Experiments 232
Sorption Isotherms 266
Natural Gradient Test 232
Single Well Pulse Test 232 Hydrophobic Sorption of Organic
Two-Well Tracer Test 233 Compounds 267
Single Well Injection or Withdrawal with Multiple &-based Approaches for Modeling the Sorption
Observation Wells 233 of Metals 269
Estimates from Contaminant Plumes and Multiparameter Equilibrium Models 269
Environmental Tracers 233 12.5 Oxidation-Reduction Reactions 272
Massively Instrumented Field Tracer Tests 233 Oxidation Numbers, Half-Reactions, Electron
Borden Tracer Experiment 234 Activity, and Redox Potential 272
Validation of the Stochastic Model of Kinetics and Dominant Couples 275
Dispersion 235 Control on the Mobility of Metals 276
Biotransformation of Organic Compounds 276
Chapter I1
12.6 Hydrolysis 277
Principles of Aqueous Geocbernistry 238
12.7 Isotopic Processes 277
11.1 Introduction to Aqueous Systems 238
Radioactive Decay 277
Concentration Scales 239
Isotopic Reactions 278
Gas and Solid Phases 240
Deuterium and Oxygen-18 279
11.2 Structure of Water and the Occurrence of
Mass in Water 240
11.3 EquilSbrium Versus Kinetic Descriptions of
Chapter 13
Reactions 240
Colloids and Microorganisms 282
Reaction Rates 241 13.1 A Conceptual Model of Colloidal
Transport 282
11.4 Equilibrium M o d e l s of Reaction 241
Occurrence of Colloidal Material 283
Activity Models 242
Stabilization 283
11.5 Deviations from Equilibrium 243 Transport and Filtration 284
11.6 Kinetic Reactions 244 13.2 Colloidal Transport in Grour Water 284
11.7 Organic Compounds 245 Sampling and Measuring 284
11.8 Ground-Water Composition 248 Studies at Cape Cod 285
The Routine Water Analysis 248 13.3 Microbiological Systems 285
Specialized Analyses 249 Biofilms 287
11.9 Describing Chemical Data 250 Sampling and Enumerating Microbial
Abundance or Relative Abundance 252 Populations 287
Abundance and Patterns of Change 253 Plate Counts 288
 Contents xi
Direct Counting Procedures 288 Carbon-I4 320
Biochemical Techniques 288 Chlorine-36 322
Rates of Microbial Reactions 288 Indirect Methods 322
Microbial Ecology of the Subsurface 290 6IHOand SD 322
13.4 Microbial Processes 291 Chlorofluorocarbons 322
Issues in Biodegradation 292
Biofilm Kinetics 292 Chapter 16
13.5 Biotransformation of Common Mass Transport in Ground-Wuter Flow:
Contaminants 293 Geologic Systems 326
Hydrocarbons and Derivatives 293 16.1 Mass Transport in Carbonate Rocks 326
Halogenated Aliphatic Compounds 294 The Approach Toward Chemical Equilibrium in
Halogenated Aromatic Compounds 294 Carbonate Sediments 327
Polychlorinated Biphenyls (PCBs) 295 The Problem of Undersaturation 329
Complex Transformation Pathways 295 Dolomitization 330
16.2 Economic Mineralization 330
Cbapter 14 Origin of Ore Deposits 331
Tbe Equations of Mass Transport 296 Roll-Front Uranium Deposits 33 1
14.1 Mass Transport Equations 296 Mississippi Valley-Type Lead-Zinc Deposits 332
The Diffusion Equation 296 Noncommercial Mineralization: Saline Soils and
The Advection-Diffusion Equation 297 Evaporites 337
The Advection-Dispersion Equation 297 16.3 Migration and Entrapment of
14.2 Mass Transport with Reaction 298 Hydrocarbons 337
First-Order Kinetic Reactions 298 Displacement and Entrapment 337
Equilibrium Sorption Reactions 299 Basin Migration Models 339
Heterogeneous Kinetic Reactions 299 16.4 Self-Organizationin Hydrogeologic
14.3 Boundary and Initial Conditions 300 Systems 341
Patterning Associated with Dissolution 34 1
Patterning Associated with Precipitation and
Chapter 15 Mixed Phenomena 34 1
Mass Transport in Natural Ground-Water
Systems 303
15.1 Mixing as an Agent for Chemical
Chapter I 7
Change 303
Introduction to Contaminant
The Mixing of Meteoric and Original Formation
Hydrogeology 344
Waters 303 17.1 Sources of Ground-Water
Diffusion in Deep Sedimentary Contamination 344
Environments 305 Radioactive Contaminants 346
15.2 Chemical Reactions in the Unsaturated Trace Metals 347
Nutrients 349
Zone 306
Other Inorganic Species 349
Gas Dissolution and Redistribution 306
Organic Contaminants 349
Weak Acid-Strong Base Reactions 307
Petroleum Hydrocarbons and Derivatives 349
Sulfide Oxidation 309
Halogenated Aliphatic Compounds 350
Gypsum Precipitation and Dissolution 309 Halogenated Aromatic Compounds 350
Cation Exchange 309 Polychlorinated Biphenyls 350
Organic Reactions 309 Health Effects 350
15.3 Chemical Reactions in the Saturated Biological Contaminants 350
Zone 310 17.2 Solute Plumes as a Manifestation of
Weak Acid-Strong Base Reactions 3 10 Processes 352
Dissolution of Soluble Salts 312 Fractured and Karst Systems 357
Redox Reactions 312 Babylon, New York, Case Study 357
Cation Exchange 3 15 Alkali Lake, Oregon, Case Study 359
15.4 Case Study of the Milk River Aquifer 316 17.3 Design and Quality Assurance Issues in
15.5 Age Dating of Ground Water 319 Solute Sampling 360
Direct Methods 319 Design of Sampling Networks 360
Tritium 319 Assuring the Quality of Chemical Data 362
X
# Contents

17.4 Sampling Methods 362 A Case Study of Gasoline Leakage 404

Conventional Wells or Piezometers 362 Hyde Park Landfill Case Study 404
Multilevel Samplers 363 19.3 Partitioning 405
Solid and Fluid Sampling 364 19.4 Fate of Organics in the Unsaturated
Cone Penetrometry 365 Zone 406
Other Sampling Methods 367 Volatilization 406
Dissolved Contaminants in the Unsaturated Gas Transport by Diffusion 408
Zone 367 Equilibrium Calculations of Mass
17.5 Indirect Methods for Detecting Distributions 409
Contamination 367 Mass of VOC in Gas Phase 41 1
Soil-Gas Characterization 367 Mass of VOC in Aqueous Phase 41 1
Geophysical Methods 368 Mass of VOC in Sorbed Phase 41 1
Electrical Methods 369 Mass of VOC in NAPL Phase 41 1
Ground-Penetrating Radar 370 19.5 Fate of Organics in the Saturated Zone 411
Magnetometry 371 Equilibrium Calculations of Mass
Seismic Methods 371 Distribution 4 12
19.6 &-Permeability Testing 412
Cbapter 18 19.7 Recognizing DNAPL Sites 413
Modeling tbe Transport of Dissolved Systematic Screening Procedure 414
Contaminants 372
18.1 Analytical Approaches 372 Chapter 20
Advection and Longitudinal Dispersion 372 Remediation: Overview and Removal
The Retardation Equation 375 Options 417
Radioactive Decay, Biodegradation, and 20.1 Containment 417
Hydrolysis 376 Shirr)' Walk 417
Transverse Dispersion 377 Sheet Pile Walls 418
Models for Multidimensional Transport 378 Grouting 418
Continuous Sources 378
Surface Seals and Surface Drainage 418
Numerical Integration of an Analytical
Hydrodynamic Controls 419
Solution 380
The Instantaneous Point Source Model 380 Stabilization and Solidification 420
18.2 Programming the Analytical Solutions for 20.2 Management Options 420
Computers 382 20.3 Overview of Methods for Contaminant
18.3 Numerical Approaches 384 Removal 420
A Generalized Modeling Approach 384 EXclvdtion and Ex Situ TrCdtmellt 421
The Common Solution Techniques 385 Pump and Treat 421
Adding Chemical Reactions 386 Interceptor Systems 421
Soil-Vapor Extraction 421
18.4 Case Study in the Application of a
Numerical Model 386 20.4 Pump and Treat 422
The Problem of Pump and Treat 422
Technical Consideritions with
Cbapter 19
Muhipbase Fluid Systems 393 Injection-Recovery Systems 423
Methods for Designing Pumpand-Treat
19.1 Basic Concepts 393 Systems 425
Saturation and Wettability 393 Expanding Pilot-Scale Systems 425
Interfacial Tension and Capillary Forces 394 Capture 7 ~ n e s426
Imbibition and Drainage 394 Analytical Approaches to Defining Capture
Relative Permeability 395 Zones 426
Solubility and Effective Solubility 397 Model-Rased Apprimches for the Design of
19.2 LNAPLsandDNAPLs 398 Recovery Systems 429
Conceptual Models for the Occurrence of Simulation-Optimization Techniques 429
LNAPLs 399 Issues in tlie Design of Capture Zones 430
Occurrence of DNAPLs in Ground Water 401 20.5 Interceptor Systems for NAPL
Secondary Contamination Due to NAPLs 401 Recovery 430
Conceptual Models and Quantitative 20.6 Soil-VaporExtraction 431
Methods 403 Components of an SVE System 432
 Contents XA

When Can SVE Systems Be Used? 433 21.3 Abiotic Chemical Destruction 449
Estimating Removal Rates 434 Reactive Barrier Systems 450
Removal Rate Calculations 435 Funnel-andGate Systems 450
Field Estimates of Soil Permeability 435 21.4 RiskAssessment 450
Heterogeneity and the Efficiency of SVE Data Collection and Data Evaluation 451
Systems 436 Exposure Assessment 451
20.7 AitSparging 437 Toxicity Assessment 453
Airflow and Channeling 437 Health-Risk Assessment 454
Designing Air-Sparging Systems 438 Types of Risk Assessments 455
20.8 Case Studies in Site Remediation 438 Environmental Risk Assessment 456
Oil Spill: Calgary, Alberta 438 21.5 Fernald Case Study 456
Gilson Road: Nashua, New Hampshire 439 Detailed Risk Assessment 457
Hyde Park Landfill: Niagara Falls, New York 440
Groveland Wells Site, Massachusetts 44 1 Answers to Problems 461
Appendix A
Chapter 21 Derivation of the Flow Equation in a
In Situ Destruction and Risk Deforming Medium 463
Assessment 443
21.1 Intrinsic Bioremediation 443 Appendix B
21.2 Bioventing and Bioslurping 445 About the Computer Disk 464
Applicability of the Technology to Contaminant Appendix C
Groups 446 Table of Atomic Weights 466
Requirements for Success with Bioventing
Systems 446 References 468
In Situ Respiration Testing 447
Progress in Solvent Bioremediation 448 Index 494
 Chapter 1

Introduction

1.1 What I s Hydrogeology?

1.2 The Relationship Between Hydrogeolugy
and Other Fields of Geology
1.3 The Hydrologic Cycle

This book was written from the perspective that the to the concepts of science than those offered by the
reader is interested in becoming a hydrogeologist. We study of hydrogeology. The discipline blends field, exper-
recognize that this is not likely to be the case generally. imental, and theoretical activities. Sometimes the experi-
However, for us to take a different perspective would mental activities take place in the laboratory; at other
require a different kind of book, one that is perhaps not times they take place in the field, where the comfort
as detailed or as rigorous. And this would be unfair to of laboratory control is lost. Additionally, the field and
all students in several different ways. First, the student experimental activities often play a major role in the
interested in hydrogeology as a career is entitled to know formulation of reliable theoretical models of processes
the breadth of the field and the fact that there is much and events. In increasingly more cases, the methods of
to study, so much in fact that it cannot all be learned in hydrogeology give us the wherewithal to study and per-
one course. Additional courses will be required, not only haps quantlfy some of nature’s experiments. These same
in hydrogeology but in supporting sciences such as soil methods help us study, quantify, and sometimes rectify
physics, soil mechanics, geochemistry, and numerical some unfortunate “experiments” of an industrialized so-
methods. Graduate training is an essential requirement. ciety. All of this, and much more, is hydrogeology.
Additionally, all students are entitled to know that hydro-
geology contains not only geology but a heavy dose of
physics and chemistry as well, and its main language is
mathematics. Thus, the entry level to the field is high,
-
1.1 What I s Hydrogeology?
but so are the intellectual and practical rewards. For earth We have chosen to call this book Physical and Chemical
scientists not interested in hydrogeology as a career, let Hydrogeology for two reasons. First, this is what the
us state outright that virtually every activity in the earth book and the field are about. Second, most other potential
sciences requires some knowledge of subsurface fluids, candidate titles have been preempted. The simple title
and rock-water interactions in particular, and that is Ground Water is appealing and was used by Tolman in
what hydrogeology and this book are about. 1937 and by Freeze and Cherry in 1979. The books are
Last, from our own biased perspective, we think it quite different in scope because during the period 1035
would be difficult to find a more rewarding introduction to the present, the field of knowledge was undergoing
2, Cbapter 1 / lntroductdon

tremendous expansion and this expansion is expertly logic cycle suited his interests well. It still remains a good
captured by Freeze and Cherry (1979). Lamarck wrote a definition, but it does not go far enough. We will offer
book entitled Hydrogeology in 1802, as did Davis and a definition after we develop an understanding of what
Dewiest in 1966. Obviously, due to the times, there is the field is today and how it got that way, and who the
no similarity between the subject matter of these books. major players were. Only then will we become aware of
A book entitled Geohydrologywas prepared by Dewiest the scope of hydrogeology.
in 1965 and one entitled Groundwater Hydrology was
written by Todd, first in 1959 with a later edition ap- Physical Hydrogeology
pearing in 1980.
Before the Early 1940s
Some people remain uncomfortable with the situation
where the subject matter of subsurface fluids can be The turn of the century was an exciting time for hydroge-
organized under a variety of titles and taught under a ologists, especially those inclined to the rigors of field-
variety of disciplines. The meaning of the terms hydroge- work. Their main tools were rock hammers, compasses,
ology and geohydrofogyin particular have caused some and some crude water-level or fluid-pressure measuring
debate. Frequently it is stated that the former deals with devices. These hydrogeologists were likely aware of two
the geologic aspects of ground water whereas the latter important findings of the previous century. First was the
places more emphasis on hydraulics and fluid flow. This experimental work of Henry Darcy in 1856 providing a
arbitrary division is no longer taken seriously by most law that described the motion of ground water, and sec-
people in the field. In fact, the term hydrogeology was ond was some work by T. C. Chamberlin in 1885 that
defined long before the modern era in hydrogeology, described water occurrence and flow under “artesian”
which differs markedly from its early beginnings, and conditions. Armed with these details, these workers were
both definitions likely reflect the special interest of their busy delineating the major water-yielding formations in
promulgators. In 1919, Mead published a book on hydrol- North America and making important measurements of
ogy where he defined hydrogeology as the study of the the distribution of hydraulic head within them. It was a
laws of Occurrence and movement of subterranean water. good time to be a field hydrogeologist, with the emphasis
There is nothing wrong with this definition. However, being on exploration and understanding the occurrence
Mead stressed the importance of ground water as a g e e of ground water and its interrelationship with other com-
logic agent, especially as it contributes to an understand- ponents of the hydrologic cycle.
ing of rivers and drainage systems. As a hydrologist inter- In addition, there was some abstract thinking occurring
ested largely in surface phenomena, this emphasis was at this time, but it had little or no impact over the follow-
well suited to Mead’s interests. In 1942, Meinzer edited ing three or four decades. Slichter (1899) in particular
a book called Hydrology, which he defined within the did original theoretical work on the flow of ground water,
context of the hydrologic cycle, that is, the march of but he was several decades ahead of his time. King
events marking the progress of a particle of water from (18991, too, attempted to provide some calculations to
the ocean basins to the atmosphere and land mdSSeS and support his field findings.
back to the ocean basins. He divided this science into The culmination of this era probably occurred in 1923,
surface hydrology and subterranean hydrology, or “geo- when Meinzer published his book on the occurrence of
hydrology. Meinzer had an illustrious career devoted
” ground water in the United States. The major water-
almost exclusively to the study of ground water as a yielding formations were described, and water supply
water supply. In 1923, he published his famous volume was the order of the day. As we shall learn later, there
on the occurrence of ground water in the United States, are generally four stages in the ordered utilization of
which essentially brought to a close the exploration pe- ground water for water supply: exploration, develop
riod that started before the turn of the century. Indeed, ment, inventory, and management. With the exploration
because of this volume and some modern supplements, stage completed and the resource already undergoing
we are no longer exploring for ground water in North development, much of the effort of the U.S. Geological
America and have not been for several decades. Shortly Survey turned toward inventory. However, at least one
after this publication, Meinzer turned to inventorying major theoretical finding resulted from this work, and
the resource, for example, measuring or estimating the this was provided by Meinzer in 1928 as a result of his
inflows and outflows and the changes in storage over study of the Dakota sandstone (Meinzer and Hard, 1925).
time within ground-water bodies (Meinzer and %earns, In his inventory it seemed that more water was pumped
1928;Meinzer, 1932). Such studies require detailed infor- from a region than could be accounted for; that is, the
mation on the interrelationships between subterranean inventory (which is really a water balance) could not be
water and other components of the hydrologic cycle closed. Meinzer concluded that the water-bearing forma-
to which it is connected. Thus, Meinzer’s definition of tion possessed some elastic behavior and that this elastic
geohydrology as the subsurface component of the hydro- behavior played an important role in the manner in which
 water is removed from storage. Although nothing was two decades. Training in well hydraulics today is ;ISneces-
made of these ideas for another seven years, this was the sary to a hydrogeologist as training in the proper iise of
start of something that would change the complexion of a Bninton compass is to a field geologist. Today. Meinzer,
hydrogeology for at least two decades, if not forever. Theis, Jacob, and Hantiish are no longer with us. How-
That “something” occurred in 1935, when C. V. Theis, ever, of all the hydrogeologists who ever lived, over 95%
with the help of a mathematician named C. I. Lubin, are still alive and still working. In spite o f this enomious
recognized an analogy between the flow of heat and the pool, it is unlikely that such an era as occurred from
flow of water (this analogy was also recognized earlier I935 to about 1960 will ever repeat itself.
by Slichter as well as by others). The significance of this
finding was that heat flow was already mathematically
Chemical Hydrogeology
sophisticated whereas hydrogeology was not. However,
by analogy, a solution to a heat flow problem, of which
Before the Early 1960s
there were several, could be used to provide a solution It would be pleasing indeed if we could trace ;I bit o f
to a fluid-flow analogy. Thus, Theis presented a solution historical development in chemical hydrogeology that
that described the transient behavior of water levels in parallels the common interests demonstrated on the
the vicinity of a pumping well. Imagine the significance physical side of things. However, no such organized ef-
of such a finding to a group of scientists who were totally fort materialized, largely because there was no compara-
committed to studying the response of ground-water ba- ble early guidance such as that provided by Darcy, Cham-
sins to pumping for water supply. For this work, Theis berlin, and the early fieldworkers. Back and Freeze ( 1 98.3)
was awarded the treasured Horton medal almost 50 years recognize an “evolutionary” phase in chemical hydroge-
later (1984). ology that started near the turn of the century and ended
Two additional major contributions came about in sometime during the late 1950s. By evo1ution;iry. we
1940. The first was by Hubbert (1940), who published mean a conglomeration of ideas by different workers
his detailed work on the theory of ground-water flow. interested in a wide variety of hydrochemical iispects.
In this work Hubbert was not interested in small-scale Several good ideas came out of this period, including
induced transients in the vicinity of pumping wells, but certain graphical procedures that are still used t o inter-
in the natural flow of ground water in large geologic pret water analyses (Piper, 1944; Stiff, 195 I ) and ;I partic-
basins. The hydrogeologic community at that time was ularly good treatment of sodium bicarbonate w;iter by
far too preoccupied with transient flow to wells, so this Foster (1 950) that is still highly referenced today. I n adtli-
contribution did not become part of the mainstream of tion to these studies, two others stand out from the per-
hydrogeology until the early 1‘XOs. spective of the chemical evolution of ground w,ater with
The second finding, and we think the most important position in ground-water flow (Chebotarev, 1955; Back.
of the first half of the twentieth century, was provided 1960). For the most part, however, the early effort on
by Jacob (1940). Now let us fully appreciate the situation the cheniicdl side of things was directed at deterniining
at the time. The Theis solution was provided by analogy water quality and fitness for iise for municipal and ;agrictil-
from some heat-flow solution that was rigorously derived tural purposes.
from some equation describing hedt flow. Is there not The first extensive guidance to the working chemical
some equation describing the flow of fluid from which hydrogeologist was provided by Hem’s (1 959) treatment
it (as well as other solutions) could be derived directly? . on the study and interpretation of the chemical character
In 1940, Jacob derived this equation, and, significantly, of natural waters. This work demonstrated, among other
it incorporated the elastic behavior of porous rocks de- things, that most of the important reactions were known.
scribed by Meinzer some 12 years earlier. Now, why all The majority of the contributions to this knowledge come
the fuss about something as abstract as a differential from chemists, sanitary engineers, biologists, and lininol-
equation? A differential equation establishes a relation ogists. Hem’s volume served as the bible for the working
between the increments in certain quantities and the hydrogeologist for several years.
quantities themselves. This property allows us to say The early parts of the 1960 decade marked sonic unifi-
something about the relation between one state of nature cation of the divergent interests of the previous 5 0 y e m .
and a neighboring state, both in time and in space. As Carrels’ (1960) book in particular focused on the equilib-
such, it is an expression for the principle of causality, rium approach in chemical thermodynamics. From this
or the relation between cause and effect, which is the point on, a main focus of much of the work in c1iemic;il
cornerstone of natural science. Thus, the differential hydrogeology shifted to understanding regional geo-
equation is an expression of a law of nature. chemical processes in carefully conducted field investig;a-
Following Theis, Jacob, Hantush (one of Jacob’s stu- tions. Interestingly, at about that same time, hydrogeolo-
dents), and several of Hantwh’s students, the transient gists became interested in regional ground-water flow ;IS
flow of water to wells occupied center stage for about described by Hubbert (1 940) some 20 !.ears earlier. Tlius.
the timing was perfect for a serious bonding of regional facilities, which eventually revealed the extent of ground-
hydrogeology and the chemical evolution of formation water contamination throughout the land. This led to the
waters. Today, some 30 years later, training in regional Comprehensive Environmental Response, Compensa-
ground-waterflow and the chemical evolution of ground- tion, and Liability Act (CERCLA) in 1980,more commonly
water basins is an essential part of hydrogeology. known as the “superfund” for cleanup. RCRA does not
address radioactive or mining wastes. Nor until recently
did it address leaky storage tanks. All wastes will eventu-
Post-1360 Hydrogeology ally be regulated. Hydrogeologists will be involved with
We have already given the reader a glimpse of post-1960 the fate and transport of contaminants in the subsurface
hydrogeology by making reference to the marriage of for many years to come.
physical and chemical hydrogeology by our contemporar- Contaminant transport is merely a special application
ies well aware of the pioneering work of Hubbert (1940) of mass transport in ground-water flow. The transport
and Garrels (1960). Basically, we may isolate three con- processes are essentially physical phenomena, for exam-
tributing factors that are more or less responsible for ple, the movement of dissolved mass by a moving ground
additional developments beyond the early l960s, one water. This movement is frequently accompanied by the
technological and two institutionally motivated. The transfer of mass from one phase (liquid or solid) to an-
technological factor was the development and eventual other. Mass transfer from one phase to another proceeds
accessibility of high-speed computers. Problems that at a decreasing rate until the two phases come to equilih
could be solved tediously by analytical mathematics (or rium with each other. Thermodynamic information re-
those that could not be solved at all) are readily handled garding this equilibrium is basic to an understanding of
by numerical methods requiring computer devices. Thus, mass transfer, but a more immediate concern is the rate
numerical methods in hydrogeology is a core coiirse in at which the transfer takes place. This is a kinetic, as
any serious curriculum in hydrogeology. opposed to an equilibrium,concern, which in many cases
Two institutionally motivating factors initiated the in- is influenced by the rate of ground-water flow. Here again
terest of hydrogeologists in transport processes, or pro- the timing is perfect for a serious bonding of physical
cesses where some physical entity such as heat or chemi- and chemical hydrogeology. Training in mass transport
cal mass is moved from one point to another in ground- and reactions in ground-water flow is an essential part
water flow. Interest in heat flow and geothermal energy of modern hydrogeology.
in particular was at its peak in the early 1970s in response Given this brief description of what hydrogeology is
to an effort to develop alternative energy sources due to and how it got that way, we are now in a better position
the oil embargo. With a return to pre-1970 prices for oil to attempt a partial description of the field. We should
and gas, the interest in geothermal resources diminished still like to hedge a bit here by making a distinction
considerably. However, the groundwork was laid for a between the principles of the science and their various
continued interest in heat transport by ground-water applications. The principles of the science reside in the
flow, especially with regard to the thermal evolution of principles of fluid, mass, and energy transport in geologic
sedimentary basins, low-temperature mineralization, the formations. The applications of the science are not so
dissipation of frictionalheat followingearthquake genera- easily enunciated, but so far we have emphasized prob-
tion, thermal pollution, and the fluid and thermo- lems in human affairs such as water supply, contarnina-
mechanical response of rocks used for repositories tion, and energy resources. There is, however, a historical
to store high-level nuclear waste. connotation to hydrogeology, and this is taken up in the
A second institutionally motivating factor occurred in next section.
the mid-1970s in the form of a federal environmental
law. Environmental laws were quite common during the
1960s and 1970s: the Clean Water Act, the Clean Air Act, 1.2 The Relationship Between
the Clean Drinking Water Act, the Surface Mining Act. Gdrogeology and Otber Fields
Many of these dealt largely with surface water and were
relatively simple to initiate and enforce. The Clean Water
of Geology
Act, for example, made funds available for treatment of The cornerstone of geology resides in four areas: mineral-
sewage disposed of in surface waters, thereby contribut- ogy, petrology, stratigraphy, and structure. The role of
ing to the cleanup of polluted rivers. In 1976,the environ- hydrogeology in geologic studies can best be appreciated
mental loop was closed with the introduction of the if, when making observations, we ask ourselves, “How
Resource Conservation and Recovery Act (RCRA), which did it get this way?” This is an historical question that is
is fundamentally a ground-waterprotection act. The pur- fair in this case because geology is an historical science
pose of RCRA is to manage solid hazardous waste from studied Iargely from the perspective of observation. Let
the time it is produced to its ultimate disposal. The act us remember further that the rock mass we are examining
had provisions for forced monitoring of waste disposal consists of a stationary phase and a mobile phase. The
 1.3 The Hydrologic Cycle 5

terms “stationary” and “mobile” refer to the solids mak-

ing up the rock body and the fluids contained within
the rock, respectively. This is essentially a long-term
“through-flow system,” with the fluids continually re-
placed over geologic time frames. Thus, the ability of a
moving ground water to dissolve rocks and minerals and
to redistribute large quantities of dissolved mass has im-
portant implications in chemical diagenesis, economic
mineralization, and geologic work in general. Addition-
ally, it has long been understood that the fluids need
not reside in a passive porous solid, but instead there
frequently exists a complex coupling among stress,
strain, and the pore fluids themselves. This coupling is
often complicated by heat sources in crustal portions of
the earth. The net effect of this coupling is hilly demon-
strated in several modern environments throughout the Ffgure 1.1 Schematic representation of the hydrologic
world where “abnormally” high fluid pressures reside in cycle.
deforming rocks. Ultimately, these pressures can lead to
rock fracture and fwther deformations. Is it possible that
such environments were equally widespread throughout areas into single components. Yet another presentation
geologic time, and what consequences can be attributed of the hydrologic cycle is one that portrays the various
to them? moisture inputs and outputs on a basin scale. This is
Even metamorphic processes cannot be fully examined shown in Figure 1.2, where precipitation is taken as
in the absence of a fluid phase. According to Yoder input and evaporation and transpiration (referred to as
(1955), water is an essential constituent of many minerals evapotranspiration)along with stream runoff are outputs.
common to metamorphic rocks, it is the primary catalyst The stream runoff component, referred to as overland
in reaction and recrystallization of existing minerals, and flow, can be augmented by interflow, a process that
it is one of the chief transport agents of material and operates below the surface but above the zone where
possibly heat in metasomatic processes. As early as 1909, rocks are saturated with water, and by base flow, a direct
Munn recognized the role of ground water in hydrocar- component of discharge to streams from the saturated
bon migration and entrapment. To discuss these aspects portion of the system. Infiltrationof water into the subsur-
further would preempt some of the main considerations face is the ultimate source of interflow and recharge to
in this book. However, it should be clear from the per- the ground water.
spective of processes that one cannot ignore the role of The ground-water component is more clearly demon-
ground water in performing geologic work. strated in Figure 1.3. In this water profile, the vadose
Bearing in mind all the factors discussed in the preced- zone corresponds to the unsaturated zone, whereas the
ing paragraphs, we venture the following definition of phreatic zone corresponds to the saturated zone. The so-
hydrogeology. Hydrogeology is the study of the laws called intermediate zone separates the saturated phreatic
governing the movement of subterranean water, the me- zone from soil water. It can be absent in areas of high
chanical, chemical, and thermal interaction of this water precipitation and hundreds of feet thick in arid areas.
with the porous solid, and the transport of energy, chemi- The water table marks the bottom of capillary water and
cal constituents, and particulate matter by the flow. the beginning of the saturated zone.
The terms saturated and unsutumted require some
clarification. Given a unit volume of soil or rock material.
-
1.3 The Hydmbgic Cycle designated as V,, the total volume consists of both solids
( V , ) and voids (K,). Only the voids are capable of con-
This book is concerned solely with ground water. Be- taining a fluid, either air or water. The degree of Sdtllra-
cause ground water is one component of the hydrologic tion is defined as the ratio of the water volume to the
cycle, some preliminary information on the hydrologic void volume, V,&, expressed as a percentage. For a fully
cycle must be introduced to set the proper stage for saturated medium, the ratio is one (or 100%).A degree
things to come. of saturation less than 100% indicates that air occupies
some of the voids. Another term in common usage is
moisture content, 8, which is defined as the volume of
Components of the Hydrokbgic Cycle water divided by the total volume (V,J V,), expressed as
Schematic presentations of the hydrologic cycle such as a percentage.
Figure 1.1 often lump the ocean, atmosphere, and land At the contact between the lower part of a dry porous
-
6 Cbapter 1 / lntroductfon

+++++&+++\\\++ Precipitation \\\+++\++++++++++\+

Watershed divide Evapotranspiration
Interception
Infiltration
Overland flow

Ground water table

Ground water flow

((1)

I
Evapotri piration I
I
I Precip;tation I

lnterflow
t- p*
gE
-p!
g%i
:=
r
0

Ffgure 1.2 The basin hydrologic cycle.

material and a saturated material, the water rises to a the water in all directions because of an unbalanced
certain height above the top of the saturated material. molecular attraction of the water at the boundary. The
This gives rise to the capillary water zone of Figure 1.3. tensional nature of these forces can be compared with
The driving force responsible for the rise is termed sur- those set up in a stretched membrane.
face tension, a force acting parallel to the surface of Capillarity results from a combination of the surface
tension of a liquid and the ability of certain liquids to
wet the surfaces with which they come in contact. This
wetting (or wetability) causes a curvature of the liquid
surface, giving a contact angle between liquid and solid
different from 90". The idealized system commonly used
to examine the phenomenon is a water-containing vessel
and a capillary tube. When the tube is inserted in the
water, the water rises to a height h, (Figure 1.4). The
meniscus, or curved surface at the top of the tube, is in
contact with the walls of the tube at some angle a,the
value of the contact angle depending on the wall material
and the liquid. For a water-glass system, a is taken as
zero.
Points A and Cin Figure 1.4 are at atmospheric pressure
(1.013 X lo5newtons per meter squared (N/m2) or 14.7
Ib per square inch (psi) or approximately 1 bar). A funda-
mental law of hydrostatics states that the pressure inten-
sity of a fluid at rest can vary in the vertical direction
Ffgure 1.3 The water profile. only. Hence, point B is also at atmospheric pressure, and
 1.3 Tbe Hydrologic Cycle 7

lating it were introduced by Thornthwaite (1948). Poten-

tial evapotranspiration is defined as “the amount of water
tliat would evaporate or transpire from a surface if water
was available t o that surface in unlimited supply.” There
is thus a clear distinction between what is actual ;ind
what is potential evapotranspiration. For example. the
actual evapotranspiration off tlie surface of Hoover Reser-
voir very likely equals the potential rate. However, during
most of the year, the actual evapotranspiration in reh. ’1011s
adjacent to the reservoir is generally less than the poten-
tial rate, simply because of the lack of unlimited water.
The actual evaporation off the Sahara desert is nil whereas
the potential evapotranspiration (or potential for evapo-
transpiration) is quite high. In fact, if we wished to irri-
Figure 1.4 Capillary rise in a glass tube. gate parts of the Sahara, we would have to supply irrigs-
tion water at about the potential rate.
From this definition and the few examples stated ear-
the pressure at point D must be less than atmospheric lier, potential evapotranspiration is a maximum water
by an amount dictated by the pressure exerted by the loss (or upper limit of actual evapotranspiration), is a
length of the water column above B , or h,y,., where y,,. temperdture-dependent quantity, and is a measure of the
is the unit weight of water. At a height of 10.3 m, the moisture demand for a region. The last point stated is
value of h,y,,is just equal to atmospheric pressure. Hence, better demonstrated by the ratio of precipitation t o po-
10.3 m is a theoretic upper bound for the height of tential evapotranspiration. In a desert or tundra region,
capillary rise, at least in glass tubes. For pure water at this ratio may be less than 0.1. For these regions, precipi-
20°C, the height of capillary rise in a glass tube is ex- tation is not sufficient to meet the water demand so
pressed as that very little grows naturally. For a successful irrigation
scheme in such a region, water must be supplied at
0.153 the potential rate. The ratio of precipitation to potenthl
h,.= -
r evapotranspiration in parts of Montana and North Dakota
ranges from 0.2 to 0.6, again indicating a need for irriga-
where h, and r are expressed in centimeters and r is tion water in crop production. The range in e.1‘stern
taken as the radius of the tube. In real systems, r is taken United States is from 0.8 to 1.6, indicating a rather well-
as tlie radius of passage. balanced situation and, in some cases. a water surplus.
Of further significance to this discussion is now an The vegetation (or lack thereof) in a region is tliiis a
absolute definition of the water table. From Figure 1.4 reflection of the precipitation-potential evapotranspim
we recognize that water above the water table is at a tion ratio. This is obvious when comparing the Sahara
pressure below atmospheric while water below the wa- desert with the Amazon basin, both of which have exces-
ter table is at a pressure greater than atmospheric. The sive potential evapotranspiration, but only one of which
water table is the underground water surface at which is characterized by an equally excessive amount ofprecip-
the pressure is exactly equal to atmospheric. itation.
The intermediate zone lies above the capilkiry fringe Figure 1.5 is a qualitative demonstration of the n;itural
and consists of water in the form of thin films adhering water demand and water supply in a region ;is nie;isuretl
to the pore linings. This water is free to drain downward
by the potential evapotranspiration and the precipit~ion.
under the forces of gravity. The soil, intermediate, and respectively. From these curves, we note three import;int
capillary zones are collectively taken as the unsaturated time periods:
o r vadose zone (Figure 1.3).
1. Precipitation equals potential eval’otr;Insl’ir;itioii.
Evapotranspiration and 2. Precipitation is less than potential evapotrmis-
Potential Evapotranspiration piration.
Evaporation can take place from both soil and free water 3. Precipitation is greater than potenti;il ev;ipotriins-
surfaces. Evaporation from plants is called transpiration, piration.
and the combined process is often referred to as evapo-
transpiration. In formulating some rules (laws) that govern the behavior
The concept of potential evapotranspiration (or the of water in the hydrologic cycle, we first stipu1;ite that
potential for evapotranspiration) and methods for calcu- the demands of potential evapotranspiration must be met
S_ chapter I / ~ntroduction

Infiltration and Recharge

I ential evapotranspiration If water is applied to a soil surface at an increasing rate,
B
-z eventually the rate of supply has to exceed the rate of
entry, and excess water will accumulate. Not so obvious
1
-P
is the fact that if water is supplied to a soil surface at a
constant rate, the water may at first enter the soil quite
readily, but eventually will infiltrate at a rate that de-
creases with time. Horton (1933, 1940) was the first to
J F M A M J J A S O N D point out that the maximum permissible infiltration rate
Months of the Year decreases with increasing time. This limitingrate is called
Figure 2.5 Relationship between precipitation and poten- the infiltration capacity of the soil, an unfortunate choice
tial evapotranspiration. of words in that capacity generally refers to a volume or
amount whereas the infiltration capacity of Horton
(1933) is actually a rate.
(if at all possible) before water is permanently allocated The concept of field capacity is another useful idea in
to other parts of the cycle. Thus, when precipitation soil science to denote an upper limit of moisture content
is equal to potential evapotranspiration, the surplus is in soils. At field capacity the soil is holding all the water
it can under the pull of gravity. In a conceptual sense,
theoretically zero, and all the rainfall for that period is
available to satisfy the evaporation needs. During such both interflow and ground-water recharge can com-
time periods the actual evapotranspiration equals the mence when the moisture content exceeds field capacity.
potential evapotranspiration. This does not mean that The proportioning of the available moisture to interflow
infiltration cannot take place. It means simply that any or ground-water recharge is dependent to a large extent
infiltrated water will be available in the soil moisture for on the nature of the soil zone. If the soil zone is thin
use by plants in the transpiration process. In addition, and underlain by rather impervious rock, the interflow
any time the rate of precipitation exceeds the rate of component may be the dominant one. For thicker pervi-
infiltration, there will be some water available for over- ous soils, downward migration may dominate, with most
land runoff so that this suggested balance of precipitation of the moisture being allocated to ground-waterrecharge.
and potential evapotranspiration is not exactly true. At
any rate, the surplus after all demands of potential evapo- Base F b w
transpiration are met will be zero. The term recession refers to the decline of natural output
When precipitation is less than potential evapotranspi- in the absence of input and is assumed or known from
ration, all the precipitation is available to partially satisfy experience to follow an exponentialdecay law. The base-
potential evapotranspiration.During such time periods, flow component of streams represents the withdrawal
the actual evapotranspiration would appear to be less of ground water from storage and is termed a ground-
than the potential amount. Such is not the case, however, water recession.As this recession is generally determined
as the part of the demand not met by precipitation may
from stream hydrographs, the hydrograph must be sepa-
be met by drawing on whatever moisture is in the soil rated into its component parts, which normally consist
zone. So here again the actual rate can equal the potential of overland flow, interflow, and base flow. Frequently,
rate. If this situation continues over a prolonged period, the interflow component is ignored so that the hydrolo-
the soil becomes depleted of its moisture (Figure 1.5), gist generally deals with an assumed twocomponent
and the actual rate of evapotranspirationwill fall below system.
the potential rate. Such periods are normally labeled
Whatever the assumptions, methods employed in hy-
droughts.
drogrdph separation are wrought with difficulties. One
Finally, when temperature drops sufficiently so that
of the most acceptable techniques is illustrated in Figure
precipitation exceeds potential evapotranspiration,a wa-
1.6,which portrays an ideal stream hydrograph that plots
ter surplus is realized. This surplus immediately goes
the discharge of a river, usually in cubic feet per second
into rebuilding the soil moisture component. This rate
(cfs) at a single point in the watershed as a function of
of infiltration will generally decrease with time as the
time. According to Linsley and others (1958), the direct
soil zone becomes more saturated with water. It is during
runoff component is terminated after some fixed time
these periods that the underlying ground water becomes
7"'after the peak of the hydrograph, where this time T*
recharged and, in response to a decreased infiltration
in days is expressed
rate, overland flows become more pronounced. It is thus
7'' = A02
seen that the soil zone is a buffer in the workings of the (1.2)
basin hydrologic cycle, taking in moisture during surplus where A is the drainage area in square miles and 0.2 is
periods and releasing moisture during deficit periods. some empirical constant. Normally, the recession ex-
 1.3 The Hydrologic Cycle 2
I 7' where, as described, Q,, is the base flow at time t = 0
and Q is the base flow t time units later. This provides
a means for determining the recession constant. Clearly
I
this constant will be some number less than one and will
I
I
be large (approach one) for flat recessions and small
I
I
(approach zero) for steep recessions.
The influence of geology on the shape of the recession
curve has been discussed and demonstrated by Farvolden
(1964). Streams in limestone regions are characterized
by flat recessions, a reflection of the fact that most of
Ease-flow component the drainage takes place in the subsurface. Granitic and
Time (days) other low-permeability regions typically have rather
steep base-flow recessions.
Figure 1.6 Determining base-flow component from stream
hydrograph. A plot of discharge versus time on semilogarithmic
paper will yield a straight line, the slope of which defines
the recession constant (Figure 1.7). For this case the
isting prior to the storm is extended to a point directly recession expression becomes
under the peak (A-B,Figure 1.6) and then extended again
to the point P after the peak (B-C, Figure 1.6). The (1.5)
depressed base-flow component under the peak is justi-
fied because high water levels in the stream have the where Qo is the discharge at time t = 0, t, is the time I
tendency to cut off or retard the ground-water dis log cycle later, and t equals any time of interest for which
charge component. the value of Q is desired. For example, for t = t ,
Consider now the point on the hydrograph T* days Q = -Qv
after the peak. From this point onward, the total flow (1.6)
10
is the base-flow component derived from ground-water
discharge into the stream, at least until the next storm. The total volume of base-flow discharge corresponding
This recession starting at time T* can be described as to a given recession is found by integrating Eq. 1.5 over
the times of interest
Q = QOe-" (1.3)
where Q,, is the discharge at time T*, Q is the discharge
at any time, k is a recession constant, and time t is the
time since the recession began. In this equation, the time where tois the starting time of interest. Meyboom (1961)
t varies from zero, which corresponds to the time (T*), has employed Eq. 1.7 to determine ground-water re-
to infinity. Thus, at the beginning of the recession t = 0 charge between recessions. For example, if ?, eqiials zero
and Q = Q,,. For later times, Q decreases and follows the and t equals infinity
exponential decay law of Eq. 1.3.
From Eq. 1.3, the recession constant can be expressed Vol = QJl
- (1.8)
2.3

k=- (f) In (g) which is termed the total potential ground-water dis-
charge. This volume is defined as the total volume of

M J J A S O N D ' J F M A M J J A S O N D ' J F M A M J J A S O

Figure 1.7 Stream hydrograph showing base-flow recession (from Meyboom, J. Geophys.
Res., v. 66, p. 1203- 1214, 1961. Copyright by h e r . Geophys. Union).
1
2 Cbapter 1 / Introduction

ground water that would be discharged by base flow

4
u
Atmosphere
during an entire recession if complete depletion takes II
I
place uninterruptedly. It follows that this volume de- I
I
scribes the total volume of water in storage at the begin- I
I
ning of the recession. The difference between the re- I
I
maining potential ground-water discharge at the end of 7 I
I
a given recession and the total potential ground-water 1
I
discharge at the beginning of the next recession is a I
I
measure of the recharge that takes place between reces- In I
.-UI I
sions (Meyboom, 1961). I
n
c IM
I
8 I
Example 1.1 0 I
I
I
Determine the approximate recharge volume between I
I
the first two recessions of Figure 1.7. I
I
The first recession has an initial value of 500 ft3/s I

and takes about 7.5 months to complete a log cycle of -7-

I
I
I
I
discharge. The total recession takes about 8 months to MI 1
I
I I
complete. Total potential discharge is calculated from I I I I
Eq. 1.8:
I
Lithosphere --
1 I
I

Qo t l
Qtp =- Figure 1.8 Elements of the global hydrologic cycle.
2.3
500 ft)/s X 7.5 months X 30 days/month
X 1440 min/day X 60 s/min Hydrologic Equatfon
2.3
The hydrologic cycle as shown in Figure 1.2 and de-
= 4222 X 106ft3 scribed throughout this chapter is a network of inflows
and outflows that may be conveniently expressed as
The ground-water volume discharged through the total
recession lasting approximately 8 months is determined input - output = change in storage (1.9)
by evaluating Eq. 1.7 over the limits t equals zero to 2
The word equation given as Eq. 1.9 is a conservation
equals 8 months
statement and assures us that all the water is accounted
for; that is, we can neither gain nor lose water. The
interconnections between the components of the hydro-
logic cycle can be demonstrated on a global scale (Figure
or about 3800 X lo6ft'. Base-flow storage still remaining 1.8) or on a basin scale (Figure 1.9). On a global scale,
at the end of the recession can be determined by merely the atmosphere gains moisture from the oceans and land
subtracting actual ground-water discharge from total po- areas E and releases it back in the form of precipitation
tential discharge, which gives 422 X 10' ft3. P. Precipitation is disposed of by evaporation to the atmo-
The second recession has an initial value of about 200 sphere E, overland flow to the channel network of
ft3/sand takes about 7.5 months to complete a log cycle streams Qc,,and infiltration through the soil F. Water in
of discharge. Total potential discharge is calculated

QJi
Qrp =-
2.3
200 ft3/s X 7.5 months X 30 days/month
T
-- X 1440 min/day X GO s/min
2.3

or about 1400 X lo6 ft'. The recharge that takes place

between recessions is the difference between this value
and remaining ground-water potential of the previous
recession, or 978 X lo6 ft'.
Figure 1.9 Elements of the basin hydrologic cycle.
 1.1 What Is Hjidrogeology?

the soil is subject to transpiration T, outflow to the chan- Eq. 1.12 becomes
nel network (the interflow component) Q,,,and recharge
to the ground water RN.The ground-water reservoir may (1.13)
receive water Qj and release water Q,, to the channel where Qp is the added pumping withdrawal. As pumpage
network of streams and the atmosphere. Streams that is new output from the system, the water levels in the
receive water from the ground-water reservoir by the basin will decline in response to withdrawals from
base-flow component Qo are termed effluent or gaining ground-water storage. The stream will eventually be con-
streams. This occurs when the ground water table in the verted to a totally effluent one. In addition, with lower
vicinity of the stream is above the base of the stream. water levels, the transpiration component will start to
Streams that lose water to the ground-water reservoir as decline and eventually approach zero. If potential re-
designated by the Qj component are termed influent, or charge to the system was formerly rejected due to a water
losing, streams. This occurs when the ground water table table at or near land surface, the drop in water levels
is well below the base of the stream. Also noted on Figure will permit R.v to increase over its steady value. Thus. at
1.8 is the potential contribution of water of volcanic or some time after pumping starts, Eq. 1.13 becomes
magmatic origin.
(1.14)
Figure 1.9 illustrates that isolation of the basin scale
hydrologic subsystem cannot exclude the lines of mois A new steady state can, at least theoretically. be achieved
ture transport connecting this subsystem to the global if the pumping withdrawal does not exceed the inputs
cycle. This connection is accomplished by accounting R,,, and Q l . If pumping continually exceeds these input
for precipitation derived from the global system and total values, water is continually removed from storage and
runoff R,, to the global system. From this diagram we water levels will continue to fall over time. Here, the
reformulate our word equation given earlier as steady state has been replaced by a transient, or unsteady,
state, where some parameter (in this case, ground-water
P - E - T - R , , = AS (1.10) storage) continually changes over time. Not only is
where AS is the lumped change in all subsurface water. ground-water storage being depleted, but some of the
Each item in the equation has the units of a discharge, surface flow has been lost from the stream.
or volume per unit time. The term inventory is generally reserved for investiga-
The hydrologic equation (Eq. 1.10) must balance for tions in which a detailed accounting of inflow. outflow.
every time period to which it is applied. This balancing and changes in storage is attempted for time intervals,
will occur provided the accounting of the inflow and the such as years or other units of time, during a period of
outflow is done over a common period of time. An exact observation (Meinzer, 1932). In the Pomperaug Basin,
balance is an unreasonable goal due to poor instrumenta- Connecticut, for example, available data were organized
tion, lack of data, and the assumptions that are gener- on a monthly basis for a three-year period of study
ally applied. (Meinzer and Steams, 1928). The hydrologic equation is
Equation 1.10 may be expanded or abbreviated, de- the basis for such studies.
pending on what part of the cycle the hydrologist is Last, our hydrologic equation may be expressed in the
interested in and, of course, depending on the available form of an ordinary differential equation
data base over the period of record. For example, for dS
the ground-water component, another form of Eq. 1.10, O(t) = - (1.15)
dt
is arrived at from Figure 1.9:
where I is input and 0 is output, both taken as a function
(1.11) of time. Equation 1.15 is also referred to as a lumped-
parameter equation, a term that is reserved for any equa-
Over long periods of time, providing the basin is in its
tion or analysis wherein spatial variations in the parame-
natural state and no ground-water pumping is taking
ters are not considered.
place, the natural inputs RN and Qj will be balanced by
the natural outputs r a n d Q,,so that the change in storage
will be zero. This gives Example 1.2

(1.12)
The following is a simplified inventory based on the hy-
RNf Q j = T + Qo
drologic equation
or, stated simply, input equals output. This means that the
{recharge from direct precipitation plus
ground-water component in the basin is hydrologically in recharge from stream flow}
a steady state, and the variables (or, more properly, the
averaged values of the variables) have not changed over {minus discharge by pumping minus
the time period over which the averaging took place. discharge by e\rapotranspiration}
On the other hand, if a pumpage withdrawal is included, {equals change in ground-water storage}
12
I
Cbapter 1 / introduction

With regard to this inventory, the following are worth value, whereas discharge by evapotranspiration
noting: seems to decrease and approach some minimum
value.
1. Previous to pumping outputs, natural recharge was 3. During the course of these withdrawals, the basin is
equal to natural discharge and the ground-water ba- in a transient state where water is continually being
sin was in a steady state. withdrawn from storage. Although not shown, this
2. With the addition of pumpage and in the course of results in a continual decline in water levels. A new
withdrawals From storage, net recharge from stream steady state can be achieved by reducing pumping
flow tends to increase and reaches some maximum to about 38 X lo', m'/yr.

Table E1.2
Recharge from Net Recharge Discharge Discharge Change in
Direct from bY bY Ground-Water
Time Precipitation Stream Flow Pumping Evapotranspiration Storage
Year m3 m3 m3 m3 m3
1 3 x 107 0 0 3 x 107 0
2 3 x 107 6 X lo5 1 x 107 3 x 107 94 x 105

7 3 x 107 1 x
106 3 x 10' 9 x 106 8X I d
8 2.8 x 107 2 x
106 3.5 x 10' 5 x 106 10 x 106
9 2.5 x 107 3 x 106 3.5 x 10' 3 x 106 10 x 106
10 3.5 x 107 4 x 106 4 x 10' 1 x 106 2x 106
11 3.5 x 107 4 x 106 4.2 X 10' 1 x 106 3 x 106
12 3.5 x 107 4 x 106 4 x 10: 1 x 106 2 x 106

Problems
1. Calculate the base-flow recession constant for the users, what length OF recession (in months) would be
first recession on Figure 1.7. required to reach this minimum flow?
2. The diagram below illustrates a hydrograph deter- 4. Ten-year averages (1981-91) for a ground-water in-
mined from flow measurements on a spring located in ventory For the San Fernando Basin are as follows: Re-
carbonate rocks. The variations in spring flow are due charge from precipitation and other sources, 79.3 X lo6
to seasonal changes in rainfall (recharge). Calculate the m', recharge from injection wells 38 X lo', m', ground-
recession constant and the recharge that takes place be- water inflow, 1.01 X 106 m', ground-water pumping,
tween recessions. 120.55 X 10' m3, ground-water outflow 0.521 X lo', m3,
and base flow to streams, 3.8 X 106 rn*. Perform a water
3. If the spring in the figure below represented a water balance for the ground-watercomponent based on these
supply and a minimum of 10 ft3/s was required by the 10-year averages.

a -
10-
~~

M J J A S 0 N D'J F M A J J A S O N D ' J F M A M J J A S 0
Problem 2.
 Chapter 2

The Origin of Porosity

and Permeability

2.I Porosity and Permeability

2.2 Continental Environments
2.3 Tbe Boundary Between Continental and Marine Environments
2.4 Marine Environments
2.5 Uplifi, Diagenesis, and Erosion
2.6 Tectonism and tbe Formation of Fractures

A porous body is a solid that contains holes. All rocks

are considered porous to some extent, with the pores
- porosity and Permeability
2.1
containing one or more fluids-air, water, or some minor- Porosity and Effective Porosity
ity fluid such as a hydrocarbon. The holes may be con- Total porosity is defined as the part of rock that is void
nected or disconnected, normally or randomly distrib- space, expressed as a percentage
uted, interstitial or planar cracklike features. The degree
of connectivity of the pores dictates the permeability of n = -v,. (2.1)
the rock, that is, the ease with which fluid can move v,
through the rock body. There is nothing of more vital where V,, is the void volume and V, is the total volume.
interest to those concerned with the subsurface move- A related parameter is termed the void ratio, designated
ment of fluids than permeability, its creation and destruc- as e, and stated as
tion, and its distribution throughout a rock body. Hence,
this introductory chapter is intended to be as complete e = -v,,
(2.2)
as possible-given our limited state of knowledge-in K
describing those processes that result in the formation of expressed as a fraction, where V, is the solid volume. As
porous bodies along with other processes that ultimately total volume is the sum of the void and solid volume,
modify the rock body, the connectivity of its pore space, the following relationships can be derived:
and its contained fluids. To help us along in these tasks
n e
we call upon simple concepts known to most students e=- or n = -1 + e (2.3)
in the earth sciences. Thus, we approach this topic from
1-n
the simple rock cycle to describe the origin and formation Figure 2.1 shows some typical kinds of porosity associ-
of porous bodies, and from tectonic and chemical pro- ated with various rocks. The term prima y porosity is
cesses to ascertain whatever mechanical or chemical al- reserved for interstitial porosity (Figures 2. la through
terations rocks have been subjected to at some point in d ) , and the termsecondatyis used for fracture or solution
their history. porosity (Figures 2.1e and f).Interstitial porosity has
-
13
14 Cbapter 2 / Tbe Origin of Porosity and Permeability

(id (a) 1.1,

Figure 2.1 Relation between texture and porosity. (u) Well-sorted sedi-
mentary deposit having high porosity; (6)pcx)rly sorted sedimentary de-
posit having low porosity; (c) well-sorted sedimentary deposit consisting
of pebbles that arc themselves porous, so that the deposit as P whole
has a very high porosity; ( d ) well-sorted sedimentary deposit whose po-
rosity has been diminished by the deposition of mineral matter in the in-
terstices; (e) rock rendered porous by solution; ( J ) rock rendered ptr
rous by fracturing (from Meinzer, 1923).

been investigated by Graton and Fraser (1 9351, who dem- recently deposited sediments whereas the former value
onstrated that its value can range from about 26 to 47% is for dense crystalline rocks or highly compacted soft
through different arrangements and packing of ideal rocks such as shales. In general, for sedimentary materi-
spheres. In actuality, the porosity of a sedimentary rock als, the smaller the particle size, the higher the porosity.
will depend not only on particle shape and arrangement, This is best demonstrated by comparing the porosity of
but on a host of diagenetic features that have affected coarse grdvels with fines, and the total gravel assemblage
the rock since deposition. with silts and clays.
Porosity can range from zero or near zero to more An important distinction is the difference between
than 60% (Table 2.1). The latter value is reflective of total porosity, which does not require pore connections,
and effective porosity, which is defined as the percentage
of interconnected pore space. Many rocks, crystallines
in particular, have a high total porosity, most of which
Table 2.1 Range in Values of Porosity may be unconnected. Effective: porosity implies some
Material connectivity through the solid medium and is more
closely rebated to permeability than is total porosity. Some
SEDIMENTARY data on effective porosity are shown in Table 2.2. As
Gravel, coarse 24-36
noted, effective porosity can be over one order of magni-
Gravel, fine 25 - 3 8
Sand, coarse 31 - 4 6 tude smaller than total porosity, with the greatest differ-
Sand, fine 26-53 ence occurring for fractured rocks.
Silt 34-61 Heath (1982) recognizes five types of porosity in domi-
Clay 34-60 nant water-bearing bodies at or near the Earth’s surface
SEDIMENTARY ROCKS in the United States and attempts to map their distribution
Sandstone 5-30 (Figure 2.2). There are some difficulties with this map
Siltstone 21-41 because of the necessity of mapping a single type of
Limestone, dolomite 0-40 opening in areas where two or more types are present.
Karst limestone 0-40 However, this is a useful presentation and one to which
Shale 0 - 10 we will refer frequently in this chapter. Each pattern on
CRYSTALLINE ROCKS Figure 2.2 is associated with one or more major water-
Fractured crystalline rocks 0- 1 0 yielding formations in the United States. Thus, the solu-
Dense crystalline rocks 0-5 tion-enlarged openings in carbonate rocks that make up
Basalt 3-35 the Florida peninsula are known as the Floridian system;
Weathered granite 34-57 the sands and gravels stretching from New Jersey into
Weathered gabbro 42-45 Texas are sediments of the Atlantic and Gulf coastal
In pan from Davis (1969) and Johnson and Morris pbains; the sand and gravel in the Midwest represents
(1962). glacial deposits; the sandstones in the northern midconti-
 2.1 Porosity atid Pentieability 15

Table 2.2 Range in Values of Total Porosity and Effective Porosity

Total Porosity ("10) Effective Porosity (YO)

Anhydrite' 0.5-5 0.Oi- 0 . 5
Chalk' 5-40 0.05- 2
Limestone, dolomite' 0-40 0.1-5
Sandstone' 5 - 15 0.i- 10
Shale' 1-10 0.5-5
Salt' 0.5 0.1
Granite' 0.1 0.0005
Fractured crystalline - 0.00005-0.0 1
rock'
' Data from Croff and others (1985).
' Data from Norton and ffiapp (1977).

nent represent several formations, including the Dakota the western United States occupies the Columbia River
Sandstone and the Cambrian-Ordovician system; the Plateau.
sands and gravels in the western part of the nation occupy
alluvial basins, whereas the occurrence of these same
sediments in the central United States represents the
Permeability
remnant of a giant alluvial apron that formed on the Permeability may be described in qualitative t e r m ;ISthe
eastern slope of the Rocky Mountains; and the basalt in ease with which fluid can move through a porous rock

iG
Ad

I
0

Figure 2.2 Map of conterminous United States showing types of water-bearing openings in the domin;unt ;iqiiit'er~(froill
Heath, 1982). Reprinted by permission of Ground Water. 0 1982. All rights reserved.
16 Cbapter 2 / Tbe Origin of Porosity and Permeability

and is measured by the rate of flow in suitable units. This fourth major aquifers in the United States, depend on
qualitativedefinitionwill suffice for this section, although fractures for the transmission of water.
later the topic is taken up in quantitative terms.
Contrasts in permeability from one rock to another
2
.
2 Continental Enuironments
and, less frequently, within a given rock unit have given
rise to a variety of terms and definitions in hydrogeology, Weathering
most of which pertain to water supply. A rock unit that
Rocks at the Earth’s surface are subjected to either physi-
is sufficiently permeable so as to supply water to wells
cal disintegration or chemical decomposition. The weath-
is termed an aquifer. Major aquifers are referred to by
ering products can accumulate in place to form a soil
their stratigraphic names, such as the Patuxent Formation
or can be transported by wind or water to accumulate
in Maryland, the Ogallala Formation throughout most of
elsewhere as sedimentary materials. Weathering is also
the High Plains, and the Dakota Sandstone. These aqui-
responsible for much of the secondary porosity illus-
fers may be considered “commercial” in the sense that
trated in Figure 2.1. This is especially so with the chemi-
they supply sufficient quantities of water for large-scale
cal decomposition of carbonate rocks, where sinkholes
irrigation or municipal usage.
and crevices can readily form due to dissolution along
Aquitards have been defined as beds of lower perme-
preexisting fractures or other pathways of fluid move-
ability in the stratigraphic sequence that contain water
ment (Figure 2.3). A similar type of process occurs in less
but do not readily yield water to pumping wells. Major
soluble igneous and metamorphic rocks, where chemical
aquitards are generally considered to be low-permeability
weathering tends to open preexisting fractures, contrib-
formations that overlie major aquifers. Examples include
uting to more pervious pathways. An excellent example
the Pierre Shale overlying the Dakota Sandstone, the Ma-
of this occurs in the Columbia Plateau, which is character-
quoketa Shale overlying high-permeability rocks in Illi-
ized by numerous individual basaltic flows, along with
nois, and the Hawthorn Formation overlying the high-
occasional sand or gravel units that mark periods of
permeability Floridian aquifer. Rocks considered to be
flooding (Figure 2.4). After a particular flow is extruded,
aquitards in one region may serve as aquifers in others.
it undergoes a cooling period accompanied by the forma-
For example, water supplies are obtained from many
tion of cooling joints, fractures, and vesicles. These rocks
low-permeability materials throughout North America,
then become subjected to weathering, a process that is
such as glacial tills in the Midwest and in Canada, and
immediately terminated when another flow is extruded.
fractured crystallines and shales in many parts of the
However, the weathering features within the flow tops
continent. These materials are not able to supply suffi-
are retained in the rock and constitute major pathways
cient quantities of water for municipal and irrigation use,
for fluid movement. The pathways are largely horizontal,
but frequently are adequate for domestic or farm usage
reflecting a widespread surficial weathering process,
in rural areas. Thus, the terms aquqer and uquiturd
with the depth of weathering restricted to a few tens of
are ambiguous.
feet, depending on the length of time before the volcanic
More recently, the term @dro.strutigru@ic unit has
been employed rather extensively. A hydrostratigraphic
unit is a formation, part of a formation, or a group of
formations in which there are similar hydrologic charac-
teristics that allow for a grouping into aquifers and associ-

.:j
ated confining layers. Thus, the Dakota Sandstone and
various other aquifers in combination with the Pierre
Shale confining unit constitute a hydrostratigraphic unit,
as do the Floridian carbonates in combination with the
Hawthorn Formation, and several Cambrian-Ordovician
formations overlain by the Maquoketa Shale. 2 60
The field occurrence of permeability is well docu-
mented by the fact that about one-half of the United States
is underlain by aquifers capable of yielding moderate to 90

- ,
rocks, sandstones are the most important from the per- ship between fracture traces and the occurrence of ground-
spective of annual withdrawals, followed by carbonate water in carbonate rocks,” p. 73-91, 1964, with kind permis
and volcanic rocks. Sandstones and carbonates are virtu- sion from Elsevier Science, NL, Sara Burgerhartstraat 25,
ally ubiquitous in sedimentary terranes. Volcanics, as the 1055 KV Amsterdam, The Netherlands.
 2.2 Continental Environments Weathering 17

Diagram showing a series Diagram showing structure Thus, either the connectivity is decreasing with depth,
of lava flows of individual lava flows
or the fractures are getting tighter due to limitations on
Broken basalt
Dense flow center Flow too .Vesicular the depth of weathering, or both. It is noted that even
with vertical joints
Interbed, flow top,
flow bottom
fee:
20
1 basalt
or cI inker the most productive wells in this region have very low
yields, commonly less than 25 gallons per minute (1.58
liters/s). However, in the Piedmont, much of the rural
Dense flow center Fanning
Entablature columns domestic supply is furnished by these fractured low
with vertical joints
permeability rocks. Once more, the weathering pattern
Flow top/bottom
Block joints produces a predictable type of hydrology.
Dense flow center Columns Weathering can take place both above and below the
with vertical ioints Colonnade Platy joints water table but is generally slower in the saturated envi-
Vesicular ronment. In addition, a primary mineral subjected to
base
weathering may simply dissolve or a portion of it may
Figure 2.4 Flow tops and flow interiors, basalt. Hanford reprecipitate to form secondary minerals. The weather-
Reservation (after Vaccaro, 1986). ability of igneous silicate minerals is directly proportional
to their temperature of formation from molten materiids,
tliat is, the high-temperature-phase olivine weathers
rock becomes inundated in another sea of lava. Below faster than do the various plagioclase families. One of
the weathering zone, the rocks remain rather dense and the more common minerals, quartz, weathers the most
are characterized by poorly connected cooling joints. slowly. On the other hand, halite, gypsum, pyrite, calcite.
Hence, this weathering pattern produces a predictable dolomite, and volcanic glass weather faster than do the
layercake type of hydrology, with some of the permeable coninion silicate minerals.
flow tops extending over great distances in the plateau.
Recognizing that rocks are exposed at the Earth’s sur-
Erosioon, Transportation, and Deposition
face for long periods of time, chemical weathering can
operate for millions of years, contributing to pervious The main surficial process in continental environments
pathways in even the most insoluble of rocks. Figure is erosion of the landscape. There are several agents of
2.5 is a conceptual presentation of the occiirrence of erosion, but rivers are the most dominant. They carve
fractures in crystalline rocks in the Piedmont region of their own valleys, transporting the eroded material down-
the Appalachian Mountains. The higher permeability re- stream to lakes, rivers, or oceans. Rivers also receive
sides where the fractures are connected. In these rocks, the products of weathering, frequently by an overland
the enhanced fracture or joint system frequently becomes flow process.
tighter with depth. The data o f Figure 2.6 were collected
from this region and show that the average yield of wells Fluvial Deposits
per foot of depth below the position where the rocks Fluvial deposits are generated by the action of streams
are saturated with water decreases with increasing depth. and rivers. Boggs (1987) recognizes three broad environ-

Figure 2.5 Topographic and geologic features of the Piedmont and Blue Ridge region
showing fr:ictiire interconnections (from Heath, 1984).
 Cbapter 2 / Tbe Origin of Porosity and Permeability

C
lll x k-5km *

Figure 2.8 Diagrammatic cross section of an alluvial fan

(from Rust and Koster, 1984). Reprinted with permission of
0.0 0.1 0.2 0.3 0.4
Average yield per foot of well below the Geol. ASLOC. of C;mada.
the water table

Figure 2.6 Decrease in well yields (gpm/ft of well below

the water table) with depth in crystalline rocks of the States- mental settings for fluvial systems: braided rivers, mean-
ville area, North Carolina. Numbers near points indicate the dering rivers, and alluvial fans. Figure 2.7 contrasts the
number of wells used to obtain the average values that de- geometry of meandering and braided streams. According
fine the curve (from LeGrand, 1944). to Cant (1 982), meandering streams generally produce
linear shoestring sand bodies that are aligned parallel to
the river course, and these are normally bounded below
and on both sides by finer materials. The shoestring sands
are many times wider than they are thick. The braided
river, on the other hand, frequently produces sheetlike
sands that contain beds of clay enclosed within them.
The reason postulated by Cant (1982) lies in the meander
width, with the braided streams capable of extensive
lateral migration, whereas the meandering streams are
considered to be more rigorously confined in narrow
channels. Braided rivers are characterized by many chan-
nels, separated by islands or bars. Meandering streams
have a greater sinuosity and finer sediment load.
Alluvial fans form at the base of mountains where ero-
sion provides a supply o f sediment (Figure 2.8). Fans
can occur in arid areas, such as those in Death Valley,
California, and in humid areas as well. The upper part
of the alluvial fan is characterized by coarse sediment
due to confinement of flow to one or a few channels.
The toe of the fan has the finest sediment, where more
than one channel persisted over long periods of time.
Alluvial Valleys The most important alluvial valleys in
the United States are shown in Figure 2.9. The sand
and gravel deposits are in hydraulic communication with
their stream systems, which provides for their continual
replenishment. The more permeable material occurs in
clearly defined deposits that normally do not extend be-
yond the flood plain. Figure 2.9 does not differentiate
between those channels cut by glacial melt water, such
as those in the Midwest, and those stream channels not
affected by glacial melt water, such as those in the south-
tat ern Appalachians. Nor is there any distinction between
Figure 2.7 Contrasting the geometry of ( a ) meandering meandering and braided patterns. Alluvial valleys are fre-
and ( 0 ) braided rivers (from Walker and Cant, 1984).Re- quently underlain by silt and clay deposits and are among
printed with permission of the Geol. Assoc. of Can;ida. the most productive aquifers in the United States. Heath
 2.2 Continental Environments Weathering 13

(1982), in his modem classification of ground-water re- extending to the basin bottoms. The basins themselves
gions in the United States, established the alluvial valley are characterized by block faulting, volcanism and intni-
system as a separate entity, independent of their geo- sion, high average elevation, and high average heat flow.
graphic occurrence. Also noted on Figure 2.2 is an extensive sand and gravel
As rivers become rejuvenated from time to time, they region in the central United States extending from the
cut through their own deposits, with the margins of the southern part of South Dakota into Texas. This is part
original valley floor left as terraces (Figure 2.10). These of an extensive alluvial apron deposited on the sloping
terraces also contain permeable materials and are consid- bedrock surface extending from the foot of the Rocky
ered part of the alluvial valley system. In Figure 2.10, the Mountains eastward. Although much of the apron has
current flood plain alluvium is the youngest deposit and been removed by erosion or otherwise dissected, the
the terrace occupying the highest ground is the oldest remaining part is known today as the Ogallala Formation.
and was part of a flood plain prior to rejuvenation. Pumping from the Ogallala for irrigation purposes is the
one factor accounting for the agricultural prosperity o f
Alluvial B a s h s Alluvial basins in the United States the High Plains. Unfortunately, this prosperity is threat-
occupy a discontinuous region in excess of 1 X lo6 km' ened because of depletion of the ground-water resource.
extending from the state of Washington to the western Other than the Ogallala, which was deposited during
tip of Texas. They are demonstrated on Figure 2.2 as the Late Tertiary time, alluvial fans are noted in stratigraphic
patterns of pores in sand and gravel. The material filling sequences of earlier ages. The Van Horn Sandstone of
most of the western basins was derived by erosion of West Texas is an example of a Precambrian fan system,
the adjacent mountains, with alluvial fan development consisting of massive conglomerates that grade into thin-

$ Mrdfm I

I Fareset
cross strata
Iicross
Trough-lrls
s?rata
, andcrasstrough-fill
I strata
I
I

Figure 2.10 Cross section showing associations between al- Figure 2.11 Facies and sedimentary stnictures in Viii Horn
luvial valley and terrace development. Sandstone (from McGowen and Groat, 1971).
20 Chapter 2 / Tbe Orfgia of Porosity atid Permeability

ner pebbly sandstones, and finally into mudstones (Figure Terminal moraine
Recession$ moraine
2.1 1). Triassic sediments on the margins of the Newark
Basin are also made up of the remnants of alluvial fans and
were noted by Meinzer (1923) for their water-yielding
characteristics.
Figure 2.13 Deposits associated with a hypothetical ad-
Eolian Deposits vance and retreat of a glacier.
Eolian is a term used in reference to wind erosion or
deposition. There are two types of wind deposits: loess,
well-sorted sand and gravel resulting from glacial melt
which is an unstratified deposit composed of uniform
water (outwash plains). Buried bedrock valleys are valleys
grains in the silt-size range, and dunes or drifts composed
of sand. Loess deposits are frequently associated with that are no longer occupied by the streams that cut them.
glacial terrane, whereas dunes and drifts are associated The till sheets are the most extensive deposit and consist
of an assortment of grain sizes ranging from boulders to
with deserts. The porosity of loess is very high, 40 to
clay and are normally of low permeability. Frequently,
50%, but it is not a good transmitter of water because of
the till is fractured and contains a higher permeability
the poor connectivity of the pore space. The best exam-
ples of ancient eolian deposits are in the Colorado Plateau than was present originally.
The advance and retreat of continental glaciers gener-
and include the Jurassic Navajo Sandstone and the En-
ally results in a complex vertical stratigraphy. This is
trada, Wingate, White Rim, Coconino, and Lyons Sand-
stones. Both the Navajo and the Wingate form what is demonstrated on Figure 2.13 for a hypothetical case. The
basal till sheet marks the advance of a glacier across a
known as the La Plata Group, noted by Meinzer (1923)
for its water supply. Other eolian deposits made up of region. The terminal moraine marks the farthest advance,
with subsequent melting forming the glaciofluvial depos-
dunes and drifts are expected to have a R!aSOndbly
its downstream from the moraine. We also note the pres-
high permeability.
ence of lacustrine deposits associated with a glacial lake
that was trapped between the retreating glacier and the
Lacustrlne Deposits
end moraine complex. The recessional moraine marks
The term lacustrine pertains to lakes, which currently
form about 1%of the Earth’s surface. Ancient lake depos- the position where the retreating glacier was temporarily
halted, with its melt water forming the lake and associ-
its may include rock types ranging from conglomerates
ated delta deposits. This is only one simple example of
to mudstones, as well as carbonates and evaporites. La-
a complex Stratigraphy.
custrine deposits formed during the glacial period are
generally characterized by low-permeability silts and
clays, with an occasional high-permeability delta associ- 2.3 Tbe Boundary Between Continental
ated with the terminus of rivers. Ancient lacustrine depos-
its include the Morrison Shale in the Colorado Plateau,
and Marine Environments
the Chugwater Series of Wyoming, and the Green River The environments that occupy the boundary between
Formation in Utah, Wyoming, and Colorado. These are the continents and the ocean have been referred to as
all low-permeability rocks. marginal marine (Boggs, 1987). Within this boundary,
the following environments may be identified: deltas or
Glacial Lkposits deltaic systems, beach and barrier island systems, estua-
The glacial environment is a composite one and incorpo- rine and lagoonal systems, and tidal flats. A delta is a
rates fluvial, lacustrine, and eolian environments (Figure fluvial deposit that is built into a standing body of water.
2.12). The high permeability resides in the glaciofluvial Deltas consist of an upper deltaic plain, with characteris-
depositional environments, including the alluvial valleys, tic braided or meandering patterns above the influence
the alluvial deposits in buried bedrock valleys, and the of tides; the lower deltaic plain, which extends from
the low-tide shoreline landward to the uppermost tidal
influence; and the subaqueous delta plain, which is sub-
Outwash alain merged (Coleman and Prior, 1982). The finer-grained
material is normally transported seaward in the subaque-
ous region. Ancient deltaic systems have been identified
in the Illinois Basin and in Gulf Coast sediments. In the
Gulf Coast region, both the Frio and the Wilcox are well-
known deltaic aquifers. The features and occurrence of
ancient deltas are discussed by Eliott (1978).
The beach and barrier island environment consists of
offshore bars or barrier islands with shallow lagoons be-
Figure 2.12 Glaciofluvial deposits in glaciated terrain. tween the islands and the mainland (Figure 2.14). Such
 Next Page

2.4 Marine Envfronments 21

Lateral and Vertical Succession of Strata

The term facies is defined as any areally restricted part
of a designated stratigraphic unit that exhibits a character
significantly different from those of other parts of the
unit. Thus, a lithofacies is a distinct unit, that is, sand,
that may grade into finer material (silt and clay) as one
moves from the shoreline toward the deeper parts of the
ocean. The concept of a facies is one that relates to the
nature of the depositional environment. As shorelines
shift over geologic time, the deposits of one environment
will eventually underlie the deposits of yet a younger
and different environment. With a transgressing sea, near-
shore deposits become progressively overlain by deeper-
Pfgure 2.24 Subenvironments of a barrier island system water environments (Figure 2.15a).With a regressing
(from Reinson, 1984).Reprinted with permission of the sea, the order is reversed. Figure 2.156 shows a complete
Geol. Assoc. of Canada. transgressive- regressive sequence. Such sequences are
common in many parts of central and southern United
States (Driscoll, 1986). Frequently, periods of erosion
coasts normally form where the offshore terrain is following the regression will remove one or more of
smooth and the seas shallow. The beach and barrier island the layers. Shifts in the shoreline bring different major
system results in a narrow body of sediments parallel to environments into contact. A seaward shift of the shore-
the shoreline. Numerous sandstones have been recog- line (regression) can result in the establishment of la-
nized as beach and barrier island sytems, including the goons and beach-barrier systems on older finer-grained,
Gallup and La Ventana Sandstone in New Mexico, the deeper-water sediments; a landward shift of the shoreline
Muddy Sandstone in Wyoming, and the Mission Canyon (transgression) can result in the formation of beach de-
Sandstone in the Williston Basin. These are permeable posits on finer-grainedsediments associated with lagoons
formations, with the Mission Canyon being part of the and marshes.
Madison Group, an important aquifer system in the Wil-
liston Basin.
Estuarine and lagoonal environments consist of the
estuary, which is the lower part of a river open to the
sea, and lagoons, which are shallow seawater bodies such
as sounds or bays. Tidal flats are the regions affected by
the rise and fall of tides. One of the better documented
tidal flat environments is the Upper Dakota Group along
the Front range in Colorado. The dominant stratigraphy
consists of fine-grained tidal flat deposits overlying
thicker coarse-grained channel deposits. The deposi-
tional environment for the Dakota Group in general
ranged from fluvial to marine in a marginal marine envi-
ronment, with the greatest amount of sediment deposited
as deltas in southward-spreading seas.

2 8 Marine Environments
The marine environment extends from the continental
shelf to the oceanic ridges. Shelf sediments normally
consist of sands or, where tropical to subtropical climates
persist, carbonates. Ancient shelf deposits are well
known in the geologic record. Examples of modern car-
bonate platforms include the Yucatan shelf and the east-
ern Gulf of Mexico off Florida. The lower part of the <Fd
Edwards Formation in Texas has been cited as a possible Figure 225 ( a ) Cross section of sediments associated with
carbonate shelf deposit (Wilson and Jordan, 1983). Deep- a transgressing sea and (b) transgressive-regressive sedimen-
water sediments can be of various kinds, but generally tary sequence (from Driscoll, 1986). Reprinted with perniis
are muds of one sort or another, along with carbonates. sion from Groundwater and Wells, 2nd Ed., 1986.
Previous Page

-
22 Cbapter 2 / The Ongin of Porosity atid Pertneabflity

Pleistocene

Ffgure216 (a) Dakota aquifer system and ( b ) associated confining kdyers (after Bredehoeft, J. D., C. E. Neuzil, and P.C.
D.Milly, 1983).

The points discussed above are perhaps best demon- ate Madison Group waters to enter the basal sands of the
strated by the Dakota aquifer system and associated con- i)akota Sandstone. Swenson (1968) considers this a major
fining layers (Figure 2 . 1 6 ~ )Here
. we note three major source of recharge to the Dakota Sandstone. The Creta-
aquifers, Mississippian carbonate rocks collectively re- ceous Shale confining layer, which contains not only
ferred to as the Madison Group, the Inyan Kara Group shale but some minor chalk and liniestone aquifers, is
Sandstones, and the Newcastle Sandstone. All of these made up largely of three Cretaceous Shales: the Mowry-
aquifers crop out on the eastern flank of the Black Hills. Belle Fourche, Carlile, and Pierre.
The confining layers separating these aquifers are largely
shales and include the Madison Group confining layer,
the Skull Creek Shale, and the Cretaceous Shale confining
Ancestral Seas and Their Deposits
layer (Figure 2.16b). The Skull Creek Shale thins eastward
and eventually pinches out, permitting the Inyan Kara We have already seen Figure 2.2, which was presented as
and Newcastle Sandstones to merge and form the Dakota a map showing the major types of openings in dominant
Sandstone of eastern South Dakota. The Madison confin- aquifers across the United States. Let us supplement Fig-
ing layer also pinches out to the east, permitting carbon- ure 2.2 with yet another map that gives the ages of the
 2.4 Marine Environnients 23

'u' Explanation
Gsctogie Age Type of rock

Figure 2.I7 Geologic age of major rock groups in the United States (after U.S. Geological Survey, 1970).

major rock groups associated with the porosity patterns Virtually all these rocks are marine or marginal marine
(Figure 2.17). Excluding the environnients previously dis- in origin and are related to ancestral seas that inundated
cussed, three major groups remain: parts of the continent during periods of geologic time.
We cannot discuss all the formations in this grouping.
1. The Paleozoic consolidated sedimentary group occu-
but we can discuss some of the major ones.
pying a large southwest-northeast trend across the
United States, with a few areally restricted occur-
The Paleozoic Rock Group
rences elsewhere.
The paleogeographic maps of Dunbar and Waage ( 1969)
2. A rather dispersed pattern of unconsolidated and show that seas occupied virtually all the southwest-north-
semiconsolidated Mesozoic rocks. east region of the United States from the Cambrian period
3. A continuous pattern of unconsolidated and semi- through the Carboniferous (sometimes divided into thc
consolidated sedimentary rocks of Cenozoic age ex- Mississippian and Pennsylvanian periods). The resulting
tending along the eastern coastline from New Jersey deposits consist chiefly of sandstone. shale. limcstonc.
to Texas along with a few areally restricted occur- and dolomite, with an aggregate thickness o f srver;il
rences elsewhere. thousand feet. During this period of time there were
24 Cbapter 2 / Tbe Origin of Porosity and Permeability

several regressions and transgressions with a substantial Tbe Cenozoic Rock Group
amount of erosion. Except for those sediments along the The Cenozoic era is noted for the rise of the Rocky
eastern margins (the Appalachian Geosyncline region), Mountains, the extrusion of lava beds in the Columbia
the formations are nearly horizontal. It was during this River Plateau, block faulting in the basin and range region,
period that the Cambrian-Ordovician aquifer was depos- and the advance of glaciers across the northern part of
ited, extending from the Northern Great Plains Williston the continent. Our interest here, however, is restricted
Basin) eastward. This formation occurs near the land to the Cenozoic band of rocks that bound the eastern
surface in the Appalachian Mountains and in the Midwest. coast of the United States, extending around most of the
Cambrian sandstones in the Midwest include the Mount Gulf of Mexico. These are marine deposits and, once
Simon and Eau Claire, whereas Ordovician sandstones again, the result of continental inundation by seas.
include the St. Peter. Important carbonate rocks of Or- Seas covered the eastern continental shelf from New-
dovician age include the Plattville and Galena Formations. foundland to the Gulf Coast of Mexico throughout the
The Maquoketa Shale marks the top of the Ordovician Cenozoic and, on occasion, spread over the coastal
in Wisconsin and Illinois. Thus, the Cambrian-Ordovician- plains. Deposits from these seas have a minimal thickness
Maquoketa Shale sequence constitutes a well-defined hy- at their western-most outcrop and thicken toward the
drostratigraphic unit. Other important aquifers above the Atlantic. The deposits were progressively tilted seaward.
Cambrian-Ordovician Group include the Silurian age Carbonate facies are not abundant in the northern parts
Monroe (restricted to Ohio) and the Niagara Groups, of the region and are restricted to Georgia and Florida.
both of which are carbonates. There are relatively few The Mississippi River was formed by Miocene time, car-
good water-producing rocks in the Devonian and Carbon- rying large volumes of sediments to the Gulf of Mexico,
iferous. Extensive deposits of Permian age occur only in building the Mississippi Delta. Thus, Cenozoic sediments
the southwest part of this region and are poor water in the Gulf of Mexico region are largely of the clastic va-
producers. Thus, from the perspective of water supply, riety.
the Cambrian-Ordovician and alluvial valley systems in Within this geologic setting, numerous sandstones
the Midwest are of considerable importance. were deposited in the Atlantic coastal plain region, giving
way largely to carbonate groups to the south. This fact
Tbe Mesozoic Rock Group motivated Heath (1984) to break from precedent in his
The paleogeographic maps of Dunbar and Waage (1 969) modern classification of ground-water regions to estab-
for the Mesozoic era show a rather dispersed seaway lish the Atlantic and Gulf coastal plain region, which
pattern, which accounts for the dispersed pattern of Me- he describes as interbedded sand, silt, and clay, and a
sozoic rocks shown in Figure 2.17. Seaways were very southeast coastal plain region, which he describes as
restricted from the Triassic period through the Early Ju- layers of sand and silt over carbonate rocks.
rassic, except for what is now the west coast of the
continent. During the Late Jurassic period, the region
that now constitutes the Rocky Mountains was inundated magenesis in Marine Environments
along with some parts of Texas and Louisiana. During
The progressive burial of sediments in depositional envi-
Early Cretaceous a major seaway inundated most of Texas
ronments is accompanied by physical and chemical
and Louisiana and the northern Rocky Mountain states.
changes that affect the sediments. These physical and
During Late Cretaceous a major seaway extended from
chemical changes occur because of increases in the over-
Mexico through the entire Rocky Mountain region, in
burden pressure, increased temperature, and the chemi-
both the United States and Canada.
cal interactions between minerals and migrating pore
The marine formations associated with the Triassic
water. From a hydrologic perspective, diagenesis is im-
seas in the western United States produced few aquifers
portant because of its effect on pore space. Some pro-
worth noting. Some of the Jurassic rocks in the Colorado
cesses act to reduce the existing pore space, and others
Plateau are good water producers, for example, the La
act to enhance it. An end result of these processes is the
Plata Group, but these have already been described as
lithification of the sedimentary pile, that is, the conver-
eolian deposits. By far, it is the Cretaceous period with
sion of sediments to consolidated sedimentary rocks.
its extensive seaway that gave rise to the major water
producers in the western United States. The Early Creta-
ceous produced the Trinity Group in Texas, the Dakota Porosity Reduction: Compaction and
Sandstone in the Black Hills, the Cloverly Sandstone in Pressure Solution
Wyoming, and the Kootenai in Montana. The Upper Cre- The term compaction refers to a diagenetic process
taceous produced the Woodbine in Texas and the Dakota where the weight of the overburden contributes to poros-
Group in the Northern Great Plains, both of which are ity reduction from some initial higher value. Compaction
noted in Figure 2.17, and the Milk River Sandstone in can take place by changing either the arrangement of the
Alberta, Canada. grains or their shape. In a fluid-saturated environment, a
 2.4 Marine Enuironments 22

Below 350 m, the relationship is a linear one, with poros-

ity decreasing from 38 to about 18%at 6000 ni.
The decrease in porosity versus depth for sandstones
shown in Figure 2.18 does not necessaril!. prove thitt
grain rearrangement in closer packings is the responsible
mechanism for porosity reduction. Another candidate is
pressure solution, where grains dissolve at grainto-grain
contacts where the stress is greatest. If the material so
dissolved is precipitated locally on the unstressed surface
of the solid exposed to the pore space, the bulk volume is
reduced at the expense of the pore volume. A conceptual
visualization of this process given by Weyl(l959) denion-
strates that the reduction in bulk volume is compensated
by a commensurate decrease in pore volume (Figure
71 I I I I I 2.19).
0 10 20 30 40 50 The deformation resulting from pressure solution is
Porosity
directly proportional to the grain-to-grain stress estab-
flfgure2.28 Porosity versus depth curves. Curve A from lished in the rock. Thus, we expect it accompanies basin
Athy (1930) for shales; ciirvr R from Blatt (1979) for santl- loading in depositional environments and is driven by
stones. Data for Blatt's curve represent 1000-ft iiverqes of the same stress as that responsible for a closer packing
17,367 porosity measurements (from an unpublished ni:inii- of the grains. There is an abundance of literature on
script by Atwater and Miller). pressure solution in rocks, including overviews by Rutter
(1983) and DeBoer (1977). However, there still remilins
some controversy regarding the driving forces. Boswortli
decrease in porosity requires an expulsion of the pore (1981) argues for a plastic deformation mechanism,
fluids from the sediments, with the total volunle of fluid whereas Green (1984) suggests that volume transfer
expelled balanced by the total volunie of porosity loss. creep during phase transforniations has many charncteris-
The term nonnal compaction is reserved for the condi- tics similar to that of pressure solution. Whateyer the
tion where the fluid expulsion takes place more or less mechanism, the reduction of porosity causes a rather
concurrently with the deformation such that there is no large reduction in permeability. Hsu (1973, for ex;iniple.
apparent increase in the pressure of the fluid. Disequilib- determined that permeability of sandstones in the Veil-
rium compaction implies a time lag between geologic tura Basin in California decreased by three orders of ni;tg-
loading of the sediments and the expulsion of their pore nitude over 3000 meters of depth.
waters, meaning that some of the load is carried by tlie Palciaiiskas and Donienico (1989) present a Sinlpk
pore water, resulting in fluid pressures higher than would constitutive law for porosity reduction based on interpen-
normally be expected in a hydrostatic (oceanic) environ- etration of spheres and the melting heat at points of
ment. As a geiieral rule, tlie finer grained the sediment, contact. Figure 2.20 shows their theoretical curve for
the greater its initial porosity, and the greater the poten- porosity reduction with depth for sandstone. Rased on
tial porosity decrease. Hence, clays will be affected by this model, they concluded that the pore volume change
compaction to ;I greater extent than sands, and carbon;ite for sandstone subjected to this inelastic deformation is
muds may not be materially affected if crystallization
occurs early in the depositional history.
Measured porosity versus depth curves for shales and Refore During Pressure Soiut:on
sandstones are shown in Figure 2.18. The curve for shale
was produced by Athy (1930) from a study of Paleozoic
shales in Oklahoma. Athy proposed the relationship
VI = n,,e-"' (2.4)
where n is porosity, n,, is the average porosity of near
surface chys, z is the depth below surface, and a is an
empirical constant, equal to 1.42 X lo-' n1-I for his data.
The data of Athy suggest that given a porosity of surface
clays on the order of 0.4 to 0.5, the porosity of compacted
clay (shale) is reduced to about 0.03 at 2000 m. The Figure 2.19 Pressure solution of identical spheres of sim-
curve for sandstone (Blatt, 1979) is based on some 17,000 ple cubic packing (from Weyl, J. Geophys. Res.. v. 64. p.
measurements of Late Tertiary sandstones in Louisiana. 2001 -2025, 1959. Copyright by hmer. Geophys. IJnion).
26 Cbapter 2 / The Otjgfnof Porosity ard Permeabilfty

3. Chemical diagenesis of sandstones is a kinetic pro-

cess involving reactions between minerals and a mov-
ing aqueous solution. The main source of water in
10 -
the depositional environment is from dewatering of
shales interbedded with sandstone.
-
I
c
0)

0 The first two statements appear to be amply supported

0
-5
2 20- by thin section and field studies (see, for example,
Schmidt and McDonald, 1977). The third statement is a
B Porosity hypothesis, albeit a reasonable one. Before accepting any
30 - of these statements, there are several questions that must
be addressed, including the responsible driving forces
for water movement from shale into sandstone in an
oceanic (hydrostatic) environment, the mass transport
40 mechanisms, and the reactions that contribute to the
0.0 0.1 0.2 0.3 0.4 0.5
Porosity diagenesis. These are major topics that will be covered
Pfgure 2.20 Theoretical upper bounds of porosity varia- elsewhere in this text. However, at this time a short
tion with depth in sandstone. The 17 data points represent statement on the driving forces for fluid movement may
looOfi averages of 17,367 porosity measurements of Late be in order in that these driving forces relate to the
Tertiary sandstones from an unpublished manuscript by At- equilibrium and disequilibrium concepts introduced ear-
water and Miller, as presented in the work by Blatt (1979) lier, as well as to the process of pressure solution.
(from Palciauskas and Domenico, Water Resources Res., v. Simply, fluid moves from the shales into the sands
25, p. 203-213, 1989. Copyright by h e r . Geophys. Union). because the shales are in a greater state of disequilibrium
compaction than the sands. Indeed, if sands are continu-
ous to the overlying oceanic waters, they are likely under-
going equilibrium compaction. Given this statement,
substantially larger than the pore volume changes associ- there are two questions that may be pursued further: (1)
ated with grain rearrangement. Also shown on this figure For a given thickness of overburden and in the absence
are the data of Blatt (1979) for sandstones. Since the of fluid flow, will the fluid in all rocks be equally pressur-
sandstone cores contain unspecified amounts of pore ized? (2) What is responsible for maintaining high fluid
filling clays and cements, the measured porosity should pressures in certain rocks and not in others over long
be lower than the theoretical curve, but ideally should periods of geologic time; that is, what contributes to
have a similar slope with depth. The theoretical curve disequilibrium compaction?
given in Figure 2.20 is in good agreement with observa- The first of these can be addressed immediately with
tions by Maxwell (1964). who stated that porosity of a statement describing the fluid pressure production in
sandstones diminished with depth to less than 5% poros- the absence of fluid flow
ity below 10 km.
dP = xdu (2.5)

Chemical Rock-Water Interactions: where P is fluid pressure, u is stress, and the factor y, is
Secondary Porosity in Sandstones called the pore pressure coefficient. If x = 0, dP = 0,
From a strictly chemical perspective, porosity can be and if x = 1, fluid pressure change equals the change in
reduced by cementation, replacement of one mineral by overburden pressure do.The pore pressure coefficient
another, and recrystallization. Conversely, porosity can expresses the change in fluid pressure with changes in
be enhanced by dissolution of grains and cements. A very stress in the absence of fluid flow; that is,
large percentage of existing porosity in sandstones is
not original intergranular porosity, but a secondary or
solution-type porosity created by some dissolution pro-
cess. The following statements by Hayes (1977) sum up where the subscript M means at constant fluid mass, that
much of modern thinking on porosity development in is, no loss or gain of mass due to fluid flow. If the rocks
sandstones: and the water contained within them are totally incom-
pressible, the pore pressure coefficient is zero, and the
1. Primary intergranular porosity is subject to almost fluid pressure would not be affected by the stress
total destruction by several mechanical and chemical changes. Here the total load is carried by the rigid matrix.
processes early in the diagenetic period. On the other hand, if virtually all the overburden is car-
2. Later in diagenesis, secondary porosity can be pro- ried by the fluid and none by the matrix, the pore pres-
duced by dissolution. sure coefficient is close to one, and the fluid pressure
 2.5 Upl@, Diagenesis, and Erosion 27

starts to approach the overburden pressure. As the indi- tional environment. In some places, the displacement has
vidual grains, the pores, and the fluids within the pores all been quite effective, with the original formation waters
exhibit compressibility to some extent, the pore pressure being completely displaced from the formation, at least
coefficient will be greater than zero but less than one. in their current continental positions. In other places,
Work by Palciauskas and Domenico (1989) suggests that the displacement process has not been effective. At any
the pore pressure coefficient is on the order of 0.99 for rate, the rocks apparently undergo yet another stage of
clays, 0.83 for mudstones, 0.67 for sandstones, and 0.25 diagenesis, rather unevenly perhaps, this time meteori-
for limestones. Thus we note that the more rigid rocks cally driven. With regard to porosity development, which
(that is, limestones and, to a lesser degree, sandstones) is our main concern here, sandstone diagenesis merely
are not capable of being excessively pressurized even in means yet another generation of secondary porosity de-
the absence of fluid flow. On the other hand, the fluid velopment by dissolution of those mineral phases that
pressure in clays and mudstones can approach the over- are not stable in the meteoric regime. For carbonates,
burden pressure because most of the incremental load again from the perspective of porosity development, our
is carried by the water in the pores. Hence, depending main interest here is the dissolution of mature carbonate
on the magnitude of the pore pressure coefficient, it may terrain uplifted into a ground-water circulation system.
be difficult to generate high fluid pressures in some rocks
relative to others, irrespective of the rate of loading and The Style of Formations
limitations on the rate of fluid flow.
Associated with Uplifl
Low-permeability sediments, such as shales, are those
in which the fluid pressures will be maintained the lon- Figure 2.21 shows three styles of development for forma-
gest because they do not drain efficiently. This statement tions associated with uplift. Figure 2 . 2 1 shows
~ the geo-
is only partially correct, but it is sufficient for our pur- logic features associated with the Rocky Mountain devel-
poses here. It follows that the driving force for fluid opment throughout the Cenozoic era. This style of
movement is the pressure differences between the fluid development gives rise to exposed outcrops in the u p
in the various rocks in the depositional environment, and lifted area, with the Dakota Group referred to earlier
this pressure difference is established because all rocks providing a classic example. Several of the beds pinch
cannot be equally pressurized due to compressibility fac- out with depth and distance eastward, whereas others
tors. Those that are pressurized the most maintain their continue across the basin and outcrop or subcrop in
pressures longer due to permeability factors. structurally undisturbed areas. The Dakota Sandstone, for
Boggs (1987) cites seven major diagenetic processes example, outcrops along the Black Hills in South Dakota
other than compaction, at least four of which have the and underlies about 90% of Nebraska, and both Iowa and
potential for effecting porosity and permeability in depo- Minnesota to some extent. The Dakota also occurs in the
sitional environments. These are replacement, where one Williston Basin in North Dakota and in the Denver Basin.
mineral is replaced by another; inversion, which is the where it contains salt water.
replacement of a mineral by its polymorph (polymorphs Figure 2.21b shows those features associated with the
are minerals having the same chemical composition but uplift of the Colorado Plateau, which was raised vertically
different crystal forms); recrystallization, or a change in along with the Front Range of the Rocky Mountains. The
crystal size and shape, generally resulting in an increase classic basin-type structure is evident, with outcrops or
in grain size; and dissolution, a selective process that subcrops circumscribing the outer rim of the basins. It
removes the most soluble minerals for a given set of is noted that water can enter this type of basin along its
environmental conditions. entire circumscribed outcrop area.
Yet a third type of association of formations with moun-
-
2.5 Uplifl, magenesis, and Erosion
tains is shown in Figure 2.21c, but the relationship here
is a passive one. This setting is supposed diagrammatically
With uplift and erosion, the sedimentary sequence we to represent the Atlantic coastal plain, where the sedi-
have been following has gone through one complete ments were derived largely from the Appalachians and
cycle. Over all geologic time, weathering and erosion act were not "uplifted" with them. As mentioned previously,
to modify any elevated landscape, with numerous rock the sediments thicken toward the sea, and, although
bodies entering the sedimentary cycle time and again. there is evidence of some eastward tilting, the association
From the perspective of permeability development, there of these sediments with tectonics is quite different from
are several important factors associated with the uplift of that mentioned earlier for the western United States.
sedimentary rocks from their marine or marginal marine However, because of their topographic relationships, wa-
environment. First, the rocks are thrust into a meteoric ter can enter the formations in the outcrop areas and
regime so that fresh water can continually enter their move eastward. Relatively freshwater zones extending
outcrop areas. The entering meteoric water mixes with several miles offshore into the Atlantic Ocean are not un-
and displaces the original formation waters of the deposi- common.
28 Cbapter 2 / Tbe Origfnof Porosity and Permeabflfty

Explanation Fmsh water DSanilslone

Figure 2.22 Style of uplift (from Heath, 1984).

Following Heath (1984), we have attempted to show gressive one that can occur over many miles. However,
the occurrence of salty or brackish waters in association the diagrams do provide a background for a general de-
with the uplift patterns of Figure 2.21. In these diagrams, scription of our observations, several of which can be
the transition from fresh water to salt water is portrayed made here.
as an abrupt interface. As noted, the transition is a pro- One notable feature of Figure 2.21 is that the younger
 2.6 Tectonism and tbe Formation of Fractures 22

permeable units can be more or less completely flushed satisfied in unconfined carbonate aquifers, which readily
of their original waters, while the deeper (older) ones allow for recharge to and discharge from the aquifer.
are only partly flushed,with the degree of flushing getting Older carbonate rocks have low porosities, serving to
poorer with depth. One possible reason for this is that concentrate flow along the fractures. A positive feeclbrick
the younger beds may have had more recent regressions loop of dissolution and transport is the logical result of
and the formations may have been subjected to long- these processes and results in a self-organized pattern of
term meteoric flushing in the depositional environment solutionallyenlarged fractures. These have been niodeled
prior to uplift. Further, they are closer to the surface and using both physical models (Ewers, 1982) and niathem;iti-
may be replenished from meteoric sources other than cal models (Dreybrodt, 1990; Palmer, 1991: Howard iind
those available at the outcrop. A second feature, as shown Groves, 1995).
in Figure 2.216, is the effect of basin perimeter recharge. The threshold, above which there is a flow in intercon-
Down-dip flushing in this case has the effect of concen- nected solutionally enlarged fractures, appears to be ;I
trating the original formation water and can be effective few tens of meters at most. At a scale of hundreds of
only for short distances from the intake areas. As the meters the patterning results in conduits, which are inter-
formation fluids become more concentrated with depth, connected solutionally enlarged fractures wider thm
there is a marked density difference between the incom 1 cm, and in which turbulent flow is likely. Finally where
ing meteoric waters and the original fluids, and density flow paths are kilometers in length, then patterning re-
stratification occurs. For these cases, the lighter fluid will sults in caves (conduits accessible to humans, thus gener-
actually attempt to move above or around the more dense ally larger than 1 m in diameter). The vast speleological
one if at all possible, or actually discharge cross-forma- literature is largely concerned with this largest scde.
tionally where further movement is impeded by the more There has been some investigation of techniques to
dense fluid. I t is difficult to conceive of any manner in use in wells to identify the extent and size of enlarged,
which flushing can be effective in such environments. interconnected fractures (Hickey, 1984;Price, 1994:Sau-
ter, 1992; Worthington and Ford, 1995). The only defni-
tive test for interconnectivity is the direct measurement
Secondary Porosity Enhancement
of flow velocity by point-to-point tracer testing. Many
in Carbonate Rocks thousands of such tests have been successfully completed
According to Stringfield and LeGrand (1966), if the depos- in carbonate terrains and almost always have revealed
its overlying a carbonate terrain are of low permeability ground-water velocities > 100 m/day.
and are thick, and ifthe carbonate was never elevated into Extensive near-surface solution in carbonate rocks re-
a ground-water circulation system, then little secondary sults in a karst landscape. Sinkholesare the primary land-
porosity will develop. On the other hand, elevation into form associated with mature karst. They can be classed
a ground-water circulation system at either an early stage generally into four types: (1) the classical solution sink,
or a late stage will lead to the development of secondary which involves a closed depression in the bedrock with
porosity and permeability.Hence, if a carbonate is uncon- or without a soil cover, (2) a cavern collapse sink that
formably overlain by a low-permeability rock such as develops when the roof falls in on a near surface conduit,
shale, there is little question that the permeability of the (3) subsidence sinks caused by upward stoping [sic]of
carbonate rock has been enhanced by solution. The same a collapsed solution cavity through substantial thick-
can be stated for carbonate rocks in direct contact with nesses of bedrock, and (4) soil piping sinkholes, which
glacial till laid down during the Pleistocene, for such represent the abrupt collapse of a soil arch on mechani-
carbonates represent long-time erosional surfaces that cally sound bedrock through soil loss into solution cavi-
were characterized by paleohydrologic drainage systems. ties (White and White, 1987). Sinkholes are a serious
On the other hand, secondary permeability development geologic hazard. Considerable damage has resulted from
cannot be assured when limestones are in conformable their formation beneath highways, railroads,dams. reser-
contact with overlying marine shales. Thus, the signifi- voirs, pipelines and vehicles.
cance of the unconformity should not go unnoticed in
the geologic record, especially when the concern is with
secondary porosity and permeability development in car- 2.6 Tectonism and
bonates.
Major solutional enhancement of permeability in car-
the Formation of Fractures
bonate aquifers is dependent on the circulation of ground A fractured rock may be regarded as intact rock bodies
water. Thus, the capability must exist for (1) chemically separated by discontinuities. The rock bodies are gener-
aggressive ground water to recharge the system, (2) the ally considered to be impermeable, but frequently some
rock unit to transmit water through the fractures, and permeability has to be attributed to these intact rock
(3) ground water to drain from the system (Stringfield blocks. From the perspective of ground-water flow. a
and LeGrand, 1966). These three conditions are best fluid-conducting medium can be described as fractured
30 Chapter 2 / The orfgln of Porosity and Permeability

if the majority of the flow takes place through discrete

fracture channels that in some fashion form an integrated,
interconnected system.
Van Golf-Racht (1982) defines a fracture as the surface
along which a loss in cohesion takes place. A fault is a (d (h)
fracture where displacement l ~ a soccurred, whereas a
joint is a fracture along which no notable displacement Figure 2.22 Schematic illustration of most common frac-
has occurred. From a purely geologic perspective, these tiires associated with a fold: (u) type I, (b) type 11. both
definitions are adequate; from a hydrologic perspective, types maintain a consistent relation to bedding, but not to
folding (from Stearns and Friedman, 1972). Reprinted by
the most important factors are fracture properties irre- permission.
spective of displacements, including orientation, density,
aperture opening, smoothness of fracture walls, and-
possibly above all else-the degree of connectivity. For
example, given a set of fractures in a rock, none of which tiires such as faults and folds. Fractures associated with
extend over the scale of the formation and none of which faults are thought to be caused by the same stresses that
are interconnected, the rock cannot transmit water via caused the faulting, and their strike will in general parallel
the fracture network. that of the fault. Stearns (1964, 1967) reported five types
of fracture patterns commonly associated with folding,
only two of which produced a significant fracture density.
Style of Fractudng As noted in Figure 2.22, both patterns consist of two
Fracture style is closely related to stress history and rock conjugate shear fractures and an extension fracture. The
type. Brittle rocks of low porosity are most susceptible pattern designated type I is thought to be earlier in origin
to fractures, given the proper tectonic setting. Van Golf- than type I1 and is common even on lowdipping folds
Racht (1982) cites three cases where stress-related frac- where type I1 is absent. The two patterns occur in the
tures may occur: same bed and represent two different stress states. Pat-
tern type I has been described as singIe zones of parallel
1. In response to folding and faulting.
fractures across the entire structure whereas type I1 frac-
2. Deep erosion of the overburden, which will produce tures may range in length from a few inches to several
differential stresses that can cause fractures. tens of feet. Type I1 frictures, however, consist of a
3. Rock volume shrinkage (shrinkage cracks) where wa- frdcture zone with three different orientations, whereas
ter is lost, say, for example, in shales or shaley sands. type I fractures appear to have a single orientation and
are not an assemblance of all three orientations. Hence,
Frequently, the terms microfracture and macrofrac-
Stearns and Friedman (1972) postulate that the type I1
ture are applied to describe the magnitude of fracture.
fractures, although smaller in length and aperture, have
These terms are imprecise, but a microfracture is on the
a greater degree of connectivity and may be the better
order of the grain size or smaller and is not readily de-
fluid conductors. On the other hand, because type I frac-
tected by the naked eye. Microfractures are frequently
tures are continuous over large regions, they may be an
caused by differential thermal expansion of individual
important regional pathway for fluid movement.
minerals. A macrofracture is a fracture that is readily
In comparing rock deformation under the Same envi-
detected without a microscope and can range from a
ronmental influences, Stearns (1967) notes that the den-
simple joint to a throughgoing discontinuity.
sity of fractures is dependent on the lithology of the rock
Stearns and Friedman (1972) use the following descrip-
(Figure 2.23). This is in agreement with DeSitter (1956).
tions in the study of fractures:
Several other studies attempt to relate fractures to li-
1. Conjugate fractures, or fractures related to a common thology or to regional and local structures. Harris and
origin that form under a single state of stress. Conju- others (1 960) conclude that fractures associated with
gate fractures are intersecting fractures, with the compressional deformation have certain features in com-
angle of intersection commonly found to be about mon, such as repetition and continuity of trend. The
60”. influence of lithology is again pointed out along with the
2. Orthogonal fractures, which are normal to each fact that thinner sedimentary rocks are more susceptible
other and are not related to a single stress state. to fracturing than are thicker ones. Regan and Hughes
(1949) discuss the productivity of fractured shales in
3. Regional fractures, which extend over large areas
California, where the fractures occur in chert zones and
and apparently are unrelated to local structures.
in zones of calcareous and platy siliceous shale.
Orthogonal fractures are independent of structure and Other types of fractures include those produced by
are generally well defined, even in flat lying rocks. Stnic- volume shrinkage, caused by cooling in igneous rocks
turally related fractures are associated with specific struc- such as basalts, and desiccation in sedimentary rocks.
 2.6 Tectonfstnand the Forntatiotr of Fractures 31

ity. Witherspoon and Gale ( 1977) review the relationship

between fracture permeability and fluid pressure. Quanti-
tative evaluations have been provided by Gangi (1978)
and Walsh (1978, 1981).

Connectivity
Consider a rock mass with a relatively impermeable ni;i-
trix that contains several fractures. This is shown i n Fig-
ure 2.24 for two cases with the same fracture orientation
and density. Assuming that all other parameters that con-
Lithology trol flow are held constant, the only difference between
these cases is the degree of interconnection between
Figure 2.23 Average fracture density for sever;tl conimon
rock types natunlly deformed in the Same physical environ- individual fractures. In Figure 2.2461, a condition is shown
ment (from Steams, 1967). in which none of the fracture clusters are connected to
other clusters. This prevents flow across the s;imple.
indicating that the unconnected clusters add nothing to
Landes and others (19m) conclude thitt water-bearing the permeability of the medium. Figure 2.246 shows
fractures are a general rule in crystalline rocks between clusters that are interconnected across the sample, sug-
the depths of 600 and 1000 ft below the surface. gesting that flow can be achieved across the region.
Weathered till is yet another material that shows a high Percolation theory deals with the idea of connectivity
fracture density. According to fklrdri and Hedges (1985), (Shante and Kirkpatrick. 1971; Dienes 1982). A percola-
studies in the Midwest and Canada show that the upper tion threshold is defined as the density of fractures that
part of most glacial tills has been weathered and contains intersect sufficiently to promote flow. A finite set o f frnc-
a dense fracture pattern. Few if any fYdctL1reS persist ture clusters can be interconnected, but the connections
in the lower unweathered portions of till sheets. The between various finite clusters may not exist. Above the
permeability for weathered tills is three to four orders percolation threshold, one cluster becomes "infinite" so
of magnitude larger than that for the unweathered tills. that flow can take place through the medium. However,
many other clusters may not be connected to the infinite
one. As the fracture density increases. isolated clusters
Fluid Pressure and Porosity become more rare. De Marsily (1985) gives several exam-
The in situ porosity and permeability of fractured rocks ples where the percolation threshold has been related
are generally considered to be a function of fluid pres- to the average number of intersections of a single fracture
sure. This statement holds for intergranular material as with all other fractures. as well as to the density mtl
well, but to a lesser extent. In discussing these concepts, length of the fractures. Robinson's (1982) results ;ire
Snow (1968) provides abundant evidence that fractures based on the latter and indicate that when the product
"breathe," or open and close in response to changes in of the fracture density and the square of their ;tvcr:ige
fluid pressure. As there is a direct relationship between length equals 1.5, the percolation threshold is achieved
fluid pressure and ilpertllre opening (and, consequently. and flow commences. Below this value, the fractures ;ire
permeability), there follows some relationship between insufficiently connected to promote flow.
fracture strength, frictional resistance to sliding, and fluid
pressure. Brace and others (1 968) discovered an increase
in permeability in fractured granites due to an increase
in fluid pressure, which caused a decrease in the stress
acting on the rocks. Serafin and del Campo (1965) sug-
gested that the percolation of reservoir water beneath
dams caused a widening of the fracture openings with
a consequent weakening o f the stnictiire. A series of
field experiments by Jouanna ( 1 972) revealed that I S the
normal stress increases, the permeability decreases, and
this process is irreversible. According to Sharp and Maini
(1972), the relationship between stress and pernieability
(n) (/I)
is nonlinear because the normal stiffness of the fracture
increases with decreasing aperture width. Thus, de- Figure 2.24 Representationsof a conceptual rock niash il-
creases in stress in fractured rock caused by increases in lustrating the importance of connectivity.(0)Ilnconnected
fluid pressure have a significant effect on rock pernieabil- clusters, (b) Most clusters connected.
32 Cbapter 2 / Tbe Origin of Porosity and Permeability

These results at this stage of the research are the conse- of fractures and that connectivity studies for the individ-
quence of numerical experiments wherein fracture pat- ual classes as well as any possible assemblage of classes
terns are randomly generated with a computer and the would give results that might be less abstract and more
individual intersections noted. Key questions and diffi- pertinent to the problems of flow in real rocks. That is
culties remain in identifying the percolation threshold to say, if classes of fractures can be identified from field
and obtaining field data on actual fracture densities. mapping, perhaps something can be said of a qualitative
Threedimensional data on fracture density and volumes or semiquantitative nature about the potential for con-
(Charlarx and others, 1984) are obviously even more nectivity and the degree of connectivity that might exist
difficult to obtain. Further, from our discussion of fracture on the basis of theoretical studies.
style, it is clear that there exists in nature different classes
 Chapter 3

Ground-Water Movement

3.I Darcy ’s Experimental Law and Field Extensions

3.2 Hydraulic Conductivity and Permeability
of Geologic Materials
3.3 Mapping Flow in Geological Systems
3.4 Flow in Fractured Rocks
3.5 Flow in tbe Unsaturated Zone

Considering the fact that the flow of fluid can be observed through sand.” His method was experimental, the results
in “real time” in laboratory experiments, it is not surpris- of which were published in 1856.
ing that laws have been developed to describe some of An experimental apparatus similar to that employed
the macroscopic details of the motion. Because “motion” by Darcy is shown in Figure 3.1. It consists of a cylinder
involves measurements of time and distance, the experi- having a known cross-sectional area A (L’) that cont;tins
mental observations are quantitative ones, so that the a porous medium (e.g., sand) and includes appropriate
laws take on a mathematical form. Our main law may be plumbing to flow water through the column. The cylin-
stated outright as der contains two manometers, whose intakes are sepa-
rated from one another by a distance A/ (L). The manome-
The velociQ, or distance traveled over some time inter- ters are nothing more than small open tubes. As will
val, is proportional to some driving force. become clear shortly, the elevation to which the water
level rises in a manometer is a measure of the energy
that the ground water possesses at the inlet of the manom-
The surprising feature of this statement is that-provided
eter. In an actual experiment, water flows into ;inti out
certain conditions are met -it holds irrespective of the
of the cylinder at a known volumetric flow r;w
complex details of the connected pore space described
Q (L3/T), and the elevation of water levels in the manome-
in Chapter 2.
ters, h, and h2(I, are
),measured relative to a local datum
(Figure 3.1).
The parameters of this experiment may be combined
33. Darcy’s Experimental Law to yield two key master variables. The specific discharge.
and Field Extensions q, has the units of velocity (L/T) and represents the
volumetric flow rate per unit surfxe area of the cylinder.
Henry Darcy was a civil engineer concerned with the It is determined from the experiment as Q / A . The tlimen-
public water supply of Dijon, France, in particular the sionless hydraulic gradient, (h, - h l ) / A f represents
, the
acquisition of data that would improve the design of filter change in water-level elevation in the manometers sepa-
sands for water purification. In search of this information, rated by Al. Over small distances, the gradient is ex-
Darcy set out “to determine the laws of flow of water pressed in familiar calculus terms as ilh/i/l.
34 Cbapter 3 / Ground-WaterMovement

per unit area of connected pore space. The expression

for v comes directly from Eq. 3.1

(3.2)

where n p is the effective area of flow and n, has been

introduced as the effective porosity. Here, the quantity
v is referred to as the linear or pore velocity of ground
water. The linear velocity v will always be larger than
the superficial velocity (specific discharge) and increases
with decreasing effective porosity. Finally, it is conve-
1 i datum 1 1 nient to designate the gradient ab/al as i so that Darcy’s
law may be expressed in the following forms:
Figure 3.1 Laboratory apparatus to demonstrate Darcy’s
law. q = Ki (3.3a)
Q=KiA (3.3b)
v = Ki/n, (3.3c)
In the terms of the variables just given, Darcy’s law
is expressed v = q/n, (3.3~0
The quantity q (as well as v ) has both a magnitude
and a direction, the latter being toward decreasing eleva-
tion of water in the manometers. This makes q and v
where q is the volumetric flow rate per unit surface area, vector quantities, a fact that shall be exploited in later
with units of velocity, and K is a constant of proportional- chapters.
ity. Because the gradient is a dimensionless quantity, the
proportionality constant K has units of velocity. The mi- Hydraulic Head. Hubbert’s Force Potential
nus sign is a convention that is employed because flow
is in the direction of a decrease in water levels in the In Figure 3.1, the water-level elevations in the manome-
manometers, that is, from where h is high to where h ters are measured with reference to a common datum,
is low (Figure 3.1). Darcy’s equation may be stated in taken as a horizontal plane beneath the flow apparatus.
words as Thus, the absolute value of the water-level elevations was
of no concern to Darcy, only the differences between
The velocity of flow is proportional to the hydraulic them. In this section, we are not interested in these
gradient. differences but in what the actual water-level measure-
ments mean. For this purpose, it is convenient to intro-
Darcy’s law is valid for flow through most granular duce a counterpart of the laboratory manometer, known
material. The law suggests a linear relationship between as a piezometer.
the specific discharge and the hydraulic gradient. This The piezometer is a tube or pipe used to measure
relationship holds as long as the flow is laminar. Under water-level elevations in field situations. It is open at
conditions of turbulent flow, the water particles take the top where measurements are taken and open at the
more circuitous paths. At the other extreme, for very- bottom to facilitate the entrance of water. A simplified
low-permeability materials, a minimum threshold gradi- version of this device is shown in Figure 3.2, where the
ent may be required before flow takes place (Bolt and common datum is taken as sea level (zero elevation). As
Groenevelt, 1969). demonstrated by Hubbert (1940), the terms elevation,
Four aspects of Darcy’s law require clarification from pressure, and total bead on Figure 3.2 can be explained
a field perspective: the specific discharge q, the water- in terms of the conventional Bernoulli equation. This
level measurements, the gradient, and the proportionality equation states that under conditions of steady flow, the
constant K . total energy of an incompressible fluid is constant at all
positions along a flow path in a closed system. This may
Tbe Nature of Darcy’s Velocity be written as
P vz
As defined, the specific discharge q is a volumetric flow gz +- +-
Put2
= constant (3.4)
rate per unit surface area of sample. Because water moves
only through the pore openings making up the surface where g is the acceleration due to gravity, z is the eleva-
area, Darcy’s q is a “superficial” velocity. We thus define tion of the base of the piezometer, P is the pressure
a more realistic velocity v that is a volumetric flow rate exerted by the water column, p,,, is the fluid density, and
 3.1 Darcy's Experimental Laic) arid Field Extensions 35

Stated simply, the total head b is the sun1 of the eleva-

tion of the base of the piezometer plus the length of the
water column in the piezometer. Hence. the total 1ie;id
at a point is found by measuring the elevation of the
water level in a piezometer (Figure 3.2). However. the
point to which this head refers is not the water level, but
P
the point at the terminus of the piezometer at elevation x.
I t

I
I

I t P w .t The total head is sometimes referred to as the energy of

Measuring point
in the flow field / position and is a scalar quantity.
Elevation head In SI units, h in Eq. 3.6 is meters (m), z is meters (in)
X
above the datum (usually sea level), P is Pascals (ki).p,,
is kg/ni3, and g is m/d. Expressed in basic SI units. ;I
+ Oaturn
Pascal is kg/m * s'. The density p,, varies as a function of
temperature and chemical composition. with fresh water
Figure 3 2 Diagram showing elevation, pressure, and total at 15.5"C having a density of 1000 kg/ni'. The gravita-
head for a point in the flow field. tional constant, g, is 9.81 m/s'. The following exnmple
illustrates the relationship among the various p;ir;inieters.

u is velocity. Dividing through by g, Eq. 3.4 becomes Example 3.1

PvL
z + - + - = constant (3.5) With reference to Figure 3.2, assume that the ground
Pld 2g surface has an elevation of 1000 m above sea level, the
where the quantity pug equals the unit weight of water. depth to water is 25 m, the total depth of the piezometer
Equations 3.4 and 3.5 describe the total energy contained is 50 m, and the water has a density of 1000 kg/nl. What
by the fluid, where the first term is the energy of position, is (a) the total hydraulic head at the measurement point.
the second term is the energy due to sustained fluid (b) the pressure head, and (c) the pressure?
pressure, and the third term is the energy due to fluid (a) b = elevation of water in the piezometer
movement. The dimensions of Eq. 3.4 are LL/T', or energy = surface elevation - depth to water
per unit mass. Equation 3.5 is expressed as energy per = 1000 - 25 = 975 m
unit weight, say, foot-lb per pound, which reduces simply (b) From Eq. 3.6, P/p,,.g = b - z
to feet. In SI units, the dimensions are newton-meters = 975 - 950 = 25 ni
per newton, or simply meters. (c) From Eq. 3.6, P = ( b - z ) . p , , g = 25 X 1000 X
The three terms of Bernoulli's equation, when ex- 9.81 = 2.45 X 10' Pa or 0.245 MPa where the
pressed as energy per unit weight, are referred to as designition M means mega or 10".
elevation head, pressure head, and velocity head, respec-
tively. The term z in Eq. 3.5 is the elevation head and
represents the elevation at the base of the piezometer. An additional expression for the energy of position of
In a theoretical sense it represcnts the work required to a fluid can be obtained from the Bernoulli expression of
increase the elevation of a unit weight of water from Eq. 3.4 where, ignoring the velocity term
datum to height z (Figure 3.2). Stated another way, every D
body at the surface of the earth has a gravitational at-
traction toward the earth's center. To raise this body
counter to this attraction requires work, and this work
where 4 is referred to as the hydraulic potential. in units
is stored in the form of a potential energy. The quantity
of energy/mass. Defining potential in this way provides
P/pu,g represents the length of the water column in the
yet another form of Darcy's law commonly found in the
piezometer. It represents the work that a fluid is capable
literature. From the statement 4 = g b . it follows that
of doing because of its sustained pressure. The sum of
these terms is called the potential energy of the fluid.
The third term, v2/2g,is the kinetic energy, o r energy (3.8)
due to the fluid motion. As the velocity of ground water
is slow, this term is ignored. Thus, the sum of the eleva- Substituting this result in Eq. 3.1 gives
tion head z and the pressure head P/p,,.grepresents the
total head b in the system; that is, (3.9)
(3.6)
where d 4 / d l is the potential gradient (Hubbert, 1940).
36 Cbapter 3 / Ground-Water Movement

Tbe Gradient and Ground-WaterF h w The gradient of h or Vh is a vector that represents the
spatial rate of change of hydraulic head (Eq. 3.11). It
Darcy’s law shows that for flow to occur there must be consists of three components x,y, and z, each of which
differences in hydraulic head, providing what has been represents how fast the head changes in that respective
referred to as a hydraulic gradient. From a field perspec- direction. The direction of Vh is that which coincides
tive, it follows that a value for hydraulic head can be with the direction in which the head changes the fastest.
defined and measured at every point within some region. For the condition K , = K,. = Kz, this direction is perpen-
The term field is generally used to describe a region dicular to lines of equal head. This direction is of particu-
where some physical quantity can be described in terms lar interest in that it coincides with the direction of
of a space coordinate system and time. In this case, the ground-wdter flow.
quantity is a scalar, giving rise to a scalar field, more Example 3.2 helps to illustrate the concept of a gradi-
commonly referred to as a potential field. There are nu- ent in a two-dimensional flow field.
merous quantities that satisfy this definition, including
temperature T(x,y,z, t) and the concentration of some
substance C(x, y, z, t). Here the Cartesian coordinate Example 3.2
system is employed to identify the value of the stated The measured hydraulic head at piezometers A, B, and
scalar quantity at a point of interest at a given time. C is 150 m, 140 m, and 130 m, respectively. Assuming
The potential field in any region can be described by that the potential field is two dimensional in x and y , it
measuring the hydraulic head at a large number of points can be represented by a dipping plane in space. The
and contouring the resulting set of data. Starting at any direction of flow, which is depicted on the figure, is
one of the piezometers, it is likely that the head will normal to the contour lines, or, in other words, the direc-
increase in some directions and decrease in others. The tion of steepest decrease in head. The gradient calcula-
gradient in x can be written as M a x , the partial deriva- tion is shown on the figure
tive indicating how h changes inx irrespective of changes
in y and z. That is, ahlax is the change in head as we
move along the x axis. Further, the gradient iny becomes
ahlay, which describes the change in head along the y
axis, and the gradient in z is ah/&. Thus, Darcy’s law
expressed as
ah
4, = -K, -
ax
(3.10a)

Equipotential line
describes the flow of fluid along the x,y, and z axes,
Figure 3.E2
where the material properties in x, y, and z are different;
that is, K, # Ky # K,.
As it is rather limiting to restrict ourselves to one of Physical Interpretation
three directions, we may ask for the rate of change in of Darcy’s Proportionality Constant
head in any direction and, most important, the direction
By experimentally varying fluid density, viscosity, and the
of maximum change. For this purpose, the gradient of a
geometrical properties of sands, Hubbert (1956) reported
scalar field h can be defined as
that Darcy’s proportionality constant K varied in the fol-
ah ah ah lowing manner
grad h = Vh = i - + j - + k- (3.11)
ax ay az
where the vectors i, j, k are unit vectors in the x,y , and
z directions, respectively, and Vh (which is pronounced
del h and is an upsidedown “delta” to remind us of
differentials)is the abbreviation for grad h. Thus, in vecto-
rial notation, for K, = K,, = K,, Darcy’s law becomes
where p is viscosity and cl is the mean grain diameter of
q = -Kgrad h = -KVh (3.12) the sand. This can be expressed as
where grad h can be read “gradient of h.” Equation 3.12
is a vectorial equation where the x, y , and z components
are given by Eq. 3.11.
 3.2 Hydraulic Conductivity and Permeability of Geologic Materials -7
3

where K’ is yet another constant of proportionality con- K , the hydraulic head h, and the gradient of the head h,
taining variables not yet evaluated. It is obvious at this Darcy’s law becomes
point that Darcy’s K is a function of both properties of
the medium and properties of the fluid. Np,.gd
q=-
Insight into the parameter K* may be obtained by EL
comparing Darcy’s law with the Hagen-Poiseuille equa-
tion long known to govern laminar flow through small- For those cases where all piezometers are bottomed at
diameter passages. Incorporating Eq. 3.13 into Darcy’s the same elevation in a flat lying bed, z is a constant so
law gives its derivative goes to zero. Taking both the fluid density
and the acceleration of gravity as constants, Eq. 3.17 be-
-K*p,.d‘ comes
q= grad h (3.14)
EL
Nd’ k
q = --gradP= --gradP (3.18)
whereas the Hagen-Poiseuille equation is EL EL

This expression is quite convenient when dealing with

(3.15) multiphase fluid systems where the permeability k by
definition is invariant with respect to whatever fluid is
where N is a dimensionless shape factor relating to the being considered.
geometry of passage and R is the diameter of passage.
Equations 3.14 and 3.15 are perfectly equivalent if the
diameter of passage R is equivalent to the mean grain Units and Dimensions
diameter so that the constant K’ is equal to the product Various units and dimensions are given in Table 3.1 in
of the dimensionless shape factor N and the acceleration both the foot-pound-second system with its FLT base
due to gravity g. Making this analogy, Darcy’s constant (force, length, and time) and the International System
of proportionality is expressed as (SI) with its MLT base (mass, length, and time).

(3.16)
3.2 Hydraulic Conductivity and
where N d z characterizes the properties of the medium i%wzeabilityof Geologic Matertals
and p J p the properties of the fluid. The parameter K is
Observed Range
referred to as the hydraulic conductivity and contains
properties of both the medium and the fluid, and the
in Hydraulic Conductivity Values
parameter k is referred to as the intrinsic permeability Virtually thousands of measurements of hydraulic con-
(equal to Nd? and contains properties of the medium ductivity and permeability have been obtained in both
only. The hydraulic conductivity with units of velocity the laboratory and in the field. Davis (1969) provides the
characterizes the capacity of a medium to transmit water, best summary of these data. Table 3.2 is taken largely
whereas the permeability with units L2 characterizes the from this review along with additional input from John-
capacity of the medium to transmit any fluid. son and Morris (1962) and Croff and others (1985). The
The term bydruulic conductivity is used most fre- values cited are given in meters per second, but conver-
quently in groundwater literature when dealing with the sions can be made to centimeters per second, feet per
single water phase. The termpermeubility is used in the second, feet per year, or gallons per day per square foot.
petroleum industry where the fluids of interest are oil, the last referred to as a meinzer unit in honor of 0. E.
gas, and water. Typical values of permeability in square Meinzer. The meinzer unit is defined as the flow of water
centimeters or square meters are quite small so that the in gallons per day through an opening of 1 ft’, under ;i
“darcy” is commonly defined as a unit of permeability. unit hydraulic gradient at a temperature of 60°F. Conver-
For a material of ldarcy permeability, a pressure differen- sion to permeability (square centimeters, square feet. or
tial of one atmosphere will produce a flow rate of 1 cc darcys) is readily accomplished with the cited conver-
(cubic centimeter) per second for a fluid with l-centi- sion factors.
poise viscosity through a cube having sides 1 cm in As noted in Table 3.2, hydraulic conductivity can range
length. One darcy is approximately equal to 10+ cm’ or, in value over about 12 orders of magnitude, with the
for water of normal density and viscosity, cm/s. For lowest values for unfractured igneous and metamorphic
tight materials, the millidarcy (md) is used, where 1 nid rocks and the highest values for gravels and some karstic
equals 0.001 darcys, or approximately cm/s. or reef limestones and permeable basalts. The mige
It is now possible to state yet another form of Darcy’s in hydraulic conductivity within a given rock type is
law commonly used in the petroleum industry. Incorpo- greatest for the crystalline rocks and smallest for the
rating our definition of the constant of proportionality sedimentary material. In general, a hydraulic conductivit!,
-
38 Cbapter 3 / Ground-Water Movenient

Table 3.2 Dimensions and Common Units for Flow Parameters

R-lb-sec Conversion
System Factor SI

Parameter Symbol Dimension Units Multiply by Dimension Units

Hydraulic head h L ft 3.048 x 10.' I, ni
Elevation head z L ft 1. m
Pressure head - L ft I. m
Fluid pressure P F/L> Wft' 4.788 x 10 M/LT' N/m2 or Pa
Fluid potential 4 L'/T' ft'/&$ 9.29 X lo-' L?/V ni'/s'
Density of water h. - - MA: kg/m '
Gravitational constant g L/T' ft/s' 3.048 x lo-' L/T' m/s'
Unit weight of water Yu,= A S F/L' Ib/ft' - -
Volumetric flow rate Q I:/T ft'/s 2.832 X lo-' L'rr m'/s
Specific discharge 9 L/T ft/s j.048 X lo-' L/l m/s
Hydraulic conductivity K L/T ft/s L/T m/s

approaching lo-' m/s and smaller can be characterized

as low-permeability material. Clay, shale, chalk, and un-
fractured igneous and metamorphic rocks fall within this
category. However, if these rocks or sedimentary accu-
mulations are fractured, the conductivity can easily ex-
ceed this limiting value by two or three orders of mag-
nitude.

Cbaracter of Hydraulic
Conductivity Distribution
Because of the way in which geological deposits are
formed, hydraulic conductivity values for a particular unit
can vary from place to place. This scatter in values can
be depicted graphically using a histogram or represented
statistically by fitting some kind o f stitisticdl distribution.
Let us explore these ideas with a set of 32 measured
hydraulic-conductivity values for a hypothetical geologic
]() ............. ........... ........-
"
unit (Table 3.3).
Figure 3 . 3 ~illustrates the frequency histogram for
these data. Like thousands of histograms constructed
from various data (Law, 1944; Bennion and Griffith,
I W ) , there is a strong positive skew in the distribution.
Statistically, this distribution is complex and quite unlike
the Gaussian or normal frequency distribution, which
characterizes many distributions. We can transform the t?
e
hydraulic conductivity data in Table 3.3 by taking the ' L 4
logarithm. For example, log(4.0 X 10-9 = -7.4. Interest- 3
ingly, the histogram (Figure 3.36) shows the lob- ' trans-
formed data to be approximately normally distributed. 2-
Such a distribution can be characterized in terms of a ~~ 9-9
0
sample mean and standard deviation, which for the distri- 1
bution in Figure 3.36 are -6.99 and 0.76, respectively. -8 I 6
1% li
In many studies, hydraulic conductivity has been
shown to be a log-normally distributed geological parame- (bi

ter. In practice, data are transformed using either base Figure 3.3 (u) Frequency distribution histogram of the
10 logarithms or natural logarithms. However, there is original 32 values of hydraulic conductivity, (6)histogram of
no hard and fast rule that hydraulic conductivity is log- the lop,-tnnsformrd hydrdulic conductivity tlata.
 3.2 Hydraulic Condrrctiaity and Permeability of Geologic Materials 32

Table 3.2 Representative Values of Hydraulic normally distributed. o r for that matter ;ire other tlow
Conductivity for Various Rock T y p e s par;inieters. For example, Woodbury and Surlicky ( I991)
show that other distributions may be ;ippropriate for
Hydraulic Conductivity
Material hydraulic conductivity dat;i ;is well. In the c;ise of poros-
(m/s)
ity, studies show that distributions from ;I single fomi;t
SEDIMENTARY tion typically have normal rather than log-normil distribu-
Gr;1vel 3 x 10 '-3 x 10 tions (Davis. 1969).
Coirse sand 9 x 1 0 -4 x I 0 '
Medium siind 9 X 10-'-5 X l o - '
Fine sand 2 x 10---2 x 10 ' Anisotropicity and
Silt, loess I x 10-'-2 x l o r Heterogeneity Within Units
Till I x 10-"-2 x lo-1'
Clay 1 X 10'"-4.7 X 10." Most rocks have a directional qu;ility to their overall
Unweathered niarine clay 8 x 10-"-2 x 1 0 'j
stnicture. IMetamorpIiic rocks are noted for their schistos-
SEDIMENTARY ROCKS
ity, most sediments for their horizontal strxitification. ;ind
Kiirst :tnd reef limestone 1 x 1 0 "-2 x 1 0 - 2 basalts for their preferred orientation of shrinkage cr;icks
Limestone, dolomite 1 x 10-"-6 x lo-(' when the cooling period is rather short (and ;I spirsit!,
Sandstone 3 x 10-1"-6 x l o - ( ' of such cracks when the cooling histon. is long). l o r
Sil tstone 1 x 10-"-1.4 x 1 0 such materials, the hydraulic conductivit!,. iis measured
Silt I x 1 0 "-1 x lo-"' from some representative sample, w i l l not be eqiial i n
Anhydrite 4 x 10'"-1 x l o - " all directions. The term niiisotrwpbic is w e d t o describe
Shale 1 x 10-"-2 x 10." materials where the permeability or conductivity ;it ;I
CRYSTALLINE ROCKS point has a directional dependency. When pernie;ibilit!,
Permcible bas;ilt 4 x lo---2 x 10 is the same in all directions, the material at that point is
Fractured igneous and isotropic. Davis (1 969) cites several c;ises for brddecl
metamorphic rock 8 X 10-"-3 X 10" sediments i n particular where the permeability is greater
Weithered grinite 3.3 X 10-"-5.2 X lo-'
in the direction of the stratification and smaller pcrpen-
Weathered gabbro 5.5 x 10=-3.8 x lo-1'
diciilar to the stratification. Although supporting dat;i ;ire
Basalt 2 x 1 0 "-4.2 x lo--
sparse, similar statements can be made for niaxiniiiiii
Unfractured igneous and
nietamorphic rocks directional permeability associated with metamorphic
stnictiires such as schistosity. Table 3.4 provides ;I siini-
To convert meters per niary of information on the anisotropic nature of some
second to Multiply by sedimentary niaterials as determined from core s m p l e s .
~~

In these tabulations, the horizontal conductivity is taken

cm/s 1OL
in the direction of the structural features. such ;IS stratiti-
(gal/d;iy)/ft' 2.12 x 10"
cation, and the vertical conductivity is t;tken ;it right
ft/s 3.28
1 x 10" angles to the stratification.
ft/yr
darcy 1.04 X 10'
The classical definition considers ;I unit t o be homoge-
fti 1.1 x10r" neous if the permeability in a given direction is the s:inie
cm' 1 x lo--' from point to point in a geologic unit. .Materials that d o
not conform with this condition are heterogeneous. A
To convert any of the Divide by the appropriate
variety o f modern studies has shown that there is no
above t o meters per + number above
second possibility of ever finding a tnily homogeneous unit i n
nature. Hydrdulic conductivity distributions i n rc.;ility

Tabk 3.3 Hydraulic Conductivity Values Measured in a Hypothetical

Geologic Unit
K (m/s)
4.0 x 1.5 X lo-" 4.0 X 10 - 2.0 x l o - - 8.0 x lo-"
1.0 x lor- 8.0 x lo-'' 5.0 x I()-- 2.5 x l o - " 4.ox 10'"
2.0 x l o - " 1.0 x lor- 2.5 x lorw 1.0 x 10~' 3.0 x lor-
4 . 0 x 10.- 1.0 x lo-- 2.0 x lo-- 1.0 x lo-'' 6.0 x 10.-
1 . 0 x lo-(' 1.0 x 10." 1.0 x lo-- 2.0 x 1 0 . " ' 2.5 x I()--
1.5 X 1 0 - 8.0 X 1 0 - 5.0 x 10'" 1.0 x 10.- 4.0 x 10
6.0x 1 0 ~ " 5 . 0 x I().."
-
40 Cbapter 3 / Ground-Water Movement

Table 3.4 The Anisotropic Character of Some Rocks

Horizontal Conductivity Vertical Conductivity
Material (m/s) (m/s)

look like the one shown on Figure 3.4, which is from dium has a statistically isotropic or a statistically aniso-
the shallow, unconfined aquifer at Canadian Forces Base tropic correlation structure. Similarly, we can examine
Borden (Sudicky, 1986). “lie contoured field is based the point-to-point variability in the expected values of
on 720 permeameter measurements along a single cross mean, variance, and autocovariance. When these are con-
section. Compared to other carefully studied aquifers, stant, the medium is considered to be stationary or statisti-
the Borden aquifer is “mildly” heterogeneous. cally homogeneous. This geostatistical concept of honio-
Geostatistical concepts have proven useful in repre- geneity is useful because it represents the reality of
senting the heterogeneous character of a hydraiilic con- natural hydraulic conductivity distributions within the
ductivity distribution. One can represent the variability context of the classical definitions. A distribution in
of the parameter in terms of the variance and spatial which there is spatial dependence is considered to be
autocovariance in the measured data. The variance for a statistically heterogeneous. Chapter 10 will include a
normal distribution represents the spread in measured quantitative approach to characterize the correlation
hydraulic conductivity around the mean. The autocovari- structure of a hydraulic conductivity distribution.
ance represents the spatial continuity in the hydraulic One source of statistical heterogeneity is variability in
conductivity distribution. What is meant by spatial conti- the expected mean hydraulic conductivity as a function
nuity? As shown on Figure 3.4, hydraulic conductivity of position in the field. An example of this kind of conduc-
values are not random, but are correlated such that two tivity field is the Milk River sandstone in Alberta, Canada.
measurements taken close together are quite similar, but The Milk River sandstone has been interpreted as the
measurements separated by a larger distance are much seaward margin of a littoral environment, some of the
less similar. Effectively, there is a decay in the correlation sand having been deposited by streams or currents run-
as a function of separation distance. ning parallel to the coastline. This depositional environ-
These geostatistical concepts have much in common ment has resulted in a northwest-southeast linear trend
with the classical definitions of isotropicity and homoge- of sand lenses (Meyboom, 1960). The trend observed in
neity. For example, depending upon how the geological Figure 3.5 indicates that the largest mean conductivity
materials are formed, the expected autocovariance at any values are associated with thick clean sands in the south-
point in vertical direction (e.g., across bedding) may be ern and central parts of the formation, with a progressive
the same or may be different than that in the horizontal decrease northward and to the west and east as well in
direction. This dependency determines whether the me- response to greater proportions of fine-grained material

Distance along section (meters)

-4.0- -In(k’)
y Location of core
a K<~O”C~IS

Figure 3.4 Hydraulic conductivity distribution at Canadian Forces Base Borden (from
Sudicky, E. A,, Water Resources Res., v. 22, p. 2060-2082, 1086. Copyright by Americdn
Geophysical Union).
 3.2 Hydraulic Conductivity and Permeability of Geologic Materials 41

- 10-3-hydrauitc conductivity (cm/s) Recharge area

Range

Ffgure 3.5 Map of the Milk River sandstone showing hy-

draulic conductivity distribution (from Schwartz and others, Figure 3.6 Schematic cross section illustmting the differ-
1982). Reprinted from Symposium on Geochemistry of ence between a coilfined and an unconfined aquifer.
Groundwater, F. W. Schwartz, K. Muehlenbachs,and D. W.
Chorley, “Flow system controls on the chemical evolution of
groundwater,” 1982 with kind permission from Elsevier Sci- portant in categorizing the occurrence of ground water
ence, NL, Sara Rurgerhartstraat 25, 1055 KV Amsterdam, The in the hydrologic cycle. If a homogeneous rock of high
Netherlands. permeability exists continuously from land surface to
some great depth, the water in this aquifer will occur
exclusively under unconfined conditions. An unconfined
in the formation. Drift in hydraulic conductivity in sedi- aquifer is one in which the water table forms an upper
mentary accumulations is often controlled by the early boundary and is at atmospheric pressure (Figure 3.6).
geologic history as established within the depositional en- In some instances, where low-permeability materials are
vironment. interbedded with higher-permeability units, downward
Research is continuing to develop alternative concep- percolating water in the unsaturated zone may become
tualizations of permeability distributions. One direction “perched” on the low-permeability units. Thus, a local-
this work is moving in is the application of fractal meth- ized zone of saturation could form above the low-perme-
ods (e.g., Hewitt, 1986; Wheatcraft and Tyler, 1988). ability unit. This condition is referred to as a “perched
Another is the integration of statistical and setlimentologi- water table,” which is a local zone of saturation coni-
cal models, where the “alluvial architecture” of a deposit pletely surrounded by unsaturated conditions (Figure
(e.g., as is shown in Figure 2.7) provides yet additional 3.7).
information with which to characterize the hydraulic Given the layered heterogeneity demonstrated in Fig-
conductivity distribution (Phillips and others, 1989). In- ure 3.6, a confined aquifer can occur as a high-perme-
formation on the alluvial architecture could come from ability unit between two low-permeabilit!. units. o r aqui-
conventional mapping approaches in the field or through tards. The aquitards in this case are referred to ;IScontin-
the application of models capable of simulating complex ing layers. The confined high-permeability unit frequently
depositional histories (Kolterniann and Gorelick, 1992). contains water under pressure due to the elevation of
the intake area in dipping beds. The water level registered
in wells tapping such a confined aquifer can be ahovc
Heterogeneity Among Units and
the Clussification of Aquifers
Hydrogeologists are often concerned with examining hy-
draulic conductivity distributions at a large scale that
could encompass several different stratigraphic units. At r__ - - - - - - ._- c _ ~
,, Perched water :able
this scale, any description of hydraulic conductivity natu- .* *J
,/’ +-.

rally rakes into account the organization imposed on the

field by the geology. One simple example relates to strati-
graphic layering where the statistical description of the
field is based on individual geologic units that commonly
--.----..cI-__c
b-.
--. - - - __ _
Regional watertable-.. - -----

have very different hydraulic conductivities. Figure 3.7 Schematic presentation of ;I perched W J I C f
This model of layered heterogeneity is historically im- table.
-
42 Cbapter 3 / Ground-Water Movement

m/s, and the shale has a horizontal and vertical hydraulic

conductivity of 1.92 X m/s. From equation 3.19

K.r =
225 m * 1 X lO-’m/s + 75 m * 1.92 X lo-” m/s
300 m
K , = 7.5 x 10-‘m/s
Figure 3.8 Layered heterogeneity. From Eq. 3.26
300 m
K, =
or below the regional water table in the overlying uncon- 225m + 75 m
fined aquifer (Figure 3.6). In some cases, the wells may 1X m/s 1.92 X lO-’’m/s
even flow at the surface, in which case they are referred K, = 7.7 X lo-’*m/s
to as artesian wells.
Thus, for horizontal flow the most permeable units domi-
nate the system; for vertical flow the least permeable
Creating Hydraulic Conductivity Averages units dominate the system. Under the same hydraulic
In many applications, it is advantageous to represent the gradient, horizontal flow is on the order of six orders of
local scale heterogeneity in hydraulic conductivity by magnitude faster than vertical flow. For this example,
some “average” hydraulic conductivity value, One case horizontal flow through 300 m of sediments with a uni-
that has been analyzed extensively is that caused by lay- form hydraulic conductivity of 7.5 X low6m/s would be
ering of high- and low-permeability units (Figure 3.8). If the same as the sum of the flows through each of the
the conductivity variation within a unit is much smaller individual layers in the example. That is, they are “equiv-
than the conductivity contrasts between beds, as might alent.”
be the case for interlayered sandstone and shale, it can
be assumed that each layer is homogeneous and iso-
For more complex media, probability theory provides
tropic. Each layer, however, is characterized by a differ-
a straightforward approach to averaging. For example,
ent hydraulic conductivity, rendering the sequence as a
Matheron (1967) shows that for uniform flow (i.e., paral-
whole heterogeneous.
lel flow lines) in a medium with a statistically isotropic
According to Leonards (1962), an equivalent horizontal
correlation structure the average hydraulic conductivity
hydraulic conductivity in the horizontal or x-direction is
for a system of any dimension (e.g., 1-D, 2-D, or 3-D)
is between the harmonic mean and the arithmetic mean
(3.19) of the set of local values. This result holds irrespective
of the spatial correlation. The harmonic mean of a set of
where K , is the equivalent horizontal conductivity, K,is numbers is given as
the homogeneous conductivity of an individual layer, and
m, is the thickness of the layer. For the direction at right
(3.21)
angles to the stratification

(3.20)
where H is the harmonic mean, N is the number of
where K, is an equivalent vertical conductivity for the values in the sample, and X is the individual values. As
layered system. an illustration, the harmonic mean of the numbers 2, 4,
These equations give the hydraulic conductivity values and 8 is 3.43 whereas the arithmetic mean for the same
parallel and vertical to the stratification for a single homo- set is 4.66. Matheron’s theoretical result is also demon-
geneous, anisotropic rock that is the equivalent of the strated by measurements of actual distributions (Cardwell
layered system shown in Figure 3.8. and Parsons, 1945; Warren and Price, 1961; and Bennion
Example 3.3 illustrates the use of Eqs. 3.19 and 3.20 and Griffiths, 1966).
in calculating an average hydraulic conductivity for a For some media, the average hydraulic conductivity is
layered sequence of rocks. a function of the geometric mean, which for a series of
numbers is given as
Example 3.3 G = ( X , X , X , . . . X,,)”” (3.22)
Consider a 300-m sequence of interbedd&fsandstone which is the Nth root of a product of N numbers. For
and shale that is 75% sandstone. The sandstone has a our simple example, the geometric mean of 2, 4, and 8
horizontal and vertical hydraulic conductivity of is 4. Hence, the geometric mean is intermediate between
 3.2 Hydraulic Conductioity and Permeability of Geologic Materials 43

the harmonic and arithmetic means. It turns out for log-

normally distributed hydraulic conductivity data and two-
dimensional uniform flow, the average hydraulic conduc-
tivity is exactly equal to the geometric mean. Similar
scaling laws for uniform flow in one- and three-dimen-
sional systems with a statistically isotropic correlation
structure are given by Gutjahr and others (1978). Dagan
(1989) develops more general relationships for statisti-
cally anisotropic media similar to that depicted in Fig-
ure 3.4.
All of these scaling laws for heterogeneous fields re-
quire that flow be uniform. If flow lines are not parallel
( e g , converging flow to a well), one cannot derive an
average hydraulic conductivity from a collection of local
measurements.
I s
I
I
(/:
Darcy's Law for Anisotropic Material
Figure 39 Array of nine components of three velocity vcc-
In most of the discussions of Darcy's law, it has been tors acting at a point on the three faces of a cube.
tacitly assumed that condrictivity was independent
of direction. For this assumption, Darcy's law was ex-
pressed as nated plane, whereas q.v,.and y.,: act tangential to the
plane in question. Thus, 'Eq. 3.24a can be expressed
Q= -Kgradh (3.23)
which is correct only when K, = K,. = K z , that is, the
material is isotropic. For anisotropic material, the follow-
ing forms of Darcy's law were given as T T T
in the direction of x axis y axis z axis
(3.24a)
(3.25)
Thus yldepends not only on gradients in s.but on gr;idi-
ents in y and x as well.
(3.24b)
In a similar fashion, the velocity vector q, of Eq. 3.24
can be resolved into three components. all of which act
ifh
4,= -K, - (3.24~) on a plane normal to thejf axis (Figure 3.9). One of these
dz will be a normal velocity component (y,,) in that it acts
which describes the flow of fluid along the x, y , and z in the direction normal to the designated plane whereas
axes where the material properties are different, that is, the others will be tangential. The same line ofreasoning is
anisotropic. However, Eqs. 3.24 are simplifications of yet followed for the resolution of qlinto three components.
a more complex form of Darcy's law for anisotropic Hence, Eqs. 3.24 can be expressed as
material. This more complex form can be examined by fluid flow
considering the fact that the velocity in each of the ex-
pressions of Eqs. 3.24 is a vector and can be resolved in the plane normal to x = q,
into components parallel to the x,y , and x axes. Consider
that these x, y , and z axes form the edges of a small cube
through which the fluid is flowing (Figure 3.9). We may
now condense and express the information in Figure 3.9 in the plane normal t o y = y,
in terms of q,,, where the first subscript indicates the
direction perpendicular to the plane upon which the
velocity vector acts and the second subscript indicates
the direction of the velocity vector in that plane. That is in the plane normal to z = 4:
to say, y,,.given in Eq. 3.24a describes the flow of fluid
along the x axis. Thus, the components of the velocity
vector q,,.on the plane of the cube normal to the x axes
become q,,x,q?,,,and q,rz.Here q.rzis a normal velocity t t t
component in that it acts along the normal to the desig- in the direction of x axis y axis I axis
 Next Page

-
44 Chapter 3 / Ground-Water Movement

In this form it is seen that there are nine components of

the hydraulic conductivity in anisotropic material. These
components can be placed in matrix form to give what
is known as the hydraulic conductivity tensor

Kw K.q K,.z

K,’., Kl:” K,,

Kz., K:v Kzz

This is a second-order symmetric tensor that has the

property Ki/ = K,,; that is, K,y = K,,,KrZ= K,,, and so
on. For the special case where the principal directions
of anisotropy coincide with x, y, and z directions of the
coordinate axes, the six components K,,., &, K,,.,., K,,,
K,, and KTrare all equal to zero. In this case, the x, y,
and z axes are called the principal axes of the porous
medium. The conductivity tensor for the principal axes (4 ((?I
then becomes
Figure 3.20 (4)Constant head and (b) falling head perniea-
meters.

Hydraulic conductivity is then determined with Darcy’s

law of the form
Stated another way, when the coordinate axes are ori-
ented parallel to the principal axes of the porous medium, K = -QL (3.26)
we recover the original expressions for Darcy’s law for Ah
anisotropic material (Eqs. 3.24) when x, y, and z are where L is the length of the sample, A is the cross-
principal axes and K , # Kr # K,. For the very special sectional area of the sample, and h is the constant head
condition where the material properties do not differ shown on Figure 3 . 1 0 ~ .
with direction, we obtain the isotropic form of Darcy’s In the falling head test, the head is measured in the
law (Eq. 3.23), where the hydraulic conductivity is taken standpipe of Figure 3.10b, along with the time of mea-
as a scalar quantity. surement. For a sample of length L and a cross-sectional
area A, the conductivity is determined by
Measurement of Hydraulic Conductivity
(3.27)
Hydraulic conductivity can be measured in a variety of
ways. In practice, only three common methods are em- where a is the cross-sectionalarea of the standpipe and
ployed: field tests, laboratory tests, and empirical or semi- (t, - to)is the elapsed time required for the head to fall
empirical methods based on grain diameter and grain- from b,,to h, . Some permeameters are designed so the
size distributions, or simple hydraulic models. Field tests sample can be brought to the original stress condition
are by far the most reliable for they permit the testing at the depth it was collected. This is important for deep
of large volumes of rock with one pumping well and samples where the measured conductivity changes as a
one or more observation wells. Pumping-in tests have a function of the applied stress.
considerably smaller area of influence but are important
in low-permeabilityrocks that do not readily yield water The Search for Empirical Correlations
to pumping wells. Field testing is an extensive topic and As hydraulic conductivity can be readily measured in the
will be covered in Chapter 6. laboratory, there have been numerous attempts to relate
the measured values to various properties of a porous
Laboratory Testing medium. Table 3.5 summarizes some commonly ac-
Permeameters for measuring conductivity are shown in cepted relationships. The first three equations relate hy-
Figure 3.10. In the constant head test, a valve at the base draulic conductivity or permeability to grain size or grain-
of the sample is opened and the water starts to flow. size distributions. The remaining two relationships are
After a sufficient volume of water is collected over the more hydraulically based.
time of the test, the volumetric flow rate Q is ascertained. In general, the permeability of porous rocks appears
Previous Page

3.3 Mapping Flow iti Geological Systems 45

Table 3.5 Examples of Empirical Relationships for Estimating Hydraulic Conductivity or Permeability Values
~~

Source
Hazen (191 1 ) K = Cdt,, K = hydraulic conductivity (cm/s)
C = constant 100 to 150 (cm/s)-' for loose sand
d,,, = effective grain size cm (lo'%!particles are finer. 90% coarscr)
Harleman et al. (1963) k = (6.54 x Io-')df,, k = perme;ibility (cm')
Krumbein and Monk k = 760 dJe-1$ 1 ~ k = perme;ibility (darcys)
(1943) d = geometric mean grain diameter (mni)
D = log standard deviation of tlie size distribution
Kozeny (1927) k = Cti'/S*' C = dimensionless constant: 0.5. 0.562. and 0.jc)? for circikir.
square, and equilateral triangle pore openings
k = permeability (L')
ji = porosity
S" = specific surface-interstitial surface areas of pores per unit
bulk volume of tlie medium
Kozeny-Carmen Bear K = Iiydmulic conductivity
(1972) p,, = fluid density
p = fluid viscosity
g = gravitational constant
d,,, = any representative grain size
ii = porosity

to be proportional to some mean grain diameter squared, in the Same unit, or in various units (Figure 3.1 1). Such
which reflects the size of a pore, along with the spread a collection of piezometers is referred to ;IS ;I piezome-
or distribution of the grain sizes (pores). Collins (1961) ter nest.
points out that the grain- (pore-) size distribution also Hydraulic head distributions vary in three spatial direc-
influences the magnitude of specific surface, which in- tions and time. The time element can be removed by
creases with decreasing grain size. Consequently, pore- assuming all the head measurements arc m;ide ;it the s;inie
size distributions are indirectly incorporated in the time (i.e., a "snapshot" of hydraulic head conditions ;it
Kozeny-type formulations. one point in time). The remaining difficulty in represent-
Hydraulic arguments have ;liso been applied to frac- ing a three-dimensional field is overcome by phicing pi-
tured rocks. For a parallel array of planar joints of aperture ezonieters in the flow field where the flow c;in he inter-
width b, with N joints per unit distance across the rock preted as largely two dimensional. In other words. one
face, the permeability may be described by (Snow, 1968) can construct two-dimensional planes (e.g.. ;I cross sec-
tion or a m;ip view) where the flow is essenti;ill!, nvo
k = CnD' = CNbb' (3.28)
dimensional. However, the resulting cross section o r ni;ip
where k is the permeability with dimensions of length must be viewed as ;I two-dimensional projection o f ;I
squared, C is a dimensionless constant, in this case three-dimensional field (Hubbert, 19.40). Some projec-
(1/12), and the porosity n is taken to be a planar type
of porosity equal to Nb. Here, N has the units L-I.

53 Mapping Flow in Geological Systems

Section 3.1 showed how the spatial variation in hydraulic
head determined the hydraulic gradient, and ultimately
the direction of ground-water flow. This concept forms
the basis of a practical technique for mapping the pattern ttM
of ground-water flow in the field. Hydraulic heads are
determined by measuring water levels in wells or piezoni-
eters installed at different locations and depths. Normally,
the piezometers are located within a study area to sample
different topographical regimes in the area of interest. Ffgure 3.22 Illustration of a piezometcr nest inclucling :I
Often, at each drilling location, several piezometers are water-tableobservation well. Piezometers ;ire emp1:iccd in
completed to sample hydraulic head ;it different depths both high- a n d low-perme;ibility units.
k6 Cbapter 3 / Ground-WaterMovement

- 1060 Elevation
(rn.a.s.1.)

(dTopographic map
Nest 4
I ,1057.5 .,,
A

I Nest 1
I
Nest 2
I
Nest 3
I A
/
0 1055 -
1050
1047.5
1038 0 1045
1041 2 1042 7e
vi 1040
e1031.5 1 1035
1035
1035.0
1030 1 e1031.1
e1027.0 e1031.2

+
e1031.9
1029.6
c

1031.9
e1029.6 e1031.1
I 1029.6 I 1 I

.r-distance (rn)
1029.6 Hydraulic head (m.a.s.1.)
1030 - Equipotential line (m.a.s.l.1

- Water table
Ground-water flow direction
( h ) Hydrogeologic cross section

Figure 312 Panel (u) shows the orienfation of the cross section in relation t o ii hill slope. Panel ( 0 ) is an exaniple of a hydro-
geologic cross section describing the pattern of ground-water tlow.

tions will contain most of the variations in hydraulic head data defines the spatial variation in hydraulic head. For
that have to be addressed. this simple example, ground-water flow as represented
by the arrows is normal to the equipotential lines.
To maximize the quantity of information along a' ,$w e n
Hydrogeological Cross Sections hydrogeological cross section, the choice is often made
Hydrogeological cross sections are vertical sections in a field investigation to install piezometer nests along
through a three-dimensional flow region. Ry aligning the lines in the field. To select the orientation of this line
section parallel to the direction of mean ground-water before installing the piezometers, one can examine other
flow, flow conditions can be represented accurately in existing hydraulic head data or, in the absence of such
terms of a two-dimensional cross section. Figure 3.12a &nu, infer that the direction of ground-water flow follows
illustrates a simple topographic map encompassing an the regional gradient in the surface topography (e.g.,
upland area and an adjacent valley. The hydrogeological Figure 3.12~1).
cross section A-A' is located parallel to the direction of Let us return t o the hydrogeological cross section in
mean flow. Normally, the section includes basic informa- Figure 3.12b because it helps to illustrate yet another
tion about the stratigraphy and variation in hydraulic two-dimensional plane that can be constructed through
conductivity, as well as hydraulic head data from nests a complex hydraulic head field. In the more permeable
of piezometers located along the section (Figure 3.126). lower unit, the eqiiipotential lines are vertical, and flow
A convenient way of presenting these data is to represent in the aquifer is essentially horizontal. Within this unit,
the measurement point (i.e., intlke of the piezonieter or there is an absence of vertical gradients, which suggests
elevation of the water table) by a dot and noting the that the gmdient in hydraulic head varies only in the
measured value of hydraulic head. Contouring these head plane of an aquifer (here x-y space).
 3.3 Mappitig Flow in Geological Sjatcnis 47

of piezometer penetration. Hence the potentiometric sur-

face is in reality a projection of vertical equipotenti;d lines
into the horizontal plane. Lastly, liead losses between
adjacent p i r s of equipotential lines are eqtial. and thc
hydraulic gradient varies inversely with distance betwwn
lines of equal head. This statement is simply ;I necess;inr
convention in the construction of potentionietric ni;ips
where the contour intern1 is constant, for example. I 0 0
ft in Figure 3.1.3. This accomplished, the second state-
ment follows automatically. From Figure 3.13, the con-
tours are closely spaced and the gradient is steep ne;ir
the Black Hills. Farther east, the contours are Farther
spaced, and the gradient is flattened out somewhat.
Nebraska Example 3.4 illustrates the calculations one can make
from potentiometric surfaces.
I I I I
0 50 100 150
Miles
Example 3.4
Figure 3.23 Potentiometric surface of the Dakota Sand-
stone, contour interval 1 0 0 ft (from Darton, 1909). The potentiometric map and intersecting flow lines given
below are for a nonhomogeneous but isotropic aquifer.
It is assumed further that there are no gains or losses in
Potentiometric Surface the total flow so that the volumetric flow rate passing
and Water-Tab&Maps one equipotential line must equal the flow rate passing
an adjacent equipotential line, that is, there can be no
As our example suggests, measurements of hydraulic gaps in the fluid. This means that
head obtained from water wells or piezonieters in a con-
fined aquifer can be contoured on a map. Such a map of
hydraulic heads is referred to as a potentiometric surfwe
map, defined by Meinzer (1 923) as an imaginary surface
that everywhere coincides with the level of water in where K, is the hydraulic conductivity of region i and A, is
the aquifer. If the aquifer is unconfined, the contoured the area of flow between adjacent flow lines. Recognizing
surface is referred to as ;I map of the water table. that A , = A? for a uniformly thick aquifer and A h , = A h ?
Figure 3.13 shows the potentiometric surface for the by construction
Dakota Sandstone. This artesian aquifer crops out along
the eastern flanks of the Black Hills in South Dakota and
dips eastward. Water enters this unit at its elevated intake
areas and moves downdip in an easterly direction. The or the ratio of the conductivities equals the ratios of the
water moves from where the head is high to where the lengths between equipotential lines. Thus, if the hydrau-
head is low, with each of the lines presumablyconnecting lic conductivity is known in one region, it is known for
points of equal head. the other regions.
In working with these kinds of maps, be aware of these
important points. First, a potentiometric map must be , Lines of equal head
related to a single aquifer. Other aquifers deeper or shal-
lower i n the section will have different potentionietric
surfaces that may exhibit heads that are higher or lower
than the one of immediate concern. For example, the
Madison Fornution underlies the Dakota Sandstone
throughout much of South Dakota and has its own poten-
tionietric surface showing more or less the same slope
and wnie direction of tlow, but higher heads than those
encountered in the Dakota. Second, it is assumed that
flow in the aquifer is horizontal, that is, parallel to upper
and lower confining layers. If a piezometer is placed in
such an aquifer, and a hydraulic head is noted, the hydrau- Figure 3.E4
lic head is presumed not to change with increasing depth
48 Cbapter 3 / Ground-Water Movement

Closing Statements prepared for the oxidized till layer and demonstrates
an average gradient of about 0.072 across the site. The
There is a theoretical point to be made in closing this materials underlying the small stream on the right-hand
section. To help this point along, let us briefly review a side of the diagram are to be excavated to the top of
few statements from the preceding pages. First, the term the unoxidized zone. This trench will be backfilled with
Jield is used whenever some quantity can be specified gravel with some sort of collection system. Assume an
at every point in space and time. A scalar field was defined effective porosity of 0.1 for the entire ttr. layer.
as a property that takes on a value (number) at any point
within a specific region of space and time. Clearly, the 1. Horizontal velocity in oxidized till layer
hydrogeological cross sections discussed above (Figure ~ ) =(gm/s)7.2X
v = K- ( Ah lo-'
3.12b) as well as the potentiometric surface (Figure 3.13)
n,
satisfies this description, with the scalar property being
hydraulic head. As mentioned previously, the time ele- = 7.2 X lo-" m/s = 2.3 Wyr
ment has been removed by assuming that all the head
measurements are made at the same time. Further, the 2. Volumetric flow into trench
gradient was introduced as consisting of components in Q = KL4 = (lO-'m/s)(7.2 X 10-2)(4572m2)
x, y,and z, each representing how fast the head changes
= 3.3 X m'/s
in that respective direction. It was also stated that for
the condition K, = K,,= K,, the gradient of the head is 3. Vertical velocity through unoxidized till layer
a vector perpendicular to the lines of equal head. Thus,
as the flow is presumed to be at right angles to the lines 1O-' m/s
of equal head, the gradient of the head, or grad h, is
colinear with the flow. The gradient of the head is thus
= m/s = 3.15 m/yr
seen to be the force that drives the flow, and the direction
of this driving force and the flow is at right angles to the 4. Travel time for waste to reach sand aquifer
lines of equal head. The velocity of this flow, so directed,
is given by Darcy's law. The connection between what t i m e t = L / v = 13m/3.15m/yr=4yr
Darcy observed and calculated at the experimental scale
and what can be mapped and calculated at the field scale
is thus complete. 54 Flow in Fractured Rocks
The following example shows the variety of hydraulic
In fractured rocks, the interconnected discontinuities are
calculations that can be made on the field scale.
considered to be the main passages for fluid flow, with
the solid rock blocks considered to be impermeable.
Example 3.5 Thus, on the scale of the field problem, one of two
approaches might be followed when dealing with the
1........................ .,................ ...... 278 m ___I_I
I flow of fluids in fractured rock: continuum or discontin-
uum (discrete). The continuum approach assumes that
the fractured mass is hydraulically equivalent to a porous
medium. The obvious advantage of treating fractured
rocks as a continuum is that Darcy's law as developed
can be applied so that no new theories are involved. If
Unoxidizedtill the conditions for a continuum do not exist, the flow
must be described in relation to individual fractures or
fracture sets.
Figure 3.65
Continuum Approach to Fluid Flow
Applying a continuum-based approach to fluid flow in
A waste disposal facility is to be constructed in the materi- fractures first requires an understanding of some of the
als given above. The facility will occupy a region 278 m basic assumptions implicit in a model for an intergranular
long (in the direction of ground-water flow) by 200 m porous medium. This understanding is essential in estab-
wide. The trenches will extend about 13 m into the lishing to what extent these conditions can be achieved
oxidized till, which has a water table about 5 m below in some fractured rocks. Let us state at the outset that
land surface, and a measured hydraulic conductivity of the conditions for a continuum approach to fluid flow
m/s. The underlying unoxidized till has a hydraulic in intergranular porous media are seldom, if ever, chal-
conductivity of lo-" m/s. A water-table map has been lenged. Such is not the case for fractured rocks.
 3.4 Flour in Fractured Rocks 9

Inteqgranular Porous Rocks

In Darcy’s own words, he set out “to determine the
laws of flow of water through sand.” And this he did,
exclusively at the experimental scale. There are, of Heterogeneous
course, other scales from which it is possible to study
certain phenomena. At the molecular scale, for example,
Homogeneous
one is interested in the behavior of molecules. This scale
of behavior probably never crossed Darcy’s mind because
it is patently impossible to fully understand the behavior I
of fluids at this level. Besides, Darcy, like most of the
Volume
rest of us, was more interested in the manifestations of
molecular motion, or things that could be expressed in Figure 3-14 Diagram illustrating the representative elemen-
measurable terms, such as viscosity, density, tempera- tary volume (from Hubbert, 1956). Reprinted with permis-
ture, concentration, and, of course, velocity. sion of the h e r . Inst. Mining, Met. and Petrol. Engrs.
If we step up a scale, we find ourselves at the pore
scale, referred to as the microscopic level. The micro-
scopic scale was clearly not for Darcy because nothing with the scale of the problem. This larger scale may be
could be observed within the individual pore in 1856. referred to as megascopic. It thus follows that in both
Darcy was interested in a macroscopic law, and his only the microscopic and megascopic regimes, there may be
recourse was to develop it on a macroscopic scale. That no single value that can be assigned to represent faithfully
is, he experimented with a volume of sand that was large any one of our material properties. The continuum ap-
with respect to a single pore but small with respect proach, then, is restricted exclusively to the macroscopic
to the space within which significant variations of the regime where properties are only a function of position,
macroscopic properties may be anticipated. That is to as defined by some appropriate coordinate system, and
say, he sought out the smallest possible sample that ex- time, and do not vary with the size of the field.
hibited an acceptable level of homogeneity. We call this
the macroscopic approach, more frequently referred to Fractured Rocks
as a continuum approach. As with porous intergranular media, a small control vol-
In this regard, it is convenient to introduce a macro- ume of fractured rock may be filled entirely with open-
scopic control volume that is large with respect to an ings or with solid rock. A larger volume will include
individual pore, but small with respect to the space cracks and solid rock, but the proportion of each will
within which significant variations of the macroscopic change as different points are sampled within the same
properties may be anticipated. Hubbert (1956) empha- small-scalefeature. This variation presumably gets smaller
sized this point with a diagram such as Figure 3.14, which as the volume increases so that the final representative
is a plot of porosity versus the volume of the sample in volume is achieved. For many fractured rocks, it is likely
which it is measured at some point. At the microscopic that as the volume of rock increases beyond some repre-
level, the value of porosity varies widely. For example, sentative volume, the parameters will start to vary again
the sample may be all solid or all pore. As the volume before becoming constant once more. Thus, the repre-
increases, a statistical average smooths out the micro- sentative volume may exist on several scales. This hierar-
scopic variations, and we find there are no longer any chy of scales is demonstrated in Figure 3.15, where a
variations with the size of the sample. Rear (1972) defines rock mass may be homogeneous on several scales. At
this limit as the representative elementary volume, de- small scales, say, the individual fracture scale of A, testing
fined as a volume of sufficient size such that there are could encounter the intact rock, or the crack openings,
no longer any significant statistical variations in the value with large variations in the measured parameter. As
of a particular property with the size of the element. larger-scale sampling is conducted over the entire basalt
As noted in Figure 3.14 for ;I reasonably homogeneous interior, some representative volume may be achieved
medium, the property of interest (porosity) is adequately in that increasing the volume of the unit tested has no
represented by some statistical average with little or no further effect on the value of the average property. This
variance at any scale larger than the representative ele- behavior is demonstrated at the B scale of Figure 3.15.
mentary volume. At larger scales in real rocks, ;I greater Including a few permeable flow tops in the testing
number of heterogeneities may be encountered, mostly scheme results in an increase in the average value of the
as a result of the regional stratigraphic framework that permeability (0.Including many flow tops provides for
includes several facies changes and resulting changes in homogeneity at yet a larger scale. If the rock mass was
the material properties. An average value can, of course, characterized by highly permeable vertical discontinu-
be obtained at any scale regardless of the degree of hetero- ities, the rock again would exhibit nonhoniogeneity until
geneity, but the variance about the mean will increase the volume was increased to include several such discon-
 ~,~
Chapter 3 / Ground-WaterMot)eement

0 l'L ........ f'?

........,
I
One last issue to be raised is whether representative
-r---- i ,,* Reprwntative :.,.,/'Representative ! elementary volumes even exist for some types of frac-
f" ctiernentary volume I elementary volume i tured rocks. Some theorists believe that as the scale of
I for entablature I far iniiltiple flows i
! mdcotonnae- I testing increases, the scales of heterogeneity expand
I 1
I right along with it. In other words, every block of rock
I
will contain a few significant heterogeneities that are not
representative of the fracturing found over the rest of
the volume. If ongoing research indeed could support
this idea that representative elemental volumes may not
exist, support for continuum approaches will diminish.
If the condition for a continuum does not exist, the
Test Volume flow must be described through the individual fractures
or fracture sets. This gives rise to several problems. First,
@ s ~ i ofe rnurtipte
Domain
fractures &main fractures
crf colonnade.
entablature,
of
art0 @muitlple
Doniarn of to characterize the flow, information is required about
the orientation, fracture density, degree of connectivity,
aperture opening, and smoothness of fractures. Second,
within entablature if the apertures are large, the flow may be turbulent
rather than laminar, and Darcy's law no longer applies.
Third, the values of hydraulic conductivity will vary with
changes in the three-dimensional stress field and the
fluid pressure.
Equivalent Hydraulic Conductivity An equivalent
hydraulic conductivity or permeability may be calculated
for a set of planar fractures with equations developed by
Snow (1968)

Figure 3.25 Schematic definition of a representative ele-

mentary volume as applied to basalt (from Department of where K is hydraulic conductivity, k is permeability, b
is the aperture, and N is the number of joints per unit
distance across the face of the rock (L-I). For this case,
tinuities. The average permeability would thus increase the product N6 is the planar porosity. The expression
again. for hydraulic conductivity given has been plotted for
There are a few points to be made here. If one wishes pure water at 20°C by Hoek and Bray (19811, where the
to apply continuum-based models to a large-scale prob- equivalent conductivity is taken as parallel to the array
lem, the data employed in that analysis should come from (Figure 3.16).As an example, for a spacing of 1 joint per
a testing program that is also large scale. This applies to meter and an opening of 0.1 cm, the equivalent hydraulic
all problems in ground water, not just in fractured rocks. conductivity is 8.1 X cm/s. This means that 1 m2
Virtually all the important parameters such as permeabil- of porous material with a hydraulic conductivity of
ity, dispersivity, and even rock compressibility are sig- 8.1 X cm/s will conduct just as much water as
nificantly smaller on the iaboratory scale than on the field will 1 joint per meter with an opening of 0.1 cm under
scale. Hence, representative volumes for both porous identical hydraulic gradients.
and fractured mediums can exist on several scales.
The second point deals with the question of when can Tbe Cubic Law The cubic law states that for a given
a fractured medium be treated within the continuum gradient in head, flow through a fracture is proportional
approach. This particular question has been the subject to the cube of the fracture aperture. For laminar flow
of continued research. First, it should be emphasized that between two smooth parallel plates, the volumetric flow
many models of fractured rock have been employed in rate can be expressed (Romm, 1966) as
certain flow problems without directly demonstrating
either the size or the presence of a representative volume. (3.30)
One practical approach is to test the rock properties over
the largest practical rock volume. This approach certainly where Q is a volumetric flow rate, punis the density of
does not guarantee success. Some theoretical papers dedl- water, g is the gravitational acceleration, p is viscosity,
ing with determination of the representative volume are 6 is the aperture opening, w is the fracture width perpen-
given by Endoe (1984) and Long and others (1982). dicular to the flow direction, and ilh/aL is the gradient
 3.5 Flow f n tbe Unsaturated Zone 51

Fracture openings (microns)

Ffgure3.2 7 Statistical frequency curve of opening width

(from Van Golf-Racht, 1982). Reprinted from T. D. Van Golf.
1982. Fundamentals from Fractured Reservoir Engineering,
with kind permission from Elsevier Science, NL, Sara Berger-
hartstraat 25, 1055 KV Amsterdam, The Netherlands.

has been demonstrated in several studies (Huitt, 1956;

Gale, 1975;Witherspoon and others, 1980) that conclude
Joint Opening & (cm)
that the law is valid where fluid pressure effects are
Ffgure3.26 Influence of joint opening b and joint spacing not important.
on the hydraulic conductivity in the direction of a set of
smooth, parallel joints in a rock mass (from Hoek and Bray,
1981). Reprinted with permission of the Institution of Min-
ing and Metallurgy, London. 3 2 F b w in the Unsaturated Zone
The vadose or unsaturated zone occurs between the wa-
ter table and the ground surface (Figure 1.3). With the
in the flow direction. This equation is of the form Q = exception of parts of the cdpillary fringe, the pores con-
KiA, where i is the gradient and the area A is (6240. tain both water and soil gases. The quantity of water in
Hence, the flow rate is related to the cube of the fracture a partially saturated medium can be represented in terms
aperture. The hydraulic conductivity for this parallel plate of the volumetric water content (O), which is defined as
model is
(3.33)
(3.31)
where V , is some unit volume of soil or rock, and V,, is
The plates employed in experimental work confirming the volume of water. If the pore is completely saturated,
Eq. 3.30 were smooth optical glass. Gale and others the volumetric water content is equal to the porosity (n).
(1985) review the several attempts to incorporate frac- Therefore, in the unsaturated zone, water contents vary
ture roughness in the experiments. We can think of frac- over the range, 0 5 O In.
ture roughness as the local or point-to-point variability
in aperture along a fracture. Roughness, among other HydraulZc and Pressure Heads
things, reduces the aperture openings, and most of the
expressions for conductivity in rough channels are of In the zone of saturation, the driving force for ground
the form water was demonstrated to be the hydraulic head, de-
fined mathematically as
K= PIl.RbZ
(3.32)
12/41 + C(Xyq h=r+JI (3.34)
where C is some constant larger than one, x is a group where z is the elevation head and JI is the pressure head
of variables that describe the roughness, and n is some (P/pll.g). Darcy’s law describes this flow. These same
power greater than one. Hence, roughness causes a de- concepts apply to the movement of water in the unsatu-
crease in hydraulic conductivity. rated zone. There are, however, several differences and
It thus follows that the influence of aperture opening complications that have to be considered. The most im-
is of most importance in the discrete flow of fluids in portant difference is that pressure heads in the unsatu-
throughgoing fractures. The frequency of such openings rated zone are less than atmospheric. We see this pattern
in a given rock, like permeability, appears to be skewed in Figure 3.18, which illustrates measurements of pres-
to the right (Figure 3.17). The validity of the cubic law sure heads below the water table (+ > 0).at the water
52 Cbapter 3 / Ground-Water Movement

flows into or out of the tensiometer to maintain hydraulic

equilibrium. Tensiometers used in field applications are
b.?= ir, - Pi somewhat more sophisticated, but the operating princi-
ples remain the same.
4 Although the water table usually serves as a major
boundary between the disciplines of hydrogeology and
soil physics, it is much less significant from a process
point of view. The total potential of the system as indi-
cated by hydraulic head is continuous across the water
table, and flow always proceeds in the direction of de-
Unsaturated creasing hydraulic head. The only concession to the nega-
rn% tive pressure heads above the water table is that the
&z = z, hydraulic head is determined as the algebraic sum of Ic,
and z.
Water table This point can be illustrated with the example repre-
sented in Figure 3 . 1 9 ~With
. a constant and continuous
rainfall on the soil surface, a flow of water will occur
through the porous medium. By making measurements
with tensiometers at a large number of points and con-
touring these data. one could discover the pressure head

Saturated Rainfall
zone
xtl =z, t. yl

(a)
Ceramic
CUP
N = hydraulic head 36
z= elevation head
y = pressure head
32
Figure 3.18 Pressure heads are negative in the unsaturated
zone, zero along the water table, and positive in the satu- 28
rated zone. As shown, the total hydraulic head is the alge-
braic sum of the elevation head and the pressure head.
24 24

- 5 20
table (I,!I = 0), and above the water table (J, < 0). Pressure u
20
0
heads less than atmospheric are also referred to as tension m
I
m
c.
heads, or suction heads acknowledging the capillary 16 $ 16
k k
forces that bind water to solids in the unsaturated zone.
It is this “negative” pressure head in the unsaturated 12 12
zone that accounts for the fact that water present in
partially saturated soils cannot flow into a borehole (pres- 8
sures less than atmospheric are considered negative).
That is to say, the pressure in boreholes is atmospheric
4
and the pressure in the surrounding rock is less than
atmospheric, and water does not move from low pressure
to high pressure. 0
0 12
A device known as a tensiometer is used to measure x-distance (m) x-distance (rn)
pressure heads in the unsaturated zone. In its simplest (6)Pressure head (m) (c) Hydraulic head (m)
form, it consists of a porous ceramic cup connected by Figure 3.19 Example of saturated-unsaturated flow in a
a water column to a manometer. The very fine pores of vertical cross section (a)under a constant rainfall of 4.0 X
the cup remain filled with water that provides a hydraulic lo-‘ m/s. Panels (b) and (c) illustrate the pressure head and
connection between the soil water and the water col- hydraulic head fields, respectively (modified from Sullivan
umn. As the pressure heads change in the soil, water and Suen, 1989).
 3.5 Fbw in tbe Unsaturated Zone 53

distribution in Figure 3.19b. Near the bottom of this field where 0,.and 6, are the residual and saturated volumetric
is the water table, which is defined by the I) = 0 contour. water contents, respectively, I)is the pressure head, and
As expected, the pressure heads above the water table (Y, m,and n are empirical constants determined by nonlin-

are negative, and those below are positive. The hydraulic ear regression. Usually, it is assumed that IIZ = 1 - 1/11.
head distribution for this example is shown in Figure The van Genuchten equation fits the pressure head-
3.1%. It is obviously continuous across the entire do- saturation curve in Figure 3.20 with 0, = 0.034; 6, =
main, and it is not really particularly obvious where the 0.391; a = 0.0298; n = 2.513. We can check this relation-
water table actually is. The gradient of hydraulic head is ship by calculating the volumetric water content for Fig-
generally downward, reflecting the predominantly down- ure 3.20 that coincides with a pressure head of - 100 cni
ward flow of water in the medium.
0.391 - 0.034
0 = 0.034 +
(1 + 10.0298. -1001‘“’)”‘x’3 (3.36)
Water-Retention Curves
e = 0.098 = 0.1
Generally, as the water content of a soil decreases, the
A quick check of the figure confirms the fit.
pressure head becomes more negative or, alternatively,
The actual shape of the water-retention curves de-
the capillary pressure increases. The capillary pressure
pends on several factors with pore-size distribution prob-
increases because the water that remains as the soil be-
ably being the most important. Figure 3.21 illustrates
comes drier finds itself in smaller and smaller voids. This
curves for sand, fine sand, and silt loam. The sand has
is the same behavior we noted in Eq. 1.1, which showed
the most uniform distribution of large pores, while the
that capillary pressure increased as the diameter of a
silt loam with a broad, grain-size distribution contains
capillary tube decreased. The relationship between pres-
small pores, which will facilitate much smaller pressure
sure head and water content is determined experimen-
heads. The shape of the water-retention curves changes
tally and given in Figure 3.20. This so-called water reten-
depending upon whether the soil is drying out or wetting.
tion curve is nonlinear whether pressure head is plotted
In other words, the relationship is hysteretic. A detailed
on an arithmetic or logarithmic scale. Typically, at both
discussion of this phenomenon is beyond the scope of
large and small water contents, small changes in water
content are accompanied by extremely large changes in
pressure head. The behavior at low water content reflects -20(
the fact that soils never completely lose all of their water.
This lower limit in water content is termed the “residual
volumetric water content.”
In modeling applications, one commonly fits these ex-
perimental data by mathematical relationships. Although
many different fitting equations have been derived, one -15(
of the most common is that proposed by van Genuch-
ten (1 980)
0, - 0,.
e = e,+ (1 + la$/$’)”’ (3.35)
-5 -lo(
-
3

- l o 5 0
I I I I I I I I
I
-lo4

I ! ! I
-5
I I I I I I
-103
-5c
3
-102

-10’

I 1
I I I I I C
.05 .10 .15 .20 .25 .30 .35 .40 0.6
Volumetric Water Content Volumetric Water Content

Ffgure3.20 Water retention ciirve for the Herino fine Pigure 3.21 Water retention curves for sand. fine sand.
sandy loam (modified from Wierenga and others, 1986). and silt loam (from Brooks and Core).. 1966).
52 Cbapter 3 / Ground-Water Movement

this introductory chapter. Readers can refer to more ad- Detailed information on flow in the unsaturated zone
vanced textbooks like Greenkorn (1983). can be found in Richards (1931) and Kirkham and Pow-
ers (1972).
Darcy’s Law for Variably Saturated Fbw
Unsaturated Flow in Fractured Rocks
Yet another complication with the unsaturated zone is
that the hydraulic conductivity is strongly dependent in The movement of fluid in unsaturated fractured rock
a nonlinear fashion upon the degree of saturation and, is o f some interest to hydrogeologists concerned with
consequently, pressure head. When the soil is near satura- hazardous and radioactive waste transport. Of immediaate
tion with a pressure head close to zero, the hydraulic concern, for example, is the proposal for storage of high-
conductivity takes on its maximum value. As the pores level nuclear waste in unsaturated fractured rocks in Ne-
become filled with air, the saturation decreases, the pres vada. Although this is a problem of recent interest, some
sure head becomes more negative, and the hydraulic progress has been made. Of major concern for flow in
conductivity decreases. This relationship is demonstrated such rocks is the conceptual model from which the phys-
in Figure 3.22. It thus follows that the form of Darcy’s ics of the flow process must be deduced. If the concep-
law applicable to the unsaturated zone requires that the tual model is correct, mathematical solutions may capture
hydraulic conductivity be expressed either as a function the essence of the flow. If the conceptual model is incor-
of moisture content or pressure (suction) head. From rect or incomplete, the mathematical models may be m i s
Hillel (1 97 1) leading.
The presence of surface-connected root channels and
q = -K($) grad H (3.37)
worm holes has historically served as an analog for frac-
where H may contain both suction and gravitational com- tures, prompting both theoretical and experimental
work. Bevin and Germann (1982) show that these chan-
nels conduct water onIy during rdinfall events where the
(3.38) infiltration rate through the soil matrix is less than the
precipitation rate at the surface. The water enters the
vertical cracks and can laterally infiltrate the matrix at
depth. Ponding of water can occur in vertically discontin-
uous fractures. Davidson (1984, 1985) has modeled such
a system, and Bollma and Dekker (1 978) have performed
experimental work using dyes. Bevin and Germann
(1982) provide an excellent review.
When the infiltration rate at the surfiace exceeds the
rainfall rate, the water at the surface is taken in by the
pores in the rock matrix. The conceptual model of Wang
and Narasimlian (1985) attempts to describe the flow
process by a capillary theory that recognizes that large
pores (fractures) desaturite first during the drainage pro-
cess and small pores (matrix) desaturate last. Thus, fric-
tures will tend to remain dry when the aperture is large,
with some water films or continuous water contained in
the micro aperture portions of the fracture. A dry fracture
has a hydraulic conductivity smaller than the conductivity
of the partially saturated matrix. The continuous air phase
thus produces an infinite resistance to flow across the
fracture except for those places that contain water in
the micro apertures. The water in micro aperture parts
of the fracture constitute pipelines for the transport of
water from one matrix block to another. Thus, the flow
1
Soil Water Content (cubic centimeter
lines avoid the dry portions of the fracture (Figure 3.23),
per cubic centimeter) which act as barriers to flow both along and normal to
the fracture.
Figure 3.22 Soil water content-suction curve for Lakeland
sdnd and calculated and experimental values of hydrmlic Peters and Klavetter (1988) recognize the relationships
conductivity (from Elzeftawy, Mansell, and Selim, 1976. Dis- among vertical flux, hydraulic conductivity, and matrix-
tribution of water and herbicide in Lakeland Pond during ini- fracture flow. If the vertical flux in an unsaturated frac-
tial stages of infiltration. Soil Science, v. 122, p. 297-307. 0 tured rock is less than the saturated hydraulic conductiv-
by Williams and Wilkins, 1976). ity of the matrix, the flow will tend to remain in the
 3.5 Flow in the Unsaturated Zone 55

far, held evidence suggests that fracture flow will com-

mence at some critical matrix saturation less than full
saturation (Rasmussen and others, 1989).
Yet another conceptual model of fracture flow in ;irid
zones has been presented by Nativ and others (1995).
This model is based on field observations at an industrial
complex site in Israel, where a variety of chemicals appar-
ently bypassed the lowpermeability matrix, contamin;it-
ing the lower aquifer via rapid fracture flow. The authors
proposed a conceptual model in which a small portion of
the rainwater percolates downward through the matrix
while a larger percentage of the percolating water moves
through the preferential fracture pathways. In arid re-
gions, during periods of zero or low precipitation, the
fractures drain rapidly due to gravity and/or imbibition
into the rock matrix. Most of the matrix remains at near-
Figure 523 (;onceptual model of partially saturdted, frac- saturation because the time it takes to c h i n the low-
tured, porous medium showing schematically the flow lines
permeability pores is large compared to the frequency
moving around the dry portions of the fractures (from Waiig
and Narasimhan, Water Resources Res.. v. 21, p. 1 8 6 1 - 1874,
of recharge events from storms or, in this case. industrial
1085. Copyright by Anier. Geophys. IJnion). effluent. Thus, to a large degree most of the matrix water
moves slowly and appears not to play an active role in
short-time scale hydrology, short time again referring to
significant precipitation events. The larger the storm
matrix. For this case, Wang and Narasimhan (1985) sug- event or sequence of events, the more water will enter
gest that the steady-state flow field of a partially saturated the fracture and penetrate to greater depths. During small
fractured rock can be understood without detailed precipitation events, significant fracture flow does not
knowledge of the discrete fracture network. If the flux occur, and infiltration takes place slowly through the
is greater than the saturated hydraulic conductivity of matrix. This conceptual model is more in tune with obser-
the matrix, the matrix will saturate and fractures or open vations made in tunnels and underground mines follow-
fault zones will then accept the flow. Although this state- ing precipitation events.
ment appears reasonable from the theory discussed thus

Problems
1. Calculate the specific discharge in m/s for a hydraulic 500 rn head
conductivity of lo-&m/s and a hydraulic gradient of
0.019.
2. Calculate the actual velocity in m/s from the informa-
tion given above for an effective porosity of 0.1 and
o.Ooo1.
3. What is one likely reason for the large differences
in effective porosity cited above?
4. Prove by dimensional analysis, using Table 3.1, that
pressure head has the units of length. For pressure use
F/Lz, for acceleration of gravity use L/Tz, and for density
use M/L?
5. Prove, by dimensional analysis, that hydraulic head 300 rn
has the units of energy per unit fluid weight (L) and
hydraulic potential has the units of energy per unit fluid Problem 6.
mass (L2/T2).
6. Determine both the direction of flow and the gradi- 7. In the following diagram, assume a hydraulic gradi-
ent of flow from the following three points as measured ent of 0.019 with the water level at point A being at an
in the field. elevation of 300 m above sea level. Assume further that
the water level in all the piezometers is at the top of the
-
56 Cbapter 3 / Ground-Water Movement

piezometers. The piezometers are located 5000 m apart. b. Which area has the lowest hydraulic conductivity?
Calculate the following: c. Approximately how many times higher is the con-
ductivity for the region identified in (a)than for the
Piezometer A B C D
region identified in (b)?
Depth of piezometer, m 200 190 50 8
14. Consider the potentiometric surface for a homoge-
Total head, m 300 - - -
neous, isotropic aquifer that, for inflow equal to outflow,
Pressure head, m - - - -
Elevation head, m - - - - would be characterized by a uniform spacing of the
equipotential lines across the flow field.
a. Describe the equipotential spacing in the direction
of flow if the area was subjected to uniform recharge
across the flow field (hint: inflow not equal to
011tflow).
b. Describe the equipotential spacing in the direction
of flow if the area was subjected to uniform natural
discharge across the flow field.
15. The aquifer in Example 3.4 is underlain by a uni-
Sea level formly thick homogeneous clay layer. Beneath this clay
layer is another aquifer. The head in this aquifer near b4
Problem 7. (which is assumed to be 70 m) is 60 m, or 10 m lower
than the measured value for b4. In the vicinity of h, ,
8. A geologic formation has a hydraulic conductivity of
which is assumed to have a measured head of 30 m, the
m/s.
head in the lower aquifer is on the order of 25 m, or 5
a. What is the conductivity in feet per second? m lower than the measured 30 m equipotential. In which
b. What is the conductivity in gallons per day per ft2? of these two regions is the velocity of vertical movement
c. What is the permeability in square centimeters? across the ckay layer greatest and why? Which way is the
d. What is the permeability in darcys? flow directed, upward or downward?
9. Draw or demonstrate the conditions under which 16. Calculate the specific discharge across the clay layer
Darcy’s law can be expressed as qx = (k/p)(dP/dx), in cm/s where the vertical hydraulic conductivity of the
where k is the permeability, p is viscosity, and P is the clay layer is lo-’ cm/s. In which direction is the flow?
fluid pressure as measured at the bottom of a borehole.
10. Which of the following statements for Darcy’s law
are incorrect (if any)?
a. Q = KiA e. v = ( K / n , ) grad b
b. q = K grad h f. v = ( k p g / p n , ) grad h
c. q = ( W g ) grad 4 g. q = ( k p g / p ) grad h
d. q = v / n e 1 Aquifer

11. Three horizontal, homogeneous, and isotropic for-

mations overlie one another, each of which is 20 m thick,
I
with horizontal conductivities of lo-’, and lo-”
m/s. Compute the horizontal and vertical components of
hydraulic conductivity for an equivalent homogeneous,
anisotropic formation. 17. Consider an upper confined aquifer overlying a
lower aquifer, separated by a low-permeability confining
12. A sample is subjected to a hydraulic gradient of db/
100 m
ax = 0.1 cm/cm in the x direction and a flux of 0.001
cm/s is measured in the y direction in response to this
gradient (there is no gradient in the y direction). The
same sample is then subjected to a hydraulic gradient
dh/dy of 0.01 cm/cm in the y direction and a flux qxis
measured in the x direction in response to this hydraulic
gradient (there is no gradient in the x direction). What
is the magnitude of the flux q. in response to dh/ay?
13. Consider the diagram in Example 3.4. Assume that
no flow is added to or lost from the system, that is, inflow
= outflow. tnfltPr;

a. Which area has the highest hydraulic conductivity? Problem 17.

 3.5 Flow in the Unsaturated Zone 57

layer. Other than the information in the diagram, you are (4) The hydraulic gradient in the near vicinity of
given the following: the inflow areas of both aquifers is on the order
(1) The hydraulic conductivity of both the upper of lo-' m/m.
and lower aquifers is on the order of m/s. a. Calculate the inflow Q (m'/s) into both aquifers.
(2) The vertical conductivity of the low-permeabil- b. Calculate the outflow Q (m3/s) out of both aquifers.
ity confining layer is on the order of m/s. c. Calculate the hydraulic gradient in the vicinity of the
(3) The heads in the lower aquifer average about outflow areas of both aquifers.
10 rn higher than those in the upper aquifer. d. Sketch the general form for the piezometric surfaces
of the upper and lower aquifers.
 Cbalbter 4

Main Equations of Flow,Boundary

Conditions, and Flow Nets

4.1 Organixing the Study of Ground-Water F l o w Equations

4.2 Conservation of Fluid Mass
4.3 The Storage Properties of Porous Media
4.4 Boundary Conditions a n d Flow Nets
4.5 Dimensional Analysis

From a field or laboratory perspective, we have pursued The essence of the quantitative approach involves r e p
the concepts of fluid movement in porous rocks about resenting information about a problem in mathematical
as far as they can be carried. We have some idea about terms and eventually solving the resulting equation. Let
the range in magnitude of porosity and permeability in us illustrate this idea with the simple ground-water flow
geologic materials, the meaning and measurement of hy- problem shown in Figure 4 . 1 ~ The. question we would
draulic head, and the determination of flow rates and like to solve is given the pattern of geological layering
direction of ground-water movement. Thus if we can (and the resulting hydraulic conductivity distribution)
map the hydraulic head, we can perform many useful and the time invariant configuration of the water table,
calculations. To go further than this requires new infor- what does the pattern of ground-water flow look like?
mation. This chapter is intended to provide this new The conceptualization of the problem mathematically
information, and the next three chapters will put it (Figure 4.1b) requires (1) finding the appropriate equa-
into practice. tion to describe the flow of ground water, (2) establishing
a domain or region where the equation is to be solved,
4.1 Urganixing the Study and (3) defining the conditions along the boundary (i.e.,
boundary conditions). The solution of the governing
ofGround- Water Flow Equations equation establishes the hydraulic head at specified
What sets hydrogeology apart from many of the other (x, z)positions within the domain. These hydraulic head
geosciences is the emphasis on treating problems quanti- values can be contoured to provide equipotential distri-
tatively. These problems might include predicting how butions from which we can deduce the patterns of flow
much hydraulic head at a point will decline by pumping (Figure 4. lc).
a nearby well for some specified time, or how the concen- This simple example is designed to highlight some of
tration of a contaminant at a point will change in response the new knowledge that is required for the quantitative
to some proposed remedial scheme. The quantitative treatment of ground-water flow. The starting place is in
approaches also provide the basis for interpreting the understanding the form and function of ground-water
results of various tests (e.g., slug tests; aquifer tests) that flow equations and how they are derived. There is a
hydrogeologists routinely undertake in the field. need to learn about boundary conditions and how they
 4.2 Conservation of Fluid Mass 59

(Ui

,Water tabk---varcabte head boundary

Sfde nc- *. -, Side 110-

flow
IlQW
boundary
boundary

x it relatiwe hydraulic
conductivity

K= 1

K = 10
I I

-
- 90 hydraulic head frn.a.s.1.)
Direction of ground-water
flow
(c)

Ffgure 4.1 A geologic problem (a)is conceptualized onto a mathematical problem (b),which is ultimately solved in terms of
the hydraulic head distribution (c).

represent the influence of the world beyond the region that captures the essence of the physics of flow. The
of interest on the specific flow problem. The develop- basis for developing such an equation is a conservation
ment of equations that apply to flow in aquifers makes statement that balances the inflow, outflow, and change
it clear that there are important hydraulic parameters in water mass within a representative volume of po-
that remain to be explained in detail. Finally, we need rous medium.
to make a start on actually solving ground-water flow In other words, a conservation of fluid mass statement
equations through the use of flow nets. may be given as
mass inflow rate - mass outflow rate =
4.2 Conservation of Fluid Mass change in mass storage with time
The mathematical treatment of ground-water flow in units of mass per time. In general, this statement may
through a porous medium depends upon an equation be applied to a domain of any size. Consider this state-
@ Chapter 4 / Main Equations of Flow, Boundary Conditions, and Fbw Nets

Zt 4.6 and dividing both sides by A x A y A z gives

which states that the net outflow rate per unit volume
equals the time rate of change of fluid mass per unit
volume. By making a further assumption that the density
of the fluid does not vary spatially, the density term on
A E the left-hand side can be taken out as a constant so that
Eq. 4.7 becomes
Figure 4.2 Representative volume.

ment as it applies to the small cube of porous material

With the simple transformation from Eq. 4.7 to Eq. 4.8
of unit volume where Ax A y Az = unity (Figure 4.2).
our equation now deals with volumes of fluid per unit
This box serves as a representative elementary volume.
volume instead of mass per unit volume, where the two
From this figure
are related through ptL,= M/V,,. For the left-hand side of
mass inflow rate through the face ABCD = p,,,q.yA y A z Eq. 4.8, ySis merely Q.JA, which expresses a volumetric
(4.1) flow of fluid per unit time per unit area. Thus, the left-
hand side of Eq. 4.8 expresses the net fluid outflow rate
The density pw has the units M/L3 whereas the specific
per unit volume. This being so, the right-hand side of
discharge qxis a velocity L/T so that p,,,qXA y A z has the
Eq. 4.8 must describe the time rate of change of fluid
units of mass per time.
volume within the unit volume, that is, a(V,,,/V,)/at.Thus
In general, the mass outflow rate can be different than
Eq. 4.8 states that the net fluid outflow rate for the unit
the input rate and is given as
volume equals the time rdte of change of fluid volume
within the unit volume.

(4.2) Main Equations of Flow

The net outflow rate is the difference between the inflow As the y’s on the left-hand side of Eq. 4.8 represent
and the outflow or, subtracting Eq. 4.2 from 4.1 Darcy’s specific discharge, we may directly substitute
a ( p u . e >A x A y A z Darcy’s law for anisotropic material, giving
net outflow rate through EFGH = -
ax
(4.3)
Making similar calculations for the remainder of the cube This expression has a positive sign because the q ’ s are
net outflow rate through = - d(Purq,) A x AY (4.4a) negative. Assuming that the material is isotropic and ho-
mogeneous
the face CDHG aY
net outflow rate through = - d(p,,,q,) A x A y A z (4.4b)
the face BCGF az
Adding these results gives
net outflow rate through all the faces =
Some clarification is required here. The term (d/dx)(ah/
ax) represents a space rate of change in the gradient
(4.5) across the unit volume. Accordingly, there must follow
velocity variations in the three component directions. If
in units of mass per time.
these velocity variations cancel each other, for example,
The right-hand side of the conservation statement is
increases in x are compensated by decreases in y , and
merely a change in mass storage with respect to time.
so forth, the fluid mass per unit volume is not changing
For the unit volume presented earlier, this becomes
with time. For this condition, the right-hand side of Eq.
4.8 is zero, which gives
Ax A y A z (4.6)
at
(4.9)
again in units of mass per time. Collecting Eqs. 4.5 and
 4.2 Conservation of Fluid Mass 61

Equation 4.9 is Laplace's equation, one of the most useful Both expressions describe the space rate of change of
field equations employed in hydrogeology. The solution some quantity, in one case the hydraulic head (or a scalar)
to this equation describes the value of the hydraulic head and in the other case a mass flux, M/L'T, which is a
at any point in a threedimensional flow field. vector. The abbreviation grad h or Vh has been used for
Yet another possibility concerning the right-hand side the gradient. To cover both cases, it is convenient to
of Eq. 4.8 is the case where the unit volume has some introduce the notation that is common to both
storage qualities. We can go no further here unless some-
thing can be said about how this storage takes place. Let
us assume that the gains or losses in fluid volume within
the unit volume are proportional to changes in hydraulic which, of course, means nothing until something is put
head, which turns out to be a reasonably good assump- into the parentheses signs. If we put in hydraulic head
tion. That is to say, an increase in head suggests that (or a scalar), this becomes the gradient, which has already
water has gone into storage, and a decrease in head been defined as a vector perpendicular to the equipoten-
suggests just the opposite, that is, a removal of water tial lines, and is given the abbreviation grad h or Vb. If
from storage. To account for these gains and losses, the we put in a mass flux (or a vector) this is called a cliver-
right-hand side of Eq. 4.8 must be of the form -
gence and is given the abbreviation div (p,,.q) or V p,,.q
so that we have
(4. 10)
-v * (PI,.@= - div (p,,.q)
where S, is some proportionality constant. Our reasoning
here is that with fluid flow, the substance of interest is
the gains or losses in fluid volume, not hydraulic head,
so that the constant S, is needed to convert head changes (4.13)
to the amount of fluid added to or removed from storage which is read "divergence of p,,.q." Now, why the term
in the unit volume. As Eq. 4.10 describes the time rate divergence?Any useful dictionary will define divergence
of change of fluid volume within the unit volume, S, as as a difference or deviation. And that is exactly what it
a proportionality constant must be a measure of the vol- is, a difference between the mass inflow rate and the
ume of water withdrawn from or added to the unit vol- mass outflow rate for the unit volume. Thus, divergence
ume when the head changes a unit amount. In this form, has the physical meaning of net outflow rate per unit
nothing can be ascertained about the physics of the pro- volume.
portionality constant s,, but it obviously must have the The left-hand side of our conservation equation is, i n
units L-' to make the equality of Eq. 4.10 dimensionally mathematical terms, a divergence, that is, a net outflow
correct. We call S, the specific storage, and we will have rate per unit volume. With this foresight our main equa-
more to say about it later. tion could have been immediately stated as
With this assumption, the conservation statement be-
comes (4.14)

(4.1 1) which is perfectly equivalent to Eq. 4.7 and which means

the net outflow rate per unit volume equals the time rate
which is called the diffusion equation. The quantity of change of mass per unit volume. In addition, Eq. 4.8
K/S,is called the hydraulic diffusivity and has the units could have been deduced immediately
L2/T. Equation 4.1 1 is another important field equation
in hydrogeology and is used to describe unsteady or (4.15)
transient flow problems. The solution to this equation
describes the value of the hydraulic head at any point in
which states that the net fluid outflow rate eqwls the
a threedimensional flow field at any given time or, more
time rate of change of fluid volume.
precisely, how the head is changing with time.
So we have a simple notation that not only is efficient
There are now a few more, perhaps subtle, observa-
in expressing long, sometimes cumbersome equ;itions
tions that can be made. For openers, let us compare
but contains physical meaning. Now, as long ;IS we ;ire
the side Of Eq' 4'7 with the expression for the on the topic of operators, there is yet another one th;it
gradient
can be used for Laplace's equation (Eq. 4.9) and the left-
ah ah ah hand side of the diffusion equation (Eq. 4.1 1). For these
i-+ j-+ k- (4.12a)
ax ay az expressions we define
62 Cbapter 4 / Main Equatfons of Flow, Boundary Conditions, and Fbw Nets

where v2 is a new operator, called a Iaplacian. Again, water obtained per unit volume drained is rather substan-
this operator means nothing until something is placed tial and is obviously equal to the volume of pore space
in the parentheses. Thus, Laplace’s equation and the dif- actually drained. For the confined case of Figure 4.36,
fusion equation can be neatly and compactly expressed as however, a drop in head is not accompanied by drainage
from storage as the aquifer remains fully saturated at all
V2h= 0 (4.16a) times. Further, the amount of water obtained in response
to the unit head drop is a small fraction of that obtained
(4.16b) in the unconfined case. An immediate question arises as
to how a fully saturated confined aquifer can release
respectively. water from storage or, equally perplexing, how it can
We have covered much ground in this section. We not take in additional water in response to head increases.
only have learned about the conservation of mass, but In what manner is space created to accommodate this
have reformulated this conservation theorem in terms of added water?
the important field equations referred to as the diffusion A similar set of questions was pondered in 1925 by
equation and Laplace’s equation. We have also intro- Meinzer (Meinzer and Hard, 1025) as a result of his study
duced a shorthand set of operators that are useful to of the Dakota Sandstone. Meinzer concluded that both
describe the threedimensional problem. These mathe- the water and the porous structure are elastically com-
matical ideas are among the most abstract that we will pressible, and changes in head are accompanied by
encounter in hydrogeology. Nevertheless, they can be changes in both water and pore volume. His calculations
understood conceptually, and the student is advised to for the Dakota Sandstone indicated that over a 38-year
study them well because they will appear time and again. pumping period, more water was pumped and removed
from the aquifer than the total recharge volume over that
time span. In addition, this volumetric withdrawal could
4.3 The Storage Properties not be accounted for by the slight expansion of water
when the pressure is lowered. Meinzer (1928) concluded
OfPorous Media that the interstitial pore space of the sandstone was re-
The specific storage was defined as a proportionality duced to the extent of the unaccountable volumetric
constant relating the volumetric changes in fluid volume withdrawals from storage. His problem, then, was to
per unit volume to the time rate of change in hydraulic demonstrate that aquifers compressed when the fluid
head. This concept is easy to grasp for the unconfined pressure was reduced and expanded when the fluid pres-
condition of flow such as depicted in Figure 4.3~3,where sure was increased. Several lines of evidence were ex-
drainage release from storage is accompanied by a drop pbred, including the response of water levels in wells
in the water table, or hydraulic head. The amount of to ocean tides and passing trains, and subsidence of the

Volume of
,4,..watw released
’ from storage .-..

Figure 4.3 Diagrams illustrating the concept of storativity in (a)an unconfined aquifer
and (b) a confined aquifer (from Heath, 1982).
 4.3 Tbe Storage Properties of Porous Media 63
land surface. These investigations established the ground- 4.20 into this result gives
work for later work by Theis (1940) on the source of
water derived from wells and by Jacob (1940), who devel-
oped most of the flow equations discussed in this section.
The bracketed quantity [pugn&] is the specific storage
Compressibility of Water and Its Relation of a rock with an incompressible matrix, defined as the
to Spec@ Storage for ConBned Aquifers volume of water released from or taken into a unit volume
due exclusively to expansion or contraction of the water
The isothermal compressibility of water may be de- when the pressure head changes by a unit amount. For
fined as a porosity of 20% and the value of water compressibility
cited, this amounts to about 9 X lo-- m3of water released
from each cubic meter of saturated sediment when the
pressure head declines 1 m. This is not much, but this
where PI,.is the fluid compressibility, with units of pres- is the nature of confined aquifers.
sure-', K,*.is the bulk modulus of compression for the
fluid, V,.is the bulk fluid volume, and P i s pressure. Here,
Example 4.1
pu,reflects the bulk fluid volume response to changes in
fluid pressure at constant temperature Tand mass M. The
minus sign is used because the fluid volume decreases as
the pressure increases. The isothermal compressibility
puscan be taken as 4.8 X lo-"' mz/N (2.3 X 10-"ft'/lb)
for ground water at 25°C.
As mass conservation requirements demand that the
product of fluid density and fluid volume remain con-
stant, that is, M = p,V,,., a decrease in fluid density in
response to a decrease in pressure is accompanied by an
increase in fluid volume. That is, as p,,.V,.equals a con-
stant, d(p,,.V,,.)= 0, or
100 m /
p,.dV,. + V,,.dp,,.= 0 (4.18)
From this equation and Eq. 4.17, Figure 4.EI

dV,,.=
v,,.dP,,.
-- The specific storage due exclusively to expansion or con-
(4.19a)
PU traction of the water has been determined to be on t h e
order of 9 X lo-- m-' for a sandstone with a porosity o f
dV,,. = - V,,.P,,.dP (4.19b)
20%. How much of the total pumpage of 110 X 10'' m'
Equating the right-hand side of these statements gives has been supplied by expansion of the water? Assume
that the total head drop over the area was on the order
of 100 m.
According to the definition of specific storage, there
The last terni in this expression is arrived at by recogniz- ;ire 9 X lo-- m3 of water removed per cubic meter o f
ing that the pressure head is only one component of total rock when the head is lowered 1 m. The total volume
head h = z + P/p,g so that if the elevation head % is of water removed from the part of the aquifer shown
invariant with respect to time, rlh/dt = (l/p,g)dP/aC. due to water expansion per unit decline in head is
Equation 4.20 states that fluid density decreases with readily calculated:
decreasing pressure (the water expands) and increases
total volume/unit head decline
with increasing pressure (the water contracts). Thus, a
= S, X area X formation thickness
confined saturated volume will release water from stor-
= 9 x lo-- m-' x 1 x 10') m' x 1 x m
age when the pressure head is lowered. This idea is
= 9 x 10' m3/m
incorporated in the concept of specific storage where,
from Eq. 4.10 For a 100-ni head drop, the total volume supplied by
expansion of the water is
9 x 10' m3/m x 1 x 10' m = 9 x 1O"m'
This leaves 100 X lo6 m' unaccounted for. which was
Assuming that the matrix is incompressible, that is, the the essence of Meinzer's (1928) dilemma.
porosity does not change with time, substitution of Eq.
64 Cbapter 4 / Main Equations of Flow, Boundaly Conditions, and Flow Nets

Compressibdlity of the Rock Matrix: in excess of hydrostatic pressure. The term excess pres-
Eflective Stress Concept sure is used in the sense that the pressure exceeds the
original hydrostatic pressure.
For a porous rock to undergo compression, there must As conservation laws of fluid mass state that no change
be an increase in the grain-to-grain pressures within the in fluid pressure can occur except by loss (or gain) of
matrix; oppositely, for it to expand, there must be a water, a hermetically sealed cylinder will maintain an
decrease in grain-to-grain pressure. Without such pres- excess fluid pressure indefinitely. If some of the water
sure changes, no volumetric changes can occur. A simple is allowed to escape, the water pressure is lowered, and
analogy of this process has been given in several places the spring compresses in response to the additional load
(Tenaghi and Peck, 1948), one modification of which it must support (Figure 4 . 4 d ) . Hence, there has been a
incorporates a spring, a watertight piston, and a cylinder. transfer of stress from the fluid to the spring. When
In Figure 4.4a, a spring under a load ahas a characteris- the excess pressure is completely dissipated, hydrostatic
tic length z. If the spring and piston are placed in a conditions once more prevail, and the stress transfer to
watertight cylinder filled with water below the base of the spring is complete (Figure 4 . 4 ~ ) .
the piston, the spring supports the load u and the water The analogy to be made is that in any porous water-
is under the pressure of its own weight, as demonstrated filled sediment, there are pressures in the solid phase by
by the imaginary manometer tube in Figure 4.46. In Fig- virtue of the points of contact, the resilient grain structure
ure 4.4c, an additional load Acr is placed on the system. represented by the spring, and there are pressures due
Because water cannot escape from the cylinder, the to the contained water, represented by the watertight
spring cannot compress, and the additional load must be cylinder. The former are referred to as intergranular pres-
borne by the water. This is again demonstrated by the sures, or effective stresses, and the latter as pore-water
imaginary manometer, which shows the fluid pressure pressures, or neutral stresses. The total vertical stress

i i
1 1

1.
....

tcn tcl
Figure 4.4 Piston and spring analogy showing the transfer of the support for the
added load from water pressure to the spring.
 4.3 Tbe Storage Properties of Porous Media dfi
acting on a horizontal plane at any depth is resolved into To grasp the full significance of the effective stress
these neutral and effective components concept, it is important to remember that any increases
to the left-hand side of Eq. 4.23 are compensated by
Cr=B-+P (4.23) corresponding increases on the right-hand side. Witness
where u is the total vertical stress, 6 is effective stress, the train example. On the other hand, a decrease in
and P is neutral stress. For the analogy cited, Au equals pressure caused by pumping a confined aquifer has no
A P at the initial instant of loading, and Au equals A D effect on the total stress. Instead, that part of the load
at the terminal condition. Intermediate between these carried by the grain structure must increase in proportion
extremes, total vertical stress is always in balance with to the decrease in fluid pressure. Expressed matheniati-
the sum of the effective and neutral stresses. Neutral cally, u is constant and, from Eq. 4.23,
stresses, whatever their magnitude, act on all sides of
d @ = -dP (4.24)
the granular particles, but do not cause the particles to
press against each other. All measurable effects, such as the negative sign indicating that a decrease in fluid pres
compression, distortion, and a change in shearing resis- sure is accompanied by an increase in intergranular
tance, are due exclusively to changes in effective stress. pressure.
A simple demonstration of the effective stress concept
is shown in Figure 4.5. This figure shows an observation Matrix Compressibility and Its Relation
well in the vicinity of a railroad station. As a train ap-
to Specific Storage of Conpned Aqurers
proaches the station, the load of the train is added to the
total stress as expressed on the left-hand side of Eq. 4.23. Earlier in this chapter, a relationship was developed be-
This added stress appears to be carried at least in part tween the compressibility of the fluid and the specific
by the ground water, as witnessed by the rise in water storage, or at least that part of the specific storage deiling
levels. The water level in this confined aquifer then de- with the fluid component. A similar approach may be
clines to a new position as the fluid pressure is dissipated taken for the component of storage due to pore volume
by diffusion of the pore fluids to areas of lower pressure. changes. That is, we should like to state that the change
The added weight of the train then appears to be carried in pore volume is proportional to the change in effective
entirely by the aquifer skeleton. Hence, there has been stress. if such is the case, effective stress can be expressed
a rather rapid stress transfer from the water to the solids in terms of fluid pressure and, ultimately, in terms of total
with a commensurate decrease in porosity. When the head. Accepting such an effective stress-pore volume
train leaves the area, the effective stress is decreased relationship requires another proportionality constant.
immediately, which now causes a vertical expansion of which in this case is an elastic property of the porous
the aquifer; that is, the porosity has been recovered. medium. This property is a material compressibility of
Owing to this sudden increase in pore volume, the water some sort, and it is important to ascertain exactly what
level in the well declines sharply, but shortly returns to compressibility we require. For example, the compress-
its original position. ibility of water as defined was expressed for the condition
of constant mass, a volumetric dilation of water where
the mass remained constant throughout. The definition
we seek must be intimately related to the manner in

+0.03 fi which pore volume changes are related to changes in

effective stress.
-c

.? +0.02 From a physical perspective, a change in effective

stress as defined so far merely means that the individual
;C
to.01
.-
a
-4 0
grains (which are assumed to be incompressible) 1i;n.e
been pushed closer together; that is, the pore volume
-
r
has been reduced due to grain rearrangement. For o u r
-;-0.01 purposes here, water must be permitted to drain freely
8
while the sample is being compressed. In addition. as
-0.03 I I I I I I the stress giving rise to this reduction in pore volume is
a vertical stress, the compression is assumed to take place
5 10 15 20 25 30 only in the vertical direction. These conditions dictate
Time (minutes) the nature of the proportionality constant relating port!
Ffgure 4.5 Water-level fluctuations in an observation well volume changes to changes in effective stress. The com-
near a railroad station (from Jacob, Trans. Amer. Geophys. pressibility we seek must be described at constant pres-
Union, v. 20, p. 666-674, 1939. Copyright by Amer. Geo- sure; that is, the water must be permitted to drain freely
phys. Union). out of the deforming sample so that fluid mass is lost.
-
66 Cbapter 4 / Main Equations of Flow, Boundary Conditions, and Flow Nets

Other constraints include incompressible grains and one- remain constant so that ail changes in head are due to
dimensional (vertical) compression. Within these con- changes in pressure head. This means
straints, the vertical reduction of the pore volume is
d P = pu,gdh (4.27)
exactly equal to the volume of pore fluid expelled. The
required expression is Equation 4.26 thus becomes
dvu*= PpVbpu@ (4.28)
Dividing both sides by vb and dh and considering a unit
(4.25) volume (vb = 1 ) and a unit pressure head decline (dh =
1 ) gives
where AVb = AVp for the special case of incompressible
grains. In this formulation, PI,is the bulk (total) compress-
(4.29)
ibility in units of pressure-', Kb is a bulk modulus of
compression, &, is the vertical compressibility, H, is a The quantity dV:, is an expression for a volume of water
modulus of vertical compression referring to the pores removed from a unit volume of rock when the pressure
only, V, is the bulk volume, V, is the pore volume, and head declines one unit, and has the units of length-'.
b is effective stress. Here, PP reflects the pore volume The quantity on the right-hand side is thus defined as the
decrease in response to changes in stress at constant specific storage of a rock containing an incompressibIe
temperature and pressure. The negative sign is a conven- fluid, or the volume of water released from or taken into
tion employed to account for a volume decrease with an storage in a unit volume due exclusively to compression
increase in stress. Typical values for the vertical com- or expansion of the matrix when the pressure head
pressibility are given in Fdble 4.1. Note that the compress- changes a unit amount. For a dense, sandy gravel with
ibility of water is of the same order of magnitude as the a vertical compressibility of mZ/N (Table 4.L), the
compressibility of sound rock. component of storage released due to compression of
As noted, the compressibility PPis perfectly equivalent the matrix amounts to about 1 X m3 of water for
to bulk (total) compressibility &, that is, ab/a@ = aVb/ each cubic meter of sediment when the pressure head
db. When the individual grains are compressible, the declines 1 rn. This volume is equivalent to the pore vol-
pore volume change is no longer equal to the bulk ume decrease. As with the volumes associated with fluid
volume change. expansion, the amount of water produced is quite small.
Following a development similar to that given by From the developments above, and Eq. 4.22, the spe-
Freeze and Cherry (1979),we now present an expression cific storage S, is given by
for the specific storage of a rock containing an incom-
pressible fluid. For the case of incompressible grains, the ss = P..g(P, + nPU.1 (4.30)
change in total volume dV, is equal to the change in The derivations above are sufficient to provide a physi-
water volume dV,,. From equation 4.25 cal understanding of the process of storage release in
elastic media but have been simplified considerably. Actu-
dVb = dV,. = -p,,Vbd@ = PpvbdP (4.26)
ally, it is not possible to obtain a correct expression for
where d @ = -dP (Eq. 4.24). From the expression for the component of storage due to matrix compression
P from Eq. 4.7 alone in that both the fluid and the solids
total head h = z + -, consider the elevation head to
P& are in motion in a deforming medium. To conserve the

Table 4.1 Vertical Cornpresslbility

Coefficient of Vertical Compressibility
Material ft2/lb m'/N bars-'
~~~ ~ ~~ ~~ ~ ~~ ~ ~~ ~~

Plastic clay 1 X 10-'-1.25 X 2 x 104-2.6 x lo-- 2.12 X 10-'-2.65 X lo-'

Stiff clay 1.25 X 10-5-6.25 X lo-" 2.6 X 1W7-1.3 X lo-' 2.65 X 10-'-1.29 X lo-'
Medium-hard clay 6.25 X 10-'-3.3 X 1.3 X 10-'-6.9 X lo-" 1.29 X 10-*-7.05 X lo-'
Loose sand 5 X 10-'-2.5 X 1 X 10-'-5.2 X lo-" 1.06 X 10-'-5.3 X lo-'
Dense sand 1 x 10-'-6.25 x lo-- 2 X 10-"-1.3 X lo-" 2.12 X lO-'-1.32 X
Dense, sandy gravel 5 X 10-'-2.5 X lo-' 1 X 10-"-5.2 X lo-' 1.06 X 10-.'-5.3 X
Rock, fissured 3.3 X 10-'-1.6 X lo-" 6.9 X 10-"'-3.3 X lo-''' 7.05 X 10-'-3.24 X lo-'
Rock, sound less than 1.6 X lo-" less than 3.3 X 1W"' less than 3.24 X
Water at 25°C 2.3 X tO-" 4.8 x lo-"' 5 x lo+
Modified from Domenico and Mifflin (1965). Water Resources Res. 4, p. 563-576. Copyright by Amcr. Geophys. Union.
 4.3 The Storage Properties of Porous Media 67

total mass, two conservation statements are required, Piezotrwtric

one for the fluids and one for the solids. This derivation
is algebraically messy and has been relegated to appen-
dix 1A.
Unlt decline
piemmetric
siirfal:e
Example 4.2
Returning to Example 4.1 determine how much of the
pumpage is supplied by compression of the matrix for
Qf
S, = m-' if one ignores expansion of the water.
total volume/unit head decline
= S, X area X formation thickness
= lo-'m-' x 1 x 10') mz x 1 x lo2 m
= lo', m3/m

For a l W m drop in head,

lo6 m3/m x 1 x lo2 m = 100 x lo', m5
Thus the problem of Meinzer (1928) is resolved. Figure 4.6 Diagram illustrating the storativity for confined
conditions (from Ferris and others, 1962).

Equation for Con@ned Flow in an Aqutfer

be useful to note that the storativity is the volume of
From developments given, the diffusion equation is ex- water removed from a prism of unit basal area extending
pressed as the full saturated thickness of the aquifer under ;L unit
head decline, whereas transmissivity is the volume of
water flowing through one face of that same prism in ;I
unit time under a unit hydraulic gradient. In this sense.
The conditions for steady flow are more readily seen both storativity and transmissivity are terms exclusively
from this expression. When both the fluid and the matrix defined for field conditions in that they relate to thr
are incompressible a,,,
= Pp = O), the right-hand side storage and transmissive properties of geologic forni:i-
of Eq. 4.31 becomes zero and the steady-state Laplace tions of a specified thickness. From Figure 4.7, hydr;iulic
equation is obtained. In addition, both S, and K may be conductivity is the quantity of water flowing in one unit
multiplied by the formation thickness m,giving time through a face of unit area (opening A) under ;I
driving force of one unit of hydraulic head change per
(4.32)

Here the product of S, and m is called the storativity Sand

is dimensionless, whereas the product of the hydraulic
conductivity and formation thickness is called transmis-
0bwrva:icn wells
sivity T, with units of Lz/T. The storativity, or coefficient
of storage, is defined as the volume of water an aquifer
releases from or takes into storage per unit surface area
of aquifer per unit change in the component of pressure
head normal to that sufiace. Stated in another way, the
storativity is equal to the volume of water removed from
each vertical column of aquifer of height m and unit
basal area when the head declines by one unit. Figure
4.6 illustrates the field concept of storativity for con-
fined aquifers.
Transmissivity, or the coefficient of transmissibility, is
defined as the rate of flow of water at the prevailing
temperature through a vertical strip of aquifer one unit
wide, extending the full saturated thickness of the aqui-
fer, under a unit hydraulic gradient. In Figure 4.7, it is Figure 4.7 Diagram illustrating hydraulic conductivity ;ind
the quantity of water flowing through opening B. It may transmissivity (from Ferris and others, 1962).
68 Cbapter 4 / Main Equatfons of Flow, Boundary Conditions, and Flow Nets

unit length. Hence, transmissivity is the product of the Table 4.2 Values of Specific Yield for Various
formation thickness and the hydraulic conductivity. Geologic Materials
Material Specific Yield (O/O)
Spedflc Yield of Aquifers Gravel, coarse 23
The discussions thus far have been concerned exclusively Gravel, medium 24
with confined aquifers. Upon the lowering of head in Gravel, fine 25
such aquifers, they remain fully saturated so that no dewa- Sand, coarse 27
tering occurs, and the water released is volumetrically Sand, medium 28
Sand, fine 23
equivalent to the volumetric expansion of the water and
Silt 8
contraction of the pore space, These processes also occur Clay 3
in unconfined aquifers, but the water volumes associated Sandstone,fine-gnined 21
with them are negligibly small compared to the volumes Sandstone, medium-grained 27
obtained from drainage of the pores. Figure 4.8 shows Limestone 14
the concept of storativity associated with the unconfined Dune sand 38
condition, which is the volume of water drained from the Loess 18
x portion of the aquifer. The storativity under unconfined Peat 44
conditions is referred to as the specific yield, defined as Schist 26
the ratio of the volume of water that drains by gravity Siltstone 12
to the total volume of rock. Specific retention, on the Till, predominantly silt 6
other hand, is the ratio of the volume of water the rock Till, predominantly sand 16
Till, predominantly gravel 16
retains against the force of gravity to the total rock vol- Tuff 21
ume. These relationships may be expressed
From Johnson (1967).
(4.33a)

space to the total volume. Johnson (1967) has summa-

(4.33b)
rized values of specific yield for a variety of rock types
and sediments (Table 4.2). The coarser the material, the
where S, is the specific yield, V,, is the volume of water
drained, VT is the total rock volume, S, is the specific more closely the specific yield approaches total porosity.
retention, and V,is the volume of water retained against
the force of gravity. It follows that the sum of K,,,, and
V,,, represents the total water volume contained within 4.4 Boundary Conditions and Flow Nets
the interconnected pore space within a unit volume of
saturated material. Total porosity is then equal to the Given the differential equations described in the previous
sum of specific yield, specific retention, and the ratio of section, there now remains one additional problem to
the volume of water contained in the unconnected pore be concerned with, namely, their solution or, more spe-
cifically, some additional information required in their
solution. To appreciate this new information, let us first
put down all the essentials of our flow theory:
Unit decline of
water table
/Unit cross-sectionalarea 1. A potential field is presumed to exist; that is, h(x, y ,
z, t) is a well-defined scalar quantity in the field
of interest.
2. The potential h(x, y, z, t ) changes over space and
time. The mapping of this potential for one point in
time constitutes the potentiometric map.
3. Because the potential changes over space, a gradient
exists. This gradient has been demonstrated to be a
vector perpendicular to the equipotential lines; that
is, it is colinear with the flow for an isotropic po-
rous medium.
4. Provided with this gradient, the water is in continu-
Ffgwre 4.8 Diagram illustrating the storativity (specific ous motion and is completely described by Darcy's
yield) for unconfined conditions (from Ferris and others, law. With such motion, the fluid must be conserved.
1962). Thus we define the divergence, which is a net out-
 4.4 Boundary Conditfonsand Flow N e t s @

flow rate per unit volume. If the net outflow rate is leave or enter a region. Expressed mathematically
zero
-ad,
= 0 (4.34)
V2b = 0 (steady flow) aN

If the net outflow rate is not zero where is the potential and N is the normal to boundary.
This means there is no gradient normal to the boundary
so that there is obviously no flow across the boundary.
If there is no flow across a boundary in a two-dimensional
flow region, then the flow must be tangential to the
These statements represent what this chapter has boundary. We conclude that a no-flow boundary must
been mostly about. In their present form, the steady be aligned at right angles to the equipotential lines and
and unsteady flow equations constitute conservation must therefore be colinear with the lines of flow. No-
principles. By themselves they mean little until they flow boundaries are called flow lines, and they intersect
are solved. To this end we add a few more state- the equipotential lines at right angles. Thus, the solution
ments. to our main differential equations is simply two inter-
5. If the flow is steady, given the head or the gradient of secting families of curves. In isotropic media, the lines
head on the total boundary of the region of interest, it are mutually perpendicular, and one family of curves
is possible to calculate h(x,y, z) for the interior. determines the other. Either may be taken as the solution.
4 If the flow is unsteady, given the head or the gradient Such intersecting families of curves are called flow nets,
of the head along the total boundary of the region, with the boundaries as well as the interior consisting of
along with additional information on the initial head flow lines and equipotential lines. For steady flow, the
before transients take place, it is possible to calculate curves do not change over time; for unsteady flow, any
h(x,y, z,t ) for the interior. given configuration represents an instantaneous condi-
tion. Unsteady flow is best viewed as a continuous succes-
Statements 5 and 6 represent some important new sion of steady states, each of which is “steady“ for short
information. To put this information in the proper con- periods of time.
text, it will be recalled that one way to obtain the head A few examples of flow nets are given in Figure 4.9.
distribution for a region is to map it. We now learn that Figure 4.9a portrays flow in the x-z plane in a pervious
it can also be calculated if certain information is known rock unit beneath the base of a dam. This flow is estab-
along the boundaries of the flow region. Slichter first
stated these ideas in a hydrogeologic cotitext in 1899,
but the impact of his effort was prdctically nil until the
late 1930s. It was during that 25-year period from the
late 1930s to the mid-1960s that hydrogeology was trans-
formed from largely a descriptive practice to a highly
quantitative and even predictive one.
Basically, there are two kinds of boundary conditions
to deal with. The first, as discussed, is a specified head
along a given boundary. An important special case for
this boundary condition is where the specified head is
constant, termed a constant head boundary. Any line
along which the head is constant must be an equipoten-
tial line. This means that there can be no flow along the
equipotential line, but instead, for isotropic material, the
flow must be at right angles to the equipotential line.
We conclude that for any region bounded in part by a
constant head, the flow must be directed at right angles
either away from or toward that boundary, and the
boundary is an equipotential line.
A second type of boundary is where the normal compo-
nent of the gradient is specified. Given the normal com-
ponent of a gradient along a boundary, there must be a (6)
flux across the boundary. Thus, we use the termflux Examples of flow nets. Panel (n) represents
F f g u r e 4.9
boundary or, more commonly, pow boundury, where flow in the x-z plane of a pervious strmim underlying a clam.
flow either enters or leaves a region across a boundary. Panel (6)shows flow toward a discharging well in the .vy
A special type of flow boundary is a no-flow boundary, plane as influenced by a line soiirce (constant head
that is, a region across which no flow is permitted to boundary).
-
70 Cbapter 4 / Main Equations of Flow, Boundary Conditions, a n d Flow Nets

lished by the head of water behind the dam, the base of head boundary so that the flow lines intersect it at right
which can be taken as an equipotential line, that is, a angles. This flow field is depicted in the x y plane and
constant head boundary across which the flow is directed represents a mathematical solution t o the diffusion
downward. The contact between the impermeable and equation.
permeable rock unit is another boundary, in this case a Figure 4.10 is from the classic study by Bennett and
nsflow boundary. Hence, this contact is a flow line. The Meyer (1952). This flow net is for the Patuxent Formation
base of the dam also functions as a flow line, whereas where the heads are below sea level (thus, when water
the stream below the dam that receives the ground-water is moving from the 20-ft equipotential to the 30-ft equipo-
discharge functions as an equipotential line. As the flow tential, it is moving in the direction of decreasing head).
pattern forms a system of squares or curvilinear squares, The numerous piezometric lows depicted on the map
the material is assumed to be isotropic and homogeneous. were caused by extensive pumping. Note that the surface
Given this information along with the boundaries, the forms a set of squares or curvilinear squares so it is repre-
interior net has been sketched and represents a graphica1 sented as isotropic and homogeneous. However, the hy-
solution to Laplace’s equation. draulic conductivity for the various pumping regions dif-
Figure 4.5% is a set of flow lines and equipotential lines fers in accordance with the differences in the size of the
in plan view around a discharging well bordered by an squares. In this projection, the head distribution was first
infinite line source. The line source represents a constant mapped and the flow lines then superposed. Figure 4.10

- -Flow
- - lines
- -- - 0 1 2 3 4 Miles
I I I I I
Scale

Ffgure 4.10 Flow net for the Pituxent Formation (from Hcnnett and Meyer, 1952).
 4.4 Boundary Conditions and Floiil Nets -
71

thus represents a field or “mapped” solution to Laplace’s I. Given tlie liead at x = 0 of 50 m and the 1ie;id ;it
equation for whatever boundary conditions are prevalent x = L of 10 m, the equipotential lines can be dr;iwn
in this area. for a homogeneous aquifer. Thus. at s = L / 2 . b =
The volumetric How rate within flow nets can be deter- 30 m, at x = 3I./4, tlie head is 20 m, and s o on.I$,
mined by summing the individual Q’s in each of the superposing flow lines on the equipotential lines. it
flow tubes of Figure 4.10. Another comnion approach is is noted that the impermeable boundaries ;ire How
derived from Darcy’s law expressed for one section of a lines.
flow channel 2. For the simple situation given, Laplace‘sequation is
Ab expressed
AQ= K - a (4.35)
L
where AQ is the flow in one flow channel, A/J is the
head drop across a pair of eqiiipotential lines, I. is the which is integrated to give
distance over which the head drop takes place, and a is
the distmce between adjacent flow lines. For the homo- b = A ~ X+ A?
geneous, isotropic system and its characteristic curvilin-
where A , and A? are constants that have to be evalu-
ear squares, L = a. As the total Q in the system is tlie
ated for the boundary conditions. Assume h = b,, at
AQ times the number of flow channels (n,), and the total
x = 0 and h = 1~~at s = I.. Thus, from this equation,
head drop across the flow system ( A H ) is the A h times
A? = h,, and A , = (17,. - h , , ) / L . The head distribution
the number of head drops (n,,),tlie total flow is expressed
is thus expressed as
n/
Q=-KAH (4.36)
n,/
The flow rate Q is expressed as the volumetric flow rate
Thus, for panel 6, h(x) = h,, at .Y = 0, and h(x) = h, ;it
per unit time per unit thickness of the flow field. For
A-= L . Given numerical values for the head at these
Figure 4 . 9 ~ this flow must be niultiplied by the length
boundaries, tlie interior may be calculated.
of the dam (measured perpendicular to the page) to give
tlie total flow in the previous statement. For Figure 4.10,
the Q must be multiplied by the saturated aquifer thick- Example 4.4
ness to give the total flow in the aquifer.
I n Figure 4.10, punipage from the Sparrows Point District
has been nieasured at 1.O X 10”ft‘ per day. The potentio-
Example 4.3
metric surface is assumed to be steady: that is, the ;iiiioiint
of water passing through tlie formation to suppl!. the
pumping wells is equal to the pumping rate. It is notctl
that there are 15 flow channels associated with the Spar-
rows Point pumping. For ;I total head drop A H of 30 ft.
the number of drops is three so that . l H / / i , / = 10. For ;I
total head drop A H of 40 ft, the number of potc‘ntial
drops is 4 so that AH/#,/ is again 10. Thus. .lH/n,, \\..ill
always be a constant for a region. The regional tr;insniis-
sivity is determined from Eq. 4.36
1 .0 x 10”ft3/day ft
T =Km = = 6666 -
(1 5)( 10) ft cia!,
m‘
= 7.16 X 10-4-
S

Graphical Flow Net Construction

Figure 4.E3 Guidelines for flow net construction have been provided
by various authors for over 50 years. with the tre;itnient
Consider the flow in a channel filled with gravel, cut into by Cedergren (1967) being among the most thoughtfiil.
an impermeable rock. Panel (6) shows a plan view of Perhaps no better tool is available for ;in ;ippreci;ition o f
tlie aquifer. the nature of flow patterns in porous media. The tirst
-
72 Chapter 4 /Main Equations of Flow, Boundary Conditions, and Flow Nets

part of this discussionwill focus on the nonhomogeneous X V h

but isotropic conditions assumed for most aquifers. It
follows immediately that, because of isotropicity, flow Original Transformed
section section
lines will intersect equipotential lines at right angles. For
every flow net considered, some equipotential lines and
(a) (b)
some flow lines will be established by the boundary con-
ditions. When the boundary of a region is of the no-flow
x
variety (impermeable), it is represented by a flow line *
and equipotential lines must intersect it at right angles.
Original Transformed
When the boundary of a region is of the constant head section section
variety, it is an equipotential line, and flow lines must
intersect it at right angles. The same quantity of flow (C) (d)
moves between adjacent pairs of flow lines, provided no
flow enters or leaves the region in the internal part of Figure 4.11 Transformation for flow nets in anisotropic
the net. It follows further that the number of flow chan- material. In panel (6) K,,> K,..
nels must remain constant throughout the net.
The guidelines stated above should all be reasonably
apparent from our previous discussions on boundary con- flow net on the original cross section will be composed
ditions and flow nets. Some “graphical” guidelines might of rectangles elongated in the direction of greater hydrau-
include selecting a practical scale for the net and a practi- lic conductivity.
cal number of flow channels so as not to clutter up
the pattern.
Not discussed in this section but demonstrated on Fig- - Dimensional Analysis
4.5
ure 4.1 is the refraction of flow lines across the boundary Solutions to the diffusion equation will in general encom-
of two formations of different hydraulic conductivity. pass a dimensionless group of terms that either dictate
This will be discussed in detail in Chapter 5. the nature of the diffusion process or demonstrate the
In anisotropic material, flow net construction is im- competition between two rate processes. One of the
peded by the fact that flow lines and equipotential lines more important uses of dimensional analysis is that it
no longer intersect at right angles. The steady-state equa- permits one to recognize the importance of parameters
tion for this flow condition is or, more precisely, groups of parameters in a fluid flow
problem. To arrive at this dimensionless group for the
(4 37)
9 diffusion equation, we first establish the following dimen-
sionless variables that are noted with a superscript “+”
where K,, and Kl,mean horizontal and vertical hydraulic
conductivity, respectively. If we keep in mind that a x+ = x/L
flow net is a graphical solution to Laplace’s equation, a y +=y / L
coordinate system transformation will yield the appro-
priate equation, where we define 2 = x/(Kb)If2and 9 = z+= z / L
y/(K,.)”?so that t + = t/t,

(4.38) ht = h/h,
v+= LV
From a graphical perspective a flow net can always be v2+= LLV2
constructed by multiplying the original horizontal dis-
tances by (K,,/K,,)”’.The hOriZOntdl dimensions of the where L is some characteristic length, t, is some charac-
transformed section will always be shorter or longer than teristic time, and h, is some characteristic head. Note
the original section, depending on whether K,, is larger that the operators are also written in dimensionless form.
or smaller than K,. (Figure 4.11). With these dimensionless quantities, the diffusion equa-
Cedergren (1967) discusses the procedures to follow tion for fluid flow expressed as
for obtaining a flow net for anisotropic conditions. First,
the cross section is redrawn to the appropriate reduced (4.39)
or extended horizontal scale as shown in Figures 4.1 l b
and 4.11d. This is referred to as the transformed section. becomes, in dimensionless form,
The flow net is then constructed on the transformed
scale. Following this, the flow net is redrawn on the
(4.40)
original sections (Figures 4.1 la and 4.1 lc). The resulting
 4.5 Dimensional Analysis 73

where all terms are dimensionless, including the quantity N1 0 - S,L’/K

--=- T*
(4.42)
in the brackets. This bracketed quantity is known as the t, 4 2

Fourier number, NF(,, and occurs in all unsteady flow

problems. where T” is the time constant for a basin. If the time t,.;it
The Fourier number is more readily expressed as which we wish to observe some transient is significantly
larger than T*, the transient will not be observable. and
(4.4 1 ) the basin would appear to be in some steady state. If the
time of observation is less than the characteristic time
where the numerator has already been introduced as the T”,the transient is readily observable. Note that the char-
hydraulic diffusivity. The quantity 1/24 that determines acteristic time T* gets large for large values of S.and /.?
the “well function” W ( u ) used in well hydraulics is of and small values for K. This form for the dimensionless
this form (Chapter 6). Another useful form for the inverse group is useful in studying problems in disequilibrium
of this dimensionless group is compaction (Chapter 8).

Problems
1. Address the following problems as they pertain to total, neutral, and effective pressures at the bottom
Eq. 4.31. of the clay layer beneath the trench? Assume the
a. Express this equation for onedimensional flow. trench is filled with water to land surface.
b. Express this equation for a porous medium in which c. Is it possible to maintain the stability of the bottom
the fluid can be considered to be incompressible. of the trench in 6 above if the trench is dewatered?
c. Express this equation for a porous medium in which d. What is the approximate minimum depth of a trench
both the fluid and the matrix are considered to be excavated in the clay layer that will cause a quick
incompressible. condition or failure at its bottom? Assume the trench
d. In what type of material (clay, siltstone, sandstone, to be filled with water.
gravel, limestone, granite) would you consider ignor- 5. If the total withdrawal from pumping center B o f
in3 the term pII.gnPI,.? Figure 4.10 is 3 X 10’ ni’/day, compute the average
2. The storativity of an elastic aquifer is 3 X lo-’. The transmissivity of the aquifer in the vicinity of pump-
porosity of the aquifer is 0.3, the compressibility of water ing center B.
is 4.8 X 10-“’m’/N, and the thickness of the aquifer is 6. Designate whether the following are equipotential
about 60 m. lines or flow lines and if they are constant head or
a. How many cubic meters of water are removed from no-flow boundaries: AB. BC. CD, DE. FG.
storage under an area of 1 X lo6 m‘ when the head 511-1- , ‘ lrnpervicus blanket
declines one meter? v w e r ievei I
Water level
b. Ascertain the storativity if the aquifer matrix is in-
compressible.
c. What is the value for the coefficient of vertical com-
pressibility of the aquifer?
3. Consider a well pumping from storage at a constant G
rate from an extensive elastic aquifer. Problem 6.
a. The rate of head decline times the storativity
summed up over the area over which the head de- 7. Redraw the flow net of Problem 6 for ( N ) ;in impen+
cline is effective equals -. ?
ous blanket one half as long, and (6) in the ;ibsence
b. At any given time, the total head decline, summed of an impervious blanket. For a pool height of 20 ni
up over tlie area over which the head change is and a hydraulic conductivity of 5 X lo-’ m/s. c;ilcu-
effective, times the storativity equals -. 1
late the volumetric flow for the redrawn flow nets
4. A clay layer extending from ground surface to a depth above and for the flow net of Problem 6 .
of 50 ft is iinderlain by an artesian sand. A piezometer 8. Consider a vertical cross section ABCD where A R is
drilled to the top of the aquifer registers a head of 20 ft the upper boundary. DC is the lower bounrlan. A13
above land surface. The saturated unit weight of the clay is the left-hand side boundary, and BC is the right-
is 125 Ib/ft.’. Assume a water table at land surface. hand side boundary. Let the side boundaries be one
a. What are the values of total, neutral, and effective half the length of the upper and lower boundaries.
pressures at the bottom of the clay layer? Draw flow nets for the following conditions:
b. If a trench is dug to a depth of 20 ft, what are tlie a. A B is a constant head of 60 m, A 5 m c l BC arc iniper-
-
74 Cbapter 4 / Main Equations of Flow, Boundary Conditions, and Flow Nets

meable, and DC is divided into two equal sections, long as side boundary AD. Here the bottom boundary
with the left-hand part being a constant head of DC remains the same whereas the upper boundary has
10 m and the right-hand side being impermeable. a slope. Consider this upper boundary to be a sloping
b. AB is as in (a), DC is impermeable, and the side water table, and draw a flow net in the region assuming
boundaries are divided into two equal segments, DC, AD, and BC are all no-flow boundaries, and the water
with the upper segments being impermeable and table (AB) is a variable head boundary.
the lower segments being constant heads of 10 m. 10. Repeat Problem 9 for anisotropic material, where
9. Alter your vertical section of Problem 8 by making the horizontal hydraulic conductivity is five times larger
the right-hand boundary BC one and one-half times as than the vertical hydraulic conductivity.
 Chapter 5

Ground Water in the Basin

Hydrologic Cycle

5.1 Topograpbic Driving Forces

5.2 Surface Features of Ground-WaterFlow
5.3 Some Engineering and Geologic Implications
of Topographic Drive Systems

~ ~~~ ~~

The driving forces responsible for ground-water move- as an aquifer, and the watertight or impervious beds
ment were known before the turn of the century and are known as confining beds, commonly referred to as
were categorized by King (1899) as gravitational (topo- aquitards. Chamberlin noted further that no stratum is
graphic), thermal, and capillary. King did not mention entirely impervious so that the aquifer can progressively
the importance of tectonic strain as a driving force, al- discharge its fluids down dip. Of special importance here
though he clearly demonstrated an up-dip migration of is the fact that the elevation differences of the outcrops,
water in response to the compaction of sediments in or the outcrop and the subcrop, provide the topographic
active depositional environments. Our interest in this drive for ground-water movement, hydraulicallydepicted
chapter is on how driving forces can be manifested in the by the potentiometric surface (Figure 5.1).
basin hydrologic cycle, and what physical consequences Other configurations of ground-water flow were also
result from these manifestations. Such consequences are being recognized at about this same time. This is docu-
important in a variety of engineering projects and in mented in the United States Geological Survey’s Nine-
understanding a host of processes operating near the teenth Annual Report (l899>,part two of which is enti-
Earth’s surface. tled “Papers Chiefly of a Theoretical Nature.“ Included
in this volume is the classic work of Slichter (1899) on the
principles and conditions for the movement of ground
water, and a theoretical-field analysis by King (1899).
5 2 Topographic Oriving Forces Of importance here is King’s observations regarding the
Tbe Early Field Studies influence of gravity on shallow ground-water flow:
“The basal principles of artesian wells are simple. The The contours of the ground water lerlel show that the
schoolboy reckons himself as their master.” This is the (water) surface presents the features of the hills and
opening statement in Chamberlin’s (1 885) classic treat- z~alleysapproximately conformable with the relief
ment on the requisites and qualifying conditions for the forms of the surface above, the water being lotil where
occurrence of artesian wells in pervious rocks. The pervi- the surface of the ground is low, and higher where the
ous stratum discussed by Chamberlin has been defined surface of the ground is high.
-
76 Chapter 5 / Ground Water tn the Basin Hydrologtc Cycle

surface of minimum potential energy (the hydrostatic

or nonmoving condition). This tendency is opposed by
continual replenishment. Elevation differences in the out-
crops areas for the confined case, or in the land surface
in the unconfined case, provide the topographic drive
mechanism for movement of the contained fluids. Contin-
ual replenishment assures that this drive persists indefi-
nitely.
Figure 5.1 Schematic cross section showing trace of the
potentiometric surface and areas of flowing and nonflowing
wells (from Hubbert, 1953). Reprinted by permission. Conceptual, Graphical, and Mathematical
Models of UnconBned Flow
In short, King stated that the water table was everywhere From what we have learned from the field studies dis-
a subdued replica of the topography, and water moves cussed in the previous section, the question arises as to
from topographically high areas to topographically low the possibility of establishing at least some features of a
areas. Figure 5.2, taken directly from King, diagrammati- conceptual model that will always be valid. One such
cally illustrates these facts. As with confined systems, the feature, for example, might be that the water table is a
topographic drive is provided (and limited) by differ- subdued replica of topography, at least for unconfined
ences in surface elevation. flow systems that are undisturbed by pumping. Thus,
The field evidence offered by Chamberlin (1885) and given information on the topography of a region-even
King (1899) still provides the basis for the categorization if nothing else is known-it is at least possible to say
of ground-water flow under confined and unconfined something about the uppermost part of the saturated
conditions. At one extreme is Chamberlin’s concept of zone. In addition, following King’s portrayal of flow, the
the confined condition, exemplified by dipping aquifers flow appears to “branch” at the topographic divides,
bounded by low but finite permeability units with contin- with the flow going in divergent directions. Thus, topo-
uous replenishment at the outcrop areas. If a we11 is graphic divides appear to be ground-water divides as
placed in such an aquifer, the water will rise in the well well, this being the obvious result when the water table
to the elevation of the water table in the intake area, is a subdued replica of the topography. Again, in concert
minus any head losses incurred from the point of intake with King, streams appear to receive all the ground-water
to the point of measurement. The pressure-producing discharge from the adjoining parts of the valley, with
mechanism is the hydrostatic weight of the body of water none of the flow passing underneath the stream. The
extending down dip from the water table in the outcrop streams, or at least the major ones occupying the major
area. The dynamic mechanism required to maintain the topogrdphic lows, are ground-water divides in that they
high-pressuresystem is the continuous replenishment by prevent the intermingling of ground water from opposite
precipitation. At the other extreme is the unconfined flanks of the same basin. This, too, is inevitable for the
condition discussed by King (1899), where the water lowest topographic points in a region where the water
table is a subdued replica of topography. The dynamic table reflects the topography.
mechanism required to maintain this flow is likewise a It follows that a useful conceptual model of a flow
continuous replenishment by precipitation. Indeed, as field should incorporate those facts that have been
water offers no resistance to deformation, its movement learned from field investigations. Such a model was pro-
from high elevations to low elevations would, in the vided by Hubbert (1940) and is shown in Figure 5.3.
absence of recharge, result in drainage of the water con- Notice in this model that King’s measurements on the
tained in the topographic highs, thereby producing a flat relationship between the water table and the land surface

Figure 5.2 Dhgrarnrnatic section illustrating ground-water flow in a watershed (from

King, 1899).
 5.1 Topograpbic Driving Forces 77

Equipotential line ter emplacement encounters higher hydraulic heads with

depth so that the water level in the well will actually be
above the water table adjacent to the well. Hubbert
(1940) noted that it is possible to obtain a flowing well
for such conditions in that all that is required is the
intersection of the piezometer with an equipotenti;il line
that has a larger value than the land surface at that point.
\ I I ‘\\
\\.J
1.
/
I
Again, the reason for this behavior is shown in Figure
5.3, where the head increases with increasing depth so
Figure 5.3 Topographically controlled flow pattern (from
Hubbert, 1940). Reprinted by permission of the Journal of that flow is directed upward. Regions where the flow o f
Geology, University of Chicago Press. Copyright 0 1940. water is directed upward with respect to the water table
are called discharge areas, and the line that separms the
recharge area from the discharge area is called a hinge
are faithfully reproduced. In addition we find that King’s line (Toth, 1963). Clearly, the concept of a discharge
observations at the topographic highs and lows are like- area must be incorporated in our conceptual model; the
wise incorporated in this model. This has the effect of field problem is to delineate such areas, and we expect
setting up individual flow cells in each of the basins them to coincide with topographically low areas.
shown in Figure 5.3, with the flow in one basin being Now Iet’s stretch our basin out a bit, giving it a flatter
more or less independent of the flow in adjacent basins. water table (topographic) expression (Figure 5.4). We
It is noted also that the symmetrical portrayal of the land note that the recharge and discharge areas are diminished
surface has the effect of producing a mirror image flow somewhat, and we notice a large region where the equi-
pattern for the flow in adjacent flanks of a given basin. potential lines are vertical or near vertical. If a piezometer
There are, of course, some differences between the is emplaced in this region, it will follow a single equipo-
presentations of King (1899) and those of Hubbert tential line. Thus, once the piezometer encounters the
(1940). In Hubbert’s conceptual diagram, the flow is water table, the water level will not change with increas-
presented by intersecting flow lines and equipotential ing depth. It will be noticed here that this is the only
lines. It is noted that the water table has a point of set of conditions in the flow field where a piezometer
intersection with each of these equipotential lines, with emplaced below the water table actually provides a direct
the value of the hydraulic head anywhere on a given measurement of the water table elevation. Such a region
equipotential line taking on this water table elevation at can be called a region of lateral flow, which corresponds
its point of intersection with the equipotential line. This to Toth’s (1963) hinge line.
is an important detail and provides information that can- As long as the water table is not fluctuating too severely
not be obtained from King’s diagram. For example, con- in response to recharge and discharge, the flow depicted
sider a piezometer emplacement somewhere in the vicin- in Figures 5.3 and 5.4 is steady, or nearly so.It follows that
ity of the topographic high. After passing through the these figures represent graphical solutions to Laplace’s
unsaturated zone, the piezometer will encounter the wa- equation in the two-dimensional region depicted. Thus,
ter table and the water will stand in the well at the they are in fact mathematical models of the flow field. A
water table elevation. The piezometer is then set deeper, question now arises as to the nature of the bounclan.
intersecting one of the equipotential lines. This means conditions so that it is possible to reproduce other con-
that the water will now stand in the well at an elevation
equal to the elevation of the water table where that
equipotential intersects the water table. Thus, the water
level in the well will be at a lower elevation than the
water table at the point of piezometer emplacement.
With continued depth of emplacement, the water level
in the well will continually occupy a lower elevation.
The reason for this is obvious from Figure 5.3, which
shows that in the vicinity of topographic highs, the poten-
tial (head) decreases with increasing depth as the flow
is directed downward. Regions where the flow of water
is directed downward with respect to the water table
are called recharge areas (Tbth, 1963).The incorporation
of recharge areas is thus part of our conceptual model;
the delineation of such areas is a field problem, but we Figure 5.4 Topographically controlled flow pirrerii (from
expect them to coincide with topographic highs. T&h, J. Geophys. Res., v. 67, p. 4375-438’. 1902. Copyright
Focus now on the topographic lows, where a piezome- by Amer. Geophys. Union).
-
78 Chapter 5 / Ground Water in the Basin Hydrologic Cycle

zt r

z = zc

-alJ
=o s
a% x=C
s=0 x=$ .u=L x
Figure 5.5 Two-dimensional region denionstrating typical
boundary conditions for regional flow. Figure 5.E1 Twtxlimensional flow region.

figurations of flow for other water table configurations mined from

and basin depth to basin length ratios. Because of the
symmetry demonstrated on either side of a stream dis
charge area, one limb of a basin will serve these purposes.
Because of the diverging flow at the topographic high and
and the confluence of the flow at the topogrdphic low,
these lateral boundaries are obviouslyof the no-flow type. n nx
all=
That is to say, no flow crosses the topographic high into L cosh (nnzf,/L) L
or out of the region, or crosses the topographic low into The equations given are general in that they apply to any
or out of the region. As no-flow boundaries, these may water table configuration. For a specific case, assume
be represented by flow lines indicating that the flow is that the equation of the water table (6(x, z,,) = A - B
tangential to them. Provided we assume there is a real cos nx/L. The coefficients then become
impermeable layer at some depth in the region, this im-
permeable layer is likewise a no-flow boundary. It, too, -B
must be represented by a flow line. These boundaries
a,,= A ; a, = . other a’s = 0
cosh (nz,,/L)’
can be placed in the general category; that is, they will
always be present for the major topographic low and the The solution for Laplace’s equation then becomes
major topographic high of a region, and for a specified
depth. The upper surface or water table is neither a flow &.x, Z) =A -
cosh (nzf,/L)
line nor an equipotential line, and is a specified head
boundary where the head varies as a function of space This case permits a good understanding of the geometri-
in the same fashion that the elevation of the land surface cal controls on the spatial distribution of the potential.
varies as a function of space. Clearly, this boundary condi- It is noted that at the points z = z,~for x = 0, x = L ,
tion will be specific for a given topographic expression. and x = L / 2 , the value of the potential is A - B, A +
If this head varies as a function of time as well, the 8, A, respectively. Further, at x = L / 2 for all z, the
problem is no longer one of steady flow, but of unsteady potential equals A; that is, a vertical equipotential line
flow. The boundary conditions so described are given in exists at x = L / 2 . This vertical equipotential corresponds
Figure 5.5. An analytical solution to this problem for to the midline of flow.
three fixed boundary conditions and a variable water
table configuration is given in Example 5.1.
Enects of Basin Geometry on Ground-WaterFlow
The mathematical description of the boundary conditions
Example 5.1 (From Domenico and Palciauslaas, 1973) given in Figure 5.5 was first stated by Toth (1962, 1963)
A solution for Laplace’s equation in the region of Figure in two modernday classic studies. This work was the
5E.1 where the upper boundary is not specified but all first significant extension of Hubbert’s (1940) ideas in
other boundaries are of the no-flow variety is over 20 years. T6th’s work is based on the following a s
sumptions:
30
nnx nrrx
&x, z) = a, + n= I a, cosh -L
cos -
L
1. The medium is isotropic and homogeneous to a spec-
ified depth, below which there exists an imperme-
where C#J is the hydraulic potential and the coefficients able basement.
a, and allare determined from the equation of the water 2. Flow is restricted to a two-dimensional vertical sec-
table for various special cases. The coefficients are deter- tion. The topography can be approximated by simple
 5.I Topographic Driwing Forces 73

where the second term on the right-hand side corre-

sponds to the regional slope and the third term to the
local relief superposed on the regional slope. In this third
term, 6 is the amplitude of the sine wave and A is the
number of oscillations, so that L / A is the number of flow
cells. As noted in Figure 5.6, E' is the height of the major
topographic high above the major topographic low and
x is a variable ranging from zero to the basin length I,.
x=o x = I. From this analysis, Toth (1963) identified a local system.
which has its recharge area at a topographic high and
Figure 5.6 Sinusoidal water table with a regional slope
its discharge area at a topographic low that are adjacent
to each other; an intermediate system. which is character-
ized by one or more topographic highs and lows located
curves, such as straight lines or sine waves, and the between its recharge and discharge areas; and a regional
water table is a subdued replica of the topography. system, which has its recharge area at the major topo-
graphic high and its discharge area at the major topo-
3. The upper boundary of the two-dimensional section graphic low (Figure 5.7). The important conclusions from
is the water table, the lower boundary is the impervi-
this study are as follows:
ous basement, and the lateral boundaries are ground-
water divides. 1. If local relief is negligible (6 sin 2nx/A = 0, Eq. 5.1)
and there is a general slope of the topography, only
Totli was not content to investigate the individual flow
regional systems will develop where ground water
cells demonstrated in Figure 5.3.Instead he considered
moves from the major topographic high to the major
a sinusoidal water table with a regional slope of the form
topographic low.
(Figure 5.6)
2. If regional slope is negligible (B'xIL = 0,Eq. 5 . l ) ,
only local systems will develop. The greater the relief
(5.1)
the deeper the local systems that develop.

Key Standard Datum

Region of regional system

Figure 5.7 Two-dimensional isotropic flow model showing the distribution of local. intermedi-
ate, and regional ground-water flow systems (from TBth. J. Geophys. Res., v. 68, p. 4795-4812,
1963. Copyright by h e r . Geophys. Ilnion).
-
80 Cbapter 5 / Ground Water in tbe Basin Hydrologic Cycle

3. If both local relief and a regional slope are negligible,

neither regional nor local systems will develop. Wd-
terlogged areas will be common, and the ground
water may discharge by evapotranspiration in re-
sponse to a flat water table near the land surface.
4. Given both local relief and a regional slope, local,
intermediate, and regional systems will develop.

Egects of Basin Geology on Ground-Water F b w

The work of Tdth demonstrates the importance of basin
geometry on ground-water flow. By geometry, we mean
the ratio of basin depth to length and the relief or topogra-
phy making up the surface. As the medium was assumed
isotropic and homogeneous, little can be inferred about
the influence of basin geology on ground-waterflow, that
is, the influence of the stratigraphic framework with its
attendant permeability variations. This problem was ad-
dressed by Freeze and Witherspoon (1966, 1967) within
the framework of the T6th model. In their approach, n
homogeneous and isotropic interlayered formations
were treated within the two-dimensional vertical section
(Figure 5.8). The water table is incorporated with a series
of straight-line segments so that any configuration can
be approximated as a function of the space variables.
The boundary value problem solved by Freeze and
Witherspoon is actually n interrelated problems. When
(cl
n equals 3, Figure 5.8 applies. The condition that h, =
h,,+,ensures a continuous potential across the contact of Figure 5.9 Diagrams of flow-line refraction and conditions
at the boundaries between materials of differing perme-
two layers. A second condition is required so that fluid
ability.
mass is conserved when flow crosses a boundary be-
tween adjacent strata of differing permeability. This is
referred to as the tangent refraction law (Figure 5.9).
Analytically, a refraction of the flow lines occurs such ratios of the tangents of the angles the flow lines make
that the permeability ratio of the two units equals the with the normal to the boundary.

-k,_ -- tana, (5.2)

k2 tan a2
In Figure 5.9b, the hydraulic gradient in the lower-
permeability unit is steepened to accommodate the flow

d
I
. .
91 =m
91 = 9 2 I I
crossing the boundary from the unit of higher conductiv-
ity; in Figure 5.9c, both the hydraulic gradient and the
cross-sectional flow area in the high-permeability unit
are decreased to accommodate the flow crossing the
boundary from the low-permeability unit.
Freeze and Witherspoon (1966, 1967) employed nu-
merical methods to solve the flow problem discussed
earlier. With analytical methods, both the problem and
the solution are expressed in terms of parameters. This
was the method used by Tdth (1962,1963).With numeri-
cal solutions, both the problem and the solution are ex-
pressed in terms of numbers. Hence, with the latter
method, each case is quite specific, and several cases
Pigure 5.8 Two-dimensional,unconfined model with n ho- must be solved to obtain some general information on
mogeneous, isotropic layers (from Freeze and Witherspoon, the subject.
Water Resources Res., v. 2, p. 641-656, 1966. Copyright by The main conclusions of "0th (1963) regarding control
h e r . Geophys. Union). of topography on the ground-water flow were verified
 5.1 Topograpbic L3riving Forces 81

with the modeling effort of Freeze and Witherspoon. ing a discharge scenario at its terminus that would not
This is demonstrated in the sequence of figures given as be predicted on the basis of topography. Shifting the
Figure 5.10.Figure 5.10~ shows that a lack of local relief high-permeabilitylens to the discharge end of the system
results in development of a single regional system. Figure has similar effects on the overlying units.
5.106demonstrates the effect of adding a slight amount The sloping stratigraphy case is demonstrated in Fig-
of topographic relief, and Figure 5 . 1 0 ~demonstrates the ures 5.12~and 5.126.If the dip of the bed is in the
local, intermediate, and regional flow system develop- direction of topographic drive (Figure 5.12a),the flow
ment when the relief is altered to that characteristic of is down dip. With the hydraulic conductivity contrasts
hummocky terrain. demonstrated, the behavior here is identical to what we
The primary effort of Freeze and Witherspoon (1967) would expect from the Chamberlin model. If the dip of
dealt with determining the role of permeability contrasts the bed is opposite the topographic drive, the flow is
in influencing various degrees of flow-line refraction at up dip, with the large conductivity contrasts causing
the permeability boundaries between adjacent strata as downward flow in the overlying low-permeabilityunit.
well as the influence of this contrast in controlling Such a system has been postulated for part of the flow
the resulting equipotential and flow-line distributions field in southern Maryland (Figure 5.13).
throughout the formations. These effects are shown in The points just discussed become important when as-
Figure 5.11. Figures 5.1 la and 5.116show the effects certaining the influence of structure on ground-water
of a high-permeability layer underlying a layer of lower flow. For example, most diagrams illustrating flow in
permeability. As the permeability of the lower layer in- folded beds consisting of anticlines and synclines show
creases, it acts like a pirating agent for the flow, forcing the flow directed at right angles to the synclinal and
near-vertical flow through the uppermost unit. This is anticlinal axes, that is, up and down the structural limbs
best demonstrated by the orientation of the equipotential (see, for example, Figure 5.1). This, of course, is appro-
lines in the lower-permeabilityunit for a 100-to-1 perme- priate for such folded rocks that obtain their topographic
ability contrast. Figures 5.1Ic and 5.1 I d show the effects drive from the elevated outcrop areas. If, however, a
of a high-permeability lens in the flow field. If the lens series of anticlines and synclines have been uplifted into
is in the recharge area, it too acts like a pirating agent, a meteoric recharge area and possess a distinct plunge to
forcing vertical flow through the overlying unit and creat- lower elevations, the flow will be parallel to the structural

(CJ

Figure 5.10 Effect of water table configuration on regional ground-waterflow through homoge-
neous isotropic mediums (from Freeze and Witherspoon, Water Resources Res. v. 3, p. 623-634,
1967. Copyright by Amer. Geophys. Union).
 Next Page

82 Cbupter 5 / Ground Water In tbe Basin Hydrologic Cyck

(d)
Ffgure 5.11 Regional flow showing the effect of permeability contrasts in adjacent layers (a)
and (6)and the effect of a high permeability lens in the flow field of c and d (from Freeze and
Witherspoon, Water Resources Res., v. 3, p. 623-634, 1967. Copyright by Amer. Geophys.
Union).

(b)
Figure 5.22 Regional flow in sloping topography (from Freeze and Witherspoon, Water Re-
sources Res., v. 3, p. 623-634, 1967. Copyright by Amer. Geophys. Union).
Previous Page

5.1 Topograpbic Driving Forces 83

s 2. A major valley will tend to concentrate discharge in

2 i the valley. Where the regional water table slope is
c
d z
m uniform, the entire upland is a recharge area. In
e hummocky terrain, numerous subbasins will be su-
perimposed on the regional system.
Sea
3. A buried aquifer of significant permeability will act as
level a conduit that transmits water to principal discharge
areas, and it will thus affect the magnitude of re-
charge and the position of the recharge area.
4. Stratigraphic pinchouts at depth can create recharge
or discharge areas where they would not be antici-
pated on the basis of water table configuration.
A field application of the ideas just presented is shown
in Figure 5.14 for the Palo Duro Basin in Texas. The Palo
Duro has been studied as a potential repository for the
Figure 5.13 Diagrammatic cross section through southern disposal of high-level radioactive waste by the U.S. De-
Maryland showing the lines of ground-waterflow (from partment of Energy. Figure 5 . 1 4 is~ a geologic cross sec-
Back, 1Na). tion demonstrating the various geologic units in the sec-
tion, where the salt beds have been considered to serve
as the repository unit. Figure 5.146 shows the flow pat-
terns as measured or inferred in the field, with the large
axes in the direction of the plunge. In this case, the
hydraulic gradient across the salt beds reflecting their
elevated rocks in the recharge area provide the topo-
low permeability. Figure 5 . 1 4 is
~ a numerical model of the
graphic drive, and the anticlinal axes serve as ground-
flow field. Note that the downgradient lateral boundary is
water divides separating the flow systems in adjoining
depicted as a constant head in that the flow presumably
synclines.
continues across this area to discharge in parts of the
Based on the information presented thus far, it is possi-
basin not depicted in this flow diagram.
ble to obtain a picture of regional flow for any basin for
which the pertinent data are available. Data requirements
include permeability distribution and geometry of the Ground Water in Mountainous Terrain
basin boundaries. Although the ideal homogeneous case
Mountainous terrain occupies about 20% of the Earth's
is contrary to field observations, it is merely a condition
land surface and presents some special problems in hy-
in the mathematical development, not an assumption
drologic analysis, mostly because hydraulic head data are
necessary to the general validity of the theory. Further,
seldom available to confirm the suggested conformation
the steady-state solutions do not deny the existence of a
between the land surface elevation and elevation of the
fluctuating upper boundary of flow. The main argument
water table. Further, mountainous regions promote deep
is that the effect on flow patterns will be small if (1) the
circulation of ground water, denying access to ground
fluctuating zone is small compared with total saturated
water outcrops that aid in regional investigation. A fur-
thickness and (2) the relative configuration of the water
ther complication arises because mountainous regions
table is unchanged throughout the cycles of fluctuation.
are frequently fractured and may be in an active state
Given that these conditions are reasonably satisfied, the
of compression or extension, suggesting that fracture
value of the information provided by the flow net can
apertures may be functionally related to the state of stress
best be appreciated if one is interested in the distribution
in the Earth's crust. Thus, the elevation of the water
of recharge and discharge areas along the water table, the
table may be intimately related to a changing hydraulic
depth and lateral extent of local systems in hummocky
conductivity of the region and the variable climatic fac-
terrain, and the degree to which the zones of high perme-
tors that influence infiltration. For a given hydraulic con-
ability act as major conductors of water and their overall
ductivity distribution, the lower the infiltration rate, as
influence on the flow pattern. In summarizing the factors
controlled by climatic factors, the deeper the water table.
that influence these items, Freeze and Witherspoon
It follows that the simple relations given by King(1899)
(1967) cite the following as the most important:
and Hubbert (1940) and substantiated in numerous stud-
1. The three most influential factors affecting the poten- ies in low-lyingsediments may not apply to mountainous
tial distribution are terrain characterized by a fracture permeability. In such
a. The ratio of basin depth to lateral extent. cases, the water table may be considered a free surface
b. The configuration of the water table. whose depth and configuration depend on the inter-
c. Variations in permeability. play between infiltration and permeability distribution.
 td U'
Figure 5.14 Geology, regionalflow, and simulation offlow in tbe Palo Durn Basin. (a) Geology and (c) simulation
offrow,from Senger and otbers (1987); (b) regional flow, from Bair (1987). Figure (b) reprinted by permission of
Ground Water. Copyright 0 1987. All rights reserved
 5.1 Topographic Driving Forces 82

Jamieson and Freeze (1983) were among the first to use ture zones. They conclude that permeability has the
a free surface technique as applied to mountainous ter- greatest impact on the mountainous flow system. Asym-
rain in British Columbia. They used a fixed infiltration metry can cause the displacement of the ground-water
rate to estimate the range of hydraulic conductivity that divides from the topographic divides and a relatively
might be expected to produce a water table at a given small increase in the vertical permeability of fractures
depth. Each hydraulic conductivity pattern resulted in a relative to the horizontal permeability causes significant
different elevation of the free water surface. This study declines in the water table elevation (on the order of
has been expanded by Forster and Smith (1988) to in- 400 m). The authors also state that high relief can pro-
clude the effects of varying infiltration rates, surface to- mote deep ground-water circulation to elevated tempera-
pography, topographic symmetry, and permeable frac- ture regions, requiring a modeling of both the fluid flow

117" 116"
I 38"

17'

p
%c
%
7,
%. Las Vegas

t 6'

- 1200

0
- Line of equal potentiometrtc level in
meters above sea level. Dashed where
inferred. Contour interval = 100 m.
Potentiometric level control point (well or spring)
-
0

0
10 20 mi

10 20 30 40 km

Figure 5.25 Potentiometric surface map for the Death Valley ground-water system encorn-
passing Yucca Mountain (modified from DOE, 1988&).
-
86 Chapter 5 / Ground Water in the Basin Hydrologic Cyck

and the thermal regime. The thermal aspects of mountain- Late Quaternary time, with the “widening” occurring in
ous regions will be discussed in a later chapter. the form of vertical fracture development. The mountain
One mountainous region that does not qualify as “data is characterized by extremely low infiltration (less than
poor” is Yucca Mountain in southern Nevada, a volcanic 1 mm/yr) and a deep (boo m), rather contorted water
tuff pile located north of Las Vegas. Yucca Mountain is table configuration. Winograd and Thordarson (1975)
currently under study as the first site in the United States have shown that the water table throughout the Death
for disposal of commercial nuclear waste. As part of the Valley flow system (which includes Yucca Mountain) is
southern Great Basin, a distinct feature of the Yucca a series of “plateaus” separated by sharp steps and ranges
Mountain region is crustal extension, which suggests that in altitude from over 1900 rn to below sea level in Death
the mountain has become “wider” during Late Miocene- Valley (Figure 5.15). On the “plateaus,” the gradients are

0 5 10 15 20

Kilometers

tl Paleozcks not mapped

Figure 5.26 A map of the Paleozoic rocks and the potentiometric surfkce at the study area [modified by
Strom (1993) from Robinson (1985)l.
 5.1 Topographic Driving Forces 87

small, measured in tens of meters per kilometer. Between Ground Water in Carbonate Tewain
the “plateaus,” slopes as large as 10 and 30%are present.
Szymanski (1989) relates this water table configuration Carbonate terrains occupy about 10%of the Earth’s sur-
to the hydraulic conductivity structure, with steep slopes face and supply ground water to about a quarter of the
occurring where the upper transmissive part is missing global population (Ford and Williams, 1989). Carbonates
or thin, and small gradients in areas characterized by provide a challenge to the hydrogeologist because the
a thick upper transmissive unit. There are, of course, rocks are soluble and thus susceptible to the formation
alternative causes such as faulted barriers to flow or of self-organized networks of solutionally enlarged fmc-
highly transmissive zones conducting a limited flux. In- tures. The result is a double-porosity aquifer. Most o f
deed, the geology underlying the volcanic tuffs may be the water moves slowly through the matrix and little-
the primary control on steep and shallow gradients enlarged fractures and is the fraction predominantly sani-
throughout the region (Fridrich and others, 1991, 1993). pled in wells. The network of enlarged fractures, o r con-
Note on Figure 5.16 that the steep gradient in the volcanic duits, makes up the second fraction. These commonly
rocks roughly coincides with those areas where the make up a tiny fraction of aquifer volume and thus ;ire
Eleana aquitard no longer overlies the Paleozoic high- unlikely to be intercepted by wells. Nevertheless, the
permeability carbonates (Fridrich and others, 1991, conduits will transmit solutes rapidly. For instance. in ;I
1993). Thus water may be “perched” in the volcanics limestone aquifer in England, Atkinson (1 977) recorded
where the Eleana is present, only to cascade into the linear velocities of 520 m to 21,200 m/day using tracers
carbonates at the contact. In the northeastern part of the in 38 tests over distances of up to 1 1 km. Calculations
region, the low-permeability Proterozoic quartzite serves showed that the conduits made up a mere 0.03%o f
as the perching layer. The flat gradient automatically fol- aquifer volume, showing that almost all of the specific
lows in that much of the water in the volcanics has been yield of 0.92%is from the matrix and from little-enlarged
siphoned off to the directly underlying carbonates. fractures through which water moves v e n slowly.
The geologic map shown is rather crude, and more In Central Kentucky flow in the unconfined Mississip-
information will be required to fully understand this pian aquifer is toward the Green River. Core samples
phenomenon. indicate that the limestone has a primary porosity of 3.3%
Lastly, it is important to recognize the water table- and a hydraulic conductivity of 2 X lo-’’ni/s ( i h w n
topographic coincidence in the model studies discussed and Lambert, 1963). Solutional enlargement of fractures
thus far is part of the boundary conditions. The free has been extensive, resulting in integrated conduit net-
surface analysis suggests that the depth and configuration works and caves. Quinlan and Ray (1981) used water
of the water table will depend on the interplay between levels from 1400 wells to produce a water table map,
infiltration and permeability distribution. A subsurface part of which is shown in Figure 5.18. The water table
topographic high on an aquitard, for example, may result is marked by a number of troughs, and each trough termi-
in a ground-water divide in an area of uniform precipita- nates in a downgradient direction at one or more springs
tion and recharge. In Figure 5.17, the aquitard acts as a along the Green River. More than 500 tracer tests were
perching layer and its topographic high supports a conducted to these springs to identify flow directions.
ground-water divide that would not be suspected from over distances up to 22 km. The tracer tests showed that
surface topogrdphy. flow routes converge on the water level troughs and that
this convergence may occur at considerable distances
from the springs. Thus tributary flow occurs throughout
the aquifer. Linear velocities along the conduits. as deni-
onstrated by the tracer tests, varies from less than 200
m/day to more than 10 km/day.
For more than 20 years sewage effluent rich in
heavy metals was discharged into the ground about
2 kni from the axis of one major water table trough.
The effluent was identified in a number of springs along
the Green River, but in none of the 22 sampled wells
between the sewage-treatment plant and the river (@in
Ian and Rowe, 1977). This demonstrates that springs can
be effective monitoring points in carbonate rocks, since
they may discharge the integrated flow from consider;tble
areas. Conversely, wells that are “downgradient“ from ;I
facility are not reliable unless they can be shown by
Figure 5.1 7 Subsurface topographic high on perching tracer tests to lie on the flow path from that facility
layer supporting ground-water divide. (Quinlan, 1990).
 Next Page

-
88 Chapter 5 / Ground Water in the Basin Hydrobgtc Cycle

Banks and Rictor (1953) suggest five methods of con-

trolling saltwater intrusion, four of which require control
of the water table or piezometric surface at or near the
coast. The five methods include artificial recharge, a re-
duction or rearrangement of the pumping wells, estab-
lishing a pumping trough along the coast, thereby lim-
iting the area of intrusion to the trough, formation of
a pressure ridge along the coast, and installation of a
subsurface barrier. These methods have been discussed
at length by Todd (1 980).
Field observations in coastal areas suggest the presence
of a mixing zone where miscible sea- and fresh water
interfuse about their boundary. This phenomenon is best
documented in the Biscayne aquifer near Miami, Florida
(Kohout, 1960; Kohout and Klein, 1967). There is ample
evidence to suggest that the mixing zone between the
fresh and salt water moves seaward during periods of
heavy recharge and inland during periods of low freshwa-
ter head. During the periods of inland flow, the cyclic
flow of salt water in the mixing zone tends to limit the
extent to which seawater invades the aquifer. That is to
say, salinity differences establish density differences that
drive the cyclic motion. A typical cross section of the
mixing zone is given in Figure 5.20, and a generalized
I theory of the phenomenon is given by Bear and Bach-

-
.egend
50"- Water level
Spring
mat (1967).

0 Spring with sewage effluent

Figure 5.28 Water table map and flow routes from tracer
tests.

Ground Water in Coastal Regions

Although frequently of limited relief, topographic drive
systems play a major role in coastal areas where mixing
of fresh water and salt water occurs. Under natural condi-
tions, it is the flow of fresh water toward the sea that
limits the landward encroachment of seawater. With the
development of ground-water supplies and subsequent
lowering of the water table or piezometric surface, the
dynamic balance between fresh and seawater is dis-
turbed, permitting the seawater to intrude usable parts
of the aquifer (Figure 5.19). This phenomenon has been
reported in several parts of the world, including the Neth-
erlands (Ernest, 1969) and Israel (Schmorak, 1967). The
best documented cases in the United States include Long
Island (Lusczynski and Swanenski, 1966), Miami (KO-
hout, 1961), and many parts of California (California De-
partment of Water Resources, 1958). Saltwater intrusion (b:
is a special case of ground-water contamination and is Figure 5.29 Hydraulic conditions near a coastline (a)not
yet another phenomenon subject to numerical simulation subject to seawater intrusion and (6) subject to seawater in-
(Pinder and Cooper, 1970; Segol and Pinder, 1976). trusion with an advancing seawater wedge.
Previous Page

5.1 Topographic Driving Forces 83

-4 __
Key
Terminus of fully cased well
By designating pf and pr as the densities of fresh and
u' Land surface
salt water, respectively, the condition of hydrostatic bal-
ance is expressed

P G = P!lg(hf + r) (5.3)
2E -20- lsochlors in ppm
or

where x is the depth below sea level to a point on the

interface. If the density of fresh water is taken as 1.O and
I I I I I I I I I I I I
1600140012001000 800 600 400 200 0 200 400 seawater as 1.025
Distance from shoreline (feet)

Figure 5.20 Cross section of a mixing zone near Miami, x = 40!, (5.5)
Florida (from Kohout and Klein, 1967). Reprinted with per-
mission of Intl. Assoc. of Hydrol. Sci. as confirmed by Ghyben and Herzberg by observation.
This means, where this condition is approximately cor-
rect, a freshwater level of 20 ft above sea level corre-
sponds to 800 ft of fresh water below sea Level. Stated
in another way, a lowering of the water table by 5 ft will
The FresbwaterSaltwater Interface cause a 200-ft rise of salt water. It is further noted that
in Coastal Regions the slope of the interface is 40 times greater than the
There are many studies that attempt to ascertain the slope of the water table.
position of the freshwater-saltwater interface, both in The length of the prevailing saltwater wedge under the
coastal areas and underlying oceanic islands, and for con- hydrostatic conditions of the Ghyben-Herzberg formula
ditions of flow as might be induced in the vicinity of may be established for the simple geometry of Figure
pumping wells. Many of these studies have one major 5.22. From Darcy's law of the form Q = KiA = KirizY,
assumption in common, namely, that the two fluids-salt where i is the hydraulic gradient, iiz is thickness, and Y
water and fresh water-are immiscible and are separated is the length of the shoreline
by a rigid interface. That is to say, the mixing zone dis-
cussed above is replaced by a line (or, more appropri-
ately, a plane) across which no flow can occur. The (5.6)
interface is common to both fluids and in some cases
only one fluid is perniitted to move and in others both where Q' is the discharge per unit length of shoreline,
fluids may move. We will discuss these theories here as and (h,.- O ) / L is the hydraulic gradient where h, is taken
applied to coastal regions, focusing on their applications at the distance L from the shoreline. The Ghyben-
and limitations.
The Ghyben-Herzberg Relation At the turn of the
century, it was generally thought that salt water in coastal
areas occurred at a depth approximating sea level. Two
/--
investigators, working independently, demonstrated that
seawater actually occurred at depths below sea level
equivalent to approximately 40 times the height of fresh
water above sea level (Ghyben, 1889; Herzberg, 1901).
The analytical explanation of this phenomenon is re-
ferred to as the Ghyben-Herzberg formula and is derived
through simple hydrostatics. For two segregated fluids
with a common interface, the weight of a column of fresh
water extending from the water table to the interface is
balanced by the weight of a column of seawater ex-
tending from sea level to the same depth as the point
on the interface; that is, the weight of the column of
fresh water of length b, + z equals the weight of the Figure 5.22 Hydrostatic conditions of the Ghyben-
column of salt water of length x (Figure 5.21). Herzberg relation.
90 C b a p t w 5 / Ground Water in the Basin Hydrologic Cycle

+L+

Figure 5.22 Idealized geometry to calculate the length of

the saltwater wedge from the Ghyben -Herzberg relation.
Figure 5.24 Idealized geometry to calculate the length of
the saltwater wedge.

Herzberg relation predicts that at a distance L from the

shoreline Glover (1 964) where the total discharge takes place be-
low sea level. The position of the freshwater-seawater
interface is determined from

Solving Eq. 5.6 for hf and equating this result to the

right-hand side of Eq. 5.7 gives

(5.8) where z is the depth below sea level to the interface, x is

the distance measured positive inland from the shoreline,
Thus, the length of protrusion of salt water under natural and Q' is the freshwater flow per unit length of shoreline.
conditions is directly proportional to the hydraulic con- The width of the freshwater discharge gap is determined
ductivity and thickness squared and inversely propor- where z = 0
tional to the flow of fresh water to the sea. As the flow
of fresh water to the sea is reduced, say by pumping (5.10)
along coastal reaches, the length of the intruded saltwater
wedge increases. whereas the depth to the interface below sea level at the
Tibe Sbape of the Interface with a Submerged Seep- coastline (x = 0) is given as
age Surface The relationships just cited are only ap-
proximately correct because of the inherent hydrostatic (5.1 1)
assumptions. Further, the coincidence at the coastline
of a zero freshwater and saltwater head, along with the Last, the height of the water table at any distance x is
assumption of a rigid interface across which no flow can given as
occur, effectively closes the system at its discharge point.
Another approach to this problem is demonstrated on (5.12)
Figure 5.23, which gives the results of an analysis by
so that at x = 0, bf = 0. Thus, the greater the flow to the
sea, the deeper the interface, and the larger the required
freshwater discharge gap.

- Land surface Seepage face

/
For the geometry of Figure 5.24, the landward protru-
sion of the saltwater wedge is determined directly from
Eq. 5.9

Fresh water

It is noted that the first term in the brackets is identical

to Eq. 5.8 based on the hydrostatic condition, whereas
the second term is equal to the depth to the interface at
lnterface x = 0 (Eq. 5.11). The role of fresh water flow to the sea
Figure 5.23 Flow pattern near the coast (modified from as the major control on retarding the advance of salt
Glover, 1964). water is clearly shown in this development.
 5.2 Surface Features of Ground-Water Flow 91

the maximum permitted pumping rate should not exceed

Water table

Q,,,,,
9 0 . 6 d2K
~ (
PI - P/ 7)
.. Dagan and Bear (1968) present other relationships deal-
ing with upconing of the interface as well as estimates
QW
of maximum discharge to assure stability.
New interface

Original interface 5.2 Surface Zeatures

Figure 5.25 Freshwater flow to a well pumping above the of Ground-Water Flow
interface.
The saturated ground-water system discussed thus far
has been isolated from the basin hydrologic cycle by
introducing the effect of its environment through bound-
Upconing of tbe Interface Caused ary conditions. In the real world there is a recognizable
by Pumping Wells interaction between ground water and the rest of the
The Ghyben-Herzberg relationship indicates that any basin hydrologic cycle. Some of the manifestations of
lowering of head results in a rise in the interface. When this interaction are readily seen at the surface, whereas
the lowering takes place by pumping wells that withdraw other types are best examined with models of some sort.
water from above the interface, the interface can rise, a Surface features of ground-water flow include all obser-
phenomenon referred to as upconing (Figure 5.25). An vations that can be used to ascertain the occurrence
approximate analytical solution for this upconing has of ground water, including springs, seeps, saline soils,
been given by Schmorak and Mercado (1969). Their solu- permanent or ephemeral streams, ponds, or bogs in hy-
tion gives the new equilibrium elevation to an interface draulic connection with underground water. The areal
in direct response to pumping occurrence of these features is invariably restricted to
areas of ground-water discharge, and their comprehen-
sion requires some knowledge of the nature of ground-
(5.14)
water outcrops (Meyboom, 1966b). The significance of
these features may best be understood by examining a
where z is the new equilibrium elevation, Q is the pump-
few studies that focus on their interpretation.
ing rate, and d is the distance from the base of the well
to the original (prepumping) interface (Figure 5.26). Both
laboratory and field observations suggest that this rela- Recharge-Discbarge Relations
tionship holds only for very small rises in the interface
and there exists a critical elevation at which the interface The prairie profile (Figure 5.27), consisting of a central
is no longer stable and salt water flows to the well topographic high bounded on both sides by areas of
(Schmorak and Mercado, 1969). major natural discharge, has been offered as a model of
Dagan and Bear (1968) suggest that the interface will ground-water flow to which all observable ground-water
be stable for upconed heights that do not exceed one- phenomena in a prairie environment can be related (Mey-
third of d given in Figure 5.26. Thus, ifz is taken as 0.3d, boom, 1962, 1966b). Geologically, the profile consists
of two layers, the uppermost being the least permeable,
with a steady-state flow of ground water toward the dis-
charge areas. The unconfined flow pattern has been sub-
stantiated by numerous borings for both small-scale sys-
tems, say, a typical knob and adjacent kettle common to
rolling prairie topography, and for a scale of the magni-
tude of the prairie profile. Recharge and discharge areas
have been delineated and are characterized by a decrease
in head with increased depth and by an increase in head
with increased depth, respectively. The occurrence of
1
d Upconed interface flowing wells is noted in parts of the discharge area.
i Original interface
As it is apparent that most of the natural discharge
occurs by evapotranspiration, considerable attention has
Ffgure 5.26 Upconing of interface in response to been given to this phenomenon and to the surface fea-
pumping. tures observed where it occurs. Included among these
92 Chapter 5 / Ground Water in the Basin Hydrologic Cycle

- Approximate boundaries
of recharge area of
+- Area of deep-
flowing wells

1
-11
deep layer

k- Area of
Discharge area of
Discharge area:
saline creeks,
flowing ;
negligible
side hillbase
valley;
flow, playa soils;
saline and
wells
I 1
saline river valley. deeo- and
Precipitation shallow-flowing

-k 50 miles _______c(
Figure 5.27 The prairie profile (from Meyboom, 1966b: Ground-water
studies in the Assiniboine River drainage, Pt. 1, The evaluation of a flow
system in southern central Saskatchewan Geol. Survey Canada Bull. 139.
Energy, Mines, and Resources Canida. Reproduced by permission of Sup-
ply and Services, Canada, 1990).

observations are (1) the occurrence of willow rings and observations in the prdirie environment of Alberta,
the chemical character of water bodies centered within Canada.
them, (2) the distribution and types of vegetation of saline The mapping techniques discussed here are only of
soils with respect to the occurrence of local and regional value in arid and semiarid regions, where surface water
flow systems, and (3) the location of ponds and bogs is not sufficientlyabundant to mask or conceal the surface
with respect to ground-water flow. effects of ground-water flow.
With regard to (l), Meyboom (1966a, 1966b) noted Investigations of the type discussed have not been
that most of the willows are located in the higher areas, restricted to the prairie environment. Extensive investiga-
which suggests that the recharge area is covered by nu- tions in Nevada, starting perhaps with Meinzer (1922),
merous discharging points, each of which is character- have focused on surface features as they relate to the
ized by a willow ring. Willows are phreatophytes, defined occurrence of ground water. In discussing basins in the
by Meinzer (1927) as plants whose roots extend to the Great Basin and the lakes that occupied them during
water table so that they obtain water for transpiration the Pleistocene epoch, Meinzer (1922) recognized three
directly from the saturated zone. As willows have a low types: (1) those in which lakes still exist, (2) those that
tolerance for saline (alkali) conditions, their occurrence no longer have lakes and are discharging ground water
is associated with waters that have not moved very far from the subterranean reservoirs into the atmosphere by
in the system. Although their occurrence in the higher evaporation and transpiration, and (3) those that do not
watershed areas is widespread, they do occur elsewhere have lakes and in which the water table is everywhere
in the basin. These other occurrences are regarded as so deep that they do not discharge ground water except
possible manifestations of local flow systems superim- by subterranean leakage out of the basin.
posed on the regional ground-water flow. The basin of type 2 is exemplified in early studies of
With regard to (2), extensive areas of saline soil occur Big Smokey Valley (Meinzer, 1917). Water enters the
in areas of regional ground-water discharge, where a net alluvial basin by influent seepage of streams of the alluvial
upward movement of mineralized ground water takes fan areas, and at the contact between the surrounding
place. A consistent transition noted is from willow vegeta- mountain ranges and the alluvium. Ground water is dis-
tion on the watershed areas to halophytic plant communi- charged by evaporation and transpiration in the valley
ties within the discharge areas. A halophyte is a phreato- lowlands. Studies of the Big Smokey Valley and similar
phyte that thrives on saline waters. A related transition environments provided the first understanding of the
is noted for local flow systems that receive water from surficial manifestations of ground-water flow, including
highly saline formations. Where the local system is not the distribution of soluble salts in discharge areas
replenished by saline water formations,the ground water (Meinzer, 1927). The concentric arrangement of phreato-
is relatively fresh, and saline zones fail to develop. In phytes was first noted in these early studies, with salt
the latter case, freshwater phreatophytes occupy the grass occupying the inner belt where the water table is
discharge area and make possible its delineation as an near land surface. Other species, such as greasewood
end point of a local system. Toth (1966) reported the in northern Nevada and mesquite in southern Nevada,
results of a similar surface mapping method based on occupy the outer belt, where the water table is farther
 5.2 Surface Features of Ground-WaterFlow 92

below land surface (Meinzer, 1927). This type of basin time period occurred regionally throughout eastern
is exemplified in Figure 5.28, where the basin alluvium Nevada and western Utah, thus suggesting that above-
is underlain by low-permeability rock. average precipitation was widespread throughout the
The basins of type 3 have been investigated by Wino- drainage area.
grad (1962). Eakin (lo&), and Mifflin (1968), resulting in Most of Winograd’s (1962) conclusions are based on
the delineation of two regional flow systems in carbonate the results of an extensive drilling program within the
terrain in southern and eastern Nevada. These systems Nevada test site and adjacent areas. Hydraulic head dat;i
are typified by large drainage areas encompassing several suggest that ground water moves from at least as far
topographic basins, relatively long flow paths, and large north as Yucca Flat to Ash Meadows, without regard to
spring areas of invariant discharge, where water tempera- topographic divides (Figure 5.31). The northern part o f
ture is several degrees higher than mean air temperature. the system is a recharge area, receiving water from the
In Figure 5.28 these basins are exemplified where the overlying alluvial reservoirs. Discharge through spring
basin alluvium is underlain by highly permeable carbon- areas occurs in Ash Meadows, the terminal point of the
ate rock that siphons off the flow so as to eliminate system. Boundaries to the system, at least to the south.
ground-water discharge at the surface. appear to be controlled by thick sequences of lower-
One of these regional flow systems occurs in eastern permeability clastic rock. Winograd and Thordarson
Nevada in the White River area (Figure 5.29). The major (1975) discuss the hydraulic details of flow in the South
discharge points occur as springs in White River Valley, Central Great Basin, with special reference to this carbon-
Pahranagat Valley, and Moapa Valley, the last identified ate flow system.
as the terminal point of the flow system. Boundaries to The examples just cited deal with discharge phenoni-
this flow system, at least to the south, appear to be ena in conspicuous topographic lows. Such areas reflect
controlled by thick clastic rocks whose permeability is the observable manifestations of the theories of ground-
considerably less than that o f the carbonates. Eakin water flow and are logical starting places for study o f
(1964) obtained closure of the water balance within a the system of flow contributing to large spring are;is i n
13-valley area, with 78% of the recharge estimated as volcanic and carbonate terrain.
occurring in the four northern valleys, and 62% of the
discharge estimated to be from spring areas in Pahranagat
Ground-Water-Lake Interactions
and MOapd Valley. A tentative analysis suggests a 15- to
20-year lag in spring discharge response to recharge from In performing the simulations leading to his conclusions
precipitation in the recharge areas (Eakin, 1964). This is on regional ground-water flow, T6th (1963) recognized
demonstrated on Figure 5.30, where precipitation about the presence of stagnation zones at the juncture of flow
1 0 0 miles north of the springs in Moapa Valley is plotted systems. A stagnation point is a point of minimum head
against spring discharge. The two curves match for the along a subsurface divide separating one flow system
sharp rise in precipitation during the period 1935- 1941 from another. This is shown diagrammatically in Figure
and the rise in the discharge graph during the period 5.32a, where the subsurface divide separates a local flow
1956- 1960. The aboveaverage precipitation for that system associated with a lake from a regional system.

Figure 5.28 Flow systems in the Great Basin (from Eakin and others, 1976).
94 Chapter 5 / Ground Water in tbe Basin Hydrologic Cycle

Long Valley
8

Jakes Valley
-
0 5
T l
N

15
e in miles
25

17

1
White River Valley Cave Valley

38 / l4
-
77
-
-37 Spring discharge White River Valley
40

Garden Valley
8

I Pahroc Valley
2
Dry Lake Valley
5

Delarnar Valley

* \$ j
60
-25
- Spring discharge
35

i
Coyote Springs and
Kane Springs Valley
2
37
-
-36
+1
Spring discharge Upper Moapa Valley
Net difference in estimates
<- - - ,,.
Ffgure 5.29 Estimated average annual recharge to and discharge from a regional
ground-water system in eastern Nevada (from Eakin, Water Resources Res., v. 2, p. 251 -
271, 1966. Copyright by Amer. Geophys. Union).

The minimum head at the stagnation point is greater than (1974). most of which were in the Midwest region, 24
the head in the lake. Thus, the lake of Figure 5 . 3 2 ~is were discharge lakes, 23 were through-flowlakes, 6 were
termed a discharge lake in that it receives ground-water recharge lakes, 4 were placed in some combined cate-
discharge from the aquifer. A recharge lake is one that gory, and 6 changed from one type to another in response
recharges the ground-waterflow, and a through-flow lake to seasonal changes (Anderson and Munter, 1981).
(Figure 5.326) is one that both receives and releases The search for stagnation points is aided by model
water t o the ground-water flow (Born and others, 1974, studies of the type initiated by Toth (1963) and Freeze
1979). For a 63-lake sample studied by Born and others and Witherspoon (1967). One example is provided by
 5.2 Surface Features of Ground-WaterFlou) 92

Ffgure 5.30 Cumulative departure from average precipita-

tion 1 0 0 miles north o f Mo;ip;i Valley versus cumulative de-
parture from average spring flow from Moapa Valley (from
Eakin. 1964).

fb!
Winter (1976, 1978). Figure 5.3361 gives results typical Figure 5.32 Ground-water flow paths in the vicinity of
of these studies. In Figure 5.33~7,;I stagnation zone exists lakes. Panel (a) is a discharge lake, where the asterisk shows
on the downgradient side of the single lake. The lake is the location of a stagnation point. Panel ( 0 ) is a flow-through
thus classified as ;I discharge lake and receives water lake (from Anderson and Munter, Witer Resources Rcs.. v.
17, p. 1139-1 150, 1981. Copyright by Amer. Geophys.
from the flow locdl system. If a high-conductivity layer
[inion).
is introduced at depth (Figure 5.336), the stagnation
point is eliminated and the pirating effect of the high-
conductivity layer converts the lake into a recharge lake.
Multiple-lake systems on a regional slope have also been investigated by Winter (1976) and can be recharge, dis-
charge, or through-flow depending on basin geometT
and geology. Steady-state models have commonly been
used to describe such behavior in the vicinity of lakes
(McBride and Pfiannkuch, 1975; Larson and others, 1976).
In addition to these model studies, other investigitors
(Meyboom, 1966a, 1967; Anderson and Munter, 1981)
1i:ive conducted field investigations in the vicinity of lakes
for the purpose of determining seasonal changes in the
flow pattern.

Ground-Water-Surface-Water Ititeractiotis
Flow studies of ground-water basins may be usefully
applied to gain an understanding of the interrelations
hetween the processes o f infiltration and recharge ;it
topographically high parts of the basin, and of ground-
water discharge via evapotranspiration and base flow ;it
topographically low points. Such studies permit ;I spatial
evaluation of the flow of ground water in the hydrologic
cycle. For example, at least sonie of the water derived
from precipitation that enters the ground in recharge
areas will be transmitted to distant discharge points. ;ind
b
\ ; o ---
5 lomi
so cause a relative moisture deficiency in soils overl!.ing
Ffgure 5.31 Regional flow systems in rocks underlying the recharge areas. Water that enters the ground in discharge
Nevada test site and adjoining area (from Winograd, 1962). ;ireas cannot overcome the upward potential gradient.
96 Cbapter 5 / Ground Water in the Basin Hydrologic Cycle

&!
Figure 5.33 Flow conditions in the vicinity of a lake demonstrating the
effect of a high-permeability layer at depth (from Winter, 1976).

and therefore becomes subject to evapotranspiration in the area immediately within the nearest topographic
the vicinity of its point of entry. The hinge line separating high, and possibly from more distant areas if the region
areas of upward and downward flow may thus serve as is characterized by both regional and local systems. If
a boundary common to areas of relative soil moisture base-flow calculations are used as indicators of average
surplus and deficiency (T6th, 1966). In nonirrigated ag- recharge, significant error may be introduced because
ricultural areas, this may be reflected by variations in base flow may represent only a small part of the total
crop yield. Further, the ramifications of human activities discharge occurring downgradient from the hinge line.
in discharge areas are immediately apparent. Some of Other interesting aspects resulting from such studies
these include waterlogging problems associated with sur- involving the basin hydrologic cycle deal with determin-
face-water irrigation of lowlands, or due to the destruc- ing the actual amount of ground water that effectively
tion of phreatophytes. participates in the hydrologic cycle. T6th (1963) calcu-
The spatial distribution of flow systems will also influ- lated that about 90% of the recharge never penetrates
ence the intensity of natural ground-water discharge. The deeper than 75 to 90 m. Similar conclusions have been
main stream of a basin may receive ground water from arrived at by tritium dating studies, which demonstrated
 5.3 Some Engineering and Geologic Implications of Topograpbic Drive Systems 97

a stratification of tritium and tritium-free waters (Carlston

and others, 1960).
As might be expected, the majority of surface-water-
ground-water studies have been conducted with model
calculations. Transient flows of water in the unsaturated Reservoir
zone have been examined by Rubin (1968), Hornberger I

and others (1 969), and Freeze (197 1). The models couple , I

saturated-unsaturated conditions. The ideas of infiltra-

--
tion capacity have likewise been studied in a transient ' Postreservoir water level I
mode with models that incorporate coupled unsaturated
and saturated behavior (Freeze, 1969). Models that cou- I
I I

ple surface and subsurface flow have been used to study I Ground-water divide Ground-water divide I
the runoff components of the hydrologic cycle (Smith Figure 5.34 Changes in water levels due to constnictioii
and Woolhiser, 1971) as well as aquifer-stream connec- of reservoirs (from Bryan, 1925).
tions (Hornberger and others, 1970).

documented by Van Everdingen (1972), who studied the

5.3 Some Engineering changes in the ground-water regime in the vicinity of
and Geologic Implications Lake Diefenbacker on the South Saskatchewan River.
of Topograpbic Dm've Systems Piezometric levels were obtained along a line crossing the
Saskatchewan River both before and after construction of
Large Reservoir Impoundments the reservoir, with the piezometers placed in the perme-
Dams, with their attendant surface-water reservoirs, are able sand units that are part of the Bearpaw shales. The
generally constructed in areas of ground-water discharge. prereservoir flow system was exactly what one would
If a reservoir is to occupy a deeply incised stream, the predict, with flow directed toward the Saskatchewan
water level in the reservoir can be higher than the re- River, where it eventually ends up as ground-water dis-
gional ground-water levels underlying the stream reach charge. After construction, there were some notable
and in the adjacent flood plains that are inundated. With changes. In the deeper unit, the transversal flow was
such an imposed head due to the filling of the reservoir, maintained toward the river valley, but at much lower
ground-water discharge to the stream and flood plain in gradients. In the intermediate unit, the transversal flow
the entire region that is inundated is no longer possible. was reversed away from the river valley. In the uppermost
This has little or no influence on the rate of recharge unit, all the transversal flow was unidirectionally away
taking place at adjacent topographic highs where the from the site.
ground-water levels are higher than the highest reservoir Frequently, an increase in spring discharge has been
level. A situation is thus created where more water is observed in the vicinity of new reservoirs. This has been
entering the system than can be discharged from the reported by Gupta and Suknija (1974). who measured
system under the prevailing hydrologic regime. The pre- increments in spring flow on the order of 0.145 litersh
vailing hydrologic regime, however, does not prevail in springs some 15 miles from a dam in India. The dam
long. What was likely a steady state in preconstruction is constructed on basalt. Gupta and Rastogi (1970) report
days now becomes a transient one, where the incoming fluctuations in four piezometers installed in limestone
water goes into ground-water storage, everywhere raising near a reservoir in Greece that were identical to those
the water table (Figure 5.34). Eventually, the total dis- associated with fluctuations in the lake level. Snow
charge from the system must once more equal the re- (1972) has also reported increases i n spring discharge in
charge. Normally this will take place after the water table the vicinity of the same reservoir.
has risen sufficiently so that new discharge regions are The impounding of reservoirs can also lead to some
created to take the place of those that can no longer stability problems. Underneath the reservoir areii itself.
function properly. Cady (I 941) describes such effects on the increased fluid pressures as shown in Figure 5.35 ;ire
ground-water levels in response to surface storage in the partially compensated by the increased weight of the
vicinity of Flathead Lake, Montana, making note of the water that must be added to the total stress. Downstream
potential impact on agricultural lands where the water from the reservoir, however, no such Compensation oc-
table is not far below the surface. curs. In Figure 5.356, two principal effects are noted.
In association with an impoundment and subsequent These include increased deformation of the valley bottoni
rising of the water table, a reversal in flow directions in and increased deformation and landslides in the valle!'
the subsurface is a distinct possibility. This has been walls.
93 Cbapter 5 / Ground Water in t b e Basin Hydrologic Cycle

tb)
Figure 5.35 Effect of dams on water levels showing (a)possible stabil-
ity problems created by reservoir blocking regional ground-water dis-
charge and (6)effects of high fluid pressures on valley bottom and wells
downstream from the reservoir (from Patton and Hendron, 1974).

Excavations: Infrows and Stability sions were criticized by Brown (1937) and other notable
experts in the profession at that time, including Thomp-
Tbe Sea-Level Canal
son, Meinzer, and Stringfield (1938).
By their very nature, sea-level canals are built in coastal In his investigation, Paige noted the general relation-
areas of low topographic relief, such as the Florida and
Panama peninsulas. The natural equilibrium position of
the freshwater-saltwater interface is already at a fairly
high elevation, although still below sea level. Excavation
below the water table permits the canal to act as a natural
drain, causing water levels to decline with an attendant
rise of the interface (Figure 5.36). In a preliminary investi-
\ \ ‘, \ ---.,
I /

gation of a sea-level canal across Florida, Paige (1936)

contended that a new ground-water rdble would evolve
and be steepest near the canal and approach the older
one asymptotically at a distance not in excess of 15miles.
As the canal would be cut approximately 10 m below
sea level, it was taken as nothing more than an extension
of the existing coast line of the Gulf of Mexico. The Ffgure 5.36 Sakwdter-frt?Shwdterrelationship due to exca-
argument here was that fresh water is currently present vation along coastal areas (from Hubbert, 1940). Reprinted
in coastal wells along the Gulf of Mexico, and similar by permission of the Journal of Geology, University of Chi-
conditions will prevail along the canal cut. These conclu- cago Press. Copyright @ 1940.
 5.3 Some Engineering and Geologic Implications of Topographic W i l e Sjwtems 93

Spring Stream Lake purportedly completely describes the projected re-

sponse.

' Discharge Water table Discharge ' Ground-Water Inflows into Excavations
Ibrahinl and Brutsaert (1 965) provide ;I simple an;ilytic;il
Figure 5.37 General relation of ground water to topogra-
phy (from Paige, 1936). Reprinted from Economic Geology,
solution for predicting inflows and a projected water-
1936, V. 31, p. 537-570. level decline in the vicinity of excavations such as the
sea-level canal just discussed. The case they considered is
shown in Figure 5.38 as a two-dimensional homogeneous
isotropic region. The trmsient response of the water
table is presented by the dimensionless graph o f Figure
ship of ground water to topography, focusing in particu- 5.38b,where
lar on the various water table highs associated with topo-
graphic highs and the fact that the outcrop areas of the (5.15)
formation to be cut (Ocala limestone) actually served as
discharge areas (Figure 5.37). He noted topographically where S,.is the specific yield and the product KH is the
high areas both south and north of the canal, between transmissivity, here obviously assumed not to be affected
which there was generally lower ground along the canal by the water-level decline, and T' is the time constant
route. Based on this information along with various bor- for the basin. The dimensionless discharge y, is given i n
ings and detailed investigations, Paige concluded tlvat the Figure 5 . 3 8 and
~ is expressed
canal route traversed the most important discharge area
of the Floridian plateau. In his reply to his critics, Paige
(5.16)
(1938) stated that this discharge area played an important
part in the selection of the canal route, as opposed to
where q' is the flow rate into the excavation per unit
many others examined by the Army Corps of Engineers.
length of excavation and has the units L?/T.
This was likely one of the first largescale engineering
To apply this method, information is required for K.
projects where the ground-water flow system with its
S,., H, and L. Verma and Brutsaert (1 970. 197 1) provide
attendant recharge and discharge areas actually figured
numerical models for similar problems.
in the decision process on site location.
One of the important issues was the potential effect
on Silver Springs, the largest limestone spring in the Example 5.2
United States and an important tourist attraction. Silver Assume that the basin parameters S,,. I.'. K, and H are
Springs is 3 miles north of the canal route and about 12 such that T' = lo', in days. For a time period t = 1 0 0
m above sea level. Paige argued that Silver Springs was days, f / T * = 1. From Figure 5.386.this means that the
an outlet of flow systems draining southward from the ground-water head at the face of the excavation (x = 0)
north so that the canal should not affect the source of is about 0.1 of the original saturated thickness H , and at
water supplied to the springs. In the final analysis, Paige the ground-water divide (x = L ) it is about 0.5 o f the
(1 936) recognized that there would be a substantial effect
original saturated thickness. The position of the water
on water levels near the canal, but these effects would table may be readily determined at other points. For t /
be of little consequence in terms of damage. As it turned T* = 1 and Figure 5.38c, the dimensionless discharge o f
out, the potential impact of the canal was probably not Eq. 5.16 is determined to be about 0.25. from which
as severe as the impact of ground-water development q' can be determined. Note that for longer times, t / T *
that took place in later years. Observations by Brown increases and the discharge rate (I' declines in response
and Parker (1945) along a completed portion of the canal to the flatter hydraulic gradients.
showed that saltwater encroachment had been facilitated
by hydraulic gradients created not by the canal but by
pumping. The Stability of Excavations
From this brief discussion, it is clear that most of the itr Ground-Water Discharge Areas
arguments for or against the canal were largely qualita- One of the most common problems in engineering con-
tive, based exclusively on the geologic controls o f the stnittion is the bottom heaving of trenches or other
hydrology. In view of the times it could be no other way. excavations constructed in areas of ground-water dis-
If such an event were to repeat itself today, the debate charge. Several case histories are available. but we shall
would not go on in a scientific journal, but in the courts, focus on only one to bring out the principles involved.
with a host of lawyers and their various consultants and This particular situation occurred during a project where
more than likely more than one mathematical model that cuts up to 20 m were required in glacial m;iterials. The
-1 Cbapter 5 / Ground Water in tbe Basin Hydrologic Cycle

/Ground surface

impermeable boundaries

-L-
(N)

1.o

0.75

0.1
$ 0.50

0.25

0
‘
- 0.25 0.50

(bl
-
X
I.
0.75 1.00
0.01
C

(C)
I
T

Figure 5.38 Prediction of ground-water inflows into an excavation (from Ibrahim and
Brutsaert, 1965). Reprinted by permission of the American Society of Civil Engineers.

major drainage in the region adjacent to the proposed materials where the overall heterogeneity prevented a
excavations occupied an intermoranic sag; such sags are good hydraulic communication between the permeable
noted for their heterogeneity, variation in grain size, and lenses. The response in the underlying limestone was
general lenticular character of the sands and gravels. The likely the result of pressure relief associated with pump-
sands and gravels that do occur in these geologic environ- ing the lake.
ments generally have a high permeability in that they Contract documents required the contractor to main-
have been well sorted by glacial melt water. tain the lake level about 10 m below its natural level with
A lake occupying a gravel pit was located adjacent to the idea that this would sufficiently reduce pressures in
the construction area. The lake was pumped prior to the glacial material so that stability problems would not
construction for a period of several months, lowering the be encountered. However, during construction, the bot-
lake level about 15 m. Five observation wells indicated tom of the excavations heaved, and the construction had
changes of 2.5 to 5 m over this period. Four of the wells to be stopped until dewatering wells were put in at an
were bottomed in the underlying limestone, which is additional cost.
highly fractured with excellent hydraulic communication One interesting point in all this is that the position of
throughout. The one well in the glacial material showed the water table was entirely overlooked in these investiga-
a head drop of about 2.5 m, and this well was closer to tions. All the observation wells in the area indicated water
the gravel pit than some of the deeper wells that pene- levels that were at or a few feet below land surface.
trated the limestone, which showed a greater response Indeed, one of the piezometers was flowing. This was
to pumping. It is likely, then, that pumping from the pit termed the “water table” in the preconstruction investi-
did not materially affect the water levels in the glacial gations. However, backhoe data from several different
 5.3 Some Engineering and Geologic Implications of Topographic Drive Systems 101
locations indicated that the sediments were dry from when u - P = 0, the overburden is in a state of floatation
land surface to in excess of 4.6 m in depth; in none (Terzaghi, 1950), and the resistance to shear is reduced
of these excavations was the water table encountered. to zero for cohesionless material, and to the cohesive
Hence, we reemphasize the fact that water levels in pi- strength for materials that possess cohesive properties.
ezometers that penetrate below the top of the zone of Terzaghi (1950) argued that seasonal variations in rain-
saturation d o not indicate the position of the water table, fall can give rise to seasonal variations in the fluid pres-
but indicate a head or potential that may be greater than sure, thereby reducing the shearing resistance indepen-
the water table, or less than the water table, depending dent of any effect on the angle of sliding friction. Thus,
upon where one is measuring in the flow system. In during periods of heavy or prolonged rainfall, slopes be-
this case, as in all discharge areas, the former is the come more susceptible to failure because of the attendant
actual situation. increases in water levels and decreases in effective stress.
A simple calculation could have removed any doubt This sometimes leads to a definite periodicity in slope
as to the need of dewatering or pressure relief prior to failures. Seepage from sources of water such as newly
construction. Assume that the water table is at 15 ft built reservoirs or unlined canals can have similar results.
below the surface, the dry unit weight of the material is Hence, if a slope starts to move, the means for stopping
on the order of 100 Ib/ft’, and the saturated unit weight the movement must be adapted to the processes that
is 110 Ib/ft3. The total vertical stress at the base of the started it (Terzaghi, 1950). As most slides are related to
glacial material (100 ft) is thus 10,850 Ib/ft’. If the water increases in fluid pressures, this means that both surface
level in a piezometer finished at the base of the glacial and subsurface drainage or dewatering is called for.
material registered a head at land surface, the neutral As with most original developments in science, once
stress is calculated to be about 6240 Ib/ft’. Hence, after the preliminary physics of a probleni are brought out
removal of about 43 ft of material the effective stress at in theory, others immediately rush in to embellish the
the base of the glacial material is reduced to zero. It original ideas. By embellish we mean to improve by add-
would appear that cuts approaching 60 ft in depth are ing detail. The detail added in this case is of a hydrologic
not likely to be stable without some pressure relief pro- and geologic nature. Deere and Patton (1967) put Eq.
vided by pumping wells. 5.17 in a hydrologic perspective. recognizing that slopes
are part of a regional flow system that is characterized
by elevation differences in water levels and in land sur-
Landslides and Slope Stability face. They recognize that small topographic deviations
In his ciassic treatment of landslide development, Ter- within a regional system produce local flow systems that
zaghi (1 950) noted the following causes for the relatively are often the most critical with regard to fluid pressures
rapid movement of slopes: (1) external changes, (2) and slope stability, especially in areas of ground-water
earthquake shocks, (3) iubrication by water, and (4) discharge. In addition, these authors demonstrated se-
ground-water level rises. External changes refer to man- quences of dipping beds susceptible to high pressures,
made or natural activities that disturb the stability of the susceptibility of weathered shales to landslides when
an existing slope, such as undercutting the toe, say, by the slope is in a discharge area, and the pore water
meandering rivers or excavations. Earthquake shocks in- pressures that can develop at the base of low-permeability
crease the shearing stresses relative to the shearing resis- materials overlying more permeable rocks in a ground-
tance as accelerations associated with earthquakes repre- water discharge area. These same ideas are carried over
sent horizontal forces. The lubrication by water was by Patton and Hendron (1974).
dismissed by Terzaghi because water actually acts as an Fractured rock masses are treated more or less along
antilubricant in that there is more friction between “wet” the same lines. There are, however, some differences.
mineral grains than between dry ones. The main empha- again, purely geological. In that the porosity of fractured
sis of Terzaghi’s paper is the effect of water-level changes rock may be two or three orders of magnitude smaller
in the interpretation of the Mohr-Coulomb equation, than that of sedimentary rock, a typical rainfall event
stated here as should result in a relatively larger water-level response.
Further, because of the prevailing shear stresses in a
T = T, + (u- P) tan Q, (5.17) body of rock forming a slope at some inclination to the
horizontal, the joints may be wider and more numerous
where T is the shearing resistance per unit area, T,,is the than those located in otherwise similar. flat-lying rock.
cohesion per unit area, u is the total normal stress at a Hence the quantity of water that can enter the rock mass
point due to the weight of the solids and fluids, P is the is greater in the vicinity of the slope.
fluid pressure at the point, and d, is the angle of sliding Numerous references are available on ground-wmr
friction. The greater the fluid pressure P,the greater is and slope stability. These include Cedergren (1967), Mor-
the portion of the overburden that is carried by the water; genstern (1970), Kiersch (1973), and Royster (1979).
-
102 Chup#er5 / Ground Water in the Basis Hydrolugk Cycle

Problems
1. Briefly discuss the meaning of the boundary condi- b. Calculate the flow rate q' into the canal for 10 years
tions given on Figure 5.5. What boundary conditions are and 100 years.
included in Figure 5.8 that are not in Figure 5.5? Why? 4. Perform the calculations for the case history given
2. Construct flow nets in the x-z plane for Figure 5.4 in the section dealing with stability of excavations.
where the left-hand boundary is a flow (as opposed to a 5. A saltwater interfxe is stable at a position of about
no-flow) boundary, and for Figure 5.27 for the case where 40 ft beneath the base of a well in a formation with a
K, = Kz. hydraulic conductivity of 1 X lo-' cm/s. Calculate the
3. Set up a problem similar to the one discussed for maximum pumping rate so as not to cause upconing.
the sea-level canal. Assume a flat water table 12 m above 6. The flow of fresh water toward the sea is 2 X lo-'
sea level, a canal that extends 9 m below sea level, and ft'/s per foot of coastline in an aquifer that is 100 ft thick
a ground-water divide (springs) approximately 3 miles with a hydraulic conductivity of 3.3 X lo-$ft/s. Calculate
from the canal. Assume a hydraulic conductivity of 8.35 the inland distance from the shoreline to the toe of the
cm/day and a specific yield of 0.01. saltwater wedge. What is the new position from the
a. Calculate the head h at the canal and at the divide shoreline if the seaward flow is reduced by one-half?
for times of 10, 20, 40, and 100 years.
 Chapter 6

Hydraulic Testing:
Models, Methods, and Applications

6.1 Prototype Geologic Models in Hydraulic Testing

6.2 Conventional Hydraulic Test Procedures and Analysis
6.3 Single-Borebole Tests
6.4 Partial Penetration, Superposition, and Bounded Aquvers
6.5 Hydraulic Testing in Fractured or Low-Permeability Rocks
6.6 Some Applications to Hydraulic Problems
6.7 Computer-Based Calculations

Hydraulic testing as used in this book is a description of environment. Three such environments are shown in
the field tests and testing procedures required to obtain Figure 6.1. Figure 6.Iu shows a typical confined aquifer
certain hydraulic properties, pressure measurements. or of large areal extent, assumed to be isotropic and homo-
indirect determinations of ground-water velocity in rocks geneous, overlain by a confining layer over which there is
of various kinds. Such testing procedures were born out a water table. There are three variations of this prototype
of the practical necessity of evaluating ground water as model. First, if the confining layer is impermeable and
a water supply. Thus the early beginnings in this field contains ;in incompressible fluid within an incompress-
dealt exclusively with pumping tests designed for perme- ible matrix (S: = O), all the water removed from the aqui-
able formationsthat were important from the perspective fer will come from storage within the aquifer, ;ind the
of water supply. In more recent times, due largely to water-level change (drawdown) versus time response ;it
concerns of hazardous waste migration and nuclear waste some observation point will be exponenti;il, that is. will
burial, interest has shifted to testing procedures suited plot ;IS a straight line on semilogarithmic p:iper (cune
to low-permeability rocks or rocks with a fracture-type A , Figure 6.2). For reasons th;it will be cle;ir later. this is
permeability. For simplicity we classify the v;irious tests referred to as the nonleaky response. If the conlining
as conventional or speciiilized. By conventional we nie;in layer has some tinite permeability but still contains ;in
those tests that virtually rill hydrologists conduct in a incompressible fluid within an incompressible matris. i t
routine fashion in relatively simple geologic environ- may be possible to invoke the transfer of water ;icross rhe
ments. Specialized testing procedures are often required confining layer. The time -drawdown obsenxtion will
in low-permeability rocks and especially in fractured reflect this additional water source with ; i n upw:ird in-
rocks. flection at some point in time (curve 8. Figure 6.2).l'liis
is referred to as the leaky response. Two things arc' worth
6.1 Prototype Geologic Models noting here. First, because of the time delay for w t e r
to enter the pumped aquifer. an ;iquifer may appc:ir non-
in Hydraulic Testing
leaky over several hours or d;iys of pumping ;incl will hr
Every hydrologic model employed in hydraulic testing so classified if the test is terminated before the Ie;ik;igr
is patterned after some specific or prototype geologic has been detected. This, however, will not ;iffect the
-
104 Chapter 6 / Hydraulic Testing:Models, Metbods, and Applications

c
U
I
E
D
storage in \\

I No leakage (A)-\

I
\
Log time
Figure 6 2 Comparison of time-drawdown response for
geologic prototypes of Figure 6 . 1 (from
~ Hantush,J. Gee
phys. Res., v. 65, p. 3713-3725, 1960. Copyright by h e r .
Geophys. Union).

horizontal conductivities in a horizontal plane (x and y )

I
are not equal. The response of such a system is shown

c3 Plan view
in the plan view given in Figure 6.3 where the transmissiv-
ity in the x direction is larger than the transmissivity in
they direction. Variations of this model include situations
where the principal directions of the hydraulic conduc-
tivity tensor are neither vertical nor horizontal. T w o and
(C) three-dimensional anisotropic models may be useful in
Figure 6.2 Prototype geologic models. the analysis of fractured rock.
This brief discussion does not exhaust the prototype
geologic model guiding the hydraulic testing of real rock.
analysis in that the leaky and nonleaky curves remain Others include a double-porosity concept where water
coincident or nearly so until the inflection point is real- can be removed from both the fractures and matrix of a
ized (Figure6.2). Second,leakage across a confining layer fractured porous medium and discrete flow in individ-
may not be the only cause for the upward inflection. If ual fractures.
pumping is near a stream or a lake and surface water is Operational hydraulic equations have been derived for
brought into the aquifer due to the established hydraulic the ideal prototype models discussed earlier. The solu-
gradients, a similar inflection will occur. tions have been obtained from some form of the diffusion
The last version of Figure 6 . 1 is~ for a permeable con-
fining layer with a finite specific storage. In this case, the
confining layer itself can contribute water to the aquifer,
Principal directions
resulting once more in an upward inflection (Figure 6 . 2 ~ ) . of anisotropy
This is referred to as leakage with storage in the confining
layer. It is noted now that the time delay is considerably
shortened (Figure 6.2). For long times, the leakage with Well,
storage case is coincident with the leaky case, but
this coincidence evolves through a different time- ---
drawdown pathway.
Yet another geologic prototype is the aquifer that is
considered homogeneous,but anisotropic. In Figure 6. lc
the aquifer is considered to be horizontal and one of the
principal directions of the hydraulic conductivity tensor Figure 63 Plan view of drawdown response in an aniso-
is vertical, that is, parallel to the well. In this case, the tropic aquifer.
 6.2 Conventfonal Hydraulic Test Procedures and Analysfs

equation, where time is a parameter. The essence of function of the pumping rate, the properties of the aqui-
hydraulic testing calls for the matching of real response fer, and time. A solution to the heat-flow problem just
data with the theoretical response expected from the stated is given by Carslaw and Jaeger (1959). The hydro-
prototype condition. In most of the conventional nieth- logic analog was given by Theis (1935) and is of the form
ods, this matching is performed with a curve-fitting pro-
cedure or by some simple graphical technique. With
some of the specialized testing procedures, the matching
is frequently conducted with the aid of a computer. where h,, is the original head at any distance r from a
fully penetrating well at time t equals zero, h is the he;id
at some later time t, s is the difference between b,,and
612 Conventional Hydraulic Test b and is called the drawdown, Q is a steady pumping
Procedures and Analysis rate, T i s the transmissivity, S is the storativity, and the
exponential integral is a well-known tabulated function.
The term conventional is used here in reference to hy-
the value of which is given by the infinite series
draulic testing procedures that are most commonly used
in formations of moderately high permeability with the
stated purpose of determining the transmissive and stor-
age properties of aquifers. In this section we will restrict
our discussions to tests that require one pumping well
and one or more observation wells or piezometers in
which it is possible to measure the response to pumping.
From an historic perspective, these procedures origi- where
nated with the work of C. V. Theis in 1935.
u = - r2S (6.3)
4 Tt
Tbe Tbeis N o n e q u i l i b r i u m Pumping Equation 6.1 is a solution to the polar coordinate form
Test M e t h o d of the diffusion equation, or
The mathematical expression for removal of heat at a
constant rate from a homogeneous, infinite slab has pro- (6.4)
vided a useful analogy for study of ground-water flow to
a pumping well. It is assumed, initially, that the slab is for the initial and boundary conditions
at some uniform temperature. An infinitesimal rod of h(r, 0) = h,,
lower temperature parallel to the z axis is then allowed
to draw off the heat (Figure 6.461).In mathematical termi- h ( m , t) = b,,
nology, the rod represents a continuous line sink. The
temperature change at some distance r from the rod is
a function of the rate at which heat is withdrawn, the
properties of the slab, and time. By analogy, the infinite, The first two of these are read: “The head at some radius
homogeneous slab is replaced by an extensive, honioge- r from the well at time zero equals the initial head I!!,,,
neous aquifer, and the rod by a well of infinitesimal and the head at an infinite distance at any time t equals
diameter (Figure 6.46).Similarly, the rate of pumping is the initial head h,,.” The second condition stipulates that
analogous to the rate of heat withdrawal, and water-level water Withdrawal may affect distant parts of the aquifer
change at any distance r from the pumping well is a but the exterior boundary is never encountered (as ex-
pected for an aquifer of infinite extent). The third condi-
tion provides for a constant withdrawal rate at r,,., which
is the radius of the well, which, in turn, is taken as infini-

n
ROd
tesimal.
Equation 6.1 is referred to as the nonequilibrium equa-
tion. It is sometimes referred to as the nonleaky well
flow equation in that all the water pumped is removed
from a single aquifer with no contributions from other
beds either above or below the pumped aquifer.
A verbal interpretation of the parameters incorporated
(0) (b) in the nonequilibrium equation can provide insight into
Figure 6.4 Schematic diagram of (a) an infinite slab and the shape and growth of a cone of depression in an ideal
(6)the well flow analogy. aquifer. It is first noted that the exponential integral of
-
106 Chapter 6 / Hydraulic Testing:Models, Methods, and Applications

Table 6 2 Values of W(u) for Values of u

U 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0

x 1 0.219 0.049 0.013 0.0038 0.001 1 0.00036 0.00012 0.000038 o.ooo012

x lo-' 1.82 1.22 0.91 0.70 0.56 0.45 0.37 0.31 0.26
x 4.04 3.35 2.96 2.68 2.47 2.30 2.15 2.03 1.92
x lo-' 6.33 5.64 5.23 4.95 4.73 4.54 4.39 4.26 4.14
x 10-4 8.63 7.94 7.53 7.25 7.02 6.84 6.69 6.55 6.44
x 10-5 10.94 10.24 9.84 9.55 9.33 9.14 8.99 8.86 8.74
x lo-" 13.24 12.55 12.14 11.85 1 1.63 11.45 11.29 11.16 1 1.04
x 10-7 15.54 14.85 14.44 14.15 13.93 13.75 13.60 13.46 13.34
x lo-" 17.84 17.15 16.74 16.46 16.23 16.05 15.90 15.76 15.65
x 10-9 20.15 19.45 19.05 18.76 18.54 18.35 18.20 18.07 17.95
x 10-1" 22.45 2 1.76 21.35 21.06 20.84 20.Ci 20.50 20.37 20.25
x 10-11 24.75 24.06 23.65 23.36 23.14 22.96 22.81 22.67 22.55
x 10-12 27.05 26.36 25.96 25.67 25.44 25.26 25.11 24.97 24.86
x 10-1' 29.36 28.66 28.26 27.97 27.75 27.56 27.4 1 27.28 27.16
x 10-14 31.66 30.97 30.56 30.27 30.05 29.87 29.71 29.58 29.46
x 10-15 33.96 33.27 32.86 32.58 32.35 32.17 32.02 3 1.88 3 1.76
From Wenzel (1942)

Eq. 6.2 is a function only of the lower limit of integrdtion As a final point, the dimensionless variable u is of
so that Eq. 6.1 can be written the form of an inverse Fourier number introduced in
Chapter 4
s=-Q W(u) (6.5)
47rT
where u is defined in Eq. 6.3. The term W(u) is termed
the well function of u, which indicates its dependence where Sr2/4T has the units of time and may be referred
on values of u expressed in Eq. 6.3. It follows that the to as a time constant T'. Thus, if the time constant T* is
value of the exponential integral may be ascertained and small compared to time t, u becomes small, W(u) be-
tabulated for each of several values of u. Such a tabulation comes large, and drdwdowns become large (Figure 6.2).
is shown in Table 6.1 for values of u ranging from lo-'' Small values for the time constant are associated with
to 9.0. If u equals 4.0 X lo-"', then W(u) equals 21.06, small values of storativity, say, for confined conditions,
which is the value of the series of Eq. 6.2 for this particu- small distances to the point of observation, or large values
lar value of u. of transmissivity. If the time constant Sr2/4T becomes
Of the variables comprising u , the storativity and trans- large with respect to a proposed time of observation t,
missivity may be considered constant for a given set of the expected drawdown at the point of observation r
conditions, and distance and time considered variables. becomes imperceptibly small. Large values for the time
Values of W(u) may be plotted against values of I/u in
such a way that for any fixed distance r, the plotted curve
reveals the exact shape of the drawdown curve as a
function of time t (Figure 6.5). Curves showing the rela-
tion between W(u) and u are termed "type" curves.
In that the value of u depends on time, distance, trans-
missivity, and storativity, u determines the radius of a
cone of depression (Theis, 1940).The radius not only
increases with increasing time but, for a given time, is
larger for decreasing values of storativity and increasing I I
I
values of transmissivity.By examining the complete state-
ment for drawdown (Eq. 6.5), drawdown at any point
for a given time is proportional to discharge and inversely
proportional to transmissivity. The lateral extent of a 0.1 1 10 103 104
cone of depression at any given time and its rate of 1lu
growth are independent of the pumping rate. Figure 65 Values of W(u) plotted against values of l/u.
 6 2 ConventionalHydraulic Test Procedures and Anafysfs 107

constant are frequently associated with large values for 6.5 can be solved for transmissivity by using the match-
the storativity, as expected for unconfined conditions. point coordinates s and W(u)and the discharge Q.Equa-
tion 6.3 can be solved for stontivity by using the match-
Tbe Curve-MatchingProcedure point coordinates l/u and t, the distance r from the
The Theis (or nonequilibrium) equation is used exten- pumped well to the observation well, and the value of
sively to determine the hydraulic properties of aquifers. transmissivity as determined earlier.
For these purposes, it will be recalled that for a fixed
distance r, drawdown versus time is of the form of the Example 61
curve in Figure 6.5.Ifdrdwdown s can be measured at one
point r for several values of time t, and if the discharge is The data of Figure 6.6 were collected at a distance of
steady and is known, the coefficients of tnnsmissivity 150 m from a well pumped at a rate of 5.43 x 10’ ni’/
and storativity can be determined by a graphical method day. From Figure 6.6
of superposition. Field data composed of drawdown ver- W(u) = 1 s = 0.2 m
sus time collected at a nonpumping observation well at
a known distance r from a pumped well are plotted on 1
- = 10’ f = 000 min (about 0.7 day)
logarithmic paper of the same scale as the type curve U
(Figure 6 . 6 ~ )The
. field curve is superimposed on the Tnnsmissivity is determined from Eq. 6.5,
type curve, with the coordinate axes of the two curves
kept parallel while matching field data to the type curve Q W ( U )= 5.43 X lO’ni’/day x 1
T=-
(Figure 6.66). Any point on the overlapping sheets is 47rs 4 X 3.14 X 0.2 m
selected arbitrarily (the point need not be on the matched = 2.2 X 10’ m’/day
curves). The selected point is defined by four coordinate
values: W(u)and s,and l/u and t (Figure 6.66).Equation = 2.5 X m2/s
Storativity is determined from Eq. 6.3,

Time (minutes) Assumptions and Interpretations

(a) The operational hydraulic equations were derived for
an ideal model aquifer, whereas time-drdwdown data
represent real aquifer response. The procedure requires
a matching of real response data with the response ex-
pected under ideal conditions. The most conspiciioiis
assumption is that the aquifer is infinite in are,i1 extent.
This means the following:
-a 1. The cone of depression will never intersect a bound-
ff ary to the system.
W(rr) = 1
. i h = lo3 2. An infinite amount of water is stored in the aquifer.
s = 0.2
t = 990 Condition 1 is likely to be met for very short times, or
for longer times in extensive aquifers of uniform material.
In that the cone of depression is expanding with time.
any geologic boundary within a short distance from the
1 ___- L _ _ _ _ L _ _ _ _I - -
lllf
pumping well will immediately disrupt the postulated
1 10 100 1000
Time ( r ) behavior. According to assumption 2, water levels will
(b) eventually return to their prepumping level once the
Figure 6 6 Graphs of (a) field dita and (b) superposition wells are shut down.
of the field data on the type curve to obtain the formation In addition, it is assumed that the well is of infinitesimal
parameters. diameter and fully penetrates the aquifer. This means
-
108 Chapter 6 / Hydraulk Testfng:Models, Methods, and Applfcatfons

that storage in the well can be ignored and that the well 0
receives water from the entire thickness of the aquifer. 0.1
Methods of treatment for deviationsfrom full penetration
have been presented in a number of papers (Muskat, -E 0.2
1937; Hantush and Jacob, 1955; Hantush, 1957). A solu- c
c 0.3
tion for drawdown in largediameter wells that takes into B
'0
consideration the storage within the well has been pre- 0.4
b
sented by Papadopulos and Cooper (1967). Yet another 0.5
assumption is that water is released instantaneously with
decline in head in the aquifer. This assumption may be 0.6
met when water is released by compression of the aquifer 0.7' I I
and expansion of the water, but fails to describe ade- 1 10 100 1000
Time (minutes)
quately the gravity flow system of the unconfined case.
Ffgure 6.7 Semilogarithmic plot of drawdown versus time
Homogeneity and isotropicity are major assumptions
in an observation well.
in the development of the descriptive differential equa-
tion of which the well flow equation is a solution. This
means that the transmissivityand storativity are assumed of time will yield a straight line (Figure 6.7).At time tl ,
to be constants, both in space and in time. Hence, the the drawdown s, is expressed from Eq. 6.9
geologic medium is assumed to be the simplest type con-
ceivable.
2.312 2.25Ttl
s, = -log-- (6.10)
47rT r2S

Modifications of the At time t2,the drawdown s2 will be

Nonequilibrium Equation 2.3Q 2.25TtZ
s2 =---log- (6.11 )
At least two important modifications of the nonequilib- 4nT r2S
rium equation can be traced to a very simple observation It follows that
made by Cooper and Jacob (1946),namely, that the sum
of the series of Eq. 6.2 beyond In u becomes negligible (6.12)
when u becomes small. This occurs for large values of
time t or small value of distance r. By neglecting the If tl and t2 are selected one log cycle apart, log ( t r / t l )=
series beyond In u, Eq. 6.1 can be expressed 1 . Equation 6.12 then becomes

.=-( Q
4nT
-0.5772 - I n e )
4 Tt
(6.7) 2.3Q
As = -
4nT
(6.13)

or where As is the drawdown per log cycle. The value of

T can be obtained from Eq. 6.13.
The storativity can be obtained by selecting any draw-
4nT down on Figure 6.7 for a given time and substituting this
As 0.5772 equals In 1.78 and In x equals 2.3 log x , Eq. value in Eq. 6.9. For convenience, s equals zero is se-
6.8 becomes lected, so that Eq. 6.9 becomes

s = -2.3Q log -
2.25Tt s=(-J=- 2.3Q 2.25Tt
(6.14)
(6.9) 4nT l o g r ' S
4nT r2S
which is presented as the modified nonequilibrium equa- This requires that 2.25Tt,,/r2S= 1 so that
tion. This equation may be applied to
S = -2.25Tt,, (6.15)
1. Time-drawdown observations made in a single ob- r2
servation well, as in the original nonequilibrium
where to is the time intercept where the drawdown line
method (modified nonequilibrium method).
intercepts the zero drawdown axis.
2. Drawdown observations made in different wells at
the same time (distancedrawdown method).
Example 6.2
Time-Drawdown Method The data of Figure 6.7 were collected at an observation
If drawdown observations are made in a single well for well 305 m from a well pumping at a rate of 5.43 X lo'
various times, a plot of drawdown versus the logarithm m'/day. Drawdown per log cycle is determined from the
 62 Conventional Hydraulic Test Procedures a n d Analysis 1 3

graph to be 0.24 m. Solving for T and S, 0

T=--2.3Q - 2.3 X 5.43 X 10' m3/day

4nAs 4 X 3.14 X 0.24 m
-E
I 1.0

= 4.1 X 10'm2/day g
'13
3 2.0
= 4.8 X m2/s e
0

s=-2 25Tt, -
- 2.25 X 4.1 X lo3m2/day X 5 min
3.0
rz 93025 mz X 1440 min/day
I I I I
= 3.4 x 10-4 1 10 100
Distance from pumped well (m)
1000

Ffgure 6.8 Semilogarithmic plot of drawdown versus dis-

Distance-Drawdown Metbod tance.
The time-drawdown analysis just described requires one
pumping well and one observation well, where a plot of
drawdown versus time on semilogarithmic paper yields where r,, is the intersection of the straight-lineslope with
a straight line. This straight line demonstrates the expo- the zero drawdown axis.
nential decline of water levels at a point in a single aqui-
fer. It is also possible to obtain the hydraulic properties Example 63
by examining drawdown at two or more points at one
Consider the following data for Figure 6.8. Observation
instant of time. At time t, the drawdown s, at a distance
wells are placed 10 m and 100 m, respectively, from a
r1is, from Eq. 6.9,
pumping well. The well is pumped at a rate of 5.43 X
2.3Q 2.25Tt l@' m'/day. At the end of 200 min (1.39 X lo-' days),
SI =-log- (6.16) the drawdown per log cycle is noted to be 1 . 1 rn. From
4nT r:S
Eqs. 6.20 and 6.21
and the drawdown s2 at a distance r2 is
T=--2.30 - 2.3 X 5.43 X 10'mm'/day
2,3Q 2.25Tt 2n As 2 X 3.14 X 1.1 m
s2 = -log - (6.17)
4nT r$
= 1.8 X 10'm2/day
It follows that = 2.1 X 10-zmz/s
2.3Q r:
s, - s, = - (6.18) S=- 2.25Tt
4nT log7 ri
Recognizing that log(r:/r:) = log( l/rf) - log( 1/ r i ) and -
- 2.25 X 1.8 X 103m'/day X 1.39 X lo-' days
that log( l/r? = 2 log( 1/ r ) , Eq. 6.18 becomes 64 X 10' d

- 2 . 3 0 log ~r = 8.8 X lo-'

sI - 2-- (6.19)
2nT rI
The graphic procedure calls for the plotting of draw- Steady-State Behavior as a Terminal Case
downs observed at the end of a particular pumping pe-
of the Transient Response
riod in two or more observation wells at different dis-
tances from the pumped well against the logarithms of The straight line in Figure 6.8 represents two points on
the respective distances (Fig. 6.8). By considering draw- a profile of a cone of depression for a given instant of
down per log cycle, Eq. 6.19 becomes time. At some time later yet another curve can be con-
structed, and it will parallel the curve shown in the figure.
(6.20) with the same drawdown per log cycle. For very long
periods of time, the drawdown in the observation wells
By extrapolating the distance-drawdown curve to its will have approached their steady-state value and no
intersection with the zero drawdown axis, the storativity longer decline with time. A plot of this drawdown on
can be determined in the same manner as described for Figure 6.8 will again form a straight line, parallel to the
the modified nonequilibrium method, other lines, with the same drawdown per log cycle. This
problem involving steady-statedrawdowns at two obser-
S=- 2.25Tt (6.21)
vation points was solved by Thiem (1906) completely
r: independent of any transient flow considerations. The
110 Cbapter 6 / Hydraulfc Testfng:Models, Metbods, and Applfcatfons

solution for Thiem's confined condition is identical to

that given in Eqs. 6.19 and 6.20. However, as time is not
a factor in Thiem's steady-state derivation, the stocitivity
cannot be determined for the steady drawdown case.
If the flow takes place under unconfined conditions,
Thiem's solution changes somewhat and may be ex- I 1
pressed as 41 I0 100 1000
Time (minutes)
Ffgure 6.9 Time-drawdown response of a leiky aquifer.
(6.22)

As in the application of the confined case, any pair of Theis solution provides some estimate of this required
distances may be employed. This equation, along with time period. Hence, for all practical purposes, the time-
its steady-state counterpart for confined conditions, is drawdown behavior previous to the inflection represents
frequently referred to as the equilibrium, or Thiem a withdrdwal of water from storage from the aquifer. no
equation. part of which was contributed from other sources. This
part of the curve, along with its straight-line extension,
may be analyzed with methods previously discussed for
The Hantusb-Jacob Leaky Aquifer Method time-drawdown behavior. That is to say, over this period
of the pumping test, the assumption that all the water
One major assumption of the Theis nonequilibrium solu-
pumped comes from storage in the aquifer is a reasonably
tion is that all the water pumped is removed from storage
valid one.
within the aquifer. There are several ways in which this
The second method requires a curve-matching proce-
condition may be compromised in field conditions: direct
dure where the drawdown for all time is given as (Han-
recharge to the aquifer from streams, direct recharge
tush and Jacob, 1955; Hantush, 1956)
across bounding low-permeability materials, and release
of water from bounding low-permeability rocks that pos-
sess their own hydraulic characteristics (i.e., specific stor-
age and hydraulic conductivity). The problem of Iedkage
s = -ZwT (u'3
- (6.23)

has been extensiveIy investigdted by Hantush and jacob where W(LI,r / R ) is the tabulated well function for leaky
(1955) and Hantush (1956, 1960, 1964). in this section aquifers. It is thus recognized that short-term drawdowns
we will be concerned only with their earliest contribu- previous to leakage are described by the Theis nonequi-
tion, which provides methods most commonly used in librium equation and the well function W(u) whereas
field problems. long-term drawdowns are described by the well fiinction
Figure 6.la will help develop the pertinent ideas. If W(u, r / B ) . Values of W(u, r / B ) are given in Table 6.2,
the artesian sand in the aquifer is pumped, and the and a type curve is presented as Figure 6.10. Note that
bounding low-permeability rocks are in fact "impernie- the family of curves converges on the nonleaky curve
able," the anticipated response at an observation well for small values of 1 / u (which correspond to small values
can be described completely by the Theis equation. How- of time) and small values o f r / B . For small values of time,
ever, if water is entering the aquifer, say, by either leakage W(u, r / B ) = W(u).
across the upper low-permeability unit or by release of The graphic procedure for solving for the formation
water from storage in the low-permeability unit, we constants is similar to that described previously. Early
would anticipate some deviation from the Theis re- drawdown data from the field curve are matched with
sponse. If we assume that the low-permeability unit has the nonleaky part of the curve, but they soon deviate
a negligible specific storage (that is, is incompressible) and follow one of the leaky r / B curves (Figure 6.11).
but a finite permeability, water can be transmitted across The match point yields values of W(u, r / B ) , l/u, t, and
it from the overlying water table. Figure 6.9 shows a s. In addition, the r / B curve followed by the field data
typical response at an observation well in such an aquifer. is noted. Trdnsmissivity and storativity are readily deter-
Note that the time-drawdown curve departs from the mined from
expected nonequilibrium behavior, resulting in less
drawdown per log cycle of time. P w 2(4);,
T=-
There are two ways to analyze drawdown curves of (6.24)
47is
the type shown on Figure 6.9, depending on how much
information is desired. The simplest method recognizes and
that it takes some length of time for water to permeate
through the upper low-permeability material, and the s=- 4uTt (6.25)
upward inflection from the straight-line behavior o f the r?
 6 2 Conventional Hydraulic Test Procedures and Analysis 111

U7
Table 6 2 Values of W(u r/B)

0.01 0.015 0.03 0.05 0.075 0.10 0.15 0.2 0.3 0.4

O.Ooo(H)l
o.oO(x)05 9.44 13
0.00001 9.4176 8.63 13
O.O(Kx)5 8.8827 8.4533 7.2450
O.O()Ol 8.3983 8.1414 7.2122 6.2282 5.4228
0.0005 6.9750 6.9 152 6.6219 6.0821 5.4062 4.8530
0.00 1 6.3069 6.2765 6.1202 5.7965 5.3078 4.8292 4.0595 3.5054
0.005 4.7212 4.7152 4.6829 4.6084 4.4713 4.2960 3.8821 3.4567 2.7428 2.2290
0.01 4.0356 4.0326 4.0167 3.9795 3.9091 3.8150 3.5725 3.2875 2.7104 2.2253
0.05 2.4675 2.4670 2.4642 2.4576 2.4448 2.4271 2.3776 2.3110 1.9283 1.7075
0.1 1.8227 1 .8225 1.8213 1.8184 1.8128 1 .8050 1.7829 1.7527 1.6704 1.5644
0.5 0.5598 0.5597 0.5596 0.5594 0.5588 0.5581 0.5561 0.5532 0.5453 0.5344
1.o 0.2194 0.2194 0.2193 0.2193 0.2191 0.2190 0.2186 0.2179 0.2161 0.2135
5.0 0.0011 0.0011 0.0011 0.0011 0.0011 0.001 1 0.0011 0.0011 0.0011 0.0011

7
U

0.0oo001
0.5 0.6 0.7 0.8 0.9 1.0 1.5 2.0 2.5

0.000005
O.o(MU)I
0.0()005
0.OMI
0.OoO5
O.O()l
0.005
0.01 1.8486 1.5550 1.3210 0.1307
0.05 1,4927 1.2955 1.2955 1.1210 0.9700 0.8409
0.1 1.4422 1.3115 1.1791 1.0505 0.9297 0.8190 0.4271 0.2278
0.5 0.5206 0.5044 0.4860 0.4658 0.4440 0.4210 0.3007 0.1944 0.1174
1.0 0.2103 0.2065 0.2020 0.1970 0.1914 0.1855 0.1509 0.1139 0.0803
5.0 0.001 1 0.001 1 0.001 1 0.001 1 0.0011 0.001 I 0.0010 0.0010 0.0009
Tmns. Anirr. Grophys. Ilnion. 37, p. 702-714. Copyright by Amer. Geophys. linion.
After Hantush (1956).

:i
0.001

0.1

0.01 I
10-1
t
Figure 610 Type ci~rvefor leahy aquifers.
 Chapter 6 / Hydraulic Testing:Models, Metbods, and Applications

Water Table Aquifers

The tests just described are designed to test the confined

Y I
r
Match I flow condition. This means that during the testing, the
Field data I aquifer undergoes no dewatering, provided the heads are
I
W(tr# =1 I not lowered below the uppermost confining layer. For
I this case water is released from storage due to elastic
= 1000 I
I compression of the matrix and expansion of the water
s = 0.2 I itself. For a water table, or unconfined aquifer, the low-
t = 990 I
I ering of water levels will actually cause some dewatering
I 10 , 100 1000 - 1 of the aquifer, and the value of the storativity should
-1 I
I,----L-”--I----1-----1 increase by a few orders of magnitude over its confined
1 10 100 1000 counterpart. In Chapter 4, the storativity for such cases
Time (minutes)
was referred to as the specific yield, S,,.
Ffgure 6.22 Graphic procedure for solving for formation Two major concerns are involved in the analysis of
constants for leaky aquifers. water table aquifers. First, dewatering of the aquifer can
actually lead to a decrease in transmissivity during pump-
In addition, the coefficient of vertical hydraulic conduc- ing in that T = Ktn, where m is the saturated thickness
tivity can be determined from the relationship of the aquifer. The second problem relates to the assump-
tion of the Theis equation, where “water is released from
(6.26) storage instantaneously with decline in head.” If this
condition is not satisfied, the time-drawdown data will
where 1/B is determined from the match, m is the thick- deviate from the expected response for the confined
ness of the bounding low-permeabilityformation, and k f condition. The mathematical description of this response
is the vertical hydraulic conductivity. has been the subject of numerous papers starting with
Boulton (1954,1955,1963), extended by Prickett (1965)
Example 64 based on the Boulton analysis, and advanced significantly
by Dagan (1967), Streltsova (1972) and Steltsova and
The data in Figure 6.1 1 are similar to those in Figure 6.6 Rushton (1973), and Neuman (1972, 1975).
except for the leakage. Assuming the data were obtained In view of the attention given to this subject, it is
from an observation well located 150 m from a well informative now to obtain a semiquantitativeperspective
pumped at a rate of 5.43 X 10’ m3/day, the coefficient of a typical time-drawdown response in an unconfined
of transmissivity is calculated aquifer. Actually, there are three distinct parts of the
T=%+,;) = 5.43 X 103m3/dayX 1 time-drawdown curve as obtained in the vicinity of a
4ns 4 X 3.14 X 0.2 m pumped well. The very early time-drawdown data
should correspond to that predicted by the Theis equa-
= 2.2 X 10’ m2/day
tion where water is released from storage due to elastic
= 2.5 X 10-zmz/s compression and expansion of the water. Thus the time -
The coefficient of storage is drawdown data should follow the Theis curve where
the storativity is on the order expected for the confined
S=- 4uTt condition. With increased time, the effects of gravity
r2 drainage take over, with vertical flow components in
-- 4 X 1 X 10-5X 2.2 X 10’ &day X 7 X lo-’ days the vicinity of the pumped well. During this period, the
22,500 m2 time-drawdown response will be a function of the ratio
of horizontal to vertical conductivity, the thickness of
= 2.7 X lo-‘ the aquifer, and the distance to the pumped well. If
Both results are identical with the numerical values ob- departures from the Theis-typecurve do occur, they will
tained for Figure 6.6. Assuming the semipervious layer be similar to those obtained when the slope decreases
is 3 m in thickness, in the leaky aquifer test. Although this is occurring in
response to vertical flow components, one may view this
in a conceptual sense as an increase in the “storativity”
over its confined value. For later time, the time-
- 2.2 X 10’m2/dayX 3 m X 5.6 X drawdown data will again follow some Theis-type curve,
22,500 m2 with the storativity now reflecting the specific yield S,,;
= 1.6 X lO-jm/day
that is, the “storativity” is no longer increasing with time.
On a logarithmic plot the curve has the shape of an
 6.2 Conventional Hydraulic Test Procedures and Analysis 113

A more exact analysis requires a curve-matching proce-

dure. According to Neuman (1975) the solution to the
flow problem is

(6.27)

where W(tr,, uB,q) is the well function for the uncon-

fined aquifer. The arguments of the well function are
Time (minutes) described as
Figure 612 Timedrawdown response of an unconfined aq-
uifer. = -r2S (for early time-drawdown data) (6.28a)
4Tt
elongated letter "s"(Figure 6.12). Given this information, rLS,.
U,rj =- (for late time-drawdown data) (6.28b)
there are two approaches to the interpretation of pump- 4 Tt
ing tests performed in unconfined aquifers. First, let us
rzK'
recognize that previous to Boulton's contribution (19541, 77 = FK (6.28~)
no distinction was made between the confined and un-
confined response. Indeed, virtually hundreds of tests where rn is the initial saturated thickness, K is the hori-
were conducted on unconfined systems using the Theis zontal conductivity, and K' is the vertical conductivity.
analysis with no apparent reference to the need for modi- Figure 6.13 shows the type curve employed in the match-
fications. Jacob (1950) recognized the difficulties with ing procedure where the type A curves merge out of the
unconfined flow to wells and recommended a procedure Theis curve shown on the left and the type B curves
to account for an apparent increase in stontivity with nierge into the Theis-type curve shown on the right. The
time. In this procedure, the trmsniissivity is first deter- following method is recornmended for the analysis. First,
mined from the early time-drawdown data by the modi- the early time-drawdown data should be matched to the
fied nonequilibrium method (Eq. 6.13). Field values of type A curves and any match point selected. This match
drawdown s for later time t are then introduced into Eq. point will have the coordinates W ( u , , v), l/u,,,t, and s.
6.9 and the equation solved for the specific yield. If the The match will provide a value for 77. The transmissivity
time-drawdown curve actually contains three segments, is then determined with Eq. 6.27 and the storativity with
as in Figure 6.12, the ultimate value of S,. can be readily Eq. 6.28a. The later time-drawdown data are then
determined. If only two segments are 'discernible, an matched to the specific type B curve of the same I) value
intermediate vattie will be obtained. determined for the A match. The match point yields

"."A

lo4 10" lo-' 1 10 lo2 103

I
-
'In

Figure 6.13 Theoretical curves of W(u,, uII, 77) versus l/u, and l/u, for an uncon-
fined aquifer (from Neuman, Water Resources Res., v. 11, p. 329-342, 1975. Copy-
right by h e r . Geophys. Union).
114 Chapter 6 / Hydraulic Testing: Models, Methods, and Applications

values of W(U8, v), l / U H , s, and t. Transmissivity is deter- rium Eq. 6.9

mined from Eq. 6.27, which should provide a value simi-
2.3Q 2.25Tt
lar to that previously determined. The storativity can h,, - h' = As' = -(log - -
4nT r2S log=) rZS
be determined with Eq. 6.28b and should yield a value
somewhat larger than previously determined. As K = T/ (6.29)
m Eq. 6 . 2 8 can
~ be used to determine the vertical hydrau-
where As' is residual drawdown, t is the time since pump-
lic conductivity. Neuman (1975) provides tables for the
ing started, and t' is the time since pumping stopped.
function W(U,, , u B ,v).
This equation reduces to

(6.30)
- Single-Borehole Tests
63
Field procedure requires a drawdown measurement at
The one feature shared by the methods described in the the end of the pumping period ( t ) and recovery measure-
previous section is the necessity of one pumping well ments during the recovery period (t').The graphic proce-
and one or more observation wells. Single-borehole tests dure is to plot residual drawdown on the arithmetic scale
offer some advantages in economy, frequently with a and the value of t/t' on the logarithmic scale. The time
certain loss of information. In spite of this, the single- t includes the interval over the pumping plus recovery
borehole test is becoming increasingly more common in period whereas the time t' includes the recovery interval
hydraulic testing associated with waste disposal sites and only. If calculations are made over one log cycle of t / t ' ,
in low-permeability material. As in the previous section,
we will be concerned here mainly with the conventional 2.3Q
As' = - (6.31)
or common methods of analysis. 4nT
where As' is the residual drawdown per log cycle. The
storativity is not determined directly with this method.
Recovery in a Pumped Well
If a well is pumped for a given period of time t and The Drill Stem Test
then shut down, the residual drawdown (the original
prepumping water level minus the water level at any time The drill stem test used in petroleum engineering is the
after shutdown) can be approximated as the numerical equivalent of the recovery test. Figure 6.15 shows the
difference between the drawdown in the well if the dis- main components of a typical drill stem test. The packer
charge had continued and the recovery of the well in assembly is set at the isolated stratigraphic interval of
response to an imaginary recharge well, of the same flow interest. The bypass valve is initially opened and the
rate, superimposed on the discharging well at the time formation flows under its head. This is equivalent to the
it is shut down (Figure 6.14). Designating original head pumping period of the recovery test. The valve is then
as h,,and the recovered head at any time as h ' , residual closed, shutting in the formation pressure. The ensuing
drawdown is expressed from the modified nonequilib- period is analogous to the recovery period of the single-
borehole test previously discussed. The method of
Horner (1951) is generally used to analyze the data

op
.-a
.PI,.- P,, = -
4nkm
(6.32)

5
a where Pll, is the pressure in the well bore, P,, is the
0
z undisturbed formation pressure, Q is the rate of produc-
tion, p is the viscosity of the fluid, k is the permeability,
m is the formation thickness, t is the time since the test
began (the sum of the production and the recovery time),
and t' is the time since the recovery portion of the
c
8 test started.
E
-0
A typical pressure buildup plot is shown in Figure 6.16.
By considering the pressure change over one log cycle,
Eq. 6.32 reduces to

Time (minutes) (6.33)

Figure 6 1 4 Arithmetic plot of drawdown and recovery
curve versus time. where AP is the change in pressure over one log cycle.
 63 Single-Borebole Tests 115

Drill pipe 2 2100

V
.-
Auxiliary valve
andlor
~
E
Circulating valve 5:
k
Air chamber gauge ~
In
U
(optional) 3
n
Y

E
Tester valve and ___ :
In
2000
bypass valve E
a
100 10 1
Flow-stream gauge __ _I - 1+61
I' At
Figure 616 DriU-stem test result.
Optional tools ____
(jars, safety joints, etc.)
others (1967). Whatever method is being employed, the
procedures for the analysis are reasonably straightfor-
ward. The original head in the borehole is first noted and
recorded. For the slug injection, water is instantaneously
added to the borehole, raising the original head above
its static level. This is termed h,, that is, the height of
Packer assembly - the slug at time equal to zero, say, for example, 2 or 3
ft. The slug will then start to decay as water enters the
formation. This change in head is noted over time, where
Perforated anchor - h is designated the height of the slug for any given time.
Thus, at time equal to zero, h/h,, = 1, and as time gets
Pressure gauge ~

large, h/h,, approaches zero, that is, the slug has been
completely dissipated (Figure 6.17) If the well or bore-
Ffgure 6.25 Main components of a typical drilI-stem test hole is screened above the water table, as is common
string (from Bredehoeft, 1965). Reprinted by permission of for investigating oiI spills or leaks, a slug withdrawal
Ground Water. 0 1965. All rights reserved. procedure is necessary.

1.
By extrapolating the pressure curve to t/t' = 1, the Water lewel immediately
undisturbed formation pressure P,, is determined; that is,

1;-.
after injection
in Eq. 6.32, log (1) = 0.
Pressure buildup tests in single boreholes have long
/
-
- - -i- -
-
Water level at time I
been of interest to petroleum engineers, with the mono-
graphs of Matthews and Russel (1967) and Earlougher
(1977) providing good reviews.

Slug Injection or Withdrawal Tests i in aquifer

rc ---Well casing
The pressure recovery in a borehole after withdrawal of
a known volume of water (slug), or the pressure decline
after injection of a known volume of water (slug), is
termed the slug test. The slug test is one of the more
common methods employed in field practice in that it can
be used in low- to moderately high-conductivity materials
and requires no pumping apparatus. Thus, small-diameter
boreholes normally used in geologic exploration may
serve double duty in a hydraulic taring program. Figure 6.27 Well into which a volume V of water is cud-
The analysis of water levei versus time data in a single denly injected for a slug test of a confined aquifer (from Coo-
borehole in response to a slug injection or withdrawal per and others, Water Resources Res, v. 3, p 263-269,
has been pioneered by Hvorslev (1 95 1) and Cooper and 1967. Copyright by Amer. Geophys. Union)
-
116 Cbapter 6 / Hydraultc Testing:Mode&, Methods, and Applications

Cooper and others (1967)present a type curve solution 1.o

0.9
for the interpretation of slug tests. Papadopulos and oth-
0.8
ers (1973) extended the range of the type curves for
0.7
greater use in practice. However, it has been noted that
0.6
it is difficult to achieve a definitive fit in the matching
procedure in that field data frequently overlap or fit parts 0.5
of several of the type curves. Consequently, the method
0.4
is infrequently used. 0.37
The method of Hvorslev (1951) is probably the most
widely used in field practice. Field data such as given in *IB 0.3
Figure 6.18 are plotted on semilogarithmic paper with
h/h, on the log scale and time t on the arithmetic scale
(Figure 6.19).The hydraulic conductivity may be deter- 0.2
mined from the relationship (Cedergren, 1967)

(6.34)

where A is the area of the borehole and F is a shape 0.1

10 20 30 40 50 60 70 80 90
factor that depends on the size and shape of the intake Time (minutes)
area. Hvorslev (1951) has evaluated several values of the Figure 6.29 Field response of slug test.
shape factor F for a variety of conditions, as has the
U.S.Navy Bureau of Yards and Docks (1961) (see, for
example, Cedergren, 1967). For a cased, uncased, or It is convenient now to take h, = h,, at t = 0 and
perforated extension into an aquifer of finite thickness hL= 0.37h, so that
for L/r > 8,

(6.35)
Equation 6.36 then becomes
where L is the length of the intake area and r is the
radius of the borehole. For a borehole area = nr*, Eq. (6.37)
6.34 becomes
where T,, is the time intercept on the field curve where
r2In(L/r)log(hl/hJ h/h,, = 0.37 (Figure 6.19). To is termed the basic time
K= (6.36)
2L (t, - fJ lag and is determined where the head ratio = 0.37. The
time lag refers to the fact that the transients in a well
1 seldom correspond exactly to those occurring in the
adjacent formation so that flows are associated with vari-
0.9 ous time lags that depend on the shape factor F, the
hydraulic conductivity, and the radius of the borehole.
0.8

0.7 Example 65
0.6 The field data of Figure 6.18 are plotted on Figure 6.19.
For an observed value of T, = 44 minutes, a casing radius
0.5 of 2.54 cm, and a well screen length of 3.8 m, the hydrau-
0.4 lic conductivity is determined from Eq. 6.37:

0.3

0.2

0.1
Response at the Pumped Wekt
"1
Specific Capacity and Well Emciency
"1 0 100 1000 10,000 100,000
Time (saconds) Well capacity, or discharge per unit time, is often used
Figure 6.28 Field response of slug test. as a measure of well yield. When comparing the strength
 Next Page

6.3 Single-Borehole Tests 117

of one well versus another, it is better to express well of the predicted drawdown, where sIr= 0 , to the actual
capacity in relation to some conimon standard. The ac- measured drawdown s,.
cepted standard of comparison is the unit drawdown. A theoretical specific capacity may be defined as the
Thus, dividing the pumping rate by the total drawdown specific capacity that would be achieved for a 100%effi-
in a well gives the specific capacity, that is, cubic meters cient well, that is, one in which the well loss s,,,= 0.
per day per meter of drawdown. Theis and others (1963) demonstrated that the theoreti-
The drawdown observed in a pumping well is com- cal specific capacity of a well can be determined from
posed of two parts: (1) driwdown due to laminar flow the abbreviated nonequilibrium equation
of water in the aquifer toward the well, referred to as
r.3
formation loss s and calculated with the well flow equa- T= (-0.5772 - In -
tions discussed in this chapter, and ( 2 ) drawdown re- 4rs 4 Tt
sulting from the turbulent flow of water in the immediate or, solving for Q / s
vicinity of the well, through the well screen or casing
openings and in the well casing, referred to as well loss 4nT
su,.Jacob (1950) stated that well loss is proportional to
Ps , 4Tt
(6.41)
In 7-- 0.5772
some power of the discharge exceeding the first power r,.S
and approaching the second. As an approximation,
The assumption here is that well loss is zero.
=
s,,, cp (6.38) From this equation, the theoretical specific capacity
of a well is directly proportional to T and inversely pro-
where s,,, is well loss ( L ) , C is the well loss constant, and portional to In t,In l/ri,,and In 1/S. Hence, large changes
Q is the pumping rate. Total drawdown in a pumping in T cause correspondingly large changes in specific ca-
well, s,, is rhen expressed pacity. Large changes in t, r,,.,and S cause comparatively
small changes in specific capacity. Given the actual spe-
s, = s + s,, = 8
4rT
W(U) + CQ2 (6.39) cific capacity of a pumped well, the radius of the well,
and the duration of pumping, and assuming some value
The well loss constant C may be determined by compar- for S, the transmissivity can be estimated with Eq. 6.4 1.
ing the actual drawdown in a pumping well with that Figure 6.20 has been prepared to aid in this determina-
predicted for a 100%efficient well (s,, = 0) as based on tion. In this figure the pumping period has been taken
the Theis equation, where r is taken as the rxlius of the as one day and the other pertinent parameters are given
well r,,,.The efficiency of the well is taken as the ratio in so-calledAmerican practical hydrology units. Here spe-

0.00001 0.00005 0.0001 0.0005 0.001 0.005 0.01 0.05 0.1

Coefficient of Storage

Ffgure 6.20 Graph showing interrelation of well diameter, specific capacit)., transmissivity. and
stontivity (from Theis and others, 1963).
Previous Page

-
118 Cbapter 6 / Hydraulic Testing:Models, Methods, and Applications

cific capacity is given in gdllonS per day per foot of

,t,
drawdown, transmissivity is given in gallons per day per
foot, and the storativity remains a dimensionless constant.
To use the figure, measured specific capacity (left-hand
--dW Piezometric level

axis) is followed across to an intersection with an esti-

mated storativity, and then the appropriate curve for I
I
Confining layer II
the well diameter is followed to the right, yielding an
approximate value for the transmissivity. If well losses
are involved in the measured specific capacity, which is
inevitable, the transmissivity so determined represents an
underestimated value. That is to say, the actual formation
transmissivity is higher than the calculated one. ( ? d l h S
per foot per day can be readily converted to m'/day by
multiplying by 1.242 X lo-'. Figure 6.22 The effect of partial penetration on tlow t o ;I
well.

Example 66
experience with vertical flow components in the uncon-
An aquifer has a transmissivity of 125 mL/day and a stora- fined case previously discussed suggests upward inflec-
tivity of 1 X lo-'. A 60-cm diameter well in this aquifer tion points in the time-drawdown response.
has a drawdown of 91 m after 24 hours of pumping Tlie topic of partial penetration has been the subject
at a rate of 5440 m'/day. The efficiency of the well is of numerous papers (Muskat, 1937; Hantush, 1961,1964;
determined as follows: Neunian, 1974). Two aspects ot the problem are of inter-
For drawdown in the pumping well in tlie absence of est here. First, solutions for partially penetrating wells
well loss have been derived and are discussed by Hantush (1961,
1964) and Walton (1970). These solutions may be applied
Q W(u) =
s =- 5440 w(u) = 3.5 W(U) directly if partial penetration is a factor affecting the
4nT (4>(3.14)(125) response. However, with such solutions, it is necessary
The value of u is determined for r equal to the well radius to know the actual degree of penetration of the aquifer,
which may not be known in many cases. A second point
is the effect of partial penetration on the water-level
response. Hantush (1964) has conducted the most work
in this area and provides some general guidelines for
From Table 6.1, W(u) = 17.2 so that s = (3.5)(17.2) = confined aquifers. The effects of partial penetration will
60 m, and well efficiency is determined as I:' = 60/91 =
not normally affect tlie pumping test result if the obsewa-
66%. tion well is located at some distmce r > 1.5m (NK')'''
The well loss constant C is determined from s,,, = CQ
from the pumped well, where K is the horizontal cotiduc-
tivity, I(' is the vertical conductivity, and ni is the aquifer
thickness. If tlie conductivities K and K' are of the same
order of magnitude, the condition I- > 1.5m is sufficient.
If this condition is not satisfied, there will he an upward
intlection in the response, similar to that obtained in the
leaky method or for some sort of recharge boundary. In
some cases, depending on tlie relative positions o f the
screens, more distant observation points can actwlly
6.4 Partgal Penetration, Superposition, show more drawdown than closer ones. For long periods
and Bounded Aquifers of time t > .Sm/2K', the tin1e-rlriiwdown slope of ;I
partially penetrating response will he the same as ex-
Partial Penetration pected if tlie well completely penetrated tlie aquifer. Tlie
When the screened or open section of a well casing does reader is referred to Hantush (1961) or Walton (1970)
not coincide with the full thickness of the aquifer it for the methodology required to treat partial penetration.
penetrates, the well is referred to as partially penetrating.
With existing wells, this is the rule rather than the excep-
Principle of Superposition
tion. Under such conditions the flow toward the pumping
well (or observation point) will be three dimensional The Theis nonequilibrium equation represents a solution
because of vertical flow components (Figure 6.21). Our to the diffusion equation for a prescribed set o f boundary
 6.4 Partial Penetration, Superposition, and Bounded Aquifers

and initial conditions. The pumping test procedure is time = 10 days

relatively straightforward where a transient response to
a steady pumping rate is observed over a limited period
of time (say, 24 hours) at some known distance from
the pumping well. Given the transmissive and storage
properties as determined over this short pumping period,
it is possible to predict the water level at later times and ~~

at other distances in response to any steady pumping 0.1 3.07 X lo-" 23.63 -4.64

rate Q. This assumes, of course, that the storativity and 1 3.07 x lo-'' 19.02 3 -i
10 3.07 X lo-- 14.42 2.83
transmissivity do not change with time, whereas the pos-
100 3.07 X lo-' 9.8 I I .93
tulated time of pumping t, the distance r, and the pump-
ing rate Q may be taken as variables.
The Theis equation (for that matter the Hantush-Jacob time = 100 days
equation) may also be used to obtain a theoretical draw- r2 X 2.7 X lo-'
u = -rzS
= = 3.07 x 10 "'r2
down for any time t at various distances r in response 4Tt 4 X 2.2 X 10' X 100
to a battery of wells pumping at various rates. This is
because the diffusion equation is linear, that is, consists
of a sum of linear terms where a linear term is first degree r, nt 14 W I ) s, t ) I
in the dependent variables and their derivatives. Linearity 0.I 3.07 X lo-" 25.93 5.09
in differential equations is synonymous with the principle 1 3.07 X lo-'" 2 1.33 4.19
of superposition, which states that the derivative of a 10 3.07 X lo-" 16-2 3.28
sum of terms is equal to the sum of the derivatives of 100 3.07 x lo-" 12. I2 2.58
the individiial terms. Expressed in terms of the pumping
response, the total effect resulting from several wells
pumping simultaneously is equal to the sum of the indi-
vidual effect caused by each of the wells acting sepa-
rately.

Example 6 7 --E 2
The principle of superposition is an important one in the C
8 3
application of hydraulic methods to real-world problems. 0
z
The purpose of this example is to demonstrate an effi- E
n
cient method with which well test data are organized to 4
examine well interference. We assume that the data of
Figure 6.1 apply; that is, a well is pumped for a relatively 5
short period at a rate of 5.43 X 10' m2/day and a transmis-
sivity of 2.2 X lO' mz/day and a storativity of 2.7 X lo-'
is determined. We assume these properties are constant -0.1 1 10 100
over a large region. The table below demonstrates a pro- Distance from pumped well (m)
cedure for tabulating values for the drawdown for times Figure 6E7
ranging from 1 to 100 days and over distances up to 100
m. It is noted that these calculations are for the specific
pumping rate of 5.43 x 102 m2/day from a single well. The graph may be referred to as a mathematic;il modcl
time = 1 day for the determination of the response to pumping in this
region of constant transmissivity and storativit).. We note
that the drawdown some 100 m from the pumping well
at the end of 100 days of pumping will be on the order
of 2.4 m, whereas at 10 m it will be 3.3 m. On the other
hand, if we are interested in the water-level response ;it
one point 100 m distant from a pumping well and 1 0 ni
0.1 3.07 x lo-"' 21.33 4.18 distant from another pumping well, the principle o f sii-
1 3.07 x 16.72 3.28
perposition states that we require the sum 2.4 + 3.3 =
10 3.07 X 12.12 2.38
100 3.07 X lo-' 7.51 1.47
5.7 m. These calculations are only valid for the "design"
pumping rate of 5.43 X lo' m3/day. However, we note
-1 Cbapter 6 / Hydraulic Testing:Models, Methods, and Applications

that the pumping rate Q is not one of the variables making

up the variable u and only occurs in the drawdown state-
ment s = @/4aT>W(u). Hence, the magnitude of d r a w A

down is proportional to the pumping rate, and so if w e

double the rate in the examples, we double the draw-
down, and if we pump half as much from each well, we
halve the drawdown.
With these points in mind, consider the following ex-
ample. Two wells are pumping for 10 days at rates of
10.86 X 10' and 2.72 X 10' m'/day. Five days after these
wells start pumping, a third well turns on at a rate of
4.07 X lo3m'/day. The pumping wells are at a distance
of 100, 10, and 1 m, respectively, from an observation
well. At the end of this loilay period, the drawdown at
the point of observation is determined from the preced-
ing figure for the postulated pumping rates
SI = 3.9 ~2 = 1.4 ~3 = 2.8 Figure 6.22 Simple two-well system for an aquifer
bounded by (a) a no-flow boundary and (6)a constant-head
giving a total drawdown of 8.1 m. boundary (from Stallman, 1952).
Suppose now that the observation well is itself a pump-
ing well pumping at a rate of 5.43 X 10' m'/day for this at right angles thereto, and at the same distance from
same 10-day period. If the diameter of the well is about the boundary as a real pumping well. If the imaginary
0.6 m, its self-induced drawdown, ignoring well loss, is well is assumed to start pumping at the same time and
about 4.0 m. Hence, the total drawdown at this point is at the same rate as the real well, the boundary will be
on the order of 12.1 m. transformed into a ground-water divide.
Similarly, if the aquifer is bounded by a stream that
provides recharge to the aquifer, the effect is to decrease
Bounded Aquvers the drawdown in a well. A zerodrawdown (constant-
An assumption employed throughout this chapter is that
head) boundary can be simukdted by an imaginary well,
the aquifer is infinite in areal extent. Geologic boundaries located as earlier, with the exception that the imaginary
limit the extent of real aquifers and serve to distort the well must recharge water at the same rate as the pumping
calculated cones of depression forming around pumping well (Figure 6.22b).
wells. The method of images, which plays an important The drawdown at any point in the real aquifer or at
role in the mathematical theory of electricity and is em- the boundary for the simple two-well system is the sum
ployed in the solution of some geophysical problems, aids of the effects of the real and imaginary wells operating
in the evaluation of the influence of aquifer boundaries simultaneously. From the principle of superposition
on well flow. This theory, as described by Ferris s = sp 2 s,= -Q [W(U)p 2 W(U),l
(1959), permits treatment of the aquifer limited in one 4aT
or more directions. However, the additional assumption
of straight-line boundaries has been added. This gives =
Q
-z W(u) (6.42)
aquifers of rather simple geometric form. 4nT
The image well theory can be explained as follows
where s is the observed drawdown at any point, con-
(Stallman, 1952): formation A is bounded by the relatively
sisting of the sum of the effects of the real well s, and
impermeable formation B , the boundary between the
the imaginary well s,,the well hmction for the real well
two located at a variable distance r from a pumping well
is W(u),,,and W(u),is the well function '.ir the imaginary
(Figure 6 . 2 2 ~ )As
. formation B is relatively impermeable,
well. If the imaginary well is a recharging well, the nega-
no flow can occur from it toward the pumping well, and
tive sign is used.
the boundary is a no-flow (barrier) boundary. The effect
By similar reasoning,
of a barrier boundary is to increase the drawdown in a
well. The problem now is to duplicate this physical situa- rjS
tion by substituting a hydraulic entity that serves this UP=- w = - r:S
- (6.43)
4 Tt 4ft
purpose. As no flow occurs across a ground-water divide,
the barrier boundary is simulated by the supposition that where r,, is the distance from the pumping well to the
formation A is infinite in areal extent and that an imagi- observation point and r, is the distance from the imagi-
nary well is located across the real boundary, on a line nary well to the observation point. From the equality
 6 4 Partial Penetration, Superposition, and Bounded Aquifers 121

expressed in Eq. 6.43

(6.44)

or
-
u, = K'u, (6.45)
where K equals the constant r,/r,,.For any point on the
boundary, r,, equals r, ( K equals l), and u, equals ti,,. It
r, .._11 ................... 4
1 ,Avemge or effective follows that W(u),equals W(tr), (Eq. 6.42), and the draw-
1---
d
mition OX line at
. down is either zero (constant-head boundary) or twice
zero flaw
(a)Real System
the effect of one well pumping in an infinite aquifer.
When a weil in a bounded aquifer is pumped, water
levels in observation wells will initially decline under the
influence of the pumping well only. When the cone of
depression reaches an exterior boundary, deviations from
the ideal response will be noted in the Observation well.
Under these conditions, the early part of the tinie-
drawdown curve unaffected by the boundary behaves ;IS
if the aquifer were infinite in areal extent, and this p;in
of the curve can be used to determine the hydraulic prop-
erties.
Figure 6.23 shows the effects of pumping near a barrier
Note Aquifer thickrress nr should be very large compared and recharge boundary and the role of image wells in
to resubnt draPvdawn mar reat wIl. simulating this pumpage.
(hl Hydraulic Countorpatiof Red System Aquifers are often bounded by two or more bound-
aries, and time-drawdown data will respond accordingly.
As mentioned, the resulting geometry must be rather
simple in order to apply the image theory. Hence, two
converging boundaries delineate a wedge-shaped aquifer,
two parallel boundaries intersected at right angles by a
third boundary forms a semiinfinite strip. and four inter-
secting right-angle boundaries form a rectangular aquifer.
A number of image wells associated with each pump-
ing well characterizes a multiple-boundary system.
Clearly, primary image wells placed across a boundary
will balance the effect of a pumping well at that bound-
ary, but will cause an unbalanced effect at the opposite
boundary in violation of the no-flow or constant-head
requirement. It is then necessary to add secondary iniiige
wells at appropriate distances to satisfi. the conditions
of no flow or zero drawdown. Figure 6.24 shows sonic
plan views of image well systems for some simple
aquifers.
Once the boundaries of a finite aquifer ha\^ been simu-
lated by means of image wells, analysis of drawdown
effects can proceed as if the aquifer were infinite in areal

Figure 6.23 Diagrams of the effect of pumping iie;ir h r -

rier and constant-head boundaries and :cppropri;ite h!~tlr:~ulir
Note Aquifer thickness ##; shaufd be very large compared counterparts for image well theory (from Ferris ;inti others.
to resultant drawdown mar real well, 1962). Note: Aquifer thickness 111 should be v e n large coni-
&I Hyd~aulicCounterpart at Real System pared to resultant drawdown near real well.
122 Cbapter 6 / Hydraulic Testing:Models, Metbods, and Applications

Example 69
The accompanying figure depicts a time-drawdown re-
sponse in an observation well affected by one pumping
well and a barrier boundary. By construction, sI= s2.It
thus follows that W(u), = W(u), and
r:S r:S
UI =- and u2=-
(a) (b) 4Tti 4 Tt,
where u , = u,. Thus
IT I 0 Discharging well

. Pattern repeats Pattern repeats

which they are
considered.
where r, is the distance from the pumping well to the
observation well, which is known, and r, is the distance
to infinity
'I'
m
to infinity
from the observation well to the imaginary pumping well
causing the inflection, which is readily calculated. How-
(C)
ever, because the exact location of the barrier is not
Figure 6 2 4 Image well systems for (a)wedge-shaped aqui- known, r, merely defines a set of points that form a circle
fer with both barrier and recharge boundaries, (6)wedge-
of radius r2with the observation well located at its center.
shaped aquifer with barrier boundaries, and (c) infinite strip.

Pump on

extent. In other words, graphic models such as given in

I I
Example 6.7 can be used to find the drawdown in re-
sponse to one or several wells pumping simultaneously.
Once the distance from real and imaginary wells to the
point of interest has been determined, drawdown (or
recovery) may be obtained from the graphs. For example,
the drawdown at any point in the real aquifer of Figure
6.24 is the algebraic sum of the effect of one real well,
one discharging imaginarywell, and two recharging imag- I I I

-
I I I
inary wells. For the infinite array (Figure 6 . 2 4 ~ 3image
, I I
I
I
I
wells are added until the most remote pair has negligible I-4- I
I I
inthence on water-level response. I* I2

Figure dE9
Example 48
A 6Ocmdiameter well is located in an aquifer with a
transmissivity of 125 m2/dayand a storativity of 5 X lo-'.
A fault (barrier) is located 300 m from the well. The
6.5 Hydraulic Testing in Fractured or
drawdown at the midpoint between the fault and the Low-Permeability Rocks
barrier at the end of one year's pumpage at a rate of Advances in hydraulic testing over the past decade have
2700 m3/day is determined been born out of the practical necessity of evaluating
rocks as hosts for chemical or nuclear wastes. In some
s = s,, + sj = -Q [ W(U),, + W(U),l cases, this requires testing in deep boreholes up to a few
4nT
thousand feet in depth, often in rocks that are of a low
permeability, and frequently in rock that is fractured. In
many cases, the emphasis has been on the single-borehole
test. These tests generally require special equipment such
as packers to isolate testing zones, feed in apparatus
whereby a slug of water may be removed from the i s a
From Table 6.1, W(u),, = 4.52, W(u), = 2.37. lated zone or the zone quickly pressurized, and pressure
transducers to measure the response. With deep bore-
2700 m3/day
S=
4 x 3.14 X 125 m2/day(6.89) = 11.85 m holes, temperature or salinity effects on fluid density
must sometimes be accounted for. Hydraulic testing in
 6.5 Hydraulfc Testing in Fractured or Low-Permeability Rocks 123

this modern era is thus a field within itself, generally where u is defined as in all previous cases as r'SI4Tt and
requiring a large degree of specialization. In this section, pis defined as
we will consider only a few of the methods in detail.
p= (L)(y2
48 S
(6.47)

Single-Borehole Tests where S' is the storativity of the confining layer and
l / B has already been defined as (k'/m'T)"'. By short
The slug test discussed earlier is well suited for low- times we mean times of observation less than m'S'/lOk'.
permeability testing. The pressure recovery or decline The quantity H(u, p) is an infinite integral whose value
in a borehole can be measured after the removal or injec-
is approximated as

*'
tion of a volume of water. The pulse test is a modification
of the slug test whereby a testing interval within a single
borehole is instantaneously under- or overpressured by
[
H(u,p) = W ( U )- - 0.2577 + 0.6931 exp
(nu)''Z
removing or adding water. The time response of the
(6.48)
pressure buildup or decay is then observed, with the
analysis being similar to that used in the slug test. where W(u) is the familiar well function of u. Some
Wang and others ( 1 978) extended the methods devel- typical curves for the well function H(u, 0) are shown
oped by Cooper and others (1967) to pulse testing in a in Figure 6.25. Hantush (1960) has tabulated relevant
single fracture. The method is based on the parallel plate values for the function H(u, p> that may be adequate
model and assumes that the aperture width is indepen- in applications. Note from Eq. 6.48 that this function
dent of fluid pressure and that no flow is derived from approaches W(u) as p approaches zero.
the matrix of the fractured medium. Barker and Black The longtime unsteady drawdown is expressed as
(1 983) extend this method to examine the effect of flow
in the matrix as well as in the fractures. Their aquifer (6.49)
model is identical to that employed by de Swaan (1976)
and Boulton and Streltsova (1977). where
The determination of vertical permeability of low-per-
meability rocks in a single borehole has also been ad-
dressed in a number of papers (Burns, 1969;Pratts, 1970;
a= 1 + (5)
HirdSdka, 1974). The monograph of Earlougher (1977)
also discusses the available methodology and technology. Note that as S' approaches zero, we recover the Han-
tush-Jacob leaky formula.
The procedure for analysis is as follows.

Multiple-Borehole Tests 1. Early time-drawdown data are fitted to the family

of curves shown on Figure 6.25 with a match point
Of the numerous multiple-borehole testing procedures producing values of H(u, p), 14, s, and t. The value
available, only two will be discussed in this section: the of /3 is also obtained as p is a parameter of the type
method of Hantush (1960, 1964). which provides infor-
mation on the vertical permeability and storativity of
confining layers, and a modification of this method by
10
Neumdn and Witherspoon (1 969a, 1969b, 1972).
The method of Hdntush (1960, 1964) requires a pump-
ing test procedure identical to the conventional pumping
well-observation well setup discussed previously. The 1.o
parameters obtained from the test include the storativity
and transmissivity of the aquifer and the storativity and
5-
vertical permeability of the confining layer. The solution :
is an asymptotic one where the water-level response in 0.1
the observation well is analyzed over a short time period
and over a long time period. A curve-fitting procedure
is involved for each of the designated time periods. For n ni
"."A

the case corresponding to Figure 6 .la, the short-term 1 .o 10 1o2 103 1o4 105
drawdown can be expressed as 1
11

Figure 62.5 Type curves for a well in an aquifer confined

by a leaky layer that releases water from storage (with per-
(6.46)
mission from William Walton).
-
124 Chapter 6 / Hydraulic Testing: Models, Methods, and Applications

curve in the same manner that r/E is a parameter of fluid into a number of isolated zones in boreholes and
the leaky curves. The transmissivity is determined measuring the response in the same isolated zones in
from Eq. 6.46 and the storativity S is determined adjacent horeholes. The response is then fitted to one
from the relationship u = r2S/4Tt. of several possible analytical solutions, generally with the
2. The usual procedures for leaky aquifers are then aid of a computer. A simple curve-matching procedure
followed for the long-term data with the type curve is given by Hsieh and others (1985).
of Figure 6.10. The match point will produce values The hydraulic testing of fractured rock is a special
of W(uS, r/B), Su, s,and t. The value of r/B is again application of flow in homogeneous but anisotropic ma-
noted. This will produce values of transmissivity with terial; that is, the hydraulic conductivity changes with
Eq. 6.49 and the vertical hydraulic conductivity from direction. With the method of Hsieh and others, the
the determined B value. As both p and B are now principal planes of anisotropicity need not be established
known, Eq. 6.47 may be solved for the storativity of in advance. Other pumping test methods designed for
the confining layer or, conversely, with known Val- anisotropic horizontal aquifers in which one of the princi-
ues of SS and S, the storativity of the confining layer pal directions is vertical have been given by Papadopulos
may be determined. Details of analysis for other g e e (1965), Hantush (1966a, 1966b), and Hantush and
metric configurations of leakage are given by Domen- Thomas (1 966). These models correspond to the geologic
ico (1972). prototype given as Figure 6 . 1 ~Much
. of the recent re-
search refers to the contribution of Papadopulos (1965)
Interestingly, the solutions just described recover the as one of the first original contributions in this field,
leaky aquifer formula as the storativity of the confining including Way and McKee (1 982), Loo and others (19841,
layer approaches zero, as well as the Theis solution where and Hsieh and others (1985). A method of determining
both the storativity and permeability of the confining vertical conductivities has been given by Weeks (1969)
layer approach zero. Unfortunately, because of the simi- and Way and McKee (1982).
larity in the shape of the family of curves of Figure 6.25
used in the analysis, it is difficult to achieve a definitive
match. Figure 6.2 shows the expected time-drawdown 6.6 Some Applications
response where leakage from storage is involved and
compares this response with the conventional leaky and
to Hydraulic Problems
nonleaky drawdown curves. There are various solutions or partial solutions to hydro-
The asymptotic short-term-long-term solution of Han- logic problems that require either the results obtained
tush (1960) has been solved over continuous time by from a hydrologic test or the utilization of the mathemati-
Neuman and Witherspoon (1969a, 1969b). These results cal model upon which the formation response was predi-
were organized into a field testing procedure whereby cated. For instance, any application of Darcy’s law re-
the hydraulic properties of the aquifer and bounding low- quires a priori knowledge of hydraulic conductivity.
permeability material may be determined (Neuman and Hence, if we wish to make some calculations of flow
Witherspoon, 1972). The test, referred to as the “ratio rates or velocity of movement, the hydraulic conductivity
test,” requires an observation well in the confining layer. must be known. If we wish to say something about travel
Testing the hydraulic response of a confining layer is not time for a contaminant that moves at the speed of ground
a new development and was part of the requirements water, some determination of hydraulic conductivity is
of earlier work dealing with low-permeability response required. At yet another level, if we wish to apply some
(Wolff, 1970; Wolff and Papadopulos, 1972). However, numerical model to ascertain the long-term response of
the so-called ratio method represents a sophisticated a p a basin to prolonged pumping, information is required
proach to testing low-permeability rock where low on the spatial distribution of hydraulic conductivity and
permeability zones bound higher-permeability aquifers storativity. Hence, the necessity of hydraulic testing.
that can be pumped over an extended period of time. On the other hand, there is also a need for the mathe-
Many fractured rocks qualify as low-permeability me- matical models employed in well testing. For example,
dia, and the testing of such rocks has been long viewed a short-term pumping test (say, 24 hours) will yield the
with skepticism. Much of the recent research was con- pertinent formation parameters T and S. As time and
ducted by the staff of the University of Arizona, where distance are variables in the mathematical model, it is
a field test facility was established in Precambrian granite possible to say something about what the drawdowns
near Oracle, Arizona. A complete description of the site will be at later times and other distances. Due to the
and the geologic, geophysical, and hydrologic investiga- principle of superposition, it is possible to do the forego-
tions are given in Jones and others (1985). The hydrologic ing for a whole battery of wells pumping at different
test theory has been given by Hsieh and Neuman (1985) rates. Such predictions are invaluable in the design of
and Hsieh and others (1985). dewatering operations, in water well interference consid-
The method of Hsieh and others consists of injecting erations, in the design of well interceptors to contain
 6.6 Some Applications to Hydraulic Problems 125
9.2 I I I 1 Two useful kinds of information may be obtained from
step-drawdown tests: pump setting requirements and
the well loss constant. Consider that the well in question
penetrates 15 m below the water table. The depth of
penetration below the water table is called the maximum
available drawdown. Thus at a pumping rate of 210 m'/
hr or 295 ms/hr, only 6.1 m or 8.8 m of the ;ivail:ible
drawdown will be utilized. With a pump setting near the
bottom of the well, either of these rates may be used i n
a long-term pumping test. At 415 m'/hr, approximately
13 m of the maximum available drawdown will be uti-
18.2 I I I I lized; that is only 2 m of water remains in the well after
10 100 1000
Time (minutes) a few hours of pumping. We thus establish a reasonable
upper bound for the pumping rate on the order o f 340
Figure 6.26 Semilogarithmic plot of time-dnwdown
mi/hr.
curve obtained during ;I step-drawdown test.
The well loss constant C may be determined directly
contaminant migration, and for predicting pressure re- from equations presented by Jacob (1950) where
sponse and contaminant migration due to deep-well in-
jection. As no new principles are involved in these appli-
cations, these points are best demonstrated with a few
examples, most of which are based on actual case histor- or, for steps 1 and 2 of Figure 6.26
ies. That is to say, we already know the theory; now we As?/AQ? - As,/A.Q, -
- (2.7/85) - (6.1/210)
need some rules for applying it. C=
AQi + AQ2 210 + 85
= 9.2 X lO-"m/(h/hr)'
Screen Diameter and Pumping Rates
and for steps 2 and 3,
The term screen is used to signify the diameter of a well
that is open to the flow of water. It may be a perforated
casing or some manufactured steel case apparatus. The
larger the pumping rate we wish to employ, the larger
= 1.2 X lo-' m/(m3/hr)'
the pumping apparatus required so that some minimum
well diameter should be associated with a given pimping for an average of 1.06 X nt/(m5/hr)'. Thus the well
rate. The following is a rule of thumb guideline. For a loss associated with each of the pumping rates used i n
desired pumping rate of 28 ms/hr, the minimum well the step-drawdown test is readily determined from the
diameter should be 15 cm. The following pairs are recom- relationship s,,, = CQ'. For each of the pumping rates
mended: 68 m3/hr, 20 cm; 136 m'/Iir, 25 cm; 272 m3/ employed, the well loss is 0.47 m. 0.92 ni. mid
hr, 30 cm; 454 m3/hr, 35 cm; 682 m'/hr, 40 cm; 1136 1.83 m, respectively.
m+/hr, 45 cm. It follows that using a 2km-diameter well
will not assure a well yield of 68 m-'/hr; the actual yield
will be controlled by the transmissivity. On the other A Problem in Dewatering
hand, a 20-cm-diameter well will not produce 682 m3/hr Figure 6.27 shows a situation where a tunnel must be
from a formation that is capable of delivering such constructed some 1 8 m below a water table. To facilitate
amounts because of limitations placed on the pumping construction, the water table must he lowered below the
equipment. tunnel during the construction period. Several questions
must be addressed:
WeU Yield.The Step-Drawdown Test 1. What pumping rates will be employed?
A step-drawdown test is one in which the well is oper- 2. What are the depth and diameter of the pumping
ated during successive periods at constant fractions of wells?
its full capacity (Figure 6.26). Such tests are useful for 3. What is the duration of pumping?
determining well yield and for establishing the depth for
4. How many wells will be required, and how should
the pump setting. The specific capacity may be deter-
they be spaced?
mined for each of the steps, that is, from Figure 6.26,
S.C. = 210/6.0 = 35 m2/hr, S.C. = 295/8.8 = 33.5 mi/ Many of these questions are interdependent. and sonic
hr, and so forth. The specific capacity decreases with have to be answered on the basis of value judgment. The
increasing pumping rates due to increased well loss. values for transmissivity, storativity, and specific capacity
126 Cbapter 6 / Hydraulic Testing:Models, Metbods, and Applications

pumping period has the benefit of a greater spacing be-

tween the wells to obtain the desired drawdown. How-
ever, the shorterduration period is the better one here

3:1
for several reasons. First, in most dewatering operations,
only a few tens of meters along the tunnel alignment are
dewatered during construction. Once the construction
- _ _Water
_ _table
_ _ _ _ _ _ _ _ _ _ _ _ _ _ - - - - - - -activities
- are completed in a certain reach, the pumps
I = 170 m'lday
S = 0.0168 are moved farther along the alignment. Thus, the whole
18 rn S.C. = 4.5 m2/hr alignment need not be dewatered in advance. In addition,
a 100-day period would require the disposal of greater
quantities of water. Thus, the 1 Wa y design period is the
6m Tunnel/ better one.
Figure 6.29 shows a plan view of one feasible design.
Clay The 10 m offset from the centerline of the tunnel was a
Figure 6.27 Schematic cross section of a proposed tunnel contractual agreement. The water levels must be lowered
through an unconfined aquifer. some 20 m all along the centerline. As a first guess,
assume the wells in each row are 30 m apart. The least
drawdown will occur at points 1 and 3, and the most at
are noted in the diagram and were obtained by hydrau- point 2. Using superposition with the aid of Figure 6.28,
lic testing.
the drawdown at points 1 and 3 is calculated to be on
The first thing to note is that the maximum available the order of 30 m. Clearly, we do not have 30 m of
drawdown for wells drilled to the top of the clay layer
available drawdown, but the model calculation is not
is on the order of 24 m. Thus, with a measured specific constrained and can produce this nonsensical result. For
capacity of 4.5 m2/hr, the maximum possible pumping a 60-m spacing, the drawdown is 24 m at points 1 and
rate is calculated to be 108 m'/hr. This, however, is
3 and 29 m at point 2. For a 90-m spacing the drawdown
somewhat unrealistic in that such a pumping rate would is approaching 23 m at points 1 and 3 and is slightly
readily remove all the water from the well. Restricting
more at point 2. Thus it appears we are slowly moving
the drawdown in a single pumping well to 21 ni, the toward an optimal spacing, and this spacing is in excess
design pumping rate is calcdated to be 95 m'/hr. From
of 60 m. However, there is need for some caution here
information provided in the previous example, this re-
in that the calculated spacings have been based on a
quires a design diameter of 20 cm for the dewatering
design pumping rate of 95 m'/hr, and drawdown is di-
wells. Thus, w e have determined that the wells will be
rectly proportional to the pumping rate. With five in-
55 m deep, their diameter will be 20 cm, and their design terfering wells, this pumping rate cannot be maintained
pumping rate will be 95 m'/hr. All that remains is the for the target 1 Ma y period. For the 90-m spacing, a
determination of the pumping duration and well spacing. decrease in the effective pumping rate from 95 d / h r to
Figure 6.28 shows some theoretical distance- 83 m"hr will produce a 20-m drawdown at points 1 and
drawdown curves for pumping durations of 10 and 100 3. For the 60-m spacing, a decrease in the pumping rate
days. Note that the difference in drdwdown between to 78 m'/hr will likewise produce a drawdown of 20 m
the two curves is only about 2 m. The longer-duration

10 rn I
A. ,Point 1
0 1
I
I
3 I
I

--E 6
I
k
+
I
I
g g I
I
-0 I
1 12
n
10 m
.*.
I --Point 2
I
15 I
I
I

'
18 I
I
P
I
I
I
I
L-

0.1 1 10 100 1000 I

10 m A. ---Point 3
Distance from Pumping Well (m) I

Figure 628 Time-drawdown data. Figure 6.29 Plan view of well alignment for dewatering.
 Next Page

6.6 Some Applications to Hjldraultc Probletns 127

at points 1 and 3. T ~ Utlie
S need for value judgments, I
with the 60-ni spacing being the more pnident one, at Boundary 2 To infinity Boundary 1
least for the preliminary design. This means that the
design pumping rate of 95 ni3/lir can fall off to as little
- Infinite array 0

as 78 m3/hr over the 1O-day pumping period and the

dewatering can still be accomplished.

A Problem in Water Supply

A power company plans to build a facility in the south-
western desert of tlie United States and requires a water
supply for cooling purposes. Tlie plant is to be con-
stnicted a few miles from ;I known ground-water source, Figure 631 Plan view of well location relative t o houncl-
and the plan is to deliver the pumped water to the plant, aries.
where it will be consumed in the cooling process (no
return flows). Tlie plant is to be built in three units,
phased in every five years, with each unit requiring ;I It is not intended to treat this entire complex problem.
continuous flow of water amounting to about 477 n1'/ but merely to show sonie elements of the use of the
hr. A cross section of the water supply is shown in Figure pertinent hydraulic equations in addressing quantitative
6.30, which is a rather narrow alluvial valley cut into questions. Figure 6.32 shows the familiar distance-
low-permeability sediments. Tlie company has access to drawdown plot for an infinite aquifer with ;I design punip-
property that extends across tlie entire valley (about 4570 ing rate of 160 ni3/hr. Let us focus on the centerniost
m) and about 3050 m along the valley reach. Three wells well. With a specific capacity of 1 1 niL/hrand ;I pumping
are shown i n the plan view o f this property, and their rate of 160 m3/hr, the drawdown is anticipated to he
locations have been idealized to fiicilitate the required about 14 ni. This will certainly increase with time, but
computations (Figure 6.3 1). let us not dwell on this detail. The influence from the
There are ;I few Facts and a few questions of importance two neighboring real wells at tlie end of a tive-year period
here. First, because of the geologic configuration of the will be on the order of 4 m, giving a total drawdown of
alluvial valley, tlie problem should be treated as an infinite 18 m. Consider now only four image wells associated
strip with image wells repeating to infinity (Figure 6 . 2 4 ~ ) . with the boundaries iiffecting this well. These will bc
Second, drawdowns in any given well should be liniited located 300 n~ and 8840 m from bound;iry 1 and 4270
t o ;ibout 30 m or the subsequent dewatering of the aqui- m and 4870 m from boundaq 2. These images ;ire thus
fer will destroy its transmissive properties. This means located the following distances from the pumped well:
that the pumping should be distributed as equally as 600 ni, 9140 m, 8590 m, and 9140 ni. Their cumulative
possible. With these constraints, can the power company effect on the pumping well is about 16 m giving ;I total
obtain sufficient water for each of the units, each of drawdown of 34 m. Yet to be considered are the other
which is phased in every 5 years? If not, can they supply images associated with this well, which have a nie;isur-
one unit for the life of the plant ( 5 0 years)? If not, how able influence from as Far away as 27,420 ni (Figure 6.31 ) .
long can they siipply one unit without excessive dewater-
ing of the aquifer? The transniissivity has been deter-
mined to he 497 m2/day, the storativity equals lo-', the
well diameters are 36 cm, and the actual specific capacity
of the wells is about 11 ni?/Iir.

Pressure level

I"
100 1000 10.000 100.000
Distance (m)
permeability Figure 632 Time-dnwdown response of ;in infinite i~cb
Figure 6.30 Cross section o f alluvial aquifer. uifer.
Previous Page

-
128 Cbapter 6 / Hydrauldc Testing: Models, Metbods, and Applications

plus the images associated with the other two pumping included at the back of the book and instructions for
wells. Based on this preliminary analysis, an attempt to loading are included in Appendix B. When the code is
develop sufficient water for one unit over a five-year run from Windows, the initial dialog box has three main
period will cause substantial dewatering of the aquifer. pull-down menus-”File,” “Editing,” and “Options.”
This conclusion must be tempered by other considera- Files-New is used when the user wishes to work a new
tions, for example, the possibility of leakage from the problem. A set of default data appears in subsequent
overlying clays and silts (which was not detected over a dialog boxes, which the user edits. Files-Open provides
mere 24-hour pumping period, but which is virtually the opportunity to load a preexisting data set saved from
assured over pumping periods measured in terms of an earlier run. The user selects the particular file-saved
years), and the possibility of some recharge to the aquifer with a *.wzfextension-and the appropriate data appear
over extended periods of pumping. Given that these in the “Input Menu” dialog box and subsequent dialog
other sources of water act to limit the drawdown, it can boxes (Figure 6.33). When a preexisting data set is
still be concluded that a water supply for more than one loaded, the first screen that is opened is the contour plot
unit is totally out of the question. The life of the water for the problem. To edit aquifer or well parameters for
supply for a single unit will depend on recharge, which a subsequent run, one would use the Edit menu. To return
may not be substantial in a desert environment, and the to the Input Menu or first screen, choose Files-New. In
rate of leakage from the overlying sediments. effect, if changes are required to the Input Menu, the
simulation data will likely require major changes, making
it a new trial.
- Computer-Based Calculations
6.7 Once the Input -Menu dialog box appears, one selects
the proper unit parameters, defines the size of the region
This section describes the operation and potential appli- over which drawdowns are to be contoured, selects the
cations of an interactive computer code, WELLz, for esti- particular aquifer type of interest, and defines the total
mating drawdown(@due to one or more pumping and/ simulation time (Figure 6.33a). For contouring, draw-
or injection wells. The present version of the code accom- downs are calculated at equally spaced data points across
modates calculations involving a fully penetrating well the region. Normally, users are advised to keep the num-
in a confined, leaky, or water-table aquifer. The equations ber of rows and columns in the calculational grid large.
for drawdown for a confined and a leaky confined aquifer The default 50, 50 is recommended, providing draw-
can be described by down calculations at 2500 separate points. In any case,
Q W(u) the number of rows and columns each cannot exceed
Confined Aquger:s = - 60. Note that the simulation region identifies a region in
4nT
infinite space where drawdowns are to be calculated and
r2S contoured. Defining this region has no impact on the
where u = - (6.51a)
4Tt calculation, as is the case with numerical model. Thus,
it is possible to define pumping/injection wells outside
of the region and their impacts may still be felt within
the sirnulation region. Similarly, drawdown may exist
where u =-r2S and - =
4Tt’ B
’ (’’~‘)’’*
- (6.51b) outside of the calculational region. The well coordinates
need to be defined so that all values are greater than zero.
The next dialog box serves to input the relevant aquifer
According to Jacob (1944), the drawdown s for an uncon- parameters. The particular dialog box that appears de-
fined aquifer can be expressed as pends upon the aquifer type. For example, Figure 6.33b
illustrates the dialog box for a confined aquifer. The next
s = m - (mZ - 2s’m)’I2 (6.52)
dialog box (Figure 6.33~)contains information on various
where s’ is the drawdown for a confined aquifer of the pumping and observation wells. One uses the mouse
same hydraulic properties, which is calculated from Eq. to add or edit well information. Observation wells are
6.52, and m is the thickness of the aquifer, before identified by a pumping rate of 0.0. After all of the neces-
pumping. sary parameters are provided through the dialog boxes,
The main difficulty in implementing these or similar a window provides the contoured drawdown over the
well-hydraulics solutions in the code comes in evaluating region of interest (Figure 6.33d). The drawdown is auto-
the well functions { W(u);and W(u,r / B ) }for the specified matically provided at the designated observation wells.
T, S, and Q. Fortunately, approximations are available to By selecting “Options- Label Contours,” one can use the
most of the frequently used functions in analytic ground- mouse to label the contour lines, or display drawdowns at
water equations. any designed point on the screen. Simply move the arrow
+
WELLz 1.O is programmed in C+ and runs under to the position where you wish the drawdown to be
Microsoft Windows. The code is provided on the disk shown and click the mouse. At this point, “File” options
 6.6 Some Applications to Hydraulic Problems 129

I JOB-TITLE 1 test 1

n
I
4. Pumping time 64

Transmissivity

Storat ivity

Figure 6.33 Example of the sequence of dialog boxes appearing in WELLz (Panels n.
6,and c) along with the printed drawdown map for the problem (Panel d ) .
130 Cbapter 6 / Hydraulic Testing:Models, Metbods, and Applications

Welt No. X Y Bumping Rate

2700

2400

2100

1800
I
1500.

1200 - \
900 '

600 -

300

0-
0 300 600 900 1200 1500 1800 2100 2400 2700 3000
X

(d)

Figure 633 Continued

 6 7 Computer-ksed Calculations 131

similar to those used previously are provided as follows:

New-rerun the simulation from the beginning
Open-open a previously saved data file with a
*.wzf extension
Save As-save the data file with a *.wzf extension
Print-print the contour map
Save Report-export the drawdowns to a file for plot-
ting, and create a report ofthe run to be stored in a file
Exit-quit the code
The number of contour lines and contour intervals
are either fixed or specified by the user by selecting
Options-Contour values. Contour lines are defined by
simply dividing the fixed minimum and maximum draw-
downs in the field into equal intervals to produce 10
contour lines. The export capabilities as File-Save Re-
port facilitate the creation of a more polished drawdown .v-distance (m)
map using other plotting packages. Provision is also in- Figure 6.34 Map showing the location o f wells within the
cluded with Export to save a summary report of the run calculation region.
in another file.

Code Demonstration 5. Add information on the pumping well and the obser-
vation well (Fig. 6 . 3 3 ~The
) . coordinates of the obser-
The following simple well-hydraulic problem illustrates vation well are (1804.8, 1500). Click on “Continue,”
how the code is used. and move to the final dialog box.
6 The drawdown due to the well is contoured in Figure
Example 410 6.33d. At the observation well. the dfitwdown is
A well that is pumped at a constant rate of 3790 m’/day indicated to be 1.6m. By selecting ”Option -Libel -
fully penetrates a confined artesian aquifer. The aquifer Contours,” one can label the contour lines and other
can be assumed infinite with a transmissivity and storativ- points of interest.
ity of 928 m2/day and 1.O X lo-’, respectively. Determine Z Finally, we save the data and exit the code.
the drawdown in an observation well located 304.8 m
from the pumping well at 64 days.
We solve the problem using WELLz, as follows. Bounded AquZfers Revisited
1. Select the WELLz icon from the Windows Programs WELLz and similar codes are particularly useful in iipplicit-
Manager by double clicking the mouse. tions to bounded aquifer systems, where the wells atid
2. Select “File-New,” because the data set does not image wells coupled with a relatively large number of
already exist. observation points make dnwdown calculations tedious.
Example 6.11 provides another, more realistic look at
3. Edit the information that appears in the dialog box image-well theory and the application of WELL2 to solve
as shown in Figure 6 . 3 3 ~The
~ . unit parameters for a problem complicated by a relatively large number of
this problem are all meters and days, as shown. The wells.
region is set up with the pumping well in the middle
having coordinates (1500, 1500) (see Fig. 6.34).The
calculations will be carried out within a region Example 6.11 (based on Walton, 1962)
3000 m by 3000 m. Note that as required by the Walton (1 962) illustrated the application of image-well
code all wells have coordinates greater than zero. For theory in the interpretation of an aquifer test by Mikels
this problem, “Theis confined aquifer” is selected. (1952) at Zion, Illinois. The test involved a single pump
Having completed editing the form, we move to the ing well (A-P), pumped at 99 gpm for 3180 minutes
next dialog box by clicking on “Continue.” and six observation wells (Figure 6.35).The unconfined
4. Edit values of Tand S on the next form (Fig. 6.3361, gravelly sand aquifer had a saturated thickness of approxi-
and continue. mately 20 ft. Walton (1962) interpreted the asymmetr).
132 Chapter 6 / Hydraulic Testing:Models, Methods, and Applications

in the cone of depression and similarities in water-level m

t Well Coordinates
responses in the wells and Lake Michigan as indications
of good hydraulic connection and a recharge boundary.
He estimated the hydraulic conductivity to be 387.7 ftl A-N-1 110 270
day the specific yield to be 0.01, and the distance to the A-E-1 151 175

effective line of recharge to be 206 ft.

Use WELLz to simulate the 3180-minute test at Zion.
See how well the simulated drawdowns match the ob-
served drawdowns shown in Figure 6.35.

1. Let us begin by formulating the problem. First, we

define a region over which the calculations are to
be conducted and a coordinate system for defining
the location of the wells (Fig. 6.36). The exact loca-
tion of the origin doesn’t matter except that the
s-distance (ft)
region should be located appropriately. The simula-
tion region we have selected stops at the recharge Pigure 6.36 Formulation of the problem for simulation.
boundary 306 ft from the origin. Nevertheless, the The table lists the coordinates of each of the wells (modified
from Walton, 1962, and Mikels, 1952).
image (injection) well is included as part of the input
data and its influence is felt within the simulation
domain.
2. We run the code as before, entering the appropriate
information. For those more curious, the data set is included as “Ex6- 1 1.wzf” on disk and can be loaded
through File-Open.
3. The simulation results at 3180 minutes are plotted in
Figure 6.37. Note how the presence of the recharge
boundary leads to reduced drawdowns between the
pumping well and Lake Michigan. The simulated
0 A-N-1 N
drawdowns at the observation wells match the ob-
served drawdowns very well.
@

(Pumped well)
A-P A-E-2
0 0 0
A-E-1 A-E-3

A-S- I

Legend

0
0.59 Drawdown (ftl

.A-S-2
0
-
Scale in feet
50

Figure 635 Map showing the location of the observation

100

wells for the aquifer test at Zion, IUiois. The drawdown in Figure 637 Output from WELLz of the calculated draw-
feet at each of the observation wells at 3180 minutes is cir- down due to pumping A-P.
cled (modified from Walton, 1962, and Mikels, 1952).
 Problems

Problems
1. After 24 hours of pumping a confined aquifer, the determined as 56%. What is the efficiency when the well
drawdown in an observation well at a distance of 97.5 is pumped at 5600 m'/day for a oneday pumping period?
m is 0.6 m and the drawdown in an observation well 8.
at a distance of 33.5 m is 1.1 m. The pumping rate is
5.43 X l o 4 m3/day. Find the transmissivity. a. List three reasons that might explain an upward in-
flection of a semilogarithmic time-drawdown plot.
2. Time-drawdown data collected at a distance of 30.5
b. List three reasons that might explain a downward
m from a well pumping at a rate of 5.43 X lo4 m3/day inflection of a semilogarithmic time -drawdown plot.
are as follows.
9. Suppose the only type curve available to you was ;I
t , min s, m plot of W(u) versus u. Now would you plot tirne-
1 1.2 drawdown data obtained from the field in order to use
2 1.6 this curve in the matching procedure?
3 1.9 10. A water well is located 300 m from a waste intercep-
4 2.1 tor trench that acts to intercept a plume emitting from
5 2.3 a uranium tailings pile. The well is to be pumped for 100
6 2.5 consecutive days at a rate of 5.43 X 10' m'/day. The
7 2.7 transmissivity of the formation is 497 m'/day and the
10 3.1 storativity is 0.1. At the end of the 100-day pumping
20 3.7 period, will the cone of depression reach the interceptor
40 4.3 trench and what will be the drawdown?
80 4.8
11. Use the figure in Example 6.7 to answer the fol-
100 5 .O
lowing:
300 5.8
500 6.2 a. A 60-cm-diameter well is located about 3 m from a
1000 6.6 fault (barrier) and is pumped at a rate of 5.43 X 10'
m4/dayfor a 10-day period. What is the approximate
Calculate the transmissivity and storativity. drawdown in the well at the end of the 10-dayperiod
3. A well 76 m deep is planned in an aquifer with a (ignore well losses).
transmissivity of 125 m2/day and a storativity of 1 X lo-'. b. What sort of shift would you expect in the curves
The well is expected to yield 2720 m3/day and will be given in Example 6.7 if the storativity was much
30 cm in diameter. If the nonpumping water level is 15 greater, say, 1 X lo-'?
m below land surface, estimate the depth to water after 12. A disposal well for liquid waste injection com-
one year's operation and after three years' operation. mences operation in a horizontal confined aquifer that
4. A 46-cm-diameter well within an aquifer with a trans- has the following characteristics: thickness = 9 m, poros-
missivity of 100 m*/day and a storativity of 7 x IO-' is ity = lo%,hydraulic conductivity = 81 m/day, specific
to be pumped continuously. What pumping rate should storage = 3.3 X m-',injection rate 5660 m'/day.
be used so that the maximum drawdown after two years The disposal well has a radius of 0.3 m
will not exceed 6 m?
a. To what distance from the well will the front of the
5. You are asked to design a pumping test for a confined cone of impression have extended after 10 days'
aquifer in which the transmissivity is estimated to be injection. (Assume, for all practical purposes, that
about 745 mz/day and the storativity about 1 X lo-'. 0.03 m of head above the regional preinjection piezo-
What pumping rate would you recommend for the test metric head marks the front of the cone of im-
if it is desired that there be a drdwdown of about 2 m pression.)
in the first five hours of the test at an observation well b. Approximately how far will the injected contaminant
30 m from the pumping well? move by the end of the 10-day injection period?
6. Replace the fault in Example 6.8 with a fully penetrat- Assume that the contaminant moves at the same
ing stream, and calculate the drawdown at the stream speed as the ground water and that the induced
and midpoint between the well and the stream for the hydraulic gradient due to the injection is imposed
same pumping rate and duration of pumping. instantaneously and is linear.
7. An aquifer has a transmissivity of 93 m'/day and a 13. Consider the enclosed figure to represent the
storativity of 1 X lo-'. A 6Ocm pumping well has a pumping history of a single well over a six-year period.
drawdown of 72 m at the end of one day's pumping This pumping caused some drawdown in a nearby obser-
at a rate of 2800 mj/day. The efficiency of the well is vation well as shown in the companion diagram. Explain
 Cbapter 6 / Hydraulic Testing:Models, Metbods, and Applications

very carefully how you would use superposition in this are required to compute the recovery in the well in
problem. response to the injection. Calculate the short-term recov-
Start out with “I would pump the pumping well at a rate ery in the injection well of radius 0.15 m at the end of
of - - for years to obtain a response; then 1 0.22 days of injection at a rate of 28 mi//day. (Hint: Figure
would pump it at a rate of -for _ _ years, and 6.25 is required to obtain the result.)
(add) (subtract) the results to or from the previous re-
sponse; then I would pump the pumping well at a rate
of- - for years, and (add) (subtract) the results
to or from the previous sum.” head boundary

Confining layer

I I
I I
I 1
injection 1 zone
I I

Problem 15
2 4 6
Time (years) Time [years)
Problem 23
16. During a field investigation of a leaky underground
storage tank for oil, a group of small-diameter observation
14. The accompanying diagram gives some pumping wells is emplaced. In the interest of interception of some
test results as measured in an observation well. You are of the oil product floating on the water table, the perford-
given the following: tions are placed both above and below the water table.
Later, it is decided that a value of the hydraulic conductiv-
ity of the saturated zone is necessary for further calcula-
tions. Explain how the Hvorslev test for a withdrawal
procedure would work. Use the data of Figure 6.19 to
obtain a value for the hydraulic conductivity as has been
done in Example 6.5, reinterpreting the datl for a with-
drawal as opposed to an injection test.
17.
a. Design a dewatering scheme for the following situd-
tion. A tunnel is to be constructed 17 m below land
surface. The water table is at a depth of 12 m, rind
Problem 24 the tunnel will reside 5 m below the water table. The
availdbk saturated thickness from the water table to
a clay layer is 12 m. The transmissivity of the material
transmissivity = 930 m*/day to be dewatered is 3730 m‘/day, and the storativity
is 0.01. The actual specific capacity of the well used
storativity = 1 X lo-’
to conduct the hydraulic testing has been measured
the distance from the observation well to the pumping at 48 m2/hr. The dewatering wells will be finished
well (rl) is 300 m 6 m below the tunnel, and they must be offset 5 ni
the time (fl) at which sI is measured is 0.1 day from the centerline of the tunnel. The wells will be
the time (fz) at which the drawdown s2 is measured is placed alternately on each side of the tunnel (see
10 days Figure 6.29). Consider four wells pumping simulta-
the drawdown sz is two times the drawdown sI neously over a l(Way period at a pumping rate of
270 mi/hr. Determine the well diameter and well
Find the distance from this observation well to the image spacing required to lower the water table by 6 m
well responsible for this behavior. along the tunnel alignment.
15. An injection test in a single well produced the b. Repeat the exercise for paired wells, that is, four
curve-fitting parameters for the transmissivity T, storativ- wells set in groups of two on the same line perpen-
ity S,vertical permeability k,,,and the storativity S’ for dicular to the tunnel alignment. Comment on the
the geometry shown. As no other facts are given, you “best” design in terms of well spacing.
 Problems 135

c. Your assignment is to type a report not to exceed possibility of obtaining 480 m'/hr for one unit for a 5-,
one page on your selected design. Include two fig- lo-, and 50-year period.
ures, one that shows your mathematical model such 19. A well was pumped at 30 gallons per minute. The
as depicted in Figure 6.28 and one that shows the drawdown response was characteristic of a leaky artesian
optimal design (Figure 6.29). aquifer. The transmissivity of the aquifer is 200 ft?/day,
18. Extend the boundaries of Figure 6.3 1 an additional the storativity is 0.0004,the thickness of the confining
5330 m on both sides and analyze this problem for the bed is 15 ft, and hydraulic conductivity of the confining
centermost well and the pumping demand of 480 m3/ bed is 0.015 ft/day. Calculate the drawdown ;it three
hr. The wells must remain in their current position; only observation wells using WELLz after 2000 minutes of
the boundaries have been moved. Consider the effect of pumping. The pumping well has coordinates (1 00 ft, 1 0 0
three real wells pumping and four image wells associated ft), and the coordinates for the observation wells are OBI
with each of the pumping wells. Include a sketch of (100, loo), OD2 (78, 5), and OB3 (340, 177).
your real and image-well location and comment on the
 CbaDter 7

Ground Water a s a Resource

7.1 Development of Ground-Water Resources

7.2 Introduction to Ground-Water Flow Simulation
7.3 Formulating a Finite-DknerenceEquation f o r Flow
7.4 The MODFLOWFamily of Codes
7.5 Case Study in the Application of MODFLOW

Thus far w e have learned about aquifers and how water abundant surface water, ground water is frequently an
contained within them can be developed by wells. This underexploited resource. Conversely, in areas without
chapter extends these ideas in addressing questions of surface-water supplies, ground water is almost always
how a ground-water resource as a whole is managed and overexploited. In areas between these extremes, the de-
some of the important tools and strategies. One cannot velopment of the total water resource depends on the
discuss issues of regional-scale aquifer development and demand for water. How the total water-resource system
management without talking about simulation models is developed and the manner in which it is operated
that are now at the core of most basin-wide assessments. depends not only o n availability of supply but on legal,
What has made aquifer simulation models indispensable political, and socioeconomic precedents and constraints.
for ground-water studies is the power to integrate the
complexities of hydrogeologic settings, hydrogeologic The Response of Aquifers to Pumping
processes, and water utilization.
Prior to the initiation of pumping, ground-water recharge
is balanced by the natural discharge of water to springs,
7. I Development of Ground-Water creeks, rivers, and lakes. Withdrawal of water by wells
is an additional stress put on the ground-water system
Resources that must be balanced by (1) an increase in recharge, (2)
A resource represents a supply of something that can be a decrease in natural discharge, (3) a loss of ground-
drawn upon for use. Like petroleum, water is a resource water storage, or ( 4 ) a combination of these factors.
that can be readily transported to achieve a better balance Theis characterized this pattern of change as a “state of
between the location of its supply and the demand for dynamic equilibrium” whereby a new balance between
its use. Unlike petroleum, ground water is not a minority inputs and outputs might be achieved at a new but lower
fluid in the subsurface environment and its value is not level of ground-water storage. A s the system adjustment
normally determined by the marketplace. However, often requires time, the basin can remain in a long-term
ground water can also be nonrenewable, at least when transient state, with water levels falling in response to
viewed within a human time frame, and its exploi- the pumping. Thus, in Theis’s words, ground water may
tation is subject to supply and demand. In areas with be classified as a renewable resource, but there are in-
 ZI Development of Ground-Water Resources 137

stances where it may not be so within a human time In addition, it is not possible to develop a safe yield in
frame. the absence of initial developmental overdraft, a neces-
sary first stage in ground-waterdevelopment where with-
drawals cause a lowering of the water table in areas of
Yield Analysis natural discharge and recharge. Further, there are sea-
Management of ground-water basins requires some kind sonal or cyclical overdrafts where water levels eventiially
of yield analysis to determine how much ground water return to their original levels during periods of limited
is available for pumping. Frequently, the management withdrawals. Thomas (195 1) and Kazmann (1 956) have
strategy is to obtain the maximum possible pumping suggested abandonment of the term because of its in-
compatible with the stability of the supply. The term definiteness. Freeze (1971) introduced a concept of maxi-
safe yield as an indicator of this maximum use rate has mum stable basin yield, determined from a tliree-dinien-
had an interesting evolution since first introduced by Lee sional saturated-unsaturated numerical model. Although
(1915). Lee defined safe yield as the concept is a good one, determining its value is not
a simple task.
The limit to the quantity of water which can be with-
drawn regularly and permanently without dangerous
depletion of the storage resenie.
Case Study: The Upper Los Angeks
This definition was expanded by Meinzer (1923), who River Area
defined safe yield as
There are ground-water basins, especially in the arid
The rate at which water can be withdruwn from an southwestern part of the United States, where a basin-
aquiferfor human use without depleting the supply to wide response to pumping on a massive scale is evident.
tbe extent that withdrawal at this rate is no longer A case in point is the Upper Los Angeles River area
economicallyfeasible. (ULARA) in southern California, which is comprised of
Thus, the “dangerous depletion” of Lee is described in one large ground-water basin, the San Fernando Basin,
economic terms by Meinzer, and both speak of pernia- and three other smaller basins- the Sylmar Basin, the
nency of withdrawals. Verdugo Basin, and the Eagle Rock Basin (Figure 7.1).
As a philosophical concept, the definition of Lee is a The San Fernando Valley area includes the cities of Los
good one. However, no guidance is provided on how Angeles, Burbank, and Glendale and is home to various
this rdte or rates may be determined and the concept Hollywood studios and industrial companies such as.
remains far too ambiguous for practical use. Conkling Lockheed, Rockwell, and 3M.
(1946) attempted to make the concept less ambiguous The ULARA is bounded by various mountain cinges
by specifying the conditions that constitute a safe yield. and hills (Figure 7.1). Ground water is extracted by wells
He described safe yield as an annual extraction of water from the thick valley-fill aquifer across much of the S;m
that does not: Fernando Valley. While the valleytill aquifer is lo~all!~
productive, rates of recharge are relatively low due
1. Exceed average annual recharge. mainly to the arid climate and seasonal nature ofprecipi-
2. Lower the water table so that the permissible cost tation. Over the past 100 years. the valley floor received
of pumping is exceeded. an average of about 14 inches (35.6 cm) of rainfall, most
of which falls from December to March (ULARA W,i-
3 Lower the water table so as to permit intrusion of
water of undesirable quality. termaster Report, 1994).
As the population of the San Fernando Valley increased
A fourth condition, the protection of existing water from the 1930s onward, it became inevitable that the
rights, was added by Banks (1953). v.~lley-fill
aquifer could not provide an inexhaustible sup-
The single-valued concept of safe yield as proposed by ply of water. Lost initially was the natural discharge to
Conkling and modified by Banks encompasses hydro- the rivers and a loss of storage in the ground-watersystem
logic, economic, quality, and legal considerations. This itself. Shown in Figure 7.2 are two observation wells
overspecification of the term is not likely what Lee or (3700A, located at 2; and 3914H, located at 5 ) . The hy-
Meinzer had in mind. The controversial nature of the drographs for these wells depict the historical decline of
concept as defined by Conkling is clearly demonstrated water levels in the valley-fill aquifer from the late 1930s
in 43 pages of discussion of his original 28-page paper on to 1968. As suggested by the hydrographs, water-level
by no fewer than 10 authorities. In practice, safe yield declines of the order of 100 to 200 feet (30.4 to 60.8 m)
has no unique or constant value, its value at any time were not uncommon.
depending on the spacing and location of the wells and Ground-water withdrawals in the basin were limited
their influence on the dynamics of interchange between by court actions in 1968 to approximately 104,000 acre-
ground water and other elements of the hydrologic cycle. feet per water year (LJLARA Watermaster Report, 1994).
-
138 Chapter 7 / Ground Water as a Resource

Figwe Z 1 Location of ground-waterbasins of the Upper Los Angeles River Area. Shown in circles are the location of ohserva.
tion wells referred to in the text (modified from Watermaster Report, 1994).

Water Surface Elevation Well 3 7 0 0 A

I City of Los Angeles G.S.E.: 743.0 ft.
720

W 620 -
3 600 -
v)
580 -

-
1 1 1 1 1 1 1 i i 1 1 1 i i 1 1 i 1 1 1 1 1 1 1 1 i 1 1 i 1 i 1 ' 1 1 1 1 1 1 1 1 1 1 1 1 i 1 i 1 1 1 1 1 i 1 1 1 1 1 i 1 1 1
X560
3 0
r?
I

8
0
P
I

8
0
r
8
0
9 I

8
e8 L
0
co
c
8
5
B
Water Surface Elevation Well 3914H
=: City of Los Angeles G.S.E.: 440.0 ft.
g 440

Figure 7.2 Hydrographs of two wells located in the San Fernando Basin (modified from the Watermaster Report, 1094).
 ZI Developnient of Ground-Water Resources

This rate of withdrawal was about two thirds of the and imported sources; and 4. protection m i l planned
average of the previous six years. This decrease is re- maintenance of ground-water quali@.
flected in the behavior of the hydrographs, which subse-
quently leveled out or recovered (Figure 7.2). Overall, the As this definition implies, the most important element in
basin is managed to keep total ground-water withdrawals any water-management scheme is a well-defined limit on
within what is referred to as the safe yield of the basin. the quantity of water that can be pumped. In its simplest
Safe-yield concepts were applied to the San Fernando form. this limit may be related to the safe yield of the
Basin to help allocate the available water. The safe yield basin as determined by the quantity of recharge from
of the basin was estimated to be 43,(50 acre-feet per natural sources. However, most management schemes
year. The entire amount is available to the City of Los are more complicated given sources of water other than
Angeles for extraction. This yield represents the long- ground water and the possibilities for various recharge
term average recharge from precipitation. Some of the schemes. Thus, in the definition above, item 1 makes
water that is imported to the basin from other sources reference to artificial recharge, whereas item 3 refers to
ends up as additional sources of recharge to the valley- conjunctive use. The use of artificial recharge implies
fill aquifer. This water is available as well water for pump- that both surface water and ground water are being used
ing by the cities of Los Angeles, Burbank, and Glendale. conjunctively, although all conjunctive-use systems do
In the 1993-94 water year, some 43,900 acre-feet of not rely on artificial recharge as the means of augmenting
R ground-water supply.
this iniported water was available for pumping (ULARA
Watermaster Report, 1994). Across the San Fernando Ba-
sin actually more water ends up in the ground from Arti@ciuC Recharge
recharge of imported water than from recharge due to Artificial recharge is defined by Todd (1980) as aug-
natural precipitation on the basin. menting the natural infiltration of precipitation or surface
The management scheme for the ULARA also provides water into tlie ground by some method of construction,
the capabilities for Los Angeles, Glendale, and Burbank spreading of water, or a change in natural conditions.
to store water and extract equivalent amounts in the Artificial recharge is often used to ( I ) replenish depleted
future. In effect, in years when the pumping of the cities supplies, (2) prevent or retard saltwater intrusion, or ( 3 )
is less than their allocated amount, the water is deemed store water underground where surface-storage facilities
to be saved and is available for hiture use in years of are inadequate to supply seasonal demands. California
high demand. has long been the leader in artificial-recharge operations,
Historically, water-management activities like those in generally for the reasons cited in ( I and (2) and, iii more
the ULARA relied on water-balance calculations for broad, recent years, for reason (3).
basin-scale estimates of inflow and outflow. In the future, Several recharge schemes have been developed, in-
there will be more reliance on calibrated mathematical cluding recharge pits, ponds, wells, and water-spreading
simulation models at the basin scale that more rigorously grounds (Figure 7.3). Literature in this field is voluniinous
account for tlie inherent variability in material properties. and includes the publications of the Ground Water Re-
recharge rates, and ground-water withdrawals. The topic Chdrg Center in California, which deal primarily with
of numerical simulation is taken up in Section 7.2. infiltration and water spreading (Schiff, 1955; Behnke and
Bianchi, 1965); the extensive work of Raumann, dealing
primarily with the theoretical aspects of recharge
Management Strategies through wells (1963; 1965); the experience 011 Long Is-
land (Brashears, 1946;Johnson 1948; Cohen and others.
The previous discussions provide the context for why
1968;Seaburn, 1970);and comprehensive annotated bib-
ground-water resources require management and some
liographies prepared by Todd for the United States Geo-
of the tools for yield analysis to establish quantities of
logical Survey (1 959a) and Signor and others (1 970).The
water available for pumping. This section explores in
potential of ground-water basins as storage facilities is
more detail some of the issues related to the management
demonstrated in the San Joaqiiin Valley, where the
of ground-water basins. The California State Department
ground-water storage capacity has been estimated to be
of Water Resources (1 980) describes “ground water ba-
nine times the capacity of surface-water reservoirs associ-
sin management” in the following terms:
ated with the California Water Plan (Davis and others.
Ground zuuter busin winitagemerit includes plunned 195%
use of the ground wuter basin yield, stcwuge space, There are at least five interrelated questions associated
transmission capability, and water in storage. It in- with artificial recharge operations:
dudes I . protection of MUtl4I’Ul recharge mid use of
1. What is tlie nature of the rechargeable water mirce?
rrrt@ciulrecburge;2.plunned tiariution in amount and
location of pupnpjizg oper time; -3. trse ofgrotdnd uJ&er* 2* Of recharge be llsed?
storage conjunctively with sutface uwter froin local 3. What are the expected injection rates?
140 Cbapter 7 / Ground Water as a Resource

Q
Recharge
- - - _basin
-
Mound during

Initial water table

, , "iI' initial water
table

((1) (b)

Figure 7.3 Schematic illustrdtion of recharge through ba-

sins and wells.

4. How will the system respond hydraulically to in-

jection?
5. How will the injected water be managed as part of Figure 7.4 Flow diagram illustrating ways proposed for the
the total water-resource system? use of reclaimed water from the Tillman Plant (modified
from Watermaster Report, 1994).
There is considerable interdependency between the first
two of these questions. For example, with water-spread-
ing methods, water is generally diverted directly from
surface-water sources to topographically lower areas. On total organic carbon, biological oxygen demand, nitrite,
Long Island, New York, abandoned gravel pits are used ammonia, and turbidity levels.
to collect storm runoff that previously discharged to the The most important design issues concerned with arti-
ocean. Moreover, in this same region, more than 1000 ficial recharge are developing and maintaining appropri-
recharge wells are in operation in response to legislation ately high inflow rates (Todd, 1959b). Table 7.1 gives
that requires direct recycling of ground water pumped some representative spreading basin recharge rates and
for air-conditioning and industrial cooling purposes. In some average well recharge rates. Recharge rates associ-
the Santa Clara River Valley, flood waters are stored in ated with spreading grounds vary from 15 m3/m2/day
reservoirs and later released at low rates so as to enhance (m/day) for some gravels to as little as 0.5 m/day in sand
their infiltration into the natural streambed. Alternately, and silt (Bear, 1979). Typically, the infiltration rates fall
municipal waste water subjected to secondary treatment off with time due to swelling of the soil after wetting
is generally recharged by irrigation or spreading methods (Figure 7.5). There is some threshold value of infiltration
(Todd, 1980). Recharge wells for waste water must be where continued recharge is no longer economical so
accompanied by tertiary treatment. that the spreading ground is either abandoned or some
Within the San Fernando Basin, the Watermaster Re- scraping and cleaning process must be put into opera-
port (1994) outlines plans to enhance the water available tion. Decreases in well recharge rates can occur in re-
for pumping by artificial recharge schemes. The pro- sponse to silt introduced in the recharge water, dissolved
posed East Valley Water Reclamation Project involves the air in the recharge water, which tends to decrease the
use of up to 40 million gallons per day of treated sewage permeability in the vicinity of the screen openings, and
from the Donald C. Tillman Water Reclamation Plant bacteria-induced chemical growth on the well screen.
to provide water for various purposes (Figure 7.4). A
significant proportion of this water will be used for
ground-water recharge via two spreading grounds.
Pilot studies have been under way in the San Fernando
Basin to examine the water-quality implications of re-
charging treated sewage. At the Headworks Spreading
Ground (see Figure 7.1), water from the Los Angeles
River, whose low flow is mainly treated reclaimed water
from the Tillman Plant, was spread over an area of ap-
proximately 30 acres. The spread water was removed
from the ground by a well 1000 feet away. The water
I I I I
removed by pumping was approximately 45% reclaimed 0'
0 20 40 60 80
water and 55% native ground water. Preliminary indica- Time (days)
tions are that transport through the ground improved Figure 7.5 Time variation of recharge rate for water
the water quality. Reductions were apparent in coliforms, spreading on undisturbed soil (from Muckel, 1959).
 Next Page

7.1 Development of Ground-Water Resources 141

Table 7.1 Representative Artificial Recharge Rates

Spreading Basins Recharge Wells

Location Rate (m/day) Location Rate (mj/day)
California California
Los Angeles 0.7-1.9 Fresno 500- 2200
Madera 0.3-1.2 Los Angeles 2900
San Gabriel River 0.6- 1.6 Manhattan Beach 1000- 2400
San Joaquin Valley 0.1-0.5 Orange Cove 1700- 2200
Santa Ana River 0.5-2.9 San Fernando Valley 700
Santa Clara Valley 0.4-2.2 Tulare County 300
Tulare County 0.1
Ventura County 0.4-0.5
New Jersey New Jersey
East Orange 0.1 Newark 1500
Princeton <0. 1
New York Texas
Long Island 0.2-0.9 El Paso 5600
High Plains 700- 2700
Iowa New York
Des Moines 0.5 Long Island 500 - 5400
Washington Florida
Richland 2.3 0rIando 500- 5 1,000
Massachusetts Idaho
Newton 1.3 Mud Lake 500- 2400

From Todd (1959b). Reprinted from Groiriiduwler Hydrology. Copyright 0 1959. John Wiley & Sons.
Inc. Reprintd by permission of John Wiley Sr Sons, Inc.

Conjunctive Use Los Angeles. This study proposes full use of underground
supplies and storage capacity together with local and
Conjunctive use involves the coordinated use of surface
imported supplies where artificial recharge plays a sizable
and ground water to meet some specified water demand
role in the project. The plan of basin operation is to
in a given area. The following questions from Buras
provide water to the consumer at the lowest possible
(1 966) provide insight into the questions associated with
cost. Yet a third type of operation is where surface water
conjunctive use:
is imported and is used directly to supplement the
1. What system has to be built to minimize the discrep- ground-water supply with little or no connection be-
ancy (in time, space, and quality) between the natural tween the two sources, other than their availability.
supply of water and the demand for it? Cochran (1968) describes such a situation in Nevada,
2. To what extent should the water-resource system be where the objective was to minimize the cost of opera-
developed, and how extensive should the region tion. For this case, Domenico (1972) derived a mathemati-
serviced be? cal decision rule for the timing of surface-water importa-
tion to supplement an overdeveloped ground-water
3. How should the system be operated so as to achieve supply. As a marginal value rule, this decision rule states
a given set of objectives in the best possible way?
that the importation should take place when the cost
Conjunctive operations may be of various kinds. The of importation equals the cost of mining. The cost of
most common type that is of interest because of the importation includes the initial investment plus operating
hydraulics is the interconnected stream-aquifer system, costs, whereas the cost of mining includes current pump-
where the development of one affects the other. This is ing charges as well as the capitalized cost of all future
the type of system addressed by Bredehoeft and Young pumping charges associated with a lowered ground-
(1970) and Young and Bredehoeft (1972). An overall water storage level.
objective of such studies might be to use the total re- In recent years, the concept of conjunctive use has
source in such a way as to maximize benefits. Here, the been extended to issues of riparian protection. In the
hydraulic connection between the two sources plays a arid southwestern states, plants and animals thrive in
major role. Yet another type of operation has been de- fragile ecosystems developed along the perennial
scribed by Chun and others (1964) for the coastal plain of streams. These systems are particularly at risk when the
Previous Page

-
142 Cbapter 7 / Ground Water as a Resource

overdevelopment of ground-water resources lowers wa-

ter tables in the riparian zones, or results in significant
water-table fluctuations (Arizona Department of Water Ground-Water
Resources, 1994). The Santa Cniz River, near Tucson, Flow Simulation
Arizona, is one of several rivers where changes in the
riparian ecosystems have raised concerns. Along the Conceptual Model Numerical Model Results
lower reaches of this river, the extensive pumping of -hydrogeologic framework -calibration criteria
-boundary conditions met?
ground water for irrigation and public supply wells have -stresses
eliminated riparian vegetation. Lower ground-water lev-
els along the lower Santa Cruz were a major cause of the
destruction of mesquite woodlands (Arizona Department
of Water Resources, 1094).The State of Arizona is pursu-
ing active programs to maintain and improve the ecologi-
cal health of riparian ecosystems through the manage-
ment of ground-water and surface-water resources.
S New Data Needs

Ffgure 7.6 Synthesis of the modeling process.

L
7.2 Introduction to Ground-Water synthesized on the shape, thickness, and hydraulic prop-
E w Simulation erties of the major geologic units (for example, hydraulic
conductivity and storativity), the distribution of hydraulic
For more than 50 years, the quantitative orientation of head within the major hydrogeologic units, and the distri-
hydrogeology has differentiated this field from other geo- bution and rates of ground-water recharge.
sciences. A fundamental tenet holds that properly de- Other elements of the conceptual model are the bound-
signed and constructed mathematical models can form ary conditions and the distribution of stresses on the
the basis for accurate future predictions. Filndamentdl aquifer due to pumping. BOiinddry conditions are defined
beliefs in the inherent usefulness of models have estab- along the edges of the simulation domain, including the
lished the direction for academic research, as well as the top and bottom. Their main function is to separate the
nature of industrial practice. Numerical approaches let model region from the rest of the world. In other words,
hydrogeologists leave behind the analytical approaches the boundary conditions account for the influence of
of well hydraulics and their limitations. They provide flow conditions outside of the simulation domain. For
tools to treat hydrologic systems as they occur in na- example, in the construction of flow nets, we saw how
ture-where transmissive properties of aquifers are not boundary conditions influenced the resulting pattern of
constant, where recharge is spatially distributed, and flow. Bounddry conditions, thus, are required for the
where many wells can operate at the same time with solution of any ground-water flow equation. In general,
differing rates. there are two commonly used boundary conditions: (1)
specified-hydraulic-head boundaries and ( 2 ) specified-
Generalized Modeling Approach flow boundaries. A no-flow boundary is a special case of
a specified-flow boundary, and a constant-head boundary
A variety of different tasks are required in the construc- is a special case of a specified-head boundary.
tion of a ground-water model. As Figure 7.6 illustrates, Boundary conditions are often difficult to define along
the main steps include: the development of a conceptual the edges of the domain because hydraulic heads or in-
model, creation of the model and execution of various flow-outflow rdtes can be poorly defined. One strategy
trials, evaluation of the model results, and either the to overcome this problem is to place model boundaries
compilation of new data or validation testing depending along “natural” hydrogeologic boundaries, or parallel to
on the success of the calibration. Once a model is success- pathlines. An example of a natural hydrogeologic bound-
fully validated, it can be used predictively. ary is a river defined by constant hydraulic-head values
equivalent to the stage of the river. Locally, such a bound-
ary condition would provide a zone of recharge or dis-
Conceptual Model
charge. Another example would be the selection of the
A conceptual model is created through the evaluation major watershed divide as a no-flow boundary. The
of the hydrogeologic data to provide a picture of the boundary is no-flow because the gradient is such that flow
hydrogeologic setting over some region of interest. One moves away from a divide. Features of the hydrogeologic
element of the conceptual model is the hydrogeologic setting also help in assigning boundaries. For example,
framework, which encompasses the main features of the a thick, low hydraulic conductivity unit is selected as the
hydrogeology normally determined from geologic investi- bottom of the simulation domain. At some depth in al-
gations and hydraulic testing. Normally, information is most any system, it should be possible to define a bound-
 7.2 Introdudion to Ground-Water Flow Sinrrilntion 143

ary of no flow, thus implying that deeper circulation

is niinini;il.
Defining boundary conditions ;dong tlie top of tlie
simulation region is more difficult bec:iuse often there
are active inflows and outflows of water due to recharge
and discharge. Experience shows that recharge-
discharge rates vary in time ;ind space ;ind are difficult
hydrogeologic parameters to measure. An example is
provided later that illustrates how boundary conditions
are defined as part of a model conceptualization.
With transient models (that is, models where hydraulic
head varies as a function of time), one also needs to
define tlie initial condition for the simulation. The initial
condition is the hydraulic-head distribution at time zero.
In other words, it reflects the state of the system before
the simulation begins.

Ground-Water Flow Simulation

The response of an aquifer system is calculated by the
numerical solution to a ground-water flow equation. The
term tzurnericul implies that numerical methods, as op-
posed to analytical mathematical approaches, are used
to solve the governing partialdifferential equation. In
tlie ground-water flow equation, hydraulic head is the
unknown. The main numerical approaches used in prac-
tice today for solving ground-water flow equations are
finite-difference and finite-elenients methods. Both are
sophisticated methods that in different ways replace the
governing differential equation for ground-water flow by
a system of algebraic equations.
Both methods require that the region of interest be
subdivided using a mesh or grid network. Sometimes,
the cumbersome term discretizutiorz is used to describe
this process. The finitedifference approaches use a regu-
lar discretization, where an aquifer is subdivided into a
series of rectangular grid blocks. In a two-dimensional
model, each model cell is assumed to have a thickness,
tn. T l i each
~ of the grid blocks represents a volume of
tlie aquifer, rn Ax Ay. I n a three-dimensional model,
consisting of aquifers and aquitards, individual units are
subdivided vertically into cells of ;I specified thickness.
Operationally, the size of a grid block in tlie xg! plane
(that is, a map view) is usually kept small relative to the
overall extent of the aquifer. The spacings between the ((.)
rows and columns vary, but for simplicity the example Figure 7.7 Examples of (CI) mesh-centered and (0)block-
grids (Figure 7.7u,0)assunie a constant spacing (that is, centered finitedifference grids. Panel (c) illustrates the dis-
Ax = Ay). Associated with the grid blocks are nodes that cretiz;ition of an irregularly shaped aquifer with linear trim-
represent the points where the unknown hydriiulic head g1Il;lr rlemrnts.
is calculated. Depending on the formulation o f the nunier-
ical model, node points can be either mesh centered
(Figure 7.761)or block centered (Figure 7.76). and to ensure that node points coincide with nionitoring
The finite-element method permits ;i much more gen- wells or with geographic features (for eximple, ;I river).
em1 armngenient o f node points. The discretization with In this and following sections, we develop preliminary
triangu1;v elenients (Figure 7 . 7 ~illustrates
) how easy it concepts of finite-difference modeling i n relation to the
is to clefine tlie bounckiries of irregularly s1i;iped aquifers industry-standard code, MODFLOW (McDonald and Har-
- Cbapter 7 / Ground Water as a Resource
144

baugh, 1988). We will leave the detailed discussion of draulic-head data are given as
modeling methods to other, more specialized texts like
Zheng and Bennett (1995) and Huyakorn and Pinder Mean Error: -1 2(h,
’I
- h,),
(1983). n i=I

One of the important features of the numerical ap- 1

proaches is that hydraulic heads are calculated only at Mean Absolute Error: - 2 I@,), - hJi1
n l=I
I’

the node points. This aspect of the formulation will be-

come clear in Section 7.3, which examines the develop-
ment of the finitedifference form of the ground-water Root-MeanSquared Error: [2 n i=I
- hJ;]
u.5

flow equation.
Applying a model requires the construction of a data where n is the number of points where comparisons are
set that replicates the most important features of the made, h, is the measured hydraulic head at some point
actual system. The basic data for this purpose come from I, and h, is simulated hydraulic head at the same point
the conceptual model. Exactly how the data are orga- (Anderson and Woessner, 1992). Of the three error esti-
nized depends on the particular model. Generally, every mates, Anderson and Woessner (1992) point to the root-
node or cell must be supplied with information on hy- mean squared error as the best quantitative measure if
draulic conductivity or transmissivity, storativity, and the errors are normally distributed. The mean error is
fluxes due to sources and sinks (that is, recharge, pump- not preferred because large positive and negative errors
ing, and evaporation). The model also needs to be sup- can cancel each other out. A small error estimate, thus,
plied with information on boundary and initial condi- may hide a poor model calibration. Other requirements
tions. are sometimes applied in addition to error estimates, such
as quantitatively correct flow directions and flow gra-
dients.
Evaluation of Model Results
The hydraulic-head prediction that comes from a flow Model Verification
model is commonly used as the basis for model calibra-
tion. Calibration is a process of selecting model parame- Once calibration is complete, a verification test is com-
ters to achieve a good match between the predicted and monly added to check that the model is a valid representa-
measured hydraulic heads, or other relevant hydrogeo- tion of the hydrogeologic system. Commonly,model veri-
logic data like streamflow changes between gauging sta- fication involves using the calibrated model to simulate
tions. Most commonly, calibration is accomplished by a hydrologic response that is known. For example, one
a trial-and-error adjustment of model parameters. This might hold back results from one or more large-scale
procedure involves the systematic variation of model pa- aquifer tests and examine how well the calibrated model
rameters like hydraulic conductivity, storativity, flows, simulates the test. Again, the errors between the ob-
or boundary conditions. Automated inverse procedures, served and simulated hydraulic-head values can be quanti-
like those in MODFLOWP (Hill, 1992) may speed up fied in terms of the error measures. If the model success-
calibration. Calibration or model tuning is required be- fully passes this last test, then it can be used for
cause ground-water systems are so poorly known. predictive analyses.
Each model run provides a set of predictions about
hydraulic-head distributions or other flow rates that can
A Note of Caution
be compared to existing field measurements. Commonly,
the comparison is not favorable-suggestingproblems in Predictions made with simulation models must be inter-
the system conceptualization or inappropriate parameter preted with caution. The “aura of correctness” (Brede-
values. The lack of calibration may force a reexamination hoeft and Konikow, 1993) attached to model calculations
of how the model is constructed or in some cases prompt often exerts much more influence than is reasonable,
additional datacollection studies. After many iterations given the typically uncertain data on which models are
around the circle in Figure 7.6, the predictions of hydrau- built. Oreskes and others (1994) believe that “models
lic heads and various flow rates should match the mea- are representations, useful for guiding further study but
sured values. not susceptible to proof.”
Common practice is to set calibration criteria in ad- Predictability becomes a problem because ground-
vance of the calibration exercise. There are no hard and water systems are often so poorly characterized. It is
fast rules for what constitutes a good calibration except usually unrealistic to find data sufficient to describe hy-
that errors should be small relative to the total hydraulic drologic processes in space and time. Thus, the model
head. Criteria may be based on (1) the mean error, (2) design depends significantlyon the “informed judgment”
the mean absolute error, or (3) the root-mean squared of its builder rather than real information. This uncer-
error (Anderson and Woessner, 1992). These measures tainty does not disappear simply because a model is con-
of the difference between measured and simulated hy- structed.
 Z3 Fomulating a Finite-DifferenceEquationf o r Flow

The calibration-verification process does not lead to a difference method, the governing differential equation
unique description of a hydrogeologic system. For poorly for ground-water flow is replaced by a difference equa-
known systems, a very large number of different models tion that embodies conservation principles of the origi-
can be developed without knowing which, if any, is nal differential equation. Three-dimensional flow is de-
correct. Stated another way, different model developers, scribed by the following equation:
given the same hydrogeologic data, will probably develop
different conceptualizations of the same system, each of
which can be calibrated and verified. An example of
the difficulty of calibrating a ground-water flow model is
A
&Y
(K.m g) + $!(K,:,,
j z) + (Krz g) - W = Ss i)t
ah
(7.1)
discussed by Freyberg (1 988). Different groups using the where Kx.,.,K.I:r,
and Krr are values of hydraulic conductiv-
same set of synthetic data developed very different pre- ity along the x, y , and z coordinate axes, h is hydraulic
dictions concerning system behavior. Sources of variabil- head; W is a flux term that accounts for pumping, re-
ity in the calibration and the resulting prediction were
charge, or other sources and sinks; S, is the specific stor-
related to (1) the lack of hydrogeologic data, ( 2 ) the use
age; x, y , and z are coordinate directions; and t is time.
of different measures of success in calibration by each The form of the equation implies that the principal direc-
group, and (3) differing strategies for Calibration (for tions of the hydraulic conductivity ellipse coincide with
example, changing local transmissivity values around the coordinate axes.
wells versus changing values over large areas).
The success in model predictions can be examined
with post-audits. This term describes the process of Description of the Finite-mnerence Grid
checking a prediction made by a ground-watermodel. For
The finite-difference solution requires that the domain
example, if one makes a prediction about the behavior of
be discretized by a grid. With MODFLOW, the cells are
a system 10 years from now, one could return after 10
brick shaped (Figure 7.8). The grid system is referenced
years, make the necessary hydraulic-headmeasurements,
in terms of a row, column, and layer-numbering scheme
and check whether the original model can be validated.
with blockcentered nodes (Figure 7.8). As the notation
Bredehoeft and Konikow (1993) commenting on results
implies (Figure 7.8), the dimensions of each cell can be
from the few available post-audits indicate that “extrapo-
varied. Thus, a dense system of nodes can be provided
lations into the future were rarely very accurate.” They
around features of interest and a sparse system in areas
identified the following problems with the models: “the of lesser concern. Overall, the variable grid minimizes
period of history match (that is, calibration) was too short
the number of nodes in a simulation.However, care must
to capture an important element of the model, or the
conceptual model was incomplete, or the parameters
were not well defined.” They concluded that the record
of “validating” models was “not encouraging.” Colunllls ( j }
If model predictions are suspect, what use are they? 1 2 3 4 5 6 7 8 ’ 3
In addressing this issue, Oreskes and others (1994)
pointed out:
models can corroborate a hypothesis by oflering eili-
dence to strengthen what may be already partly estab-
lished tbrougb other means, Models can elucidate dis-
crepancies in other models. Models can also be used
for sensitivity analysis- for exploring “what if” ques- 1

tions-thereby illuminating which aspects of the sys- 3 2

I

tem are most in need offurther study, and where more E 3

?*
empiricul data are meded. f4

In summary, one must use the power provided by com- 5

4-’
puter models carefully.These are useful tools that should q
be used with full knowledge of their limitations. Explanation
Node
As Dimension of cell along the row direction Subscript (1)indicates
the number of the column
7.3 Formulating a Finite-mnerence Ac, Dimension of cell along the column direction Subscript (I) indicates the
Tuationfor F ~ W number of the row
A I Dimension
~ ~ of the cell along the vertical direction Subscript Uz)indicates
This section develops finite-differenceequations for aqui- the number of the layer
fer simulations. The theory here relates specifically to Figure 7.8 Discretization of a three-dimensional system
the industry-standard code, MODFLOW. In the finite- (modified from McDonald and Harbaugh, 1788).
146 Cbapter 7 / Ground Water as a Resource

cally as

(7.2)

where Qi’s account for flow into the cell from adjacent
cells (for example, see Figure 7.10) through six sides,
and for water added or withdrawn (for example, recharge
(4Aquifer cross section
pumping), S, is the specific storage, Ah is the change in
head over a time interval At, and AV is the volume of
the cell. The LHS of Eq. 7.2 can be expanded in terms
Grid Layer 1 of flows through the six sides of the cell, the sources
-cc) and sinks, and the RHS in terms of the cell dimensions
Grid Layer 2
Cell contains material to give
from only one
Grid Layer 3 stratigraphic unit. Qi,i- I/?.k + QIJ+112.k + QI-112,j . k Qi+ 112.j . k + Q1.j. k - 112
Faces are not reclaiigles (7.3)
W Aquifer cross section with Ahi.1,k
deformed grid superimposed i- Qt.j,iz+ 112 + QSi./.k = ssi,l,c 7 hrj hcl A V k
Figure 7.9 One possible approach for representing where Q,,,- is the volumetric fluid discharge through
(a) complex layering by (b) a deformed grid (modified from
the face between cells I , j, k; and i , j - 1, k , QSl,l,kis
McDonald and Harhaugh, 1988).
the sum of all other inflows to and outflows from the
cell; and A t , , Auk, At-, are the dimensions of the cell I,
j, k (Figure 7 .lo).
be taken to change cell sizes gradually. The rule of thumb Inflow through the cell faces can be expanded with
is that the dimensions of adjacent cells in a given direction Darcy’s equation as
should not differ by more than a factor of 1.5.
The cell sizes in the row and column directions are
specified explicitly in the input data. The vertical dimen-
sion is specified implicitly by specifying trdnsmissivity where hi,,,kis the hydraulic head at node i , j ,8; h i , , - l . k is
values for individual cells. McDonald and Harbaugh the hydraulic head at node i,j - 1, k; KRf,j-l/z,kis the
(1988) discuss the variety of ways of representing the hydraulic conductivity along the row between nodes
vertical layering in relation to the hydrogeologic units. i , j , k and i , j - 1, k; Aci Auk is the area of the cell face
It is beyond the scope of this brief overview to discuss normal to the row direction; and Atj-.l,2 is the distance
all these options. The most common strategy has layers between nodes I,j, k and i,J - 1, k (McDonald and Har-
conform with readily identifiable geologic units. Consider baugh, 1988). The minus sign is neglected to simplify
the simple example depicted in Figure 7.9a with a coarse the development. The form of this and similar equations
sand, a silt, and a sand and gravel layer. Vertical discretiza- for the other faces can be simplified by defining a “hy-
tion is accomplished using three model layers that coin- draulic conductance,” in this case CR where
cide with the three stratigraphic units (Figure 7.90).The
changing vertical thickness of individual cells, for exam-
ple within layer 1 would be accommodated by changing
the transmissivity of each cell.
With the model layers so defined, individual cells are
no longer bricks. They have an irregular slyape where Cell Isi-1, k Cell f , i,k
the cell faces are no longer rectangles. Although this
representation of a cell gives rise to errors, they are usu-
ally insignificant in relation to errors caused by poor
estimates of transmissivity,storativity, and recharge rates.

Deriuation of the Finite-Dflerence Equation

The development of the finitedifference flow equation 1

Arrl A’i
is based on the same ideas of continuity that were devel- , ,
oped in Chapter 4. In words, the equation of continuity *rj-1/2
states that the sum of flows into and out of any cell is Figure 7.20 Example of inflow through the face of cell
equal to the time rate of storage plus or minus additions i,j , k from cell f,j- 1, k. The dimensions of cell i , j ,k are
of water from sources or sinks. It is written mathemati- Ac,, Ar,, and Auk (from McDonald and Harblugh, 1988).
 7.4 The MODFLOW Family of Codes 147

Making this substitution, the six inflow equations can be where

written as
Qi.1-112.k = CRi.j-t/z.k(bt.j-i.k - b1.j.k) (7.6)
Qi.j+in.k = cRi,j+iiz.k (ht.j+i,r - h t . j . k ) (7.7)
Qi-l12,j.k =C~~-~/~.].k(bi-i.j,k-hl,j.k) (7.8)
Qi+l/z,j,k = cci,t l / 2 , j , k ( b l + l , j . k - ht.j,k) (7.9) SCI,I 1, = .y,, I k b,ACl A V k

Q/.j,k-1/2 - hi.j.k)
= C&.j.k-l/.2 (hi,).&-! (7.10) Writing one of these equations for each of the nodes
in the system yields a system of equations
Qi,j~k+l/z = C K , j , k + I / 2 ( h i , j , k + l - hi,,,&) (7.11)
[A104 = I 4 (7.16)
where the hydraulic conductances in the column and
layer directions are given by CC and CV, respectively where [A] is the coefficient matrix, {b} is the vector of
(McDonald and Harbaugh, 1988). unknown head values, and {q}is a vector of constant-
Following the notation of McDonald and Harbaugh head terms. The mathematical solution of this system of
(1988), the combined flows due to N sources or sinks is equations provides the hydraulic head for the given time
given as QSi,j,k where step. Given the complexity of these approaches relative
to the introductory character of our book, we must leave
QSi.j.k = p i , ] , k b i . ] ,k Qi.j. k (7.12) readers on their own to examine the details.
where P 1 , j . k and Q,.j.k represent the sum of constants re-
lated to N different inflow or outflow processes like
pumping, induced recharge, etc. -
7.4 The MODFLOW Family of Codes
Combining these equations yields the close to final MODFLOW has emerged as the de facto standard code
form of the finite difference equation or for aquifer simulation. The original code was developed
with an extensive User's Guide and released in 1984.
-
c R i . j - ~ / ~ . k ( b , . j - ~ . k bi.j.k) CRi.j+~/Z.k(bi,/+i,k- b i . j . k )
This version was superseded in I988 (McDonald and
+ c c i - I I Z , ~ . k ~ ~ l - l . kj - h1.J.k) + ~ c l + U 2 . j , k ( h i + l . j . k- h i . j , k ) Harbaugh, 1988). The popularity of this code can be
attributed to four factors. First, the code has proved to
+ C ~ . j . h - I / 2 ( h t . j , k - I - h1.j.k) C K , j , k + ~ / ~ ( h i . j . k +-
, hi,j,k)
be a powerful, robust, and well-crafted product. Early
users had great success in applying the code to a wide
variety of practical problems. Second, the User's Guide
is extremely detailed and provides clear descriptions of
One final step in the derivation replaces the Ah/At how various code options are used. Third, the code has
term on the RHS of JQ.7.13 by a backward time dif- been supported strongly by the U.S. Geological Survey
ference and is readily available. In recent years, more specialized
versions of the code have been marketed by several pri-
(7.14) vate companies. Finally, the success of the original code
has spawned an extensive array of training courses and
where m designates the present time at which the heads a large number of related products and calculational mod-
are unknown and m - 1 designates the previous time ules. Several companies market software that assists in
step. Ail of the h terms on the LHS of Eq. 7.13 would the preparation of MODFLOW data sets, such as, Ger-
carry the superscript "m." Essentially, the head terms aghty & Miller's ModelCad product, Waterloo Hydrogeo-
carrying the superscript "m" are unknowns, and those logic's Visual MODFLOW, and the Department of De-
with the superscript "m - 1" are known from the previ- fense's Groundwater Modeling System. Other utilities
ous time step. facilitate the contouring of hydraulic-headfields (SURFER;
The final form of the finite difference equation can Golden Software, Inc.). The examples presented later
be expressed assume readers have access to one of these versions of
MODFLOW.
C~J,k-t/2h/.j.k-~
111
+ C c i - 1 1 2 . j . k b ~ ~ ' i l . j+
.k CRi./-~/?,k~::"j-I.k There also has been a variety of follow-on products
related to the code. Pollock (1989) has developed MOD-
(- CVt,j.b-1/2 - C c t - ~ / ~ . j .-k C R i , j - ~ / z , k- C R i , j + I / r , k
PATH-a post-processing package that takes output from
- cci+I/>.j.k- CF:.j,k+1/2 + HCOFi.j.k)h::"j,k (7.15) steady-state simulations with MODFLOW and computes
threedimensional pdthhes. This package has found ap-
+ CRI./+l/2,hh?j+l.k + c c i + I l , L j , kby:I,j,k
plications in the simple modeling of contaminant trans-
+ CVi,j.k+112h?j.k+i = R H s i , j , k port and in the location of wells for pump-and-treat sys-
-
148 Cbapter 7 / Ground Water as a Resource

terns for the recovery of dissolved contaminants. Hill her of nodes because in most cases the number of model
(1992) has developed MODFLOW, a code that includes layers is small. One iteration is complete when all of the
capabilities for estimating various parameters required in slices are processed. After a large number of iterations,
a MODFLOW simulation. This code provides a tool for the solution converges.
automatic calibration. Other enhancements include the The Strongly Implicit Procedure (SIP) is a more compli-
ability to simulate ground-water and surface-water inter- cated approach. This procedure involves solving the un-
actions (Prudic, 1989), a new equation solver~ill,l990), knowns for the entire grid simultaneously (McDonald
approaches to accommodate the rewetting of cells in and Harbaugh, 1988). More recently, Hill (1990) imple-
the model that have become dry (McDonald and others, mented a preconditioned conjugate gradient procedure
1991), the ability to model aquifer compaction (Leake for use with MODFLOW. In general, this and similar
and Prudic, 19911, techniques to represent a transmissiv- solution techniques are extremely fast and robust and,
ity field that is smoothly varying (Goode and Appel, for these reasons, are now often used in solving systems
1992), and a capability to treat narrow horizontal barriers of linear equations.
(e.g., faults) that may impede ground-water flow (Hsieh
and Freckleton, 1993). These new features make the Modular Program Structure
MODFLOW code tremendously powerful and capable of
handling a variety of ground-water conditions. MODFLOW is built with a modular design that consists
of a main program and “packages.” The packages are
groups of independent subroutines that carry out specific
Solving Systems simulation tasks such as accounting for flow into or out
of Finite-LRifferenceEquations of a river, adding recharge at the top surface, or invoking
Procedures for solving systems of algebraic equations a specific calculational procedure for solving the finite-
can be broadly categorized as direct and iterative. Direct difference flow equations like slice-successiveoverrelaxa-
approaches involve rearranging the system of equations tion (McDonald and Harbdugh, 1988).
to a form that can easily be solved. The methods that This modular design is useful in several ways. It pro-
most of us learned in high school to solve simple systems vides a logical basis for organizing the actual code with
of equations are examples of the direct methods. The similar program elements or functions grouped together.
iterative approaches involve making some initial guess Such a structure facilitates the integration of new pack-
at the unknowns and refining these guesses through a ages to enhance the code’s capabilities. The modular
series of repeated calculations until an accurate solution structure also provides a convenient way to organize the
is obtained. user’s manual (McDonald and Harbaugh, 1988). Finally,
The original MODFLOW contained two iterative the modular structure simplifies the preparation of data
schemes. The simpler is Slice-SuccessiveOverrelaxation for a simulation trial. Typically, data are read by a subrou-
(SSOR). Instead of solving the entire system of unknowns tine within a package. Thus, for simple problems, where
at the same time, the equations are formulated for a two- only a few packages are used, the input data will be
dimensional slice (Figure 7.1 1) with the assumption that modest.
the heads in the adjacent two slices are known. The Table 7.2 lists the packages in the basic code (McDon-
resulting system of equations (actually formulated as the ald and Harbaugh, 1988) along with a brief description
change in hydraulic head) is solved by Gaussian elimina- of their function.
tion. The slice will usually contain a relatively small num-
Illustrative Example
Let us consider again Example 6.10, which was solved
analytically. A well fully penetrates a confined and infinite
artesian aquifer having a transmissivity of 928 m2/day
and a storativity of 1.0 X The initial head in the
aquifer is 0.0 m. The well is pumped at a constant rate
of 3790 m3/day. Determine the drawdown versus time
for 64 days at an observation well located 304.8 m from
the pumping well. Determine the drawdown at this node
at 64 days.
The first step in solving this problem is to construct a
Full grid Slice 4 grid. To keep the problem simple, we will use a grid
Figure 7.11 Division of the threedimensional model array with fixed spacings. MODFLOW provides for variable
into vertical slices for processing in the SSOR package (from grid spacing to increase the density of nodes around
McDonald and Harbaugh, 1988). wells. Let us begin by placing a node to represent an
 7.4 Tbe MODFLOW Family of Codes

Table 7.2 A Summary of the Packages Contained In MODFOW

Package Name Package Description

Basic Handles those tasks that are part of the model as a whole. Among those tasks are specification o f bound-
(*.BAS)' aries, determination of time-step length, establishment of initial conditions, calculating a water budget.
and printing of results.
BlockCentered Required in all model calcuhtions. Reads basic information on the grid spacing, pattern of layering. aqui-
Flow fer type, and hydraulic parameters to calculate terms of finite-difference equations that represent tloar
(*.BCF) within porous niediiim, specifically flow from cell to cell and flow into storage.
Well Designed to add effects o f injection or withdrawal wells. Reads information on the well numbers. their
(*.WEL) location, ;ind pumping schedules. The package adds terms representing flow to wells to the finite-differ-
ence equations.
Recharge This package is designed t o simulate the addition of natural recharge from precipitation to the model.
(*.RCH) Adds terms representing distributed recharge to the finitedifference equations.
River This package provides the capability of niodeling ground-water, surface-water interactions. Accounts for
(*.RJv) both the size of the stream and the possibility of a discrete low-permeability streambed. Adds terms rep-
resenting How t o rivers to the finite-difference equations.
Drain Siniilar to tlie river package except considers only inflow to the drain. The package adds terms repre-
(*.DRN) senting flow t o drains to the finite-difference equations.
Evapotranspiration Adds ternis representing evapotranspiration to the finitedifference equations. Accounts for the effects
(*.EVT) of plant ev;ipotraiispir;ttioii and direct evaporation of ground water. Rate varies as a function of tlie WJ-
ter-table depth.
General-Head Provides the possibility of variable flow across the model boundaries. The extent of flow depends on
BOUnddlkS the hrad difference between the cell and some constant head at a source away from the boundary.
(*.GHB) Adds ternis representing general-head boundaries to the finite-difference equations.
Strongly Implicit Iteratively solves the system of finite-difference equations using the Strongly Implicit Procedure.
Procedure
(*.SIP)
Slice-Successive Iteratively solves the system of finite-difference equations using Slice-Successive Overrelaxation.
Overrelaxation
(*.SOR)
Output Control Provides user with the ability to control what calculations are printed as output and at what time steps.
(*.OC)
' Shown in parentheses are the commonly used extensions for ixiniing liles o n a PC. For example, EXI.BAS is the tile containing the basic input d;it;~.
Modified from McDonald and Harbnugh (1988).

Observation well 304.8 m from t h e pumped well. For the 15 columns) appear poorly resolved. Extremely large
regular grid of this problem, the grid spacing must be grids (for example, 200 X 200) impose a significant com-
304.8 m or some fraction thereof (for example. 1/2 of putational burden and usually require graphics software
304.8, 1/3 of 304.8, etc.). Second, the aquifer must be to interpret the results. A recommended starting number
infinite, which from a modeling point of view means that of rows and columns would be between 25 and 50. The
the boundaries should be located far from the well. This simulation region in this case is subdivided by 30 grid
consideration with constant grid spacings pushes us to blocks along the row and column directions (Figure
select the largest grid spacing or 304.8 m. With a variable 7.12). The numbering scheme for the rows and columns
grid, it would be n o problem to locate the edges of the is always the same as shown in Figure 7.11. with column
model far from t h e well. By default, the four sides of the 1 and r o w 1 located at the top left corner of the grid.
models are no-flow or inipermeable bound;uies. Thus, The pumping well is located in the middle of the grid
during the 64-cIay simulation, as Iong as t h e cone of de- (row 15, column 15). The observation well is located
pression does not reach t h e boundaries, t h e model will one node or 304.8 m away. For this symmetrical problem,
represent an infinite aquifer condition. If significant draw- the actual direction does not matter. In real problems,
down occurs along the boundaries, the aquifer is no the actual locations of the observation wells will be
longer infinite. This detail should be checked once pre- known. This simple problem involves a single confined
liminary runs are completed. aquifer having a constant transmissivity. It is represented
Another question is h o w many grid blocks are required in MODFLOW as a single layer.
in the row and column directions to simulate an infinite Running this problem requires four packages-hisic,
aquifer? There are n o hard-and-fast guidelines in this re- blockcentered flow, well, and strongly implicit proce-
spect except that small grids (for example, 15 rows and dure. The first t w o packages provide basic infomation
150 Cbapter 7 / Ground Water as a Resource

Columns vided in the basic package that tells the code to prompt
for the particular four files. The input data for each of
the four packages are listed in Table 7.3, along with
1 explanatory comments.
Having prepared the necessary data files, the user runs
2 MODFLOW. With this output, one will notice that the
code echoes back most of the data provided in the various
input files. Incidentally, when debugging a new set of
data, errors in the input data are found by scanning these
output summaries. In looking at the predicted head calcu-
lations, you will see that only results from time step 18
or the last step are written out. This style of output is
the default. To obtain a more complete set of data one
would need to include the “output control,” which gives
the user extensive control over how the model data
are written.
One check on the success of the model calculation is
to examine the mass balance results on the output. The
results are shown in Table 7.4.
Figure 7.12 MODFLOW grid for the calculation involving The left-hand column summarizes the cumulative in-
the illustrative Example. flows to and outflows from the modeled region over all
18 time steps. The right-hand column is the water balance
for time step 18. In both cases, inflows match outflows
well, and from a calculational point of view the model
about the problem, the grid design, and time stepping. seems to have run appropriately. Normally. the percent
The well package gives details concerning the well. Fi- discrepancies should be less than 1%. The percent dis-
nally, the strongly implicit package gives information for crepancies can be lowered by reducing the convergence
the iterative solver that is required. Detailed specifica- parameter in EX1.SIP. In some cases, data errors may
tions on what information needs to be provided in each prevent the solution from converging. Beyond its diag-
of these files and what format is required is provided in nostic uses, the mass balance in complicated problems
the MODFLOW User’s Guide (McDonald and Harbaugh, provides information on how much water comes from
1988). Operationally, the data for each one of these mod- a particular source. The calculated hydraulic heads at
ules is included in a separate file with information pro- time step 18 in the output are similar to those calculated

Table 7.3 Annotated Summary of Data Required for MODFLOW to Run the Example Problem
File: EX1.BAS-Basic Package
RERUN OF W P L E 6.10 WITH MODFLOW title up to 80 clydricters
title continued, 80 characters, here blank
1 30 30 1 4 no. layers, rows, columns, and stress periods, unit, see Note 1
1112 0 0 0 0 0 0 1 9 0 0 0 switch to turn on indicated packages, see Note 2
0 0 memory allocation, mostly used as shown
0 1 description of boundary conditions, see Note 3
999.99 value of heid assigned to inactive cells, see Note 4
0 0. initial head value, 0.0assigned to each node, see Note 5
64.058 18 1.2 length of stress period, number of steps in the stress
period, time step multiplier I .2. see Note 6.
Note 1: A stress period detines a specified period of time when sources and sinks (e.g., well withdrawals, river leakage. evapo-
ration, etc.) are defined by a fixed set of parameters. It is this frlture that enables one to adjust model parameters with time.
For example, when a new well starts up, one would begin a new stress period. The well in this problem is pumped at a con-
stant rate; therefore, only 1 stress period is required.
The last number on this line specifies the consistent time unit for the input data. For this panmeter, the following codes apply
0-undefined; 1-seconds; 2-minutes; 3-hours; 4-days; 5-years. We have selected days as the time units, which means all
time units will be days (for example, transmissivity is m2/day).
 Next Page

7.4 Tbe MODFLOW Family of Codes 151

Note 2: The row of numbers on this line coincides with the following packages; RCF, WEL, DRN, RIV, EVT, XXX, GHB, R C H .
SIP, XXX, SOR, OC. The XXX are packdges that remain to be defined. The other package names are defined in Table 7.2. To
turn on a package requires that a unique nonzero parameter b e included. Here, we have turned on the RCF, WEL. and SIP.
When the code is run, it will prompt the user for the corresponding file names for all of these files. Following coninion pcic-
tice, each file is named so as to make the particular package name clear (for example, M1.BCF).
Note 3: This line applies the boundary condition to every node on a given layer of the model. The 0 indicates that the same
boundary condition is assigned to every node on the given layer. Providing a second number greater than zero meins the he;id
is unknown and must be determined, as is the case here with 1. A zero means that the cell is inactive (that is, no-flow). h VJIW
less than zero applies a constant-head boundaly. We are fortunate with our simple problem that the boundary array is filled
with 1's and can be accomplished in the shorthand manner shown in the data set. When values are not the same for all nodes.
the user would need to specify every value in the grid (see User's Guide). For our problem, letting the values be different
would mean an array of integers 30 X 30.
Note 4: On occasion, hydraulic heads are not calcdated for some cells. When model results are presented, the head vilue of
999.99 will be printed for such no-flow cells-essentially
making it obvious that the head has not been calculated for these
cells. Any number could he used for this purpose.
Note 5: This section of the input data file provides the initial value of hydraulic head for all the nodes in a given layer. Each
layer would need a separate set of input data. Again, because all nodes in layer 1 will receive the same initial value for hydraulic
head (that is, 0) w e can use the handy feature where the first value of 0 tells the code to fill the 30 X 30 array of nodes with
the second value or 0. If values had varied, one would input the entire matrix according to the instructions in the User's Guide.
Note k One of these lines is provided for each of the stress periods. For this exercise, there is only one stress period t h x is
64.058 days long. We ask there to be 18 time steps with a time step multiplier of 1.2.
File: EX1 .BSF-Block-Centered Flow Package
0 0 0 = transient simulation, 0 = no cell-bycell ternis priiitetl
0 layer type = 0 or confined; see Note 7
0 1. anisotropy factor each hyer; 1 .0here means isotropic
0 304.8 cell width along rows; notation here makes each cell 304.8 ni
0 304.8 cell width along the columns. again 304.8; see Note 8
0 .o 1 storage coefficient, the value 0.01 is assigned to a11 nodes
0 928. transmissivity along the rows: 928 ni'/day is assigned t o :ill
nodes; see Note 9.
Note 7: MODFLOW supports a variety of different aquifer types. Briefly, 0 = confined aquifer, 1 = uncoidned aquifer, 2 =
confined-unconfined constant transmissivity. and 3 = confined -unconfined variable transmissivity. One value is provided for
each layer (4012 format).
Note 8: The anisotropy factor is the ratio o f transmissivity or hydraulic conductivity along a column to transmissivity or hydrau-
lic conductivity along a row. Read one value per layer.
Note 9: This simplified input format works because c a d i row dimension is the same. For a variable grid, the input dat:i would
be more complicated.
Note 10: The transmissivity along rows is provided. The same value is assigned here to every node. The code determines the
column values by multiplying this transmissivity value by the anisotropy factor. Note that the sequence of required input data
with respect to transmissivity and storativity changes depending on the type o f aquifer. For example. with an unconfined q u i -
fer, it is necessary to specify the elevation of the bottom of the aquifer.
File M1 .WEL-Well Package
1 0 niax number of wells at any time, print no cell-bycell terms
1 for each stress period-the number o f wells active. here 1
I 15 15. 3790. for each well-location-layer, row, column. and
withdrawal rate (negative for withdrawal)
File EX1 .SIP-Solver for Finite Difference Equations
50 5 niax no. iterations, no. of iteration parameters,
I. .001 0 ,001 1 acceleration parameter (usually 1 .), convergence criterion
for iteration (.001), and other default parameters. see Note I I .
Note 11: It is generdlly difficult for the casual user to decide what parameters to use in this subroutine. As an iiiiti:rl estimate.
we could simply set the number of iterations around 50, and convergence criteria of . 0 0 1 or less. and accept the values sug-
gested in the IJser's Guide for the rest of the parimeters.
Previous Page

152 Cbapter 7 / Ground Water as a Resource

Table 7 4

RATES F O R T H I S T I M E S T E P L**3/T
......................
IN:
---
STORAGE =O.OOOOO STORAGE=0.00000
CONSTANT HEAD=0.00000 CONSTANT H E A D = 0 . 0 0 0 0 0
WELLS = 0 . 2 4 2 7 8 E + 0 6 WELLS = 3 7 9 0 . 0
TOTAL I N = 0 . 2 4 2 7 8 E + 0 6 TOTAL I N = 3 7 9 0 . 0
OUT :
---- ----
OUT:

STORAGE =0.24274E+06 STORAGE= 3788.1

C O N S T A N T H E A D =O.OOOoO CONSTANT HEAD =O.OOOOO
WELLS=O.OOOOO WELLS=0.00000
TOTALOUT = 0 . 2 4 2 7 4 E + 0 6 TOTALOUT= 3 7 8 8 . 1
I N -OUT= 43.359 I N -
OUT = 1 . 9 3 6 0
P E R C E N T DISCREPANCY= 0.02 PERCENT DISCREPANCY = 0.05

earlier with the analytical model. At 304.8 m away from problem in Section 7.4. Now, however, the drawdowns
the pumping well, the calculated drawdown is 1.641 m. at the pumping well will be compared. For the analytic
calculation, the observation point is assumed to be 0.1
m away (that is, the radius of the well). As expected, the
Operational Issues drawdown determined analytically is much greater than
that predicted by the model (Figure 7.13).
Time-StepSize
Techniques are available to provide better estimates
Normally, care must be exercised in selecting a time-step
of the actual drawdown that can be expected at a node
size to avoid errors that can occur due to large time steps.
These errors are most pronounced in the first few time representing a pumping well. One of the simplest in-
steps when pumping begins or pumping rates change. volves a variant on the distance-drawdown theory. In
The most preferred way of controlling errors is to use effect, one takes relatively accurate model predictions of
relatively small time-step sizes. Overall, it is prudent to drawdown at nodes away from the well, plots them on
a semilog distance versus drawdown plot, and obtains
place less confidence in results coming from the first few
the required estimate by extrapolation of the curve. In
time steps after a significant change in withdrawal or
injection rates. MODFLOW lets the user increase the size the illustrative example at 64.1 days (9.23 X 10‘‘minutes),
of the time step as the simulation proceeds. The given
the calculated drawdown at points 304.8 m and 3048 m
time-step size is multiplied by a number greater than 1
between time steps. A time-step multiplier of 1.2 works
well in most cases. Thus, an initial time step of 1 day 01 1
becomes 1.2 days for the second time step, and 1.2 X
1.2 or 1.44 days for the third time step, and so on. A
variable time-step size will reduce the computational ef-
fort compared to a constant step size and yet preserve --E Numerical solution
o Analytic solution
the accuracy of the calculation.

Drawdowns at “Pumping” Nodes

In numerical models, predicting drawdowns at “pump-
ing” nodes is usually a problem. The head or drawdown
at a pumping node does not represent drawdown in the
well because the hydraulic head is a “cell average.” This -8I
10‘
I I I I I IIII
103
I I I I 1 1 1 1 1
104
I I I I I l l
lo5
value is not the same as that calculated with an analytical Time (minutes)
model (for example, the Theis equation), which is the Figure 7.13 Comparison of time-drawdown curves at the
exact drawdown at the point of interest. pumping well computed using the numerical code and the
To illustrate this point, we will consider the example analytical solution.
 7.4 The MODFLOW Family of Codes 153

away from the well are - 1.64 ni and -0.26 ni, re-
spectively. Having created the approprinte distiincc-
- \-

drawdown plot (Figure 7.14),the projection of the

straight line provides ;in estimate of the drawdown 0.1
m away. The estimated value of -6.4 in compares to the
analytic value of -6.8 in.
This method is applicable only to confined aquifer with
no local changes in the tcinsniissivity. It is not genemlly
useful for the full range of cases that might be encoiin-
tered in a numerical modeling exercise. Fortunately.
there are a variety of other approaches in the literature
-- 4
/1>1,.
I '
that can be used to make these estimates. Head reference 'eve1 4 ' I

Figure 7.15 Parameters for the problem of flow to :I well

with water-table conditions (from Prickett and 1.onnquist.
Water-Table Conditions 1971).
With unconfined aquifers, the transmissivity changes ;is
if HNEW, I' 2 TOP, (aquifer is confined)
a function of drawdown in the vicinity of the well. For
example, examine Figure 7.15, where drawdown in ;in then TRI, = (TOP,I 1. - BOT, c) HYR, I r.;
unconfined aquifer causes the water table to fall. The
if TOP, s1 > HNEW, I I. > BOT, I 1,,
decrease in the aquifer thickness causes the transniissiv-
ity to decrease as well. (aquifer is unconfined)
The way the niimerical methods accommodate the
varying transmissivity is to recalculate the trnnsmissivity
theti TR, 1, = (H,\'EW, I 1, - BOT, J HYR, ,; ,
at the beginning of each iteration cycle within each time if HNEW, L. 5 BOT, I (aquifer ha3 drained)
step. Thus, every time that the hydraulic head at a node
then TR,I =0
changes, the transmissivity is recalculated ;is a product
of the hydraulic conductivity and the saturated thickness. where HYR,, is the hydraulic conductivity of cell i, j .
Within ;t time step, as the change in hydraulic conductiv- k in the row direction, TOP,, I. is the elevation of the top
ity becomes smaller after successive iterations, the of the cell i, j , k. and BOT, I is the elevation of the
changes in transmissivity are also reduced. When the bottom of cell i , j , k.
iterations are complete, both hydraulic head and trans-
missivity values will have converged to correct values.
Boundary Conditions
In MODFLOW, a comparison is made at the start of
each iteration to determine the magnitude of the hydrau- The six boundaries of the three-dimensional domain in
lic head relative to elevations of the top (TOfJiI J and MODFLOW (that is, top, bottom, and sides) are no-flow
bottom (BOT,,,,) of the unit of interest. Transniissivity boundaries by default. Creating flow across these bound-
for the cell is calculated using one of the three follow- aries requires that a different type of boundary condition
ing equations: be created. For example, one could make one of the side
boundaries a constant-head boundary or a flux boiindap
by adding water to particular nodes along the houndan..
There is a place in the Basic Input package to designiite
specific nodes as fixed head or inactive. A const;int-tlux
boundary may be simulated by simply adding or with-
drawing water from boundary nodes as if a well is pres-
ent. These boundaries are specified with the Basic Input
and/or Well Packages.
Fluxes into or out of the top and other 1;iyers of the
model could occur due to the presence of lakes or rivers.
In the model region, these features can he representecl

E
I
hp constant-head boundaries or variable-flux bound;iries
using the River Package. Similarly, recharge from precipi-
-8
I I I iilili 1 I 'Ii1" i ' iiilll I I I illill I I
tation can be added by applying a small flux :it a11 nodes
-10
lo-' 100 10' 1o2 1o3 1o4 present in recharge areas. It is this ability to adjust the
Distance (m) boundary conditions and treat various groups of intern;il
Ffgurt. Z14 Application of a distance-drawdown extripe nodes that make numerical models so powerful in repre-
lntion to estimate the drawdown at a "pumping" node. senting complex hydrogeologic systems.
154 Cbapter 7 / Ground Water as a Resource

7.5 Case Study Through the years, a great deal of hydrogeologic data
c t b e Application of MODFLOW have been collected in relation to this aquifer system
(see Springer, 1990). The data base includes extensive
Springer (1990) and Springer and Bair (1992) have a p streamdischarge measurements, hydraulic conductivity
plied MODFLOW to a buried valley aquifer complex near measurements of the streambed, lithologic logs from vari-
Wooster, Ohio. Figure 7.16 shows the geometry of the ous test holes, hydraulic information from aquifer tests
buried valley, Killbuck Creek, and its tributaries Little with a variety of wells, production rates for municipal
Killbuck and Apple Creeks, as well as the location of two and industrial wells, and measured hydraulic head data.
major municipal wellfields. From 1984 to 1988 the North The model is supported by extensive hydrogeologic data,
and South Wellfields produced from 3.5 to 4.4 million which are discussed in detail by Springer (1990). See
gallons per day. More recently, both wellfields may be Breen and others (1995) for a more recent USGS modeling
abandoned due to contamination. study of the same system.
Cross section A-A' shows the bedrock valley and the
distribution of geologic units. Immediately overlying bed-
rock is glacial till. This unit is overlain in turn by sand
Model Development
and gravel outwash, which comprise the aquifer. The
outwash unit may be separated by a silt unit up to 21 The finite difference model is composed of 28 rows, 46
feet thick (Figure 7.17~). The silt unit is relatively discon- columns, and 4 layers. The active portion of the grid
tinuous and, for example, does not occur in the area (Figure 7.18) is determined by the buried valley and is
of the South Wellfield. The uppermost geologic unit is composed of cells having dimensions Ar, = 250 ft and
comprised of siltyclay lacustrine and floodplain deposits Aci ranging between 250 and 500 ft. Figure 7.176 shows
ranging in thickness from 10 to 30 feet. Shown on the how the observed stratigraphy at the site relates to the
map (Figure 7.16) is the location of alluvial fans along 4 layers of the grid. The siltyclay unit that subdivides
Little Killbuck and Clear and Apple Creeks that directly the outwash along A-A' is not explicitly included as a
overlie the outwash aquifer. A variety of hydrogeologic separate layer. Its hydraulic effects, however, are in-
studies and careful streamdischarge measurements have cluded by lowering the vertical conductivity between
shown that the alluvial fans provide the most important units 2 and 3. For multilayer problems, it is necessary in
hydraulic connection between surface water and ground the Block-Centered Flow Package to input a matrix called
water (Springer and Bair, 1992; Breen and others, 1995). Vcont that reflects the vertical hydraulic conductivity

Explanatmn
Walk lxatton and identifier
0 spring
A croas-spctwxslrne

- A~luvialfan
Limits of aqurfer
Ffgure 7.26 Map showing the location of the Wooster study site, the extent of the buried valley aquifer, and the region en-
compassed by the MODFLOW model (modified from Springer and Bair, 1992). Reprinted by permission of Ground Water. Copy-
right 0 1992. All rights reserved.
 Z5 Case Study in tbe Application of MODFLOW

west East

Killbuck Creek

siltstone, and shak

- I
V&ical Exameration: 15x

A'
900;
\
/--

-\L Actual boundaries Ktllbuck Creek

/

-.
/

=860-
%
1-
rI =&-,r. __ _- I
F ,
\ I
"
,\ -
/

-
f I 0 ' /
r - - - -

i'
% ' 0 /

820- \
u

2
- - \

v
- \
\
0
0,

= L\ \
\
0
Z 780- /
\\
\
@ /
/

- --
'-- /
/

--+-/
. .-
Vertical Exaggeration: 15x

ExDlanation Scale
@ Layer number 0 800 1600 Feet
Confining layer
0 (not included as a layer)
(6)
Figure 7.17 Panel (a) depicts the geology along the cross section A-A' (from Springer and Bair, 1992). Reprinted by permis-
sion of Ground Water. Copyright 0 1992. All rights reserved. Panel (b) illustrates how the various units are conceptualized for
MODFLOW.

between layers divided by the distance between nodes. Figure 7.18 shows the boundary conditions for layer
McDonald and Harbaugh (1988) explain in detail how 1. The various streams are represented with the River
Vcont is cdlculated when a confining layer of some thick- Package. Thus, with pumping, induced infiltration from
ness and hydraulic conductivity is present. the creeks occurs with the quantity of infiltration deter-
156 Cbapter 7 / Ground Water as a Resource

0 4000 Feet
Figure 7.18 Location of active grid cells and a summary of boundary conditions for layer 1 (modified from Springer, 1990).

mined by the head gradient between the stream and the due to precipitation and water from melting snow. In
aquifer, as well as the area and hydraulic conductivity of general, the recharge rates are not constant everywhere.
the streambed. The relatively low hydraulic conductivity Recharge rdtes on the alluvial fan are greater than those
of the lacustrine and floodplain deposits restrict the quan- for the middle part of the valley, where lower permeabil-
tity of induced infiltration. The alluvial fans are important ity materials and agricultural drainage limit infiltration
because much more induced infiltration is possible (Springer and Bair, 1992).
through their higher permeability streambeds as the This conceptualization directly determines what MOD-
streams pass over them. Inflows to the alluvial aquifer FLOW packages would need to be included in the model.
farther up or down valley (that is, drift underflow) are Beyond those packages required for all simulations-
represented by constant fluxes created by adding wells Basic, Block-Centered Flow, and a solver (here the
along the end boundaries of the model using the Well Strongly Implicit Procedure)-the Well Package ac-
Package (Figure 7.18). Inflows to the active cells due to counts for the municipal and industrial withdrawal wells
flow through the low-permeability valley walls (that is, and some boundary fluxes; the Recharge Package adds
bedrock underflow) is simulated using the General-Head natural recharge to the system; the River Package pro-
Package (Figure 7.18). In effect, the General-Head Pack- vides for the possibility of induced infiltration from the
age is a variable-flux boundary condition that lets the streams; the General-Head Boundary Package represents
inflow through the boundaries increase as the drawdown inflow from the bedrock laterally adjacent to the various
in the aquifer increases. By being able to specify a conduc- layers; and the Output Control specifies in detail how
tance (proportional to hydraulic conductivity) between the results are to be presented.
the aquifer and the boundary, the inflow through the Overall, the potential sources of inflow to the aquifers
boundary can be controlled. In the Wooster example, are more than capable of balancing the water withdrawn
29%of the water pumped from the aquifer is thought to by pumping. Therefore, the flow system will be at steady
originate from the inflow of bedrock water through the state. The simulation model then is configured to provide
side boundaries (Breen and others, 1995)-The boundary a steddy-state result. The drawdown around the North
conditions for the deeper units (that is, layers 2, 3, and and South Wellfields is sufficiently large that layer 1 will
4 ) are essentially the same except that the streams and desaturate locally and portions of layer 2 will behave
recharge are not included. There are also minor changes as an unconfined aquifer. Thus, layer 1 is considered
around the alluvial fans, but generally the boundary con- to be an unconfined aquifer, layer 2 is considered cap-
ditions are set up to provide bedrock and drift underflows able of being converted from a confined to an un-
as was the case with layer 1. confined aquifer, and layers 3 and 4 are considered to
The Recharge Package provides a component of inflow be confined.
 7.5 Case Study f n tbe Application of MODFLOW 157

Ffgure 7.19 Map of the simulated distribution of hydraulic head for layer 3 (from Springer, 1990).

Data Preparation and Model Calibration ria involved (1) a good visual comparison between the
measured and simulated potentiometric surfaces, (2) ;I
In the Wooster case study and most every model study, root-mean square error between measured and simulated
there will never be sufficient data to provide necessary heads of less than 2.5 feet, (3) simulated water losses from
parameters (for example, transmissivity, storativity, re- streams that compared favorably to measured losses, and
charge rates) for every node in every layer. When hydro- (4) simulated leakage from bedrock that agreed with
geologic data are available, they form a basis for estimat- estimates from chemical mixing and isotopic studies
ing model parameters required for every node. For (Springer and Bair, 1992).
example, Springer (1990), through a detailed interpreta- The enclosed computer disk contains the complete
tion of geologic data, was able to develop isopack maps set of files required to nm the Wooster case study {WOO.
BAS, WOO.BCF, etc.). We would encourage readers to
that together with lithologic and hydraulic test data pro-
vided a detailed description of transmissivity distributions examine these data files to see just how complex the data
for the layers representing the outwash aquifer. Thus, files can be for a realistic MODFLOW problem. Shown in
transmissivity values for layer 2 varied from 500 to 6000 Figure 7.19 is the simulated, steady-state potentiomctric
ft'/day and in layer 3 from 20 and 90,000 ft'/day. The surface for layer 3 under the imposed conditions of pump-
transmissivity distributions for the lower hydraulic con- ing. The cone of depression due to large withdrawals ;it
ductivity units (layers 1 and 4) are less well known and the North and South Wellfields is obvious. The model
W;IS validated by removing the pumping stress and siic-
are not represented in detail. Transmissivities of layer 1
range between 0.01 to 200 ft'/day (Springer and Bair, cessfully simulating the historic, nonpumping potentio-
1992). Layer 4 is assigned a constant transmissivity of metric surface (Springer and Bair, 1992).
0.0005 ftz/day. Streambed conductance values are based
on measured hydraulic conductivities for various reaches Example 7.1
of the streambed (Springer, 1990). Recharge rates of 4
One of the main benefits of using a code like MODFLOW
in./yr for the valley and 10 in./yr are assigned to alluvial
is that the mass balance calculations provide a \'en useful
fans. These values are based on estimates of the U.S.
way to examine the source of water provided to ;I system
Geological Survey (Breen and others, 1995). Pumping
of pumping wells. With the help of the MODFLOW code
rates for the individual wells are quantified through a
and the data files supplied for the Wooster case study.
synthesis of existing records. determine the proportion of water pumped from the
Calibration involved making minor adjustments to the
system due to natural recharge, leakage from the creeks.
transmissivity field, vertical conductance values (be-
drift underflow, and bedrock underflow. To obtain the
tween vertical nodes), streambed-hydraulic conductivity
answer to the problem, it is necessary to plow through
values, and bedrock leakage rates until the steady-state
the output to find the mass balance summary.
model was successfully calibrated. The calibration crite-
- Cbapter 7 / Ground Water as a Resource
158

1. The purpose of this problem is to illustrate the im-

pact of boundary conditions on a simulation. Reconsider
the illustrative example in Section 7.3 and assume that
the two different sets of boundary conditions shown in
the figures apply, that is, neflow boundaries on four
sides, and no-flow boundaries on three sides and a con-
stant-head boundary along the top. Use MODFLOW and k?)
run simulations for 1474 days (35 time steps with a 1.2
time-step multiplier) with the two different boundary
conditions to compare the drawdowns at a point 304.8 m
away from the withdrawal well.
Hint: The original problem was actually set up with
four no-flow boundaries (see Figure 7.20~).
!b)
Thus, the first part of this problem is solved simply by
running the simulation for another 17 time steps (35 in Problem 1
total) to produce the drawdown shown on the figure. A
constant-head boundary is simulated in MODFLOW by
adjusting the information provided on boundary condi-
tions that are provided in Exl.BAS. The existing line
describing the boundary conditions is (9 spaces} 0 (9
spaces} 1. The 0 tells the code that the army describing
the boundary conditions has a constant value. The 1 is
that constant value. Adding a constant-head edge requires
that the constant-head nodes be designated by - 1. Thus,
it becomes necessary to include the complete boundary
condition array. The first few lines of this piece become

1 l(30I2)
-I-1-l-l-I-I-I-l-I-l-L-L-L-I-I-I-I-I-I-L-l-l-l-l-l-l-l-l-l-l
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 l l l l l 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 l l l l l l l l l l
l l l l l l l l l l l l l l l l l l l l l l l l l l l l l l
(continue for a total of 30 rows)

The 1 next to the format (3012) is a multiplier for all the

integers in the following array.
 Chapter 8

Stress, Strain, and Pore Fluids

8.1 Deformable Porous Media

8.2 Abnormal Fluid Pressures in Active
Depositional Environments
8.3 Pore Fluids in Tectonic Processes

As fluids are ubiquitous within the crust. more and greater between the loading of a sedimentary pile and the dissipa-
emphasis has been placed on the physics that describe tion of the fluid pressure caused by the loading. This
fluid pressures in deformable porous rocks. Outstanding type of transient behavior is frequently associated with
problems include the generation and dissipation of abnor- clays and other low-permeability sediments that I i c,ve i a
mally high fluid pressures in areis of active sediment relatively high specific storage and low hydraulic conduc-
deposition, the generation of earthquakes due to fluid tivity; that is, the hydraulic diffusivity is small. In addition
injections and in the vicinity of man-made reservoirs, to these fluid pressure transients, there are a host of
landslides and slope failures caused by high fluid pres- others associated with high-permeability miterials sub-
sure, and the role of fluid pressures in faulting and fractur- jected to natural loading. such as earth and oceanic tides.
ing. Thus in this chapter we are obliged to learn about earthquakes, and even changes in atmospheric pressure.
driving forces for fluid movement other than those associ- To examine the nature of these transients, it is necessary
ated with topographic relief. To gain an understanding to reexamine Darcy's law and the concept of effective
of such forces we start with the premise that water does stress and to reformulate these ideas into a different form
not reside in a passive porous solid, but instead there of a diffusion-type equation.
exists a complex coupling between the moving fluid,
entities that might be carried by the fluid, and the solid Development of tbe Flow Equation
matrix itself. In this chapter we focus on the couplings An excess fluid pressure has been defined as a pressure
among stress, strain, and pore fluids. The moving in excess of some preexisting hydrostatic value Within
entities (heat energy and mass) are discussed in later the concept of the effective stress statement, this may
chtlpters. be stated as
u + A u = a + (P, + P',,) (8.1)
- Deformable Porous Media
8.1 where the total pore water pressure consists of two parts,
P, and P,, , where P, is a hydrostatic pressure and P,,, is ;I
One-Dimensional Consolidation
transient pore water pressure in excess of the hydrostatic
The concept of disequilibrium compaction was discussed value, and is presumed to be caused by the increment
in Chapter 2 and was described in terms of a time lag in total stress Au.In a similar fashion, total head may be
160 C&apter8 / Stress, Strain, and Pore Fluids

restated as this companion equation may be stated as

For onedimensional vertical flow, Darcy’s law becomes where pu,gnP.,is the specific storage due exclusively to
expansion of the water and E is the one-dimensional
volume strain, or dilation, taken as the pressure-produc-
ing mechanism.
For the problem under consideration, the flow occurs
in response to a gradient in excess pressure, not total The Undrained Response of Water Levels
head, giving to Natural Loading Events
Numerous cases of water-level fluctuations in wells in
(8.4) response to atmospheric pressure, ocean and earth tides,
earthquakes, and passing trains have been reported (Ja-
as the appropriate form of Darcy’s law. cob, 1939; Robinson, 1939; Parker and Stringfield, 1950;
The conservation equation required for further analysis Bredehoeft, 1967;van der Kamp, 1972;van der Kamp and
has already been given as Eq. A.6 in Appendix A, reex- Gale, 1983). These fluctuations are generally accepted as
pressed here for one-dimensionalvertical flow of a homo- evidence that confined aquifers are not rigid bodies, but
geneous fluid are elastically compressible, an idea fostered by Meinzer
(1928) and advanced by Jacob (1940). Currently, it is a
routine exercise to compare the water-level response to
the loads that cause them to calculate certain hydraulic
where properties of the medium.
In discussing and interpreting these fluctuations, two
factors must be considered. First, there are loads that act
not only o n the rock matrix and its contained water, but
(8.6) also on the water level in an open observation well, and
there are loads that act only on the rock matrix and its
In this development, a d a t is taken as the vertical stress contained water. Examples of the former include changes
change that gives rise to the excess pressure develop in atmospheric pressure, and for the latter there are earth
ment. It is noted that such stress changes were not re- and oceanic tides. Second, it is important to consider
quired in our previous development of the flow equation the boundary conditions under which the loading takes
for confined aquifers. SubstitutingDarcy’slaw along with place. Two types are possible. For a field deformation that
Eq. 8.6 into the conservation equation gives the one- is slower than the characteristic times for the diffusion of
dimensional consolidation equation the pore fluid, the fluid pressure may rise above its ambi-
ent value and then dissipate rather quickly. Hence, for
all practical purposes, the fluid pressure remains constant
during the deformation, and such boundary conditions
Except for the one-dimensional flow condition and the are called constant pressure, or drained. This behavior
inclusion of the stress term that gives rise to the fluid is an expected one in high-permeability materials. The
pressure production, the assumptions here are identical effect of passing trains on water levels as demonstrated
to those employed in the previous development of the in Figure 4.5 is an example of a drained response, and
+
diffusion equation. The quantity pwg(nPu, PP) is the the parameter &, given in Table 4.1 provides an example
specific storage, whereas the quantity pu,gPpis the com- of a drained compressibility. Conversely, when the load
ponent of specific storage due exclusively to pore com- variations are rapid in comparison to the diffusion time
pressibility. for the pore fluid, the local fluid mass remains essentially
Equation 8.7 incorporates the time rate of change of constant during deformation. This response thus occurs
vertical stress as a pressure-producingmechanism, which at constant mass, or under undrained conditions. The
is of some importance to many loading problems in hy- undrained deformation is elastically stiffer than its
drogeology. As there is a relationship between stress drained counterpart so that the respective deformations
and strain, one would expect that there is a companion are chdracterized by different coefficients. Hence, when
equation that incorporates the time rate of change of a coefficient is sought to characterize a prescribed type
pore volume strain as a pressure-producing mechanism. of response, it is important to know the boundary condi-
For the simple case we are examining, where the fluids tions under which that response occurs.
are compressible but the individual solid grains are not, All water-level fluctuations as observed at a point that
 &l Deformable Porous Media 161
are caused by rapid loading phenomena (tidal and atmo- - 1

spheric pressure variations) are treated as constant mass

phenomena. As constant mass implies an absence of fluid
flow, all the information we require about these deforma- Note Water levels are
corrected to a constant
tions may be obtained by setting the fluid flow term in atmospheric pressure
assuming a barometric
Eqs. 8.7 and 8.8 equal to zero, which gives, after minor a
Time of the elficiency of 75%
moon's transit
rearrangement at upper B G
1.7 culmination 5
U

13 15 17 19 21 23 25 27 29 31 2 4 6 8 10
March 1939 April 1939
Days of the Month
Figure 8.2 Water-level fluctuations in a well caused by
earth tides (from Robinson, Tnns. Amer. Geopliys. Union. v.
20, p. 656-666, 1939. Copyright by Amer. Geopliys. Union).
(8.9~)

In these developments, the minus sign has been ignored

This dilation can cause water-level changes of about 1
because we are interested only in absolute values.
to 2 cm. As noted in this expression, the volume change
Equation 8.9a is termed the pore pressure coefficient
of a dilated rock-water system is represented by a volume
and is defined as the change in fluid pressure at constant change of water in the pores, where n & = (Ss- p,,.gP,,>/
fluid mass as the stress increases. interpreted literally,
pll.g.Figure 8.2 demonstrates the effect of earth tides on
this coefficient describes the percentage of an incremen-
water levels.
tal load that is carried by the pore fluid provided the
Barometric pressure has an inverse relationship with
pore fluid is not permitted to drain. The pore pressure
water levels in wells (Figure 8.3). The barometric effi-
coefficient has been referred to as the tidal efficiency by
ciency (B.E.) is used to describe how faithfully a water
Jacob (1940), an unfortunate choice of terms in that
level responds to changes in atmosphere pressure, and
this equation describes the change in fluid pressure at
is given as
constant mass in response to any blanket load, not just
ocean tides. As used by Jacob (1940), tidal efficiency
(8.11)
(T.E.)is defined as the ratio of a piezometric level ampli-
tude as measured in a well to the oceanic tidal anipli-
where P, is atmospheric pressure. The response of water
tude, or
levels to changes in atmospheric pressure may be rea-
soned as follows. Incremental increases or decreases in
(8.10)
atmospheric pressure acting on a column of water in ;I
well are, respectively, added to or subtracted from the
where H is the height of the tide. Figure 8.1 shows an pressure of the water in the well. These same stresses
example of the correspondence between ocean tides acting on any part of the confined aquifer. however. arc
and water levels in wells located in aquifers with an
oceanic suboutcrop.
Equation 8.9b has been used by Bredehoeft (1 967) to
examine the effect of earth tides on water levels in wells.

-
g $ -21.0 Ground-water level
Eg
3 -21.5 -
-
*A X
.?
0- m
L3 March 1939 April 1939 2
Days of July 1938 Days of the Month
Figure 8 1 Tidal fluctuations and induced piezometric sur- Figure 8 3 Water-level response to changes in atmospheric
face fluctuations observed in a well 30 m from shore at Mat- pressure (from Robinson, Tnns.met. Geophys. IJnicin. v.
tawoman Creek, Maryhnd (from Meinzer, 1939). 20, p. 656-666, 1939. Copyright by Amer. Geopliys. Union).
-
162 Cbapter 8 / Stress, Strain, and Pore Fluids

supported by both the grain structure and the fluid. As of 0.178. Specific storage in this case is calculated to be
long as the atmospheric pressure is undergoing change, 2.8 X lO-'ft-'. An interesting application for a composite
there must exist a pressure difference between the water rock section has been reported by Carr and van der Karnp
in the well and that in the aquifer. When atmospheric (1969). These authors measured several values of both
pressure is increasing, the gradient is away from the barometric and tidal efficiencies in sediments consisting
well, which causes the water level to decline. When of mostly sandstone and siltstone. The barometric effi-
atmospheric pressure is decreasing, this decrease is fully ciency averaged about 0.37 and the tidal efficiency aver-
removed from the water column in the well and only aged 0.57. The difference of 0.06 between the sum of
partially from the water in the confined aquifer. Hence, these values is reported to be within experimental error.
the gradient is toward the well, and the water-level rises. Forty rock samples provided a mean porosity of 0.177.
A mathematical statement for barometric efficiency The average specific storage was calculated to be about
may be arrived at by reconsidering the statement of effec- 6.3 x 10-7 ft-1.

tive stress and our interpretation of the pore pressure Barometric fluctuations are not commonly observed
coefficient. The effective stress concept may be re- in wells tapping unconfined aquifers. The reason for this
stated as is that changes in atmospheric pressure are transmitted
equally to the column of water in a well and to the water
(8.12) table through the unsaturated zone. Thus, there are no
pressure gradients.
where the pore pressure coefficient aP/aa represents The last item to be concerned with is to demonstrate
the percentage of the load that is carried by the fluid, that the tidal and barometric efficiencies as described
provided the fluid is not permitted to drain. Clearly, thus far-although defined as an undrained response-
da must represent the percentage of the stress that is consist of a combination of drained and undrained param-
carried by the solid matrix, again under undrained condi- eters. It will be recalled that the pore compressibility P,,
tions. This is the barometric efficiency, that is, could be expressed in terms of a modulus of compres-
sion l/Hp where the subscript p means we are dealing
1 = T.E. + B.E. (8.13) with pore compressibility only. It is convenient now to
so that, from Eqs. 8.10 and 8.11 introduce other notations for some of the parameters.
These are

Thus, as Pp approaches zero for a rigid aquifer, B.E. ap-

proaches one, and T.E. approaches zero. A barometric 1
- = nPU. (8.16b)
efficiencyof one suggests that the well is a perfect barom- QP
eter, which is possible only for a perfectly incompressible
aquifer (Pp = 0).On the other hand, as &, gets large, where the subscript p indicates compressibilities for the
B.E. approaches zero and T. E. approaches one. The case of incompressible solid grains. The relationship be-
relationships among the tidal efficiency, barometric effi- tween these constants is given as
ciency, and specific storage may be stated as
(8.17)
(8.15a)
With this new designation, it is clear that the tidal effi-
ciency or pore pressure coefficient is equal to R,,/H,, and
(8.15b) the barometric efficiency is equal to Rp/Q,,. In addition,
Eq. 8 . 9 becomes
~
As the sum of the barometric and tidal efficiencies is
unity, the percentage of storage attributable to expansion
of the water is B.E., and that attributed to compression
(")aff '$1
-- --
1
R/J

Qp Hp
-
--1
HI,
[ g]-- g1 [B.E.]
1 - RP
of the matrix is T.E. (8.18)
There are several examples of the application of the
results just stated. Jacob (1941) was the first to employ As a e / M is a measurement at constant mass, it too is an
these methods for a well in Long Island with a reported undrained parameter and may be designated as l/&,
tidal efficiency of 0.42 and a porosity of 0.35. Given an where K,,is an undrained modulus of compression, still
estimated value of 0.58 for the barometric efficiency, the for the incompressible solid case. It thus follows that
specific storage is readily calculated to be 8.6 X 10-7 ft-'. B.E. = H,,/K,, which is a ratio of a drained to an undrained
Bredehoefi (1967) reported a barometric efficiency of modulus, and T.E. = 1 - H,,/Ku. Further details on these
0.75 for a well in Iowa, along with an estimated porosity relationships may be found in Domenico (1983).
 8.1 Lkfontiable Porous Media 163

The Drained Response of Water Levels coefficient, the greater the percentage of the stress
to Natural Loading Events change that is initially carried by the tluid ;ind. conse-
There are numerous examples of the drained response quently, the greater the initial fluid pressure production.
o f water levels to natural loading events, including the Thus, for a pore pressure coefficient close to one. virtu-
change in head in response to changes in river stage ally all tlie stress change is initially carried by the tluid.
(Cooper and Rordbaugh, 1963), earthquakes (Cooper and As mentioned in Chapter 2. the pore pressure coefficient
others, 1965), and the inland propagation of sinusoidal decreases with increasing rock rigidity and is therefore
fluctuations of ground-water levels in response to tidal rock dependent.
fluctuations of a simple harmonic motion (Ferris, 1951;
Werner and Noren, 195 1). These problems are generally
Land Subsidence as a One-Dimensional
treated within the framework of the conventional one-
Drained Response
dimensional diffusion equation given as Eq. 8.11 where
the pressure-producing mechanisms are normally formu- Subsidence of the land surface in many areas of tlie world
lated within the boundary conditions. In other instances, has been ascribed to several causes: tectonic niovenient:
it is frequently assumed that the stress causing the excess solution; compaction of sedimentary materials clue to
pressure is applied rapidly and then held constant so static loads. vihnitions, or increased density brought
that the stress term of the one-dimensional consolidation ;thout by water-table lowering; and changes in reservoir
equation (Eq. 8.7) is taken ;IS zero. A typical example pressures with loss of fluids. Noteworthy GISCS o f apprr-
of this is the so-called Terzaghi consolidation equation ciable subsidence attributed to the last cwse 1i;ive been
(Tenaghi and Peck, 1948). In the application of this reported in Long Beach Harbor, California (Gilluly and
equation to consolidation problems, it is frequently as- Grant, 1949); the San Joaquin Valley, <hlifornia (Poland
sumed that the pore compressibility is considerably and Davis, 1956; Poland, 1961): the upper Gulf coastal
larger than the compressibility of the water so that the region, Texas (Winslow ;ind Wood. 1959: Gabnsch and
latter may be ignored, giving from Eq. 8.7 Bonnet, 1975); the Savannah area, Georgia (Davis and
others, 1963); and Lzs Vegis Valley. Nev;ida (Donienico
and others, 1966). Other localities include Mexico City
(Cuevas, 1936) and London (Wilson and Gr;ice, 1942).
where the hydraulic diffusivity C,.is termed the coeffi- The geologic requisites and qualifying conditions for the
cient of consolidation and is given as occurrence of subsidence are so well adapted to ;illiivi;il
basins that it is likely far more occurrences of this phe-
nomenon are taking place than have been reported. l h e
(8.20)
chief reason for this is the lack of close control of bench-
marks necessary to detect small c1i;inges in 1;incl-
In Eq. 8.20 the specific storage for the case of inconipress-
surface a1tit ucle.
+
ible fluids is a,,p,,g/(l e), where e is the void ratio and
Serious problems can be caused by land-surfiice subsi-
u,,is a coefficient of compressibility, defined as the rate
dence. The normal upward force of skin friction acting
of change of void ratio to the rate of change of effective
on piles or well casings may be reversed, which will
stress causing the deformation.
subject these elements to "downdrag." This ma)' result
Problems in basin loading or crustal deformation itre
in failure or protrusion above land surface. Gradients o f
normally described in terms of stress or strain rates ap-
c;inals m;iy be reduced, or even reversed. and so affect
plied over geologic intervals of time. For these purposes,
the normal flow of water. Cracking o f concrete or brick
Eqs. 8.7 and 8.8 may be restated here as
structures is common i n subsiding arcis. Subsidence i n
coastal ;ircas results in the conversion of pastures to
tidelands, where areas previously elevated become sub-
ject t o flooding from hurricanes (Winslow and Wood.
and 1959). Kreitler (1977) estimated that Hurricane C i r h
which struck the Houston-Galveston region in 1961.
would have flooded 146 square miles of Imd adjacent to
Galveston Bay if it had stnick in ly76. This is 25 sqime
where R , , H/,,and Q,, have been previously defined. It miles more than the arc;i flooded i n 196 1.
is noted that 1/R,, is incorporated within the specific Many of the observations of tlie r:ite of water-level
storage for problems involving stress rates and l/Q,, is decline and the rate of subsidence exhibit ;i fair degree
incorporated within the specific storage for problems of linearity. This linearity has been observed hy Carrillo
involving strain rates. Additionally, the pressureproduc- (1948) in Mexico City and by the Tokyo Institute of Civil
ing stress term in Eq. 8.21 is multiplied by the pore Engineering (1 975). Figure 8.4 denionstrxtes ;I rc;isonahle
pressure coefficient. Thus, the larger the pore pressure linear trend between water-level decline ;ind suhsidence
164 Cbapter 8 / Stress, Strain, and Pore Fluids

4-

1
c I I I I I I 1 [I
p 200 0

'f 240
v)
p 280
4%
4 =
320 % a
5 360 g9 g8
-
m

2
400
440
10 2
11
12
B 480

Figure 8.4 Subsidence and water-level decline in the San

Joaquin Valley (from Poland and others, 1975).

in the San Joaquin Valley over the period 1930 to mid-

1940sand then again to the late 1950s. From this diagram
it is noted that subsidence actually stopped in response Maximum Decline in Water Level, Ab (meters)
to the rise in water levels. Figure 8.5 Rekdtion between maximum settlement and
To investigate this linear trend further, consider Table maximum decline in water level.
8.1 and Figure 8.5 constructed from it. It is important
here to note the dates associated with the investigations
reported in Table 8.1. As subsidence has continued be- decline). One possible reason for this may be related to
yond the period reported, it is not certain that the rela- a component of subsidence caused by tectonic factors
tionships demonstrated in Figure 8.5 still hold. At any that have not been filtered out of the calculations. In Las
rate, there appear to be two well-defined divisions, one Vegas Valley, Nevada, for example, there is a component
where the subsidence-water-level decline ratio is larger of subsidence due to the elastic yield of the bedrock in
than 0.09 (9 m of subsidence to 100 m of water-level response to the load of Lake Mead (Figure 8.6). This
decline) and the other where the ratio is equal to or less regional deformation has affected 8000 square miles in
than 0.025 (2.5 m of subsidence to 100 m of water-level southern Nevada and adjoining states. Yet another possi-

Table &l Summary of Cases of Subsidence

~ _ _ _ _

Subsidence
Max. Settle. Area Head Decline Date of
Name of Place (m) (sq. miles) (m) Investigation

United States
San Joaquin Valley, CA 8.8 5212 90 1972
Santa Clara Valley, C A 4.0 25 1 25 1972
Houston-Galveston Area, TX 2.3 4710 100 1974
Eley-Picacho Area, AZ 1.1 - 30-60 1972
Las Vegas, NV 0.75 200 30 1972
Baton Rouge, LA 0.3 250 60 1970
New Orleans, LA 0.5 - 20 1968
Savannah, GA 0.1 19 27 1963
Japan
Tokyo 4.6 77 30 1972
Nagoya 1.5 38 - 1976
Mexico
Mexico City 8.5 58 20' 1964
Taiwan
Taipei 1.35 Approx. 48 20 1969
United Kingdom
London 0.1 - 90' 1942
Italy
Venice 0.14 Approx. 3 5 1974
 &l DeJormable Porous Media

-
f? pends on the thickness, configuration, and distribution
of the interbedded clay. For the assumption of Jacob
(1 940) that water is released from the clay beds instanta-
neously with no time lag, c is indeed a constant and will
Subsidence due to
be equal to Hlm where H is the thickness of the clayey
ground-water withdrawal beds and m is the thickness of the aquifers. Where this
condition is not satisfied, c is no longer a constant but
300 will approach a maximum value of H l m with time so
.c
0 that the storativity will increase with increasing time.
Figure 8.6 Subsidence profile, Las Vegas Valley, showing These points will be cleared up later.
differential subsidence due to pumping superposed on re- The first two components of storage in Eq. 8.23 are
gional subsidence (from Malmberg, 1960). already familiar to the reader, at least in terms of what
they mean with regard to the source of water in confined
aquifers. Lohman (1961) was the first to q u a n t e their
bility for the two divisions demonstrated in Figure 8.5 meaning in terms of the elastic compression of aquifers.
resides in the possibility of variations in the compressibil- As the original head and the change in head in the aquifer
ity of the geologic units as well as in their thickness. of Figure 8.7 is invariant with respect to depth, the ratio
of the change in thickness A m to the original thickness
Matbematfcal Treatment of Land Subsidence m may be stated as
There are two related questions of general interest in the
study of land subsidence. The first is concerned with the (8.24)
total amount of subsidence that might occur in a given
hydrostratigraphic unit in response to a lowering of water where AP is the change in fluid pressure. The storativitg
levels. From what we have learned earlier, the amount for the elastic response may be written as
can vary from point to point depending on the amount
of water-level decline, the thickness of the more com- (8.25)
pressible units, and their compressibility. Thus, differen-
tial subsidence will be the general rule. The second ques- Combining Eqs. 8.24 and 8.25 and solving for A m gives
tion deals with the time rate of subsidence in response
to the time rate of head change within the aquifer. This
(8.26)
problem is not unlike the problem of disequilibrium com-
paction described in Chapter 2. At one extreme, the
subsidence can occur concurrently with the head de- This equation (or Eq. 8.24) gives the amount of vertical
cline; at the other extreme, the rate of subsidence can shortening of an aquifer of thickness ni in response to
Iag considerably behind the rate of water-level decline
so that the basin can be in a state of progressive subsi-
dence for decades.
Vertical Compression We will focus on the first of
these questions with an ideal depth-pressure diagram as
given in Figure 8.7. Here an aquifer has an original head
h that is lowered by the amount Ah, with water being lsochrones
removed from storage from the total system. Jacob (1 940)
was the first to attempt some quantification of this phe- Confining
nomenon when he postulated that the water so removed Final fluid layer
is derived from three sources: expansion of the confined pressure

water in the aquifer, compression of the aquifer, and

I
compression of the adjacent and included clay beds. Ja-
cob concluded that the third source is probably the chief
one, a fact demonstrated by Poland in 1961. Jacob (1940)
described the storativity of such systems as

(8.23)
where all terms should by now be familiar to the reader Figure 8.7 Depth-pressure diagram of the development o f
except for p;, which is the compressibility for the clayey isochrones in a confining layer in response to lowered squi-
beds, and c, which is a dimensionless constant that de- fer pressure.
166 Cbapter 8 / Stress, Strain, and Pore Fluids

a pressure change AP. As aquifers ;ire elastic, o r nearly The result just given may be generalized for head
so,this elastic component of compression is recoverable changes in aquifers both above and below a compressible
if the pressure is allowed to recover to its origin;il (pre- confining layer (Figure 8.8).The equation describing the
pumping) value. Further, this compression takes place head change in the confining layer is given as
instantaneously with water-level decline.
Examination of the clay layer in Figure 8.7 shows that
the original head distribution varies with depth z and is
h(z)= 4hi ):( + Ah2 (I)
H-Z (8.30)

actually equal to the head within the aquifer ;it the con- Substitution of Eq. 8.30 into Eq. 8.28 and performing the
tact between the two units. A rapid lowering of head in integration gives
the aquifer is not accompanied by an immediate lowering
of the liead in the clay layer because of different hydraulic
diffusivities. Thus, the head in the clay layer is converted (8.311
to an “excess” head, at least with respect t o the dimin-
ished head at its lower boundary. In response to this Examination of Eq. 8.31 and Figure 8.8 shows that
excess head, the clay layer will drain, with the head when the transient flow process is terminated. the verti-
distribution at any time being shown by the so-called cal shortening o f the confining layer equals the product
isochrones, which mean “same time.” Eventually, the of the specific storage and the final area described by the
heads in the aquifer and in the clay layer are once more increase in effective stress on a depth-pressure diagram.
matched at the boundary between the two, and the drain From Figure 8.8 and Eq. 8 . 3 1, this area is again a triangle
age process comes to a halt. During this period ofadjust- for 4h2 = 0, or a rectangle when Ab, = AhI, or ;I
ment, the effective stress in the clay layer has increased trapezoid for Ah, # Ah2(one-half the sum of the bases,
in response to the drainage, resulting in a commensurate Ah, + Ah2, times the height H).
volume reduction in the layer itself, For the one-dimen- The concepts described pertain to the permanent com-
sional case we are considering where the solids and the pression of low-permeability layers in confined basins,
fluids in the clay layer are assumed to be incompressible, the volume of water being a one-time reserve associated
the volume of fluid removed is equal to the volume of sub- with permanent compression. This process does not af-
sidence. fect the aquifer storage capacity in tlrat the pore volume
As the head change in the clay layer ranges from zero
at the top to 4h at the bottom (Figure 8.7), the volume
of water produced from each element of the clay layer
will be different and can be expressed as

dq = Slb (2) dz (8.27)

where S: is the specific storage of the clay layer (ignoring

fluid expansion) and h(z) is the actual head change at
some point z . The total volume of water extruded from
a column of height H with a unit basal area is then 4

q = s: 1;;
h(z)dz (8.28)

The solution of this equation is a general expression for II

the vertical shortening of a column of confining layer
when steady flow conditions are reestablished within it.
It is expressed as a volume of water passing through a
unit surface area and has the units of length. As h(z)

-II
‘I

varies linearly across the layer, h(z) = Ah(z/H), where I Original aquifer
z varies from zero at the top to H at the bottom (Figure Final aquifer
pressure
8.7).Substituting this expression into Eq. 8.28 and integ-
rating gives
Dressure
Ah, -I
AhH
q==s&!- (8.29)
2
Figure f f 8 Depth-pressure diagram fbr a confining layer
Here we note that the quantity AhH/2 represents the in response to lowered pressure in adjacent aquifers (from
area of a triangle on the depth - pressure diagram of Figure Domenico and Miftlin, Water Resources Res., v. 4, p. 563-
8.7 and is termed the “effeccive pressure area.” 576, 1965. Copyright by Amer. Geophys. Union).
 dl Dej'omiable Porous Media 167

Aquifer
50
-A]>-
-
I Final head \
a
X
u
'
0)

30 Confining layer
'-Initial head
-
r
)

3
I

u
SI
Final ,daeh
n

10
Aquifer l - - l n i t i a l head

10 30 50
Liquid Limit (LL)
70 90 Finalhead -4
Figure 8.9 Plasticity cliart for soil classification. Figure 8.I 0 Schematic illustration of isoclirone develop-
ment in confining layer in response to instantaneous low-
ering of head ;it its tipper and lower boundaries.
reduction takes place in tlie confining units. According
to Poland (1961), the process of subsidence makes avail-
able ;I volume of water that would otherwise not he The Time Rate of Subsidence The isochrone devel-
recoverable. Unlike elastic compressions, the lost pore opment on Figure 8.7 represents the degree of consolida-
volume is nonrecoverable. tion at wrious depths as a function of time. The terminal
Figure 8.9 in combination with Table 8.2 can be used isochrone represents fill1consolidation for the given head
to obtain approximate values of /3, , and the 1iydr;iulic decline in the aquifer. Hence. the rapidit), of the drainage
diffusivityfor various soil groups subject to consolidation. from tlie clay layer controls the rate of subsidence.
Figure 8.9 demonstrates the relationship between the Figure 8.10 illustrates these points for a confining layer
LJnifiedSoil Classification System and tlie plasticity chart. separating two aquifers. The schematic isoclirone tlevrl-
These Same soil groups are represented in Table 8.2, opnient for the confining layer is produced for ;I r;ipitl
which gives the results of over 1500 laboratory determi- (instant;meous) lowering of heiid in the du;tl aquifer sys-
nations by the Bureau of Reclamation (1960). The values tem, here for simplicity titken to be JO. At time equal
for relative compression A H / H , , are from drained tests to zero, the head distribution through the tliree-la!wcd
on samples that were laterally confined; that is, tlie results system is assumed to be invariant with depth. When the
indicate vertical compression. pumping stwts. the heads in the ;iquifers itre immedi;ttely

Table 8.2 Laboratory Determination of Physical Properties of Sediments

soil Hydraulic Conductivity At Effective Stress At Effective Stress

Classilkation Void Ratio, e (cm/s) of 2880 Ib/ft' of 7200 Ib/ft'
W
<i - 2.6 X lo-' -C 1.25 X <0.014 -
GI' - 6.2 x 10 2 3.3 x 10-2 <0.008 -
GM - - C0.012 <0.03
Gc: - - <0.012 <O.O2i
sw 0.37 - 0.014 5 -
sI' 0.5 t 0.03 >4 x lo-" 0.008 2 0.003 -
SM 0.48 t 0.02 7.2 X 10-l' t 4.6 X 1 0 " 0.012 5 0.001 0.03 t 0.00-i
SM-SC 0.41 t 0.02 7.7 x lo-- -c 5.8 x 10.- 0.014 5 0.003 0.029 t 0.01
SC 0.48 t 0 . 0 1 2.9 X 10 - t 1.9 X 1 0 - 0.012 -c 0.002 0.024 2 0 . 0 0 5
ML 0.63 t 0.02 5.7 x I ( ) - - 2 2.2 x I()-- 0.01 5 2 0.002 0.026 t 0.00.3
ML-CL 0.54 t 0.03 1.26 x -
10 t 6.7 x 0.01 5 0.001 0.012 2 0.0
cL 0.56 t 0.01 7.7 X 10-H 2 2.9 X 1 0 P 0.014 t 0.002 0.016 It 0.004
OL - - - -
MH 1.15 t 0.12 1.5 x lor- 2 9.6 x 1 0 " 0.02 2 0.012 0.038 2 O.o()H
CH 0.8 t 0.04 4 . 8 x lo-" t 4.8 x lorx 0.026 2 0.013 0.059 2 0.015
OH - - - -
~~

From Bureau of Reclamation (1 960)

-
168 Chapter 8 / Stress, Stratn, and Pore Fluids

lowered an identical amount, converting the head in the tion so that the drainage process approaches termination
confining layer to an “excess head” with respect to its shortly after the transient started. The quantity H in Fig-
boundaries. This starts the drainage process, both up- ure 8.11 represents the distance to a drainage face and
ward and downward, with the head (or pressure) decline is taken as the thickness of the unit for single drainage
in the confining layer described by the diffusion equation (that is, the triangular effective pressure area of Figure
8.10) or H/2 for the doubledrainage case given in Figure
8.10. Thus we notice that the time lag between the low-
(8.32)
ering of aquifer pressure and the consolidation of the
interbedded clay layers contributing to subsidence is en-
subject to the boundary and initial conditions
hanced for thick clay layers of low permeability and high
P ( z , 0 ) = P, compressibility (storativity). Conversely, the time lag di-
minishes for thin units of moderate permeability and
P(H, t ) = 0
compressibility.
P(-H, t ) = 0 Last, we may now speculate that Jacob’s c in Eq. 8.23
is of the form
The initial condition stipulates a constant value of pres-
sure P,with depth. The two boundary conditions corre- t H
c = -- (8.33)
spond to decreased pressure (assumed 0, with respect T’ m
to P,) at both the top H and bottom -H of the confin-
ing layer. so that the storage component associated with the clay
The solution to Eq. 8.32 for the conditions stated is iayer actually increases with time, reaching its maximum
well known in consolidation theory (Tenaghi and Peck, value when t/T* = 1.
1948) and in heat-flowtheory (Carslaw andJaeger, 1959).
This solution provides for the pressure decline at differ-
Example a1
ent points in the confining layer, which for any time
takes on maximum values near the two drainage bound- Consider a confining layer between two aquifers. The
aries and minimum values in the center. This is too much specific storage of the confining layer is on the order of
detail for this problem as we are here only interested in 1 X 10-j ft-’, and its thickness is 40 ft. The heads in
the average pressure decline (or consolidation) of the the aquifers are lowered about 50 fi. Maximum vertical
clay layer. This average is demonstrated in Figure 8.11. shortening of the confining layer is calculated
The ordinate represents the degree of consolidation,with
values less than one indicating the presence of residual
pressures. The abscissa is the familiar ratio of observation
y = s:H(*b’ ; 100
Ah,) = 1 x 10-5 x 40 x -= 2ft
2
time and the time constant of the unit. As T’ gets small,
corresponding to a small value for H and S, in combina- The value given ( 2 ft) is taken as the maximum subsidence
tion with a large value for K,,,the steady state is rapidly in response to the 50-ft decline in water levels. The time
approached; that is, the unit approaches full consolida- to complete 50% of this subsidence is determined from
Figure 8.11 . Here, however, some additional information
is required, namely, the vertical conductivity. Assume
this value to be 1 X lo-$ft/ddy. A consolidation of 50%
(1 ft) is associated with a value of t / P = 0.196. Thus

t - t
- 0.196
T’ (H/2)’S:

0.196(40/2)*(1 X lo-))
t= = 78 days
1 x lo-’
For 95%consolidation

-T*_t - r For a water-level decline of 5 ft/yr in each of the aquifers

(H?$)/K
Ffgure8.11 Average degree of consolidation versus the ra-
tio of time to the basin time constant.
q = S,!H( f ’”) = 1 X 10-3X 40 X -
10 = 0.2 ft
2
 8.1 Deforniable Porous Media

This represents an approximate measure of the subsi- to introduce a proportionality constant between the bulk
dence to be expected with each year's pumping. Note and pore volume change. This may be acconiplished with
now that it still requires 78 days to achieve 50% of the the following expression
annual subsidence (0.1ft) and 400 days to achieve 95%
of the subsidence. If benchmarks are releveled every (8.34)
year, the results would give the appearance that there is where Pb is the bulk compressibility, P, is the grain coni-
no appreciable time lag between the lowering of the pressibility, and the difference PI, - P, is taken ;IS the
head and the attainment of full consolidation. pore compressibility. In this expression, ( is a constmt
of proportionality and may be defined as the ratio of the
Simulation of subsidence Frequently, subsidence in water volume squeezed out of a rock to the total volume
an area is sufficiently detrimental so as to warrant some change of the rock if the deformation takes place ;it
detailed model study for prediction purposes. The first constant fluid pressure, that is, the rock is free to drain.
such study was conducted by Gdmbohti and Freeze Thus, if cequals one, the bulk volume changes are equal
(1973), who utilized a finite element model to obtain to the pore volume changes, an assumption that has
predicted drawdowns, and a finite difference model to been inherent throughout our discussions of deformable
determine the time rate of subsidence. The drawdowns media. From Eq. 8.34, the constant of proportionality
obtained from the flow model were used as time-varying may be described as
boundary conditions to solve a set of onedimensional
consolidation models. The model was applied to the sub-
sidence problem in Venice, Italy, with the results re-
ported by Gambolati and others (1974a, 1974b).Helm where K,, is the bulk modulus of compression of the rock
(1 975) has also developed simulation models for applica- body and K, is the bulk modulus of compression of the
tion to land subsidence in California. Although the data polycrystalline grains making up the porous rock. In this
demands for such models are large, the measured subsi- form, we note that if /3&is m a l l with respect to &. 5 is
dence provides a means by which the models can be approxiniately unity, and the volume of water removed
calibrated and improved over time. Such models are be- from the rock is equal to the total volume change; that
coming more common where subsidence represents real is, the incompressible grain case is recovered.
as opposed to nuisance problems, for example, the HOW With the introduction of solid compressibilities. the
ton-Galveston area. elastic coefficienrs discussed in the previous section
change considerably. These coefficients are given in Ta-
ble 8.3. For the three-dimensional results. the stress term
Three-Dimensional Consolidation uand the strain term E refer to the mean confining stress
Ehstic Properties in Defonnational Problems and the volumetric dilation; that is,
In our treatment of the consolidation equation, two as-
sumptions stand out that may now be relaxed somewhat. (8.364
First, it was assumed that the stress increment producing
the abnormal fluid pressure was vertical only. Second,
the individual solid components making up the rock ma-
(8.36b)
trix were incompressible. This last assumption means
that the bulk volume changes are exactly equal to the where ulllis the mean stress, 1.' is volume. and E,/ is
pore volume changes. Or, stated another wziy, the volume the volume strain or dilation. In addition, the following
of water squeezed out of ;I rock is equ;il exactly to the relationship now holds
pore volume change of the rock. We have exploited
this assumption, not only in our developnient of the (8.37)
consolidation equation, but in the development of the
concept of specific storage. which reduces to Eq. 8.17 for 5 equal to one. Indeed, all
The most complete model from which to obtain the expressions for the three-dimensional coefficients b' wen
important material properties is one that is fully com- in Table 8.3 reduce to the onedimensional coefficients
pressible, that is, one in which a11 the coniponents are already cited for the case where 5 = 1. This implies, in
compressible, including the solid grains PA,the fluids addition, that the testing for the otie-diniensioti;ilconcli-
the pores p,,, and the bulk volume PI,. When gmin tion is conducted under constrained conditions where
compressibilities are introduced into the material proper- the concern is with vertical deformations only.
ties, the water volume squeezed out of a rock is invariably Biot and Willis (1957) discuss the measurement tcch-
less than the bulk volume change. To incorporate such niques for the various coefficients described in Table
compressibilities in the material properties, it is necessary 8.3, whereas a liniited set of values have been reported
170 Cbapter 8 / Stress, Strain, and Pore Fluids

Table a3 Definitions of the Important Elastic Coefficients

and Their Relation to the Coefficlents Defined by Blot (1941)

From Pakidilskas and I)ommico (1 989).

Water Resources Res., v. 25. p. 203-213. Copyright by Arner. Geophys.
Union.

by Palciauskas and Domenico (1982), Domenico (1 9831, measure of the changes in fluid mass in response to
Zimmerman and others (1986), and Green and Wang changes in fluid pressure at constant stress. Except for
(1986). Table 8.4 presents an overview of these values. a slight inconsistency with the generic limestone, this
Laboratory data for both the Kayenta Sandstone and the same statement holds for 1/Q as a measure in the change
Hanford basalts are reasonably good, whereas the clay in fluid mass with respect to fluid pressure at constant
and the limestone tabulations are “generic,” that is, crude strain. Also shown in the table are combinations of param-
calculations based on tables of physical constants. The eters, R / H and 6Q. The quantity R / H is already known
rocks are arranged from left to right in order of what is as the change in fluid pressure with stress at constant
expected to be decreasing compressibility. Hence, 5 as a fluid mass. As noted in the table, the more rigid the rock,
measure of the water volume squeezed out to the volume the greater the proportion of the stress that is carried
change of the material decreases from unity to 0 . 2 3 as by the grain structure rather than the pore fluid. The
rock rigidity increases. For clays, there is virtually a one- combined parameter 6Q is defined as the change in fluid
toone relationship between volume changes and associ- pressure with respect to strain at constant fluid mass.
ated water volumes released. The coefficients 1/H, 1 / R , Here, this fluid pressure appears highest for rocks in the
and l/Q are compressibilities of sorts and, in general, intermediate range and decreases toward both the soft
decrease in the direction of rock rigidity. Thus, 1 / H as and rigid-end members. The symbol K, is used for an
a measure of the pore compressibility decreases consis undrained modulus.
tently in the direction of rock rigidity, as does 1/R as a The coefficients described are for elastic, or reversible,
 8 1 Deformable Porous Media 171
Table 8 4 Typical Values for the Elastic Coefficients

Parameter Units Clay Mudstone Sandstone Limestone Basalt

5 1 0.95 0.76 0.69 0.23
1
- (Kbrtr)- 1.6 0.045 0.0079 0.0024 0.00052
H

-1 (Kbdf)-’ 1.62 0.054 0.012 0.0095 0.0044

R
1
- (Kbar)- ‘ 0.02 0.011 0.0059 0.0078 0.0043
Q
-
R
0.99 0.83 0.67 0.25 0.12
H
6Q (Kbar) 50 83 128 88 54
-1 (Kbar)-’ 1.6 0.047 0.01 1 0.003 0.0022
K

(Kbar)-’ 0.02 0.01 0.0052 0.0024 0.0022

From Palciauskas and Ihnenico, 1989.

Water Resources RKS., v. 25. p. 203-213. Copyright by Amcr. Geophys. Union

deformations. A reversible compression is one that disap-

pears completely upon release of the stress that caused
it. With geologic materials, many of the substances we
treat within an elastic or reversible framework are not where the coefficients R, Q, H , and 6 are described in
elastic at all. The point is that for some problems, the Table 8.3. These equations may be compared with Eqs.
materials need not be elastic in terms of reversibility, but 8.21 and 8.22.
the strains have to be proportional to the stresses. That As with the one-dimensional consolidation equation,
is, Hooke’s law must hold so that a constant ratio exists the equations just given can be used to describe both
between the stress and the strains. The degree of varia- the drained and undrained response of water levels to
tion in this ratio dictates how far the real behavior departs natural loads. Expressions for the undrained response to
from the ideal one, with clays and other plastic materials various loads were developed independently by van der
generally falling into this nonideal category. Reversibility Kamp and Gale (1983) and Domenico (1983). For the
also implies that any compressions occurring under load case of compressible solids, the tidal efficiency (pore
with the pore fluid free to drain are completely recover- pressure coefficient) and the barometric efficiency may
able by injecting fluid back into the pores. Indeed, the be described as
“coefficient” describing the volume changes associated
with both compression and expansion should, in theory,
be the same. Again, this reversibility does nor coincide
with our observations for many materials. and

Flow Equations for Deformable Media

The flow equations in a completely deformable media
B.E. = 1 - ‘5 ( g)
1- =1 - ( 1 - 2) (8.4 I )

are of an identical structure of those presented for the

respectively. Thus, these efficiencies are nothing more
one-dimensional consolidation equation, with the major
than partition coefficients that account for the distribu-
differences residing in the material properties, or coeffi-
tion of incremental stress between the pore fluid and the
cients. These equations may be expressed as (Pdlciauskas
solids under undrained conditions. For vertical strains
and Domenico, 1989)
only, the moduli Kb and K,, are described for the con-
strained condition. Similarly, Bredehoeft‘s (1 967) tluid-
pressure response to earth tides stated as dP/ik at con-
stant fluid mass is given simply as 5Q, where Q is defined
and in Table 8.3.
1 3 Cbapter 8 / Stress, Strain, and Pore Fluids

8.2 Abnomzal Fluid Pressures rained fluids become subjected to a thermal field. This
in Active Depositional Environments is referred to as the thermobaric regime (Figure 8.12),
where the fluid pressures may be caused by the thermal
t3rigin and Distribution expansion of water in a low-permeabilitv environment
or by trapped waters of mineral dehydraL.on.Phase trans-
Sediment deposition and compaction occur concurrently formations such as gypsum to anhydrite or smectite to
in depositional environments, with a resultant porosity illite are generally temperature dependent and will pro-
loss. Prior to porosity reduction, the added sediment load duce free water that can take on the pressure of the
is carried at least in part by the water contained in the overburden provided the permeability is sufficiently low
sediments, causing the fluid pressure to rise above its so as to inhibit movement to area:.. of low pressure. In
normal value. This has been schematically demonstrated addition, the sedimentary pile m q undergo horizontal
with the spring analogy in Chapter 4 where, in the ab- compression, as might occur because of uplift adjacent
sence of fluid expulsion, the spring (rock matrix) cannot to the deposits undergoing burial. Such compression or
compress, and the added load is borne by the water. As tectonic strain is yet another pressure-producing mech-
demonstrated further with the example of water-level anism.
changes in response to passing trains, there is a diffusion The environments just discussed will persist as long
of pore water to areas of lower pressure so that the as the basin continues to take on added sediment. Once
sediment load is ultimately transferred to the matrix of the depositional period ends and tectonic forces contrib-
skeletal grains. From the perspective of abnormally pres- ute to uplift, the flow system will evolve eventually to
sured environments, the main concern is with the length a topographic drive system. Thus, in the evolution of
of time required for this stress transfer to take place, for hydrologic basins, it is useful to think of two end mem-
this is the time period over which anomalously high fluid bers (Coustau and others, 1975): the subsidence-induced
pressures may persist. abnprial pressure environment, as might be expected
Bogomolov and others (1978) and Kissen (1978) have in young basins undergoing rapid deposition, and the
established the framework for categorizing abnormal more mature topographically driven systems that eventu-
pressure zones in depositionally active basins (Figure ally evolve from them after uplift. Figure 8.13 represents
8.12). The meteoric regime shown in the figure is the these two end members. As noted in the abnormal pres-
familiar topographic drive system where meteoric water sure environment (Figure 8.13u), the lower water zone
enters the formations in the higher portions of the basins contains original formation waters, whereas the upper
with discharge taking place at sea level. Hence, the shore- zone is continually invaded by downward-moving mete-
line is an active area of discharge, and this discharge oric water. Following uplift and erosion and the evolution
area can recede or transgress depending on shifts in the of a topographic drive system, there is one distinct hydro-
shoreline. The compaction regime is characterized by logic system that is characterized by recharge at the out-
upward migration of pore waters, as suggested in 1899 crop areas, cross-formational discharge, and a definite
by King. The fluids so affected may have been buried transition from meteoric water to highly sitline water or
with the sediment, and the abnormal fluid pressures are brines with distance from the outcrop areas.
due in part to the added sediment load in parts of the The relationships between normal fluid pressure and
basin undergoing active deposition. geostatic pressure are demonstrated in Figure 8. 14. At
To accumulate a thick sedimentary pile such as shown any given depth I,the difference between the geostatic
in Figure 8.12 the basement rock must be in some state pressure (vertical stress) and the normal fluid pressure
of subsidence. Whatever the driving mechanism, such (neutral stress) has been defined as the effective stress
subsidence assures that the sediments and their con- in accordance with the effective stress law. In this figure,
i’

Figure 8.12 Conceptual dia-

gram illustrating various hydro-
logic regimes in an active deposi-
tional environment (from
Galloway and Hobday, 1983.
Copyright 0 1983 Springer-Ver-
lag, New York.Reprinted with
permission).
J.
Subidewe
 8.2 Abnormal Fluid Pressures in Active Depositional Environments 173
&lxl.sitional Load

0 4000 8000 12,000 16.000

- Flow due to compaction

--eFlow due to gawty
Basement

Sand
water
Pressure (pounds per square inch)
Figure 815 Relation of fluid pressure and depth in some
excesspressure oil pools (from Watts, 1948). Reprinted by
(h) permission of the h e r . Inst. of Mining, Engrs, Petroleum
Figure 813 Hydrology of (a) compacting and (6)mature Div.
basins (modified from Coustau and others, 1975).

the fluid pressure P at any given depth is 0.43 times the According to Deju (1973, the occurrence of pressures
vertical (geostatic) stress. In some geologic environ- in excess of normal is worldwide. In Europe, abnormal
ments, fluid pressures have been noted to be as high as pressure has been noted in the Aquitaine Basin, France,
90%of the geostatic pressure. In such cases, the effective in the Alps, Italy, offshore Norway (0.8psi/ft, 1.8 X 10'
stress is reduced considerably, which is generally re- Pa/m), Holland, northwest Germany (0.8 psi/ft, 1.8 X
flected by higher than normal porosities and lower than 10'' Pa/m), the Carpathians (0.96 psi/ft, 2.17 X 1 0 ' Pa/
normal bulk densities of the water-saturated rock. Figure m), and in the Soviet Union (0.8 psi/ft, 1.8 X 10' Pa/
8.15 shows the relationship between pressure and depth m). The sediment types include dolomite, evaporites,
in some abnormally pressured environments reported by carbonates, marine shales, and sandstone. In Eastern Can-
Watts (1948). At relatively shallow depths, many of these ada, an offshore zone was noted at about 3650 m with
areas are characterized by normal or near-normal fluid a pressure gradient of 0.6 psi/ft (1.36 X 10' Pa/m). The
pressures. With increasing depth, fluid pressures are seen occurrences in the United States are well known and
to be in excess of normal, in some cases approaching include California (both off- and onshore), the Green
the total overburden pressure. River Basin in Wyoming, the Uinta Basin in Utah (0.8
psi/ft, 1.8 X 10' Pa/m), and several locations in Texas
and Louisiana. Some detailed studies are provided by
Watts (1948) for the Ventura oil field in California and
Dickinson (1953) for the Louisiana Gulf Coast. Other
occurrences have been reported in and around Australia,
Japan, West and East Africa, the Arabian Peninsula, South
America, Southeast Asia, and the Indian subcontinent.
On the basis of this worldwide survey, Deju (1973)
presented the following conclusions:
1.Two mechanisms appear to be responsible for most
of the abnormal pressures observed: tectonic stress
and rapid deposition of massive shales.
2. High fluid pressures of tectonic origin are llsudll)'
associated with the major mountain systems of the
Pressure (pounds per square inch) world (Pyrenees, Alps, Himalayas, Andes).
Figure 8 1 4 Relationship between normal fluid pressure 3. High fluid pressures are often associated with mud
and geostatic pressure. volcanoes, diapirs, piercement folds, and dikes.
 Next Page

-
174 Cbapter 8 / Stress, Strain, and Pore Fluids

Excess pore-fluid pressures

attributed to rapid loading 0 Possible excess
pore-fluid pressures

Figure 8.16 Geographic areas where geopressures have been attrihuted t o continuous loading and compaction, or where sedi-
ment deposition rates are such that geopressures would be predicted (from Sharp and Donienico, 1976, <ieol. Soc. h e r Publi-
cation, from Geol. Soc. Amer. Bull., v. 87, p, 390-400).

4. Rapidly deposited bentonitic shales and thick evapo- mented by the thermal expansion of the fluid, which is
rite beds seem to form the best seal for fluid in most effective in low-permeability sediments under high
the subsurface. confining pressures. With water flow t o areas of lower
5. Most high fluid pressure occurrences are restricted pressure, some redistribution of the heat occurs by con-
to Tertiary sediments, or younger. Older sediments vective trdnsport. Such convective transport can be of
have apparently had sufficient time to dissipate the importance in the thermal evolution of the sedimentary
pressure. basin. Given that the rdte of pressure dissipation by fluid
flow is slower than the rate of pressure genemion, some
Modern depositional environments associated with ab- sort o f fracture is likely. This is demonstrated in Figure
normal pressures (or the potential for abnormal pres- 8.17 with the coupling of pore pressure t o the stress-
sures) are shown in Figure 8.16. It is obvious that such strain characteristics of the rock. If the inelastic behavior
environments are rather common and were likely equally is in the form of an extensive microfracture zone, the
widespread throughout geologic time. enhanced pore volume can accommodate the volumetric
expansion o f the heated water, providing pressure relief.
Matbematical Formulation of the Problem
From a physiomathematical viewpoint, the study of ab-
normal pressure development is a study of competing

,
rates of pressure production and dissipation. The com-
plexity of the problem is well illustrated in Figure 8.17.
Gravitational loading and tectonic forces act to compress
the pore fluid by compressing or otherwise reducing the
pore itself, thus leading to the generation of abnormal
fl; Phase t r a n -
fluid expansion
h

fluid pressures. With continued burial in a thermal field, Water Flow

phase transformations produce free pore water that can Temperature distribution

take on the pressure of the overburden, and this magni- Pressure Producing Pressure Dissipating
tude of pressure cm be continually regenerated in the Gravitational loading Fluid Flow
Tectonic forces Fracture (?)
“window” that defines the critical temperature range Fluid expansion
over which the transformation takes place. The pressure Phase transformations
produced by either of these mechanisms can be aug- Figure 8.1 7 Coupling in abnormal pressure development.
Previous Page

8 2 Abnormal Fluid Pressirres in Active Depositional Etioirorir~tetils 175

If the microfracture zone coalesces to form macrofrac- Isothermal Basin Loading and
tures, the permeability may be enhanced and the pres- Tectonic Strain
sures dissipate more readily. Thus, fracture generation
In virtually all problems associated with basin loading, it
may serve as a reserve mode of pressure dissipation that
is generally assumed that one of the principal stresses is
comes into play when the rdte of pore water diffiision
oriented in a vertical direction, commonly expressed ;IS
is inadequately slow compared to the rate of pressure
generation. flf = p/Jp (8.4i)
The complexity and coupling of the processes de-
scribed in Figure 8.17 are such that ;I complete physical where the subscript v designates that the stress is vertical.
description is not possible. First, the physics of many of p. is the density of the water-saturated rock, where the
the processes described are not sufficiently well known saturation is often assumed to persist from land surface
from a quantitative perspective. For example. if one downward, g is the acceleration of gravity, and z is depth.
wishes to describe the pressure production associated The product p,,g is taken as the unit weight of the overbur-
with phase transformations, the rate of transformation den, where tlie density can vary with lithology chvigcs.
from layered or bound water at the low-temperature The stress a,.is the pressure exerted by the total weight
phase to free water at a higher-temperature phase must of the overburden (solids plus water) and is conimonly
be known. If one wishes t o examine friicturing as ;I pres- referred to as the geostatic pressure. The overburden
sure dissipative mechanism, some threshold pressure pressure niay be a major all or a minor a,, principal
must first be identified and incorporated in such a way stress, depending on whether the rocks are in a state o f
tliat describes the rate at which the fracture dissipates compression or extension. The geostltic pressure gradi-
pressure. Given that these mechanisms and rates can be ent a,./x is normally taken as 1 psi/ft (2.26 X 1 0 ' Pa/ni).
described, they can in general be incorporated into a There are a few other designations for tlie state o f
flow equation o f the form of the diffusion equation. H o w stress in the Earth's cnist. many of which are useful in
ever, at this stage of our knowledge, we must focus on the formulation of theoretical models. One assumption
these pressure-producing and dissipating processes that is that the stresses at depth are hydrostatic, that is. the
can be well described. principal components aI, , a,, a,, are all equal. This
For stress-related problems, the appropriate equation implies that principal stress differences do not increase
is (Palciauskas and Domenico, 1989) with depth, which is genefiilly not realistic from the field
perspective. Such a state of stress may exist for very soft
rocks such as shales or salts where plastic or viscoelastic
deformation permits a partial or complete equn1iz;ition
(8.42) of the principal stresses.
Yet ;inother assumption pertains to a latefiilly con-
where P,, is the abnormal or excess fluid pressure, a is strained situation where horizontal displacements are n o t
the mean confining stress, and R and H are Riot's (194 1) permitted. In the laterally constrained case, the effective
coefficients introduced in Table 8.3. A one-dimensional horizontal stress is determined from the relationship
isothermal version of this equation was developed earlier.
This three-dimensional version includes the fluid pres- V
sure changes due to the thermal expansion of water, a,,= Tr<- -i5 )
(8.
1-v
where T is temperature and a,,,is tlie change in fluid
mass content at constmt stress and pressure when the where v is Poisson's ratio. For most rocks. Poisson's r;itio
temperature varies. ranges from 0.25 to 0.33. I n accordance with Eq. 8.45.
When bulk volume strain E is taken as the pressure- the ratio of horizontal to vertical stress should lie between
producing mechanism, the appropriate expression be- 0.33 to 0 . 5 . This is contrary to most in situ mei:wrenients.
comes which lie between 0.5 and nearly one. with tlie Iower-
range characteristic of hard rocks and tlie upper-r;inge
more appropriate for soft rocks such ;ISs1i;tle and s;tlt.
Thus the horizontally constrained model niay not a1w;iys
(8.43) describe the field situation, possibly because of plastic
or viscoelastic deformation of soft rocks. or because tec-
where Q and 6 are Biot's (1941) coefficients described tonic forces can affect the horizontal stress but not the
in Table 8.3. The coefficients associated with the thermal vertical stress so that the condition D ~ , / @ >, 1 can be ob-
effect include a,]which is the thermal expansion of the tained.
rock body at constant stress and pressure. The combined As both horizontal and vertical stress generally incre;isr
term a,,,Rrepresents the change in pressure with temper- with increasing depth (McGarr and Gay. 1978). it is sonic'-
ature :it tonsrant mass and constant strain. times possible to follow the suggestion o f Jaeger and
-
176 Chapter 8 / Stress, Strain, and Pore Fluids

Cook (1969) and assume

a,,= Mu,, (8.46)
Here, M is some constant that ranges from greater than
zero to in excess of one. Thus, if M ranges between 0.3
and 0.5, we recover the laterally constrained case of Eq.
8.45. If M is taken as one, the hydrostatic case is recov-
ered. On the other hand, if M is greater than one, the
horizontal stress becomes the maximum principal stress,
as might be the case in areas of tectonic compression.
By tectonic compression, we mean the presence of
stresses that are induced by other than the simple weight 0.001 0.01 0.1 1 10 100 1000
of the overburden. The relationship given as Eq. 8.46 is Dimensionless Time -T*
.L
an oversimplification, but does permit an increase in the
Figure 829 Graphic solution to the one-dimensional con-
deviatoric stress (u,, - qJ with depth, one of the few
solidation problem, where L is the thickness of the sedimen-
reliable facts known about principal stress differences tary pile, z / L is some dimensionless depth ranging from zero
(McGarr and Gay, 1978). at the bottom of the pile to one at the top, t is time and T*
is the time constant for a basin, and P/y’L is the dimen-
One-Dlmensional Basin Loading sionless excess pressure (modified from Bredehoeft and Han-
Consider the following assumptions applied to Eq. 8.42. shaw, 1968, Geol. SOC.Amer. Publication, from Geol. SOC.
Only fluid flow and gravitational loading will be consid- Amer. Bull., v. 79, p. 1097-1106).
ered in an isothermal environment. Both the flow and
the compression take place in a vertical direction, and
the fluids and the solids are incompressible so that R/H transmitted totally and instantaneously to the water in
= .l. In addition, the stress rate a d a t is assumed to be the pores. With fluid flow, the pressure that develops at
vertical and of constant value oy’, where o is a constant any time will depend on the rate at which fluid flow can
rate of sediment supply and y’ is the unit weight of the dissipate this pressure.
submerged sediment, that is, the difference between the Equation 8.47 was developed by Gibson (1958) and
buIk weight (sediments plus water) and the unit weight applied by Bredehoeft and Hanshaw (1968) for the condi-
of water. Conceptually, the situation is depicted as a tions shown in Figure 8.18. As noted, fluid is allowed to
growing sedimentary pile in an oceanic environment (Fig- escape from the top only, with the side and bottom
ure 8.18). For these assumptions, Eq. 8.42 becomes boundaries being of the no-flow type. Further, as men-
tioned previously, the basin is laterally constrained so
(8.47) that no horizontal extensions are permitted. The solution
to this problem is shown graphically in Figure 8.19.
Prior to discussing Figure 8.19 and the conclusions
where y ’ o is the stress rate or pressure-generating term.
that might be reached from it, it is worthwhile to see if
It is noted that the specific storage S, equal to p,,.gPp
any useful information may be obtained from the govern-
is now described in terms of the vertical pore compress-
ing differential equation. From Chapter 4 , the inverse of
ibility only in that the fluid and the solids were assumed
the dimensionless group known as the Fourier number
incompressible and the deformation vertical. In the ab-
may be restated for this problem as
sence of fluid flow, the rate of pressure development aP/
at would equal y’w, suggestkg that the added load is

where T* is the time constant for a basin and is equal to

K/Sswz,and t is the depositional time to accumulate a
given thickness of sedimentary pile. Note that the dimen-
sionless group t/T* is the variable making up the abscissa
of the graphical solution given as Figure 8.19. Thus, we
see that for times t < < < P (e.g., typically less than lo-’,
Figure 8.19), there is no significant abnormal pressure
production and the basin is in the normal pressure re-
gime. The smaller the specific storage and the sedimenta-
tion rate, or the larger the hydraulic conductivity, the
Figure 8 28 Conceptual diagram of continuous sedimenta- more likely the occurrence of normal pressure. For depo-
tion in a fluid environment. sitional times t > > > T* (e.g., typically greater than
 8 2 Abnormal Fluid Pressures in Active Depositional Environments 177

100, Figure 8.191, the pressures will be approaching the 100,000,, I I I I

overburden pressure.
The conditions described here are reflected in Brede- ,000
hoeft and Hanshaw's (1068) analysis. These authors as-
sumed a Gulf Coast rate of loading of 500 m/106 years 000
for a period of 20 x 10" years, giving a sedimentary
accumulation of 10,000 m. The results of this study are 100
20 kin
shown in Figure 8.20. For an upper limit of hydraulic 10 km
diffusivity K / S , = 3.3 X lo-' cm'/s, there is insignificant 10 5 km
excess pressure production, and the fluid pressure re-
1
mains near-normal throughout the sediment accumula-
tion period (curveA). This corresponds to a time constant
P equal to 4 X years, whereas the depositional time
n .0
-. i
100 200 300 400 500
t is on the order of 20 X lo6, that is, T' > > t. On the Sedimentation Rate (meters per my)
other hand, for a lower limit of hydraiilic diffiisivity of Ffgure 821 Time constant versus sedimentation rate. Solid
3.3 X loL5cni?/s, the fluid pressures approximate the lines represent the time constant as a function of sedimenta-
overburden pressure (curve H). This corresponds to a tion rate for various values of K/S,.The dashed curves give
time constant T* equal to 4 X lo5, that is, t > > T*. the time required to accumulate 5, 10, and 20 km of sedi-
Curve C lies intermediate between these extremes, with ment for the various sedimentation rates (from Palciauskas
and Donienico, Water Resources Res., v. 25, p. 203-213,
a hydraulic diffusivity of 3.3 X lo-' and a time constant 1989. Copyright by Amer. Geophys. Union).
7+' equal to 4 X 106years.
From the graphical solution of Figure 8.19, one can
see that the limiting dimensionless numbers t/T* of lo-'
and 10' apply to any case of loading irrespective of the 5, 10, and 20 km of sediments at the given sedimentation
hydrdulic parameters used in the Bredehoeft-Hanshaw rate. Thus, we note that at a sedimentation rate of 500
(1968) study. Values of t/T' between 10 and 100 are m per million years (m/my) over a time period of 10 my,
associated with pressures approaching the overburden, the sediment accumulation is 5 km. A depositional rate
whereas values on the order of lo-' are associated with of 100 m/my over a time span of 100 my results in a 10
normal compaction. km deposit. It is noted that the time constant for sedi-
To illustrate the relationships between the variables in ments with a diffiisivit). of lo-' cm?/s is significantly
the basin lctading problem flirther, depositional time and larger than the time it takes to deposit 20 km of sedi-
the time constant have been plotted against various sedi- ments, that is, the K/S, = lo-' crn'ls line lies above the
mentation rates for a wide range of hydraulic diffusivities 20 km line. As T* > > t, these sediments will be normally
(Figure 8.21). Also plotted is the time it takes to deposit pressured. For a hydraulic diffusivity of lo-' cni'/s, the
time constant is greater than the depositional time only
for sedimentation rates less than 100 m/my, whereas for
0 larger sedimentation rates, t and T' are approximately
Curve 11 A'/s, = 3.3 10-2 cm2/s
---curve n AW, = 3.3 x 10.~cmZ/s -
equal, suggesting modest abnormal pressure develop-
Curve c KlS, = 3.3 x cmZ/s ment. Hydraulic diffusivity values on the order of 10-'
and cm'/s are characterized by time constants that
I I I I are considerably less than depositional times for most
sedimentation rates and are expected to result in appreci-
ably overpressured sediments.

Extensions of the
One-Dimensional Loading Model
There are at least three extensions to the one-dimensional
model presented that may provide further insight into
the physics pertinent to abnormal pressure development.
The first of these is for fluids and solids considered to
be compressible. This seemingly insignifrcant addition
'OO 5 10 15 20 25
Pressure Head (103m of water) has some important implications in areas of active deposi-
Figure 8.20 Pressure versus depth, Gulf Coast loading con- tion, especially with regard to the establishment of verti-
ditions (from Bredehoeft and Hanshaw, 1968, Geol. Soc. cal gradients between units of differing lithology. The
Amer. Publication, from Geol. SOC.Amer. Bull., v. 79, p. second of these extensions examines the basin loading
1097- 1 1 0 6 ) . problem within the framework of a mean stress problem.
178 Chapter 8 / Stress, Strain, a n d Pore Fluids

This implies the presence of deviatoric stresses and pro- eral percent for some rocks, it is just not as likely an event
vides an opportunity to examine the possibility of rock unless the rate of loading is very fast or the hydraulic
fracture. The third extension is the one-dimensional con- diffusivity very small.
stant strain problem, which is the other extreme of the
Horixontai Extension and Compression The ana-
one-dimensional constant stress problem.
log field condition of the one-dimensional loading prob-
Vertical Compression witb Compressible Compo- lem is the basin that is constrained laterally. This means
nents Given all the assumptions of the 13redehoeft- that the basin cannot relax tectonically in response to the
HanShdW (1968)analysis with the exception that both sediment load, nor can any form of tectonic compression
the fluid and solid components are compressible ( R / H occur during the sedimentation period. Normally in a
# 1) gives the result stated already as Eq. 8.47: tectonically relaxed state, there is little reason to expect
significint horizontal resistance at the basin borders so
that the sedimentary accumulation may react passively
to movements of the subadjacent basement, with normal
For clays, the pore pressure coefficient K / H will be suffi- faults forming along the margins. At a later stage, crustal
ciently close to unity (Table 8.4)so that there is virtually stress can become compressional, or give way to those
no difference between this formulation and the pre- that are epidermal in origin. The latter can be the result
viously described one. However, the coefficient R / H can of large differences in elevation between the bordering
range from 0.83 for mudstones to 0.25 for limestones. lands and the bottom of the subsiding basin. This may
This suggests that any incremental load in a depositional promote compressional settling or large-scale slumping
environment containing various lithologies will not be from each side of the subsiding basin, leading to compres-
totally borne by the fluid, but some part of it will be sional folding.
permanently carried by the solids, and the part carried The problem of tectonic extension or compression
by the solids will be different in different rocks. De- during loading may be examined within the framework
pending on the magnitude of this parameter, it may be of a mean stress problem (Palciauskas and Domenico,
difficult to generate high fluid pressures in some rocks 1980). For these purposes, the stress rate a a / a t is as-
irrespective of the rate of loading and limitations on the sumed to be the result of a mean principal stress rate, not
rate of flow. For example, geologists have long noted that solely the vertical one. For the case where the horizontal
ancient limestones often show little sign of compaction stresses are equal and are designated a,, the mean princi-
during their didgenetic period. This may be the result of pal stress expressed in Eq. 8.3h becomes
a pore pressure coefficient R / H considerably less than
one if the lime mud crystallizes early in its depositional = VI
a,,, -
3
+-
2a3
3
(8.51)
history so that most of the b d d h g is carried by the grains.
The increased stress on the grains can enhance pressure The relationships between the horizontal and vertical
solution in such rocks. stresses are given IS Eq. 8.46
The solution to Eq. 8.49 is likewise given by Figure
=Ma, (8.52)
8.19,where the vertical axis is replaced by P/y'l. ( R / H ) .
As R / H is less than one for rocks more rigid than clays, Here, as discussed previously, M can range from slightly
the initial fluid pressure will always be less than that greater than zero for horizontal extension to greater than
predicted by the Bredehoeft-Hdnshaw model. As the one for crustal or epidermal cornpression. For these con-
lithologies vary within the sedimentary pile, we antici- ditions, Eq. 8.47 becomes
pate large differences in vertical gradients as some units
become more highly pressurized than others.
For compressible components, the inverse Fourier
number is modified accordingly and can be expressed
where
(8.50)
C= (;) + (y) (8.54)
which indicates that the pressure production is not nearly
as pronounced for situations where R / H is less than one. It is seen that C acts as a multiplier for the effective
It is emphasized, however, that the limits esvdblished for rate of loading ( R / H ) y ' o , which is the effective rate of
the Fourier number from the results of Figure 8.19 still application of effective vertical stress. The larger C, the
hold. That is to say, it still requires a number greater than greater the abnormal pressure development. It is noted
1O-' to generate abnormal pressure due to gravitational that the graphical solution given in Figure 8.19 applies
loading; however, with the numerator decreased by sev- here also where the ordinate is expressed as P/y'L(ZM/
 &2 Abnormal Fluid Pressures in Active Depositional Environments
173
3 + 1/3)(R/H). Thus for R / H = 1 and M = 0.6 (us= Thus, for a sedimentation rate of 500 m/my and 1/R =
0 . 61%~the
~ mean stress problem yields 73%of the abnor- 1.62 Kbar-' for clays Gable 8.4)
mal pressure that would result from the laterally con-
strained case. If M = R / H = 1, the laterally constrained = - =y ' o
-ae Kbars (0.5%)
(0.1301011)
case of Bredehoeft and Hanshaw (1968) is recovered. at [R
An interesting feature of this development is that the
smaller M, the larger the deviatoric stress, and the smaller
the fluid pressure production. Later developments will
(1.62$-) (3 ;J
; = 3 x 10-'5s-'

show that failure can occur at low fluid pressures in

Thus, given the information in Table 8.4 along with
association with large deviatoric stresses, o r at higher
Figure 8.21, the relationship between the source terms
fluid pressures in combination with smaller deviatoric given as Eq. 8.56 may be used to determine strain rates
stresses. Hence, some apparent conditions for failure that develop excess pressures equivalent to those pro-
would appear to be present for virtually all states of duced by basin loading. For the example above, strain
differential stress in compacting sediments. rates on the order of 3 X s-' will produce significant
Equation 8.53 is well suited for obtaining an approxi- abnormal pressure, provided the hydraulic diffusivity is
mate idea of the effects of horizontal compression and on the order of lo-' cmz/s and smaller, and no abnormal
indicates that the pressure production can be quite large. pressure for diffusivities larger than 10-3 cm2/s (Figure
If tectonic forces are involved, the multiplier C can ex- 8.21). For this type of material (clay) with hydraulic diffu-
ceed one. Thus, for R / H = 1 and M = 1.5 (a,= 1.5u,), sivities approaching cm'/s, strain rates as low as 7 X
the pressure developed is 33%greater than one would lO-"s-l can produce abnormal pressures. For mudstones
expect from the laterally constrained case. Such pres- with small hydraulic diffusivities and 1/R = 0.05 (Table
sures would never be achieved because of some pressure- 8.4), strain rates of less than 1 X lo-'" s-' can produce
induced inelastic rock failure during the compression. abnormal pressure.
Tectonic Strain as a Pressure-Producing Mecha-
nism The other extreme of the one-dimensional con- Thermal Expansion of Fluids
stant stress problem is the problem of constant one-
dimensional strain, visualized as occurring in a layer When rocks undergo burial in a depositional environ-
whose lateral extent is very large compared to its ment, they are subjected to increases in temperature that
thickness. The layer is overlain and underlain by high- promote a thermal alteration of the component parts of
permeability material that is likewise undergoing defor- the sediment, that is, the fluid, the solids, and the pores.
mation, but remains at normal or hydrostatic pressure If the heating takes place without expulsion of the pore
due to large values of the hydraulic diffusivity. The water, the pressure changes with temperature increases
main fluid migration is taken as vertical due to the are classified as constant mass phenomenon. This as-
horizontal compression of the bed. Clearly, such a one- sumption is at the heart of much of the research in this
dimensional strain problem is only an approximation of area. The concept was introduced originally by Barker
reality, but should have some insightful features as the (1972) and advanced by Bradley (1975) and Magara
other extreme of the one-dimensional stress problem. (1975). A review of this work is given by Gretener (1 981).
The governing equation in this case may be stated as In Barker's (1972) theory, a geologic formation must
first become "sealed" at some depth and thereafter bur-
ied in a thermal field with no expulsion of its pore water.
Additionally, the assumptions of Barker (1 972) require
that the thermal expansion of water acts against a rigid,
Palciauskasand Domenico (1989) provide a solution to nonyielding matrix that remains at constant pore volume
this problem for the boundary conditions of hydrostatic throughout the burial. The pressure change with temper-
pressure at the upper and lower boundaries of the layer ature is expressed as
and an initial condition of P,, = 0 at time t = 0. The
details of this solution are not important if we are merely (8.57)
concerned with the magnitude of strain rates required
to produce fluid pressures equivalent to those associated
with basin loading as demonstrated in Figure 8.21. This where q i s the thermal expansion coefficient of the water
may be done by equating the source terms of modified and Pu.is the isothermal compressibility of water, both
forms of Eqs. 8.55 and 8.49 of which are temperature dependent (Figure 8.22). On
the basis of these arguments, a temperature increase of
25°C over a change in depth of approximately 1 km
(8.56)
(25"C/km) would cause the fluid pressure to increase by
180 Cbapter 8 / Stress, Strain, and Pore Fluids

, - ." the fluid pressure response to a geothermal gradient in

0 10.0 20.0 30.0 a granite under a large confining stress is no different
0
from that in a shale in a tectonically relaxed state. If
-0 8
c
5.4
constant mass considerations are to have any realism as
.- upper-bound calculations for pressure production, they
2 must include the expansive properties of the medium
? 16 10.8 p responsible for accommodating some of the pressure, as
k 9
N well as the boundary conditions under which the heating
2 takes place. Three possible boundary conditions present
3 24
0
16.2 x

m
-a s
Y)
themselves: heating at constant stress, at constant strain,
or under conditions that are neither constant stress nor
2 32 21.6 g
X
E
constant strain.
2
When addressing the pressure changes at constant
3 40 27.0 stress or constant strain, two avenues are open to us.

2.88
Q

We may derive the appropriate expressions directly, inde-

32.4 pendent of fluid flow considerations, o r they may be
obtained by simply ignoring fluid flow in Eqs. 8.42 and
560 100 200 300 400 37.8 8.43 and treating these expressions within the framework
of constant stress or constant strain. The end result is
a,/& (psi/"C) the same, giving the expressions
Figure 8.22 Values of a,/&. as a function of hydrostatic
pressure for various geothermal gradients (modified from (8.58)
data of Knapp and Knight, J. Geophys. Res., v. 82, p. 2515-
2522, 1977. Copyright by Amer. Geophys. Union). and

300 bars (4400 psi). This corresponds to an incremental (8.59)

pressure gradient of 0.3 bars/m (1.34 psi/ft).
In examining Eq. 8.57, both Chapman (1980) and respectively. Equation 8.58 gives the change in pressure
Daines (1982) suggested that the thermal expansion of with temperature increases at constant mass and constant
water in depositional environments is probably not a stress, whereas Eq. 8.58 is the constant mass formulation
very important pressure-producing mechanism. Their for the boundary conditions of constant strain. The coef-
statements were not based on the physics of Eq. 8.57, ficients in these equations have been discussed earlier.
but on heuristic arguments that discounted the possibility Table 8.5, taken from Hastings (1986), gives some rep-
of maintaining constant fluid mass over long periods of resentative values for various rock types. For each of the
geologic time. According to these authors, fluid flow cases shown, the change in pressure with temperature
would likely dissipate the pressures. Notwithstanding the (bars/"C) was multiplied by a geothermal gradient of
importance of fluid flow, these arguments missed the 25"C/km to give a pressure change with depth. As ex-
main point of debate concerning the appropriateness of pected, the constant pore volume case of Barker (1972)
Eq. 8.57; that is, the equation depends solely on fluid is rock independent, whereas the constant stress case
properties and is medium independent. Thus, on the yields the least fluid pressure production and the constant
basis of Barker's (1972) result, one might conclude that strain case the most. In the constant stress case, the rock

Table &5 Pressure Changes with Depth for a Geothermal

Gradient of 25"C/km for Various Rock Types and Boundary
Conditions
Rock Constant Pore Constant Constant
Type Volume stress Strain
~ ~ ~

Shale 0.423 bars/m 0.0045 bars/m 0.475 bars/m

(1.87 psi/ft) (0.02 psi/ft) (2.1 psi/ft)
Sandstone 0.423 bars/m ' 0.249 bars/m 0.479 bars/m
(1.87 psi/ft) ( I . 1 psi/ft) (2.12 psi/ft)
Limestone 0.423 bars/m 0.362 bars/m 0.471 bars/m
(1.87 psi/ft) (1.6 psi/ft) (2.08 psi/ft)
 8.2 Abnormal Fluid Pressures in Actiue Depositional Environments 181
is free to expand elastically as well as thermally to Table 8.6 for a variety of rock types. The fluid properties
partially accommodate the fluid pressure buildup. In the in these calculations were taken as averages over an ap-
constant strain case, volumetric expansion of the porous proximate 80°C rise in temperature. The elastic coeffi-
medium is not permitted. Thus the importance of the cients for these rocks are the same as those employed
boundary conditions. in Table 8.4. In all lithologies, the fluid pressure increases
Before dismissing the change in fluid pressure with with temperature at a faster rate than do increases in
changes in temperature at constant stress in soft rocks stress caused mainly by the thermal expansion of the
such as shale as insignificant, let us examine the nature solids. The greater the rock rigidity, the greater the net
of the coefficient (l/R). In elastic theory, 1 / R is a com- decrease in effective stress. For the sandstone or basalt,
pressibility that describes how the mass changes with the net decrease in effective stress is about 4 bars "C',
pressure changes at constant stress (Table 8.3). In Eq. compared to 1.54 bars O C - ' for the mudstone and 0.09
8.58, l/R represents an expansivity of the porous me- bars OC-' for the clay. Further, the elastic strain with
dium. Thus we are trapped in our elastic theory whereby temperature is obviously greater for the soft highly po-
reversibility is assumed a priori. Although this assumption rous rocks, an indication of pressure accommodation hy
of reversibility may be appropriate for most crystalline elastic volume increases. The small decrease in effective
rocks and perhaps some sandstones and limestones, it stress for soft rocks suggests that failure at some point
hardly applies in the case of soft unconsolidated sedi- map not be a likely event even under the maximum
ment. It is noted further that the constant stress case is pressure-producing assumption of constant fluid mass. It
part of the differential equation for basin loading (Eq. is noted, however, that the concept of reversibility for
8.42). Thus solutions to this flow equation or similar ones the elastic coefficients is embedded in this analysis.
where temperature effects are included, such as those It is now useful to compare these results with other
presented by Domenico and Palciauskas (lC)79), Sharp calculations for the Gulf Coast region. The calculations
(1983), and Shi and Wang (1986), focus exclusively on for the mudstone will suffice here. For a geothermal
the constant stress case and incorporate the assumption gradient of 30°C km-I, the fluid pressure increase of 2.77
of reversibility. bars "C-' translates into a thermal pressure production
The boundary conditions for constant stress or con- of 0.0831 bars m-' (0.4 psi f t - I ) . The fluid pressure pro-
stant strain are not strictly valid in field situations where duction per meter of depth would obviously be much less
rocks at depth are being heated. Rather, the displacement for clays. Gretener (1981) notes that for the prevailing
field and the stress distribution should be derived from geothermal gradient, the thermal pressure development
the stress equilibrium equation. This is generally a formid- in the Gulf Coast can range between 0.255 bars m-' (1.25
able task, but there is one situation that has been solved psi ft-I) and 0.66 bars m-I (3.2 psi ft-I). MagarJ (1975)
for the assumption that displacements far from a source cites a value of 0.285 bars m-I (1 ..Ipsi ft-I). These C d k l i h -
of heat are zero. This permits the derivation of expres- tions indicate predictions of thermal pressure develop-
sions that describe the change in fluid pressure P, the ment that are between three and eight times greater
volume dilation E , and the changes in hydrostatic stress than determined from Table 8.6 for mudstones. As noted
u purely as a function of temperature and the properties earlier, the calculations of Gretener (I 98 1) and Magara
of the medium. Some typical results are presented in (1975) are based on the constant pore volume assump-

Table 8 6 Pressure, Stress, and Strain with Temperature Changes for Various Rock Types
Clay Mudstone Sandstone Limestone Basalt

(g) 114.1r
(bar "C') 0.0956 1 .?J 5.9 9.13 8.'2

(%),, ,= (bar "C-9

11 11.23 15.5 12.58 10.1.)

(bir 0 ~ ' ) 0.18 2.77 9.1 10.39 9.34

du
-
dT
(bar "C-') 0.09 1.23 4.7 5.51 5.4

d&
ec-9 1.88 X lo-' 9.47 X lo-' 5X 2.34 X lo-' 1.5 X 10-'

From Domenico and Pdlciauskas (1988).

Gcol, Soc. Amer. Pub. Geology of N. America. 0 - 2 , p. 435-445.
-
182 Chapter 8 / Stress, Strain, and Pore Fluids

tion. We conclude that thermal pressure development Thus, if the horizontal stress is one-third of the vertical,
in very soft rocks is not very significant (unless they Eq. 8.60 predicts a fracture pressure
are heated at constant strain), but does become more
a,.+ 2P
p* = -
significant with increasing rock rigidity. (8.62)
The constant mass case gives an upper bound for the 3
development of thermal pressure in depositional environ- This states that the fluid pressure required to initiate
ments. When fluid flow occurs, we once more have com- hydraulic fracture is one-third of the overburden pressure
petition between rates of pressure production and pres- plus two-thirds of the ambient (prefracture) fluid pres-
sure dissipation. The reader is referred to Domenico and sure. AsMapproaches one, that is, as the deviatoric stress
Palciauskas (1979), Sharp (19831, and Shi and Wang approaches zero, a greater fluid pressure is required to
(1986) for details on this topic. initiate fracture.
In addition to the tensional failure described, earth
materials dilate in response to shearing strains, undergo-
Fluid Pressures and Rock Zracture ing volume changes that may be positive or negative.
The Coulomb theory of failure given in Eq. 5.17 describes This behavior has been observed in soils and both crystal-
the shear failure mode exclusively in terms of total and line and water-saturated sedimentary rocks. For crystal-
fluid pressures. This is not the only failure mode, nor are line rocks, dilation begins at a deviatoric stress of one-
other failure theories restricted to a simple effective stress to two-thirds of the stress difference ultimately required
relationship. Irrespective of the failure mode, abnormally to fracture the rock (Brace and others, 1966). Fracture
high fluid pressures can lead to some sort of rock fracture. is accompanied by a porosity increase that ranges from
On one scale is the inelastic behavior referred to as hy- 0.2 to 2 times the elastic volume change the material
draulic fracture, the accepted condition for which is fluid would have had were it perfectly elastic. This statement
pressures in excess of the least compressive stress in holds for both brittle and ductile rocks and for fracture
a regional stress field. The complete theory states that at atmospheric pressure as well as under a large total
incipient fracture propagation will occur when the fluid confining pressure. For sedimentary rocks containing
pressure exceeds the least principal stress by an amount pore fluids, dilatancy has been observed when the ratio
equal to the tensile strength of the rock. The direction of the fluid to the confining pressure is equal to about
of fracture propagation has a preferred orientation, nor- 0.8 (Handin and others, 1963). This result is reported as
mal or near normal to the least compressive stress. In the being approximately correct for sandstone, limestone,
tectonically relaxed (extensional) state, the horizontal siltstone, and probably shale.
stresses are smaller than the vertical stresses, and the Some of the features of these tests are shown in Figure
fracture growth is vertical; in the tectonically compressed 8.23. In order to interpret these results, the effective
state, they are horizontal.
At yet another scale are the pore volume increases that
materialize in the form of microfractures, commonly at
fluid pressures well below the least compressive stress
(Handin and others, 1963). As with hydraulic fracture,
this dilatancy has a preferred orientation, normal or near
normal to the least compressive stress, and is a likely
precursor in the formation of large throughgoing frac-
tures. In this section we will examine a few of these
failure theories.
The first and best understood is hydraulic fracture as
described by Hubbert and Willis (1957). According to
these authors, the pressure required to initiate fracture is

P* = u,.M - P(M - 1) (8.60)

where o,, is the vertical stress, M is the ratio of horizontal I

1 2 I
to vertical stress and is some constant less than or equal Effective Confining Pressure (lo3bars)
to one, and P is the ambient (prefracture) fluid pressure. Figure a23 Relation among effective confining pressure,
Dividing both sides by the depth z to obtain a critical deviatoric stress, and failure. Dilatancy reportedly initiated at
pressure gradient gives or previous to failure. Sandstone data from Handin and oth-
ers (1!%3); dolomite data from Brace and others (196h)
- = ( a , . M ) - P(M - 1)
P*
(8.61) (from Palciauskas and Domenico, 1980. Reprinted by pcr-
z z mission).
 8 2 Abnormal Fluid Pressures in Active Depositional Enuironrnents

confining pressure is defined as where the bracketed quantity is tlie Huhbert- Willis
(1957) fracture criteria. It will be noted here that as ‘32
a, = 0;.- P (8.63) gets large (approaches one), the deviatoric stress de-
where q. is the effective confining pressure, a,. is the creases and the fluid pressure required for the onset o f
confining pressure in the triaxial apparatus, and P is tlie dilatancy increases. This is in accord with the experinien-
fluid pressure in the pores of the sample. As noted in tal observations given in Figure 8.23. This frature is ;ilso
the figure, when fluid pressure is low compared to the incorporated in the mean stress formulation for ;tbnorni;il
experimental confining pressure (that is, a high effective pressure development (Eq. 8.53). PalciatIskils ;ind Do-
confining pressure), the onset of dilatancy requires a nienico (1980) utilize this fracture criterion to ex;iniinr
kirge deviatoric stress. As fluid pressure increases (that the initiation of microfractures in oil-bearingsource rocks
is, a lower effective confining pressure), tlie onset of at different stages of matuntion. Last, Eq. 8.66 suggests
dilatancy requires a smaller deviatoric stress. The im- that an upper limit of fluid pressure is ;ibout 0.8 u,. That
portant feature here is that fdure is once more associated is, niicrofracture generation at fluid pressures on the
with high fluid pressures at small deviatoric stresses or order of 0.8 u,. may represent an instability from which
low fluid pressures at large deviatoric stresses. there may be no way to further increase fluid pressure.
There are several other factors that dictate the onset
of dilatancy, including the density of the sediments at Phase Transformations
the time of failure (Zoback and Byerlee, 1976) and the
degree of packing as reflected by the porosity (Scott, Included among the main phase transforniations that
1963). Ideally, one might expect a physically based con- presumably lead to excess pressure development are
stitutive relation expressed in terms of a critical porosity the gypsum-anhydrite transformation (Heard and Ruby.
for a given deviatoric and mean stress. Unfortunately, 1966) and the montmorillonite-illite transform;ition
such a relationship is not available. There are, however, (Powers, 1967; Burst, 1969; Prichett, 1980). Both trans-
other appraoches to the problem. formations are thermally driven and result in the produc-
One approach is to consider tlie dilatant behavior in tion of pore fluids. With niontmorillonite. the expulsion
an empirical form suggested by Scott (1963). The basic of one interlayer of water decreases the 1:tyer spacing by
assumption is that the inelastic volume change is a fiinc- about 2.5 Angstroms, resulting in a possible 20%decrrase
tion only of the deviatoric stress. Assuming u12and q,,, in the mineral grain volume per layer. Hence, the poten-
are equal and designating the horizontal stress as u j r . tial water volume release is considerable. In the Gulf
this may be expressed as Coast, the alteration of niontmorillonite to a mixed layer
clay begins at temperatures on the order of 60 to 70°C
(dV/Qicbhtic = -Dd7* (8.64) Although it has not yet been possible to qllilntify the rate
where V is volume so that d V / V is dilatant strain, D is at which pressure is produced in this transformation.
a dilation coefficient that can be positive or negative it can be speculated that the race of transformation is
depending on the material, and 7* is the deviatoric stress. proponional to the decrease in effective stress per tinit
The coefficient D is not likely to be a constant for a given time interval.
material and will be somewhat strain dependent. When Figure 8.24 illustrates this last argument where the
D is negative, the inelastic volunie change acts to increase contained water goes from the interlayered state in the
porosity with increasing deviatoric stress. However, to lattice to a state of free witer. Bound water in a solid
perform calculations, some threshold pressure has to first structure supports a finite effective stress that goes t o
be identified and incorporated in such a way that de- zero in the free water state. Assuming the pore water
scribes the rate at which the fracture dissipates pressure. produced is not free to drain, it should immediately t;tke
A useful limiting expression for the initiation of frac- on tlie weight of the overburden. It appears further that
ture is the experimental observation of Handin and ochers this magnitude of pressure can be continually regener-
(19631, where the onset of dilatancy occurs when tlie ated in tlie “window” that defines the critical pressure
fluid pressure approaches 0.8 of the confining pressure, range in a subsiding basin, If it can be assumed that the
thiit is, transformation is described by first-order kinetic> o f thc
Arrhenius form, the size and depth of this window is ;I
Pcdlica, = 0.80, (8.65) function of certain kinetic factors along with temperiiture
where the critical designation implies ;i fluid pressure and the rate of burial. In Figure 8.25. cunr I indicatc~
that tlie reaction occurs at near surface teniper;itures ;rnd
necessary for the onset of dikatancy. If the confining pres-
sure is taken as equivalent to the horizontal stress, the the transformation is completed in the upper 1000 ni of
fluid pressure required for the initiation of microfrac- burial. In this development, N represents the number of
tures become expandable layers after time t, IV,, is the original amount,
A is the frequency factor, E is the activation energy. and
= 0.8[u,,.M- P(M
Pcn,ic2, - l)] (866) w is the rate of burial. This rapid depletion of smectite
184 Cbapter 8 / Stress, Strain, and Pore Fluids

pressible than the solids, we anticipate a net increase in

-
-0
c.
W
pore space. This will act to lower fluid pressure. At the
a8 same time, the fluid will expand due to the reduced
na
ai
0 stress, but will also contract somewhat due to a lowered-
0
8 temperature environment. Both these processes will act
Montmorillonite Before Diagenesis
to lower the fluid pressure even more. The major ques-
((1)
tion, as yet not satisfactorily answered, is whether the
increased pore space is sufficient not only to reduce the
c. pressures to their normal value, but to create a state of
W

-0
c
suction wherein the pressures are less than atmospheric.
ig If such a state can indeed be produced, the driving forces
"48 that take over are dominantly those that we associate
with the unsaturated zone. For such cases, we anticipate
After Diagenesis to lllite movement or drainage of the fluids downward due to
(0)
gravity and some sort of redistribution within the unsatu-
rated portions of the rock. Ultimately, with time, the
W
I
upper parts of the rocks will be drained of their fluids
c

8 (except for those held by capillary attraction) and-due

9
7SO
to the accumulation of fluids at depth-the formation of
Volume Lost I a deep saturated zone with a water table significantly
below land surface.
After Diagenesis and Compaction Neuzil and Pollock (1983) presented an analysis that
(C)
demonstrates that subatmospheric pressures can develop
due to the unloading of hydraulically tight rocks. It is
-
0Unit layer of clay
cWater noted that an initial condition for this analysis was hydro-
Figure 8.24 Schematic diagram of clay diagenesis showing static conditions; that is, the process did not start from
(a) virtually all bound water with no porosity or permeabil- the abnormally pressured state, but €rom a normally pres-
ity, (b) the generation of free water at the expense of the sol- sured one. Senger and Fogg (1987) and Senger and others
ids, (c)free water squeezed out upon completion of compac- (1987) discuss the effects of uplift, deposition, and ero-
tion (from Powers, 1967. Reprinted by permission). sion on the development of underpressured basins.

is contrary to observations of smectite-illite occurrences

in Gulf Coast sediments. Curve 11, on the other hand,
shows the reaction delayed to about 3000 m and then
progressing to near completion over the next 1500 m of
burial. Irrespective of the uncertainty regarding the exact
location of the critical temperature range, Figure 8.25 I II
suggests the presence of a rather narrow window associ-
ated with it. ?2-
A(SP
E(kcal/mole)
1
19.2
3 . 3 1014
53.0
~
-80
ua-
m
w(m/Io6yr) 100 100
e
2
v
L

Subnormal Pressure - 110 j

If geologic loading can give rise to abnormally high fluid
pressures, erosional unloading and eventual uplift may
result in abnormally low pressures, referred to earlier as - 140
subnormal pressure (Russel, 1972). The mechanics can
be explained by viewing the compaction process in re-
verse. With erosional unloading and uplift, the stress is
removed from the porous matrix and from the solid
grains. Hence, the pores may expand somewhat, as will Reaction Rate, - -
d("N,) ( 1 0 6 yr)-l
dt
the solid grains. As the horizontal stresses are generally Figare 8.21 Transformation rate versus depth for smectite-
increased relative to the vertical ones during erosional illite (from Domenico and Palciauskas, 1988, Geol. SOC.
unloading, it is difficult to describe the details of this Amer. Publication, from Geol. of N. Amer., 0-2, p. 435-
process. However, as the pores are generally more com- 445).
 83 Pore Fluids in Tectonic Processes 185

Irreversible Processes - Pore Fluids in Tectonic Processes

8.3
Except for our section on rock fracture, the discussions Walder and Nur (1984) posed three important questions
thus far have been exclusively concerned with elastic regarding fluids in the earth’s crust:
deformations characterized by elastic niaterial properties.
1. To what depths in the earth’s crust is free w;iter
It lids been noted by Neuzil (1986) that different time
present?
scales of deformation are characterized by different mod-
uli. Short-duration phenomena such as atmospheric pres- 2. What is the value of tlie fluid pressure ;it these
sure variations or tidal disturbances are described by depths?
elastic models that contain undrained moduli. Intermedi- 3. What is tlie permeability of tlie rocks ;it these depths?
ate-duration phenomena are likewise described with elas-
tic models, but as there is sufficient time for fluid flow In that actual measurements of crustal properties exist
to take place in response to the deformation, the appro- only to depths of 2-3 km, any answers to these questions
priate moduli are the drained coefficients. The defornia- must be based on indirect evidence. Such evidence col-
tions associated with the drained parameters are signifi- lected by these authors suggests that free water is pres-
cantly larger than in the tindrained case. It is noted that ent, at least episodically, at upper and niidcrust;il depths.
tliese same elastic coefficients have been used to describe The pressure of this fluid is thought by many to be ;ibnor-
phenomena that operate over geologic time scales. How- nial, and by others to be at normal pressure. Brace (1980).
ever, we are aware that over time periods measured in for example, states that zones of permeability on the
terms of tens of thousands of years, irreversible nonelastic order of m2 must exist to depths of at least 1 0 k m .
processes will occur at the same applied stress that is He therefore concludes that this permeability is too
driving the elastic response. These irreversible processes high for tlie maintenance of abnormal fluid pressures.
will result in considerably greater deformations, for tlie However, uniform values of permeability to midcrustal
same applied stress condition, than would be expected depths-high or low-are not to be expected, and not
from their elastic counterparts. For the most part, it ap- ;ill rocks have direct permeable pathways to the Earth‘s
pears that we have ignored these greater deformations surface. Further, from what has been learned in the sec-
and have focused on the lesser ones, and some explana- tion on active depositional environments, it is clear t1i;it
tion is in order. if some process or processes can generate fluid pressure
The reason for this apparent shortcoming is that it is at a faster rate than fluid flow can dissipate it, the pres-
most difficult to determine tlie microscopical laws gov- sures will be maintained at abnormal proportions. We
erning irreversible deformations. On the other hand, Neu- already discussed those processes that are likely responsi-
zil(1986) suggests that such new laws may not be neces- ble for such generation. Whether or not they or different
sary, and only different coefficients are required. These ones operate at midcrustal depths is another question.
arguments suggest that both short- and long-term defor-
mations can be described within elastic theory and the
Fluid Pressures and Tbrust Faulting
concept of effective stress, but that the constants applica-
ble over long time frames will be different from those that A thrust fault is a discontinuity in a rock mass character-
apply to the short term. There still remains the problem of ized by a low dip and a net slip generally measured in
obtaining the constants for the long-term deformations. miles, sometimes approaching 100 miles. The nirc1i:i-
it is obvious that they cannot be measured in tlie 1,‘1bO M - nisni for overthrusting presents a problem in mechanics
tory but may be determined from field studies; for exam- in that tlie force necessary for displacements of such
ple, porosity versus depth determinations in thick se- magnitude far exceeds the strength of the rock itself.
quences of rock. This has been the method of Slii and Thus, in the words of Smoluchowski (1 909). ”we m;iy
Wang (1986) with the use of Athy’s (1930) relation for push the block with whatever force we like: we may
porosity versus depth in shales (Chapter 2). Yet another eventually crush it, but we cannot succeed in moving
way is to devise a new microscopic law that “predicts” it.” With this constraint, two lines of reasoning were
porosity variation with depth in response to some as- followed to explain such displacements: gr~ivit;ition;il
sumed process that is operating. This was the method sliding down an inclined slope, and the lubrication o f
of Palciauskas and Domenico (1989). Their calculations tlie thrust plane, usiially by water. In other words, these
indicate that an irreversible “compressibility” coefficient propositions were concerned with reducing the sliding
for such deformation is almost two orders of magnitude frictional resistance between two rock masses. As dis-
greater than its elastic counterpart. This means that on cussed by Terzaghi (1950) in our earlier treatment of
a macroscopic scale, the irreversible deforniational char- landslides, water may not act as a lubricant but ;IS;in ;inti-
acter of a rigid sandstone is similar to the elastic deforma- lubricant.
tion of a rock whose stiffness lies intermediate between Hubbert and Rubey (1 959) invoked the Molir-<~ouloml~
a soft clay and a mudstone. theory and the same failure criteria as Terzaghi (1950)
3 Cbapter 8 / Stress, Strain, and Pore Fluids

to present the case that high fluid pressures can suffi- be realized at progressively lower fluid pressures over
ciently reduce the shearing resistance of a given rock geologic time. Hence, the process may be a decaying
mass so that large-scale displacements are possible. In one that ultimately dies out when the compression finally
their overthrust concept, fault blocks could be moved takes place at or near constant fluid pressure. The analogy
large distances by relatively small forces, provided the to this terminal situation is the normal compaction con-
fluid pressures were sufficiently high to put the rocks cept put forth earlier.
in a state of floatation. The special geologic conditions Whatever the final consensus-if any-there is little
required for the attainment and maintenance of such doubt that the Hubben and Rubey hypothesis as devel-
pressures were cited as (1) the presence of clay rocks; oped from the original work of Tenaghi (1 950) generated
(2) interbedded sandstone; (3) large total thickness, and a flurry of interest in the mechanics of the overthrusting
(4) rapid sedimentation. These conditions represent the and the strength of rocks in general.
essence of the depositional environments discussed in
the previous section.
Seismicity Induced by Fluid Injection
Since this original effort by Hubbert and Rubey, the
embellishments added by others have taken two rather Among the many thousands of fluid injection wells in
distinct lines. The first is a search for other causes of the world, there are a few that have drawn attention as
high fluid pressures needed in the overthrust concept a result of the earthquakes that occur in conjunction
within the same physical framework of the Mohr- with them. Three such wells or well sites in particular
Coulomb theory. Thus, Hanshaw and Zen (1 965) propose have been studied rather extensively: a deep waste injec-
that osmotically induced pressures across shale beds are tion well at the Rocky Mountain Arsenal near Denver,
an important cause of abnormal pressure where high- Colorado; experimental injection wells in the Chevron
and low-pressure zones are induced on opposite sides of Oil Field near Rangeley, Colorado; and a high-pressure
a shale bed. From a purely mechanistic viewpoint, it injection well used for hydraulic mining of salt in the
is not difficult to conceive of several other causes of Attica-Dale region of western New York.
abnormal pressure. In 1961, a deep disposal well (3638 m) was constructed
The second line of research is a search for mechanisms to penetrate highly fractured Precambrian schist and
other than the fluid pressure hypothesis or, in some cases, granite gneiss bedrock underlying the Denver Basin. In-
a modification of the original concepts. In this regard, jection was initiated in March 1962. During the first
Chapman (1981) exhumes lubrication, and Guth and oth- month of operation over 15,000 m3 (4 X 10' gallons)
ers (1982) argue that there are limits to the amount of were injected with well head pressure ranging from 0
fluid pressure development before some sort of fracture to 72 X lo5Pa (72 bars, 10.4 psi). During the first month
begins and releases the pressure. Gretener (1972) sug- of operation, Denver experienced its first earthquake in
gests that high fluid pressures could not be maintained 80 years (Simon, 1969). The injection period from 1962
under a thrust for long periods of time and argues that to 1966 can be divided into four stages (Hsieh and Brede-
the thrust does not move all at once, but like a caterpillar, hoeft, 1981). From March 1962 to September 1963, the
one segment at a time. The suggestion here is that abnor- injection took place under pressure. Between October
mal pressure may be episodic. 1963 and September 1964, no injection took place. From
It may be noted further that compressional forces asso- October 1964 to March 1965, injection took place by
ciated with thrusting are sufficient to generate high fluid gravity flow. Injection was stopped in April 1965, after
pressures that will inevitably fracture the rock, causing D. Evans publicly suggested a relationship between fluid
large-scale displacements. Thus, the origin of the fluid injection and over 700 earthquakes generated in the Den-
pressures may be tectonic compression as opposed to ver area over this short three-year period (Figure 8.26).
basin loading, which has the added attraction of con- Evans (1966) invoked the Mohr-Coulomb failure the-
taining the means by which the overthrust occurs. If ory of the Terzaghi (1950) and Hubbert and Rubey (1959)
the fracturing temporarily acts to relieve the pressure, hypothesis to explain the fault movements and induced
continued compression may eventually build it up again, seismicity along historically dormant faults. The effect
causing a repetition of the thrust and the caterpillar type of increased fluid pressure is to reduce the frictional
of motion. In each event the pore volume that is gener- resistance to fracture by decreasing the effective normal
ated at fracture may be just sufficient to equal the volume stress across the fracture (Byerlee, 1967).The continuous
necessary to return the rock to its ambient pressure when seismicity after injection stopped was explained by Healy
the tensile strength of the rock was first exceeded. As and others (1968), who suggested that near vertical frac-
long as the rocks remain in a state of compression, this tures, preeanhquake in origin, have long been subjected
process wiil continue, with the fractured state of the to tectonic stress. This stored strain was reledsed by the
rock at any time representing an instability from which propagation of built-up fluid pressure throughout the
it is not possible to increase fluid pressures further. It is reservoir after injections had ceased.
expected that with increased fracture, this instability will The association of the Denver earthquakes with fluid
 83 Pore Fluids in Tectonic Processes 187

Pumps off
- 60

30
#3 injected - 50 -
-30
e
- 20
10
10
5
z
0
46810122468102830135791113151718
Aug. 1971 Oct. Nov. 1971
Figure 8.28 Number of earthquakes per Ghr interval plot-
ted with top hole pressure in injection well, western New
York (from Fletcher and Sykes, J. Geophys. Res., v. 82, p.
Figure 6126 Comparison of fluid injected and the fre- 3767-3780, 1977. Copyright by Amer. Geophys. Union).
quency of earthquakes at the Rocky Mountain Arsenal. Up
per graph shows monthly volume of fluid waste injected in
the disposal well. Lower graph shows number of earth- Raleigh and others (1972, 1976). Raleigh (1971) provides
quakes per month (from Hsieh and Bredehoeft,J. Geophys. a good overview on this topic.
Res., v. 86,p. 903-920, 1981. Copyright by Amer. Geophys. Fletcher and Sykes (1977) reported a rash of earth-
Union).
quakes of small magnitude in the vicinity of an injection
well in western New York. This well is only 430 m deep.
injection prompted a renewed interest in the response The relationship between the fluid pressures and the
of earth materials to abnormal pressures. An experiment earthquakes are shown in Figure 8.28.
was quickly established at the Rangeley Field where In examining the role of water in rock strength, Rojs-
Chevron had been injecting fluids to enhance produc- taczer and Bredehoeft (1988) note that failure occurs at
tion. It was noted that a prevalent zone of earthquake the three sites just cited when the fluid pressure is a
activity persisted near the edge of the field. Four wells mere one-third to three-fifths of the greatest principal
were chosen for the experiment, and testing started in stress. Such modest fracture pressures in rocks suggest
1969 and continued through 1973, in which the fluid that these rocks already possess a large deviatoric stress,
pressures were alternately raised by injection and low- perhaps to the extent that they were v e n close to failure
ered by pumping (Figure 8.27). As noted, the frequency at ambient fluid pressures. Hence, the perturbations nec-
of earthquake activity was dramatically increased when essary to induce failure were not large. On the other
the fluid pressure within the reservoir exceeded 275 X hand, with permeable materials, the pressure-producing
lo5 Pa (275 bars). The experiments are described by mechanisms must operate rather quickly lest the fluid
pressures drain off. Injection wells are perfectly suited
for the establishment of a rapid fluid pressure response to
200 350 increases in water volumes, especially in rocks of modest
RA l l earthquakes
Quakes within 1 km of bottom
U of experimental wells
permeability and low compressibility.
300 5
g 150 2;
Z
In-E Seismicity Induced
x
3
.c
U
250 zz in the Vicinity of Reservoirs
5 gx
0
100
Threshold needed to
initiate earthquakes
200 ;
23
Since the fluid pressure mechanism of induced seismicity
was first postulated in association with the Denver earth-
2 5-
5 150
5 E
gh quakes, fluid pressure contributions to fault movements
50 sg have been inferred at tens of sites throughout the world
r (Woodward and Clyde, 1979). The first such observations
100
were at Lake Mead, where seismic activity started shortly
0 70 after the filling period commenced. In 1937, with the
1969 1970 1971 1972 1973
lake 80%filled, 800 separate earthquakes were recorded
)-Fluid ---.(+Fluid+/+- Fluid injection 4 over a two-month period, one of which had a magnitude
injection withdrawal
Figure 827 Earthquake frequency at Rangely Oil Field and greater than 5. As of 1975, 10,000earthquakes have been
its relation with induced reservoir pressures (from Wallace, recorded (Anderson and Laney, 1975). Induced xis-
1974). micity of a magnitude of 5 or greater has the potential
-
188 Cbapter 8 / Stress, Strain, and Pore Fluids

to cause failure of the dam itself. In 1967, the Konya (1988) summarize those factors upon which failure at
Reservoir in India induced an earthquake of magnitude such depths will depend: (1) the rate and duration of
6.5, which caused extensive damage and claimed 180 the pressurization mechanisms, (2) the permeability and
lives (Carder, 1970). Schleicher (1975) claims the Pali- compressibility of the rock in which it operates, (3) the
sades Reservoir in Idaho also triggers earthquakes, with degree to which the process is isolated from the surface
the number of occurrences increasing in the spring when of the earth, ( 4 ) the orientation of the fault planes relative
the reservoir is at its highest level and in the fall when to the principal stresses, and (5) the degree of difference
it is at its lowest level. Numerous other occurrences are between the greatest and least principal stresses. In this
cited by Carder (1970). regard, items 1 and 2 dictate how close the pressure
In examining earthquake activity in the vicinity of sev- generation will take place to constant mass or constant
eral dams, Roth (1969) drew the following conclusions: pressure; item 3 refers to the ability of deep rocks to
dissipate fluid pressures due to a hydraulic communica-
1. Seismic activity seems to be more commonly associ- tion with the Earth’s surface, item 4 is purely a geometric
ated with reservoirs deeper than 100 m, with the argument, and item 5 puts constraints on the pressures
total water volume of less importance than the required to promote failure.
height.
2. Seismic activity generally reaches a maximum after
the first filling of the reservoir and generally dimin-
The Pbreatic Seismograph:
ishes in the course of a few years.
Earthquakes and Dihtancy Moriels
3. In many cases the strongest shocks occur after many
foreshocks, with the frequency of the foreshocks Earthquakes occur when elastic strain in a deforming
being greater in areas of least tectonic activity. medium is released, resulting in intense seismic vibrd-
tions. When such earthquakes occur, seismic waves are
Seismic inducement at reservoirs has generally been released, some of which (so-called P and Rayleigh waves)
associated with the following conditions: have compressional and dildtional components, which
causes both compression and dilation of the materials
1. The rocks at the site are under a large deviatoric
through which the waves pass. Such waves can travel
stress and very near their failure strength or on the great distances and can cause water-level fluctuations in
verge of sliding on preexisting fault planes.
wells located far from the source of the earthquake.
2. The rocks may be associated with potentially ac- Blanchard and Byerly (1935) were the first to suggest
tive faults. that a float connected to a recording instrument would
3. The triggering force is most likely an increase in record the dilational components of passing seismic
fluid pressure. waves. These authors recognized that different wells
would produce different responses to the same wave,
Most authors again call on the Mohr-Coulomb theory to depending on the character of the well itself, and the
explain the phenomena. Some advanced reading on this hydraulic properties at the point of recording. Their solu-
topic is provided by Bell and Nur (1978) and Gupta and tion was to correct these responses empirically by com-
others (1 972). paring records from various well responses and seismo-
graphs. In this regard, Edton and Takasaki (1959)
Seismicity and Pore Fluids established empirical relations between the maximum
amplitude of a water-level fluctuation, and the type, wave-
at Midcrustal Depths
length, and amplitude of the waves producing the re-
The information obtained from the “experiments” on sponse. They noted that the response of the water level
induced seismicity suggests that failure along preexisting decreases rapidly as the wavelength of the passing seis-
faults occurs at rather modest increases in fluid pressure. mic waves decreases.
These “experiments” also suggest that in most if not all Cooper and others (1965) were the first to address
cases, the rocks are subjected to large deviatoric stresses, this problem quantitatively with an analytical solution.
which accounts for the modest fluid pressure require- This solution describes the manner in which a water
ments. Given the same or similar states of deviatoric level will respond to a passing seismic wave, and this
stress in the crust, fluid pressures that promote failure response is a function of the properties of the well, the
must be elevated at least to the levels as suggested by hydraulic properties of the aquifer, and the type, period,
the induced seismicity studies. Given smaller deviatoric and amplitude of the wave.
stresses, even greater fluid pressures are required. As little The work of Cooper and others (1965) was extended
is known about the fluid pressure and stress conditions at by Bodvarsson (1 970), who attempted to examine con-
crustal depths, little more can be stated about what is fined fluids as strain meters. Bodvarsson noted, like Coo-
actually occurring there. Rojstaczer and Bredehoeft per and others (l956), that a well’s seismograph cannot
 Problems 189
be compared to true seismographs unless the well-aquifer is manifested by an increase in the mean confining stress
system is well known. A few years later, Nur (1972) acting on a saturated unit volume. There is little question
suggested that water-level fluctuations to such waves may that the stress contributing to strain is increasing at a
be helpful in predicting the occurrence of earthquakes. faster rate than the fluid pressure within the unit volume.
Nur based his theory on the observation by Sadovsky and This is demonstrated from Eqs. 8.42 and 8.43 for constant
others (1969) as well as others that prior to earthquakes, fluid mass
the ratio of the seismic velocity of the primary and sec-
ondary waves decreased dramatically and, following the (8.67)
earthquake, resumed a normal level. The velocity of the
primary wave is greatly affected by the degree of water where R/H is generally less than one. This suggests that
saturation, but the velocity of the secondary wave is not. the effective stress is undergoing a net increase. How-
Thus the intuitive conclusion that prior to an earthquake, ever, if we speculate that the minimum principal stress
some mechanism must be operative that causes a de- is undergoing little change, the overall effect of this pro-
crease in saturation of the rocks so affected. Preearth- cess is to increase the deviatoric stress progressively at
quake dilatancy is one mechanism that can account for a faster rate than the fluid pressure increase. When the
this, where just prior to deformations that produce eanh- critical deviatoric stress is achieved for the fluid pressure
quakes, the rocks undergo inelastic volume increases that is generated, the dilatant behavior is triggered. In this
produced by the formation of new fractures. As men- model, we have imposed the same conditions riorm:illy
tioned previously, dilation of crystalline rocks begins at employed with the triaxial apparatus used in the 1,'I bora-
a deviatoric stress of one- to two-thirds of the stress tory. It is expected now that the water level in the well
difference required to fracture the rock (Brace and oth- will fall as the fluid drains to fill the newly opened cracks.
ers, 1966). This fracture is accompanied by a porosity It is not possible to state the volume of new openings
increase that ranges from 0.2 to 2 times the elastic volume that are created during the dilatant period. One obvious
change the material would have had were it perfectly estimate is that the pore volume that is generated is equ;il
elastic. Hence, the flow of water from the preexisting to the volume needed to return the fluid to its itmbie11t
pores or cracks into the newly formed pores or cracks can pressure when the tensile strength of the rock WI: ' 4 ex-
cause the rock to become temporarily undersaturated, ceeded. With continued tectonic strain, the fluid pressure
which effects the velocity of the primary waves. Nur starts to rise again, but it can never rise to a level higher
(1972) argued that as water moved into the dilatant re- than it was when the rocks first became dilatant. That
gion, the pore pressure would rise, lowering the effective is, the fluid pressure that initiates dilatancy represents ;I
confining pressure so that continued strain could cause theoretical upper limit to fluid pressure development lest
the failure that triggers the earthquake. A further modifi- the rocks become dilatant again. However, this is 110
cation of this model is given by Scholz and others (1973). longer the same rock with the same strength characteris-
It thus follows that if we had a water monitoring point tics that it possessed previous to its dilatant behavior.
in a rock mass during the sequence of events leading to For one thing, it appears likely that diliatancy represents
an earthquake we might notice the following behavior. a loss of cohesive strength, thereby reducing the resis-
First, we assume the rocks are undergoing some tectonic tance to shear. Hence, this second generation o f fhiicl
strain and if the deformation is taking place at or near pressure rise may result in failure with conseqLlent earth-
constant mass, or at least faster than the fluid can be quake production.
expelled, the fluid pressure will rise accordingly. Thus Observations of the relationships between w m r well
we would notice a rise of the water level in the monitor- fluctuations and seismic events date back se\.eraI years.
ing point. Now to go further we must speculate on the Modern observations have been repofled b!. Vorhis
dilatancy model. We presume that the increased strain (1 955). Gordail ( 1 970), and Wesson ( 1 98 1 ).

Problems
1. An elastic aquifer is overlain by 100 ft of clayey c. Ascertain the bulk modulus of compression of the
material. The aquifer has a storativity of 5 X lo-' and a aquifer.
porosity of 0.3 and is 200 ft thick. The clayey material
has a specific storage of 4 X lo-' ft-'. 2. Perform the calculation in Example 8.1 for ;I cl;~!~
a. Calculate the elastic compression of the aquifer in layer 60 ft thick. If the benchmarks are releveled even'
response to a 100-ft decline in head. year, would the results give the appearance that there is
b. Calculate the inelastic compression of the clayey no time lag between the lowering of the head and the
material for a 100-ft decline in head. attainment of full consolidation? Discuss your conclu-
-
190 Cbapter 8 / Stress, Strain, and Pore Fluids

sions in terms of the time constant T' and the observation c. Coniparing (a) and (b), which has the largest time
time t. constant T'?
3. The barometric efficiency of a well in an elastic aqui- d. What is the value of T*/t for K/S, = lo-' cm'/s and
fer is 0.4. If the porosity is 0.1 and the thickness is 91 a sedimentation rate of 500 m/my for a 5-km deposit?
m, estimate the storativity. e. For problem (d), will this situation be characterized
by excess pressure? Why?
4. Reduce Eq. 8.38 to the more familiar form of' tlie
diffusion equation given as Eq. 4.31. State carefully all 9. What assumptions are required to reduce Eqs. 8.42
the relationships and assumptions required. and 8.43 t o Eqs. 8.7 and 8.8?
5. Derive Eq. 8.9 starting from Eqs. 8.7 and 8.8. 10. Reproduce curves A, B, and C of Fig. 8.20. Take y
as 2.3 g/cm5. Then reproduce curve B if R/H = 0.8.
6. Considering sandstones, limestones, and basalts for
the assumption of incompressible grains, prepare a chart 11. A repository for nuclear waste is under construction
that shows which of these rock types have the largest at the 3000-ft level in low-permeability basalts. The total
response to atmospheric pressure changes and ocean vertical stress at this depth is 248 bars. For a water table
tides and which have the smallest. Considering the com- about 200 ft below land surface, the neutral stress is
pressible grain case, which of these rocks have the largest about 83 bars at the 3000-ft level. What temperature
response to earth tides and which have the smallest? increase due to radioactive decay will reduce tlie effec-
tive vertical stress to zero at the 3000-ft level? Assume a
7. A sedimentary layer is 10,000m thick and was depos-
constant fluid mass situation. Everything you are required
ited at a fate of 500 m / l d yr. The hydraulic diffusivity
to know about this basalt is given in Table 8.4.
is 1 m'/yr. Calculate t/T'. Determine the excess pressure
as a percentage of the overburden pressure at 2000 m 12. Prove the relations given in Eq. 8.67. Use Eqs. 8.42
and 10,000 m below the surface of the sediment. and 8.43 as your starting point.
8. Answer the following from Figure 8.21. 13. For basaltic rocks such as those in Table 8.4, calcu-
a. For K / S , = lo-' cni'/s for a 20-kni deposit, what is late the change in total stress, fluid pressure, and effective
the largest sedimentation rate required to develop stress with increased strain. Hint: See Eq. 8.67. Make
excess pressure? similar calculations for the sandstone and mudstone of
b. For K/S, = lo-' cm'/s for a 10-km deposit, what is Table 8.4. Make some general statements regarding rock
the lowest sedimentation rate required to develop rigidity and the distribution of the total stress between
excess pressure? the pore fluids and the matrix (effective stress).
 CbaDter 9

Heat Transport in Ground-Water Flow

3.I Conduction, Convection, and Equations of Heat Transport

9.2 Forced Convection
9.3 Free Convection
9.4 Energy Resources
9.5 Heat Transport and Geologic Repositories for Nuclear Waste
Storage

In the previous chapter, we discussed abnormal pres- assumed that an equilibrium exists between the solid i i d
sures that result from thermally expanded fluids and the liquid phases so that there is little or no tempecitiire
role of temperature in phase transformations. For the difference between them.
most part, however, our interest has been in isothermal As the conductance of heat is directly proportional to
flow, simply because temperatiire effects were not im- the gradient in temperature, the higher the geot1ierm;il
portant to the discussions. Our interest in this chapter gradient, the higher the conductive heat flow ;it the sir-
shifts to hydrogeologic investigations where the noniso- face. Heat flow on the shield areas has been deternlinecl
tliernial aspects are the major ones that concern us. to be on &heorder of 1 calorie per square centimeter per
Prior to getting to the main points of this chapter, a second (cal/cniLs), whereas at the midoceanic ridges. it
few well-known Pacts should be established. Lovering is on the order of 8 caI/cm?s or greater. Cogiiel (1076)
and Goode (1963) show that the effective perturbation reports that the avenge heat flux by contluction ;it the
depth of temperatiire fluctuations at the Earth's surface Earth's Surface is on the order of 1 . jcal/cni's. Vari;itions
is on the order of 10 m. Thus, only very shallow ground in surface heat flow and geothermal gr;iclients h;~vebeen
water exhibits appreciable temperature changes in re- attributed to differences in the thermal conduct;ince of
sponse to seasonal variations in the amount of solar en- various media, geologically recent intrusions of ni;igni;i,
ergy reaching the Earth's surface. This is reflected by and the influence of ground-water flow o n the t1ierni;il
the fact that the temperature of ground water at depths regime. Lachenbruch and Sass (1077) consider this Isst
ranging from 10 ni to 20 m will generally be 1" to 2°C factor to be the most important in niocliljing the he;it
higher than the local mean annual temperature (Figure flow originating at great depths in the Earth.
9.1). Below this depth. the temperature increases more
or less steadily. In general, the older and more compact
the rock, the lower the geothermal gradient. The geother- 9.1 Conduction, Convection, and
mal gradient in the Canadian Shield has been reported
as 9. 1"C/km, whereas gradients as high as 36"C/km have
Equations of Heat Transport
been reported for the Mississippi Embayment in Louisi- Heat can be transported from point to point in a porous
ana. In performing these measurements, it is normally medium by way of three processes: conduction. convec-
-
192 Cbapter 9 / Heat Transport in Ground-WaterFbw

Figure 9.2 Temperature of ground water in the United States at depths of 10 to

20 meters (from Collins, 1925).

tion, and radiation. Conductive transport may be de- to introduce an effective thermal conductivity so that Eq.
scribed by a linear law relating the heat flux to the temper- 9.1 is expressed
ature gradient. Convective heat transport is the
H, = grad T (9.2)
movement of heat by a moving ground water. Radiation,
better known as thermal electromagnetic radiation, is the where H, is an effective energy flux vector, K , is an
radiation emitted because of the temperature of a body. effective thermal conductivity, and T is understood to be
the average temperature of the solid and fluid mass.
In two phase mixtures consisting of water and solids,
Fourier's Law both of which are conductors, an effective thermal con-
Fourier's law describes the conduction of heat from re- ductivity can be described by considering the volume
gions where the temperature is high to where it is low. fractions and conductivities of the individual phases. For
This process is called heat conduction, or simply conduc- a parallel arrangement of fluids and solids, the accepted
tion, and is described for a solid or a liquid as relationship is
H= -K grad T (9.1) K, = n K l + (1 - YZ ) K , (9.3)
where H is the heat flux, T is temperature, K is the where K , is the effective thermal conductivity and the
proportionality constant relating the two, referred to as subscriptsfand s refer to the fluids and the solids, respec-
the thermal conductivity of the substance, and grad T is tively. In discussing this relationship, Slattery (1972) sug-
the temperature gradient. In the metric system, the unit gests that it must be amended to account for tortuosity
of heat is the calorie, with temperature expressed in (or a nonparallel arrangement of fluids and solids) in the
centigrades and distance in centimeters. The thermal con- porous medium. He distinguishes between two cases,
ductivity has the units cal s-) cm-'"C-l. In the English one where the fluid is stationary and one where conduc-
system, the unit of heat is the BTU (British thermal unit), tion is accompanied by fluid movement. For a nonmoving
and temperature is measured in degrees Fahrenheit. The fluid, the effective thermal conductivity becomes
thermal conductivity is expressed in (BTU) hr-'ft-'"F-l.
In SI units, the unit of heat is the joule with temperature
K, = BK, + (1 - B ) K , - K,K* (9.4)
expressed in degrees kelvin K so that thermal conductiv- where K* accounts for the reduction in free transport
ity has the units joules rn-Is-l"K-'. because of the tortuosity of the porous medium. For a
When dealing with fluids within solids, it is sometimes moving fluid, a convective effect referred to as thermal
necessary to distinguish between energy transport in dispersion is noted. Thermal dispersion is a microscopic
both the fluid and solid phase. We will assume that the dispersal of heat due to the convective transport through
solids and their contained fluids are at the same tempera- the porous structure, and has the same effect, when
ture so that we can define T in Eq. 9.1 as the average viewed macroscopically, as an increase in the effective
temperature of both phases. In addition, because both thermal conductivity. Dispersion will be taken up rather
the fluids and the solids are conductors, it is necessary completely when we discuss mass transport. Here it is
 9.1 Conduction, Convection, and Equations of Heat Transport

sufficient to say that this effect is normally incorporated Table 9.1 Thermal Conductivities of
in the transport parameter K,. Rocks
The special cases investigated by Slattery (1972) have
Thermal Conductivity
been verified experimentally. Experimental data summa- Material (cal/m s "C)
rized by Green and Perry (1961) demonstrate that the
value of the effective thermal conductivity is greater with Quartz 2
a flowing fluid than with a stagnant one and, in general, Sandstone 0.9
increases with increasing velocity. Yagi and Kunii (1957) Limestone 0.5
indicate that the effective thermal conductivity can be Dolomite 0.4- 1
Clay 0.2-0.3
separated into two terms, one independent of fluid flow
Water 0.11
and the other dependent on fluid flow. In assessing the
Air 0.006
effect of several mechanisms on the thermal conductivity,
Singer and Wilhelm (1950) include (1) molecular conduc-
tion through the fluid phase, (2) solid particle-to-particle
lower bounds of the effective Conductivity. For macro-
molecular conduction, and (3) particle-to-particle radia-
scopically homogeneous and isotropic materials, siich
tion. None of these mechanisms is affected by fluid flow,
limits may be determined by the method of Hashin and
and may be plotted as horizontal lines in a plot of thermal
Shtrikman (1962). For the upper and lower limits, respec-
conductivity versus velocity (Figure 9.2). Two additional
tively,
mechanisms control the dependence of the thermal con-
ductivity on fluid velocity, (4) a series mechanism from
r -.
fluid to solid to fluid, and so on, and (5) convection
through the fluid phase. These become predominant at
large velocities (Figure 9.2). and
Some measured values for the effective conductivity
in rocks are given in Table 9.1. In general the more
porous the medium, the lower the effective therm,.I I con-
ductivity, mostly because water has a lower thermal con- If porosity approaches zero, the upper and lower limits
ductivity than most solid minerals. approach the conductivity of the solids. As porosity ap-
Because conductance can take place through both the proaches one, both limits go to the conductivity of the
solid and fluid phase, it is possible to propose upper and fluid. Equations 9.5 hold equally well for the electrical
conductivity as well as other two-phase conductive prop-
erties of porous media.

Convective Transport
In systems where the fluid is moving, there is a convective
transport by the fluid motion. When the flow field is
caused by external forces, the transport is said to occur
by forced convection. Such is the case where ground
water movement takes place in the absence of density
gradients such that Darcy's law applies. A second type
of transport, called free convection, occurs when the
motion of the fluid is due exclusively to density variations
caused by temperature gradients. Free convection is
probably the dominant type of fluid motion i n sonic
hydrothermal systems where the bulk of liquid discharge
Velocity - 1 is in the form of steam or hot water. A similar phenome-
non can occur in mass transport, say. along fresh wiiter-
Figure 9.2 Relationship between effective thermal conduc- salt water interfaces in coastal areas where density difftr-
tivity and the mechanisms that contribute to it. (a)Conduc-
ences reflect the various salinity differences. For these
tion by fluid, (6)conduction by solid particles, ( c ) particle-to-
cases, there no longer exists a scalar potenti;il so th;it
particle radiation, ( d ) conduction by a series mechmisni
from solid to fluid, and so on, (e) convection through fluid Darcy's law does not describe the motion. It is empha-
phase. (Modified from Schuler, Stallings, and Smith, Heat and sized that forced and free convection represent two lim-
mass transfer in fixed bed reactors. Chem. Eng. Prog. Symp. iting conditions. In the case of the former, buo).ancy
Ser., v. 48, p. 19-30, 1952). Reproduced by permission of forces are assumed to be negligible. For the latter. fluid
the Anier. Inst. Chem. Engrs. motion must be described entirely in terms of huo)wncy.
 Cbapfer9 / Heat Transport in Ground-WaferFlow

When buoyancy forces dominate, the velocity field and

the energy field (temperature) are interdependent, and
H,,= -K,.
" dy
+ np,.cl,.Tvr (9.8b)
the equations must be solved iteratively.
dT
The simplest system to demonstrate forced and free H, = - K,,: -
ar
+ npl,.cll,Tuz (9.k)
convection is shown in Figure 9.3, which is a liypotheti-
cal model for flow in hydrothermal areas. Note the transi- where u is the velocity and c,,,is the specific heat of the
tion from forced convection at shallow depths to a mixed fluid, defined as the heat necessary to raise the tempera-
convection pattern, and ultimately free convection. Ac- ture of 1 g of fluid 1"C, with the units cal/g°C. The
cording to Bear (1972) the equation for ground water product pf1.c,.is the heat capacity per unit volume of
motion in such environments may be described in terms fluid. For isotropic conditions
of two driving forces, fluid pressure changes and buoy-
ancy, and is of the form H = -K, grad T + ?2plf,clf2V (9.9)

-k 8P @~gpo[l- al(T - Kill -

d~
where v is a vector with components u,, v,,,vz.
zI=--- (9.6)
nfi ax w t3X

where p, is some reference density, p is the viscosity,

Equations of Energy Transport
and 0, is the coefficient of volume expansion for The starting place for development of conservation state-
the water. If the pressure distribution is hydrostatic, ments pertaining to enerky transport is a word equation
aP/ax = -p,gaz/z/ax, and Eq. 9.6 becomes of the form
kgfia,u/(T- Ti) -
dz energy inflow rate - energy outflow rate =
V = (9.7)
np ax change in energy storage with time (9.10)
where, for vertical flow, W 8 x = 1. The quantity in units of energy per time. In general, this statement
pogtu/(T - 23 is the driving force per unit volume of fluid. can be applied to a domain of any size. When applied
The rate equation combining conduction with a Darcy to a representative unit volume with inflows through
type of fluid motion (forced convection) may be ex- three sides, the left-hand side of this word equation is
pressed readily expressed as the divergence of the flux of interest
(see Chapter 4):
z/aT
H.r = - K,? -
ax
+ np,.c,~Tv,,. (9.W -div H = net energy outflow rate per unit volume

High surficial
heat flow area

Meteoric
flow systems

Mixed convection

Magnetic fluids (3

Magmatic body (heat source)

Ffgure9.3 Forced, free, and mixed convection in a hydrothermal system (from Sharp and Kyle, 1988, Geol. Soc. Anier. Pub.,
from Geol. of N. America, 0-2, p. 461-483).
 3.1 Conduction, Convection, and Equatiooris of Heat Transport

It remains to describe the nature o f the Rux, which could Table 9.2 SpecGc Heat of Substances
be conduction (Eq. 9.2) or conduction combined with
convection (Eq. 9.8). The right-hand side of the word
equation represents the gains or losses in energy per unit
time for the unit volume. It will be recalled that for the Air 50 0.2iX
conservation equation for fluid flow, it was assumed that Water vlpor 100 0.w
the time rate of change of fluid volume per unit volume Methane I5 0.528
was proportional to the time rate of change of hydraulic Ucnzene 20 ( 1. i( )6

head, thereby introducing a proportionality const;int Ethyl alcohol 25 0.581

Hasalt (dry) 20- 100 0.20
termed the spciJicstoruge. For the equations of energy
Chalk ( d r y ) LO- 100 0.214
transport we assume that the loss or gain of heat inside
CI;1). (dry) 20- I 0 0 0.22
the unit volume is proportional to the temperature in Granite (dry) 12-100 0 .I92
the unit volume. Some proportionality constant is now Quartz (dry) 12-100 0 ,IXH
needed to convert temperature changes to the gains or
losses o f heat. This proportionality constant is the heat
capacity per unit volume. defined for water alone as the introduced in Chapter 4. The solution t o this equ;ition
product of the fluid density p,,. and the specific heat per describes the value of temperature at ;in!. point in ;I three-
unit volume of water c,,,. The heat c a p c i t y per unit dimensional field, or more precisely. how tlie teniperii-
volume must express the quantity of heat gained or lost ture is changing with time.
from the unit volume when the temperature changes by If the net outflow is zero
a unit amount (cal/L,'"C). For the fluid component in the
unit volume, this becomes npt,.c,,, . For the solid compo-
nents, the expression becomes (1 - n)p,c,, where the
subscript s stands for solids. Thus w e define an effective
This equation is called Laplace's equation, as WAS the
lieat capacity for the unit volume
steady-state fluid flow equation introduced in Chapter 4.
p'c' = np,,.c,,, + (1 - n)p,c, (9.11) The solution of this equation describes the value of tlie
temperature at any point in a three-dimensional field.
The conservative statement of Eq. 9.10 then becomes
dT The Conductive-ConvectionEquation
-div H = p'c' -
dt
(9.12) For conduction and convection described by Eqs. 9.8
i)T
where p'c' is the heat removed or gained from the unit -div H = -div[ -K,, grad T + ~ z p T~v J~= cp'c'
~ ~-
volume when the temperature changes a unit amount.
rlt
Equation 9,12 states that the net energy outflow rate per (9.17)
unit volume equals the time rate of change of energy or. with the usual assumptions of constant K,., 1 1 , p,#, c,, .
within the unit volume. Table 9.2 gives some values for
the specific heat of various substances.

The Heat Conduction Equation For steady ground-water flow, v v =0

For a conductive system defined by Fourier's law of the
form of Eq. 9.2 K,. -
V'T - I1pI,.c,,V V T = p'c' -
ij T
il t
(9.19)

-div H = div[K,, grad TI (9.13) Note that when the velocity is zero we recover our trans-
Here, K, has been defined as an effective thermal conduc- port equation for pure conduction (Eq. 9. IS). When the
tivity. The conservation statement then becomes temperature is steady
il T K,. V ' T - lrp,,c,,v * VT = 0 (9.20)
diV[K,grdd TI = p'C' - (9.14)
dt The equations developed here are called the conduc-
By assuming h a t the effective thernial conductivity is tion-convection equations. They are consen'at ion expres-
constant sions that describe the manner in which energy is moved
from one point to another by means of bulk fluid motion
and by conduction. The one-dimensional form o f Eq. 9.I9
is written
This equation is called the heat conduction (diffusion)
equation and is analogous to the unsteady flow equation
-
196 Cbapter 9 / Heat Transpot? in Ground-WaterFlow

Table 9.3 Thermal Diffusivities of Some Common Rocks and Soils

h cm2/s
-~ ~

Soils and Unconsolidated Material

Quartz sand, medium, dry
Quartz sand, 8.3%moisture 0.0020
Sandy clay, 15%moisture 0.0033
Soil, very dry 0.0020-0.0030
Some wet soils 0.0040-0.0100
Wet mud 0.0022
Soil,Lexington, Kentucky 0.0021
Soil, Lexington, Kentucky (average 0- 10 fi in place) 0.0072
Gravel 0.0057-0.0062
ROC&
Shale 0.0040
Dolomite 0.0080
Limestone 0.0050- 0.0 10
Sandstone 0.01 13-0.0140
Granite 0.0060-0.0130

From Cartwright (1973).

where v, is the mean ground-water velocity in the x

direction. The first term on the left describes energy
transport by conduction, and the second describes en- (9.23)
ergy transport by convection. In all cases, it is assumed
that the temperature of the fluid and the solids are equal.
where y is Darcy’sspecific discharge and L is some char-
Slattery (1 972) describes the case where distinctions are
acteristic length. The Peclet number expresses the trans-
required between the fluid and solid phases.
port of energy by bulk fluid motion to the energy trans-
Last, the combined term K , / ~ ’ chas ’ the units of
port by conduction. In a practical sense this number
LZ/Tand is referred to as the thermal dispersivity A. It is
reflects a competition between two rate processes,
understood that the convective effect of thermodisper-
forced convection and conduction. Large numbers mean
sion is incorporated in the parameter K , . In the absence of
that convective transport dominates over conductive
fluid movement, this parameter (A) becomes the thermal
transport. The second dimensionless group on the right-
diffusivity. Table 9.3 gives some values for the thermal
hand side of Eq. 9.22 is recognized as the Fourier number
diffusivities of rocks.
for heat transport.
Suppose now that we assume a buoyancy-driven fluid
Dimensionless Groups
system. A dimensional analysis of the steady-state trans-
The equations governing the transport of heat in ground-
port Eq. 9.20 in combination with the buoyancy-driven
water flow have been called the conduction-convection
velocity of Eq. 9.7 gives
equations. The solution to these equations for certain
boundary and initial conditions provides us with the tem-
perature distribution of a region. By applying the nondi-
mensionalizing procedures introduced in Chapter 4, it
Vl’T‘ - gp&:.p,:.Lkq(T - K 3
[ PKc
] V’T’ = 0 (9.24)

is easily seen that the behavior of a thermal system is where the Rayleigh number for energy transport is
controlled by a few combinations of dimensionless vari-
ables. We have already seen this in our discussions of
the Fourier number. (9.25)
To the cited dimensionless quantities in Chapter 4 we
add now a dimensionless temperature T+ = T/T,, where The Rayleigh number expresses t h e transport of energy
T,is some characteristic temperature and the superscript by free convection to the transport by conduction and
+ indicates a dimensionless quantity. The conduction- is generally used to establish the conditions for the onset
convection Eq. 9.19 becomes of free convection. In most one-dimensional cases, the
characteristic length is taken as equal to the thickness H
of the formation over which the temperature difference
AT is measured. In two-dimensional problems, it will be
(9.22) some length associated with fluid movement.
 9.2 Forced Convectfon 197

In problems of mixed convection, both the Peclet num- Fluid flow

ber and the Rayleigh number are important in the analy-
sis. As we will see later, convective rolls at high Peclet 1111111111
numbers (substantial forced convection) take on a differ- Hydraulic
ent geometry than those occurring at low Peclet
numbers.
Undisturbed
conductive
gradient
9.2 Forced Convection
Smith and Chapman (1983) recognize three main classes (0)

of forced convection problems of interest to hydrogeolo-

gists. The first of these is the vertical steady-state flow of
ground water and its effect on a purely conductive verti- Hydraulic
cal temperature distribution. If the resultant temperature gradient
Isotherms
profile can be measured, it can be matched against some
curves representing the mathematical solution of the one- conductive
dimensional transport equation, and the Darcy velocity gradient
may be extracted. The second of the problems treats a
two-dimensional vertical cross section of regional
Fluid flow
ground-water flow where the flow field alters the conduc-
(b)
tive temperature profile. Two versions are possible here.
First, for reasonably shallow systems where it can be
assumed that the fluid density and viscosity are not af- Hydraulic
fected by temperature, the velocity field can be deter-
mined independently of the transport problem and subse- Isotherms
quently used in the solution to the transport problem. If
the density or viscosity is affected by temperature, the conductive
gradient
velocity field and the temperature field are “coupled,” I 1
and the equations must be solved iteratively. The third (c)
class of problems is for the two-dimensional geometry, Figure 9.4 Three ideal cases where a fluid flow field is im-
but the temperature field or the velocity field is assumed posed on a conductive thermal gradient.
outright and is not part of the problem. This was the
method of Kilty and Chapman (1980), who were in-
terested in conceptual models to account for heat-flow creasing depth, whereas in Figure 9.46, it will decrease
variations in certain geologic settings. To this threefold with increasing depth. In addition, the heat flow at the
classification, we may add yet a fourth, whereby the surface will be greater for Figure 9.48. In Figure 9.4c-,
temperature field is affected by a three-dimensional flow the hydraulic gradient and the temperature gradient are
field, and yet a fifth concerned with temperature distribu- normal so that the streamlines of fluid flow are collinear
tions in evolving geologic basins. with the isotherms of heat conduction. Here convective
transport is eliminated as it is impossible to transport
Temperature ProJiles and Ground-Water heat along an isotherm. Mathematically, the convective
term in our transport equation becomes
Velocity
v grad T = 0
Consider the three ideal cases shown in Figure 9.4, where
a one-dimensional velocity field is imposed on a one- For this case there is no alteration of the conductive
dimensional purely conductive thermal field. In Figures gradient. In between the extremes of streamlines normal
9.4a and 6, the direction of ground-water movement is to isotherms and streamlines collinear with isotherms,
taken as normal to the conductive isotherms. That is to some alteration of the conductive gradient will occur.
say the hydraulic gradient and the conductive tempera- with the maximum alterations associated with the former
ture gradient are collinear in the vertical direction so that situation. In addition, either of these extremes can be
the streamlines of fluid flow are normal to the isotherms viewed from the perspective of a one-dimensional trans-
of heat conduction. For these cases, a convection flux is port problem. This is the basis of the Bredelioeft-
established that will produce the greatest alteration of Papadopulos (1965) model for determining ground-water
the conductive temperature distribution. In Figure 9.4a, velocity from temperature profiles.
the resultant temperature gradient will increase with in- Figure 9.5 gives the conditions under which the model
198 Cbapter 9 / Heat Transport tn Ground-WaterFlow

Ground surface cal model just described. The experiment involved the
measurement of the temperature distribution in a column
of glass beads and sand through which fluid was flowing
counter t o an imposed temperature gradient. Sore).
(1971) constructed the dimensionless plot shown as Fig-
ure 9.7, which matches the N,,equals 3.3 curve of Figure
-f- T'T,, - 9.6. For the data reported by Kunii and Smith (1961) and
I
L J.T: Semiconfining the methods described, the mass flow rate was calculated
I layer to be pi,y = 2.3 X lo-'g/cm's, which compares favorably
+ with the measured mass flow rate of 2.6 X lo-' g/cm's.
*T,
Field results have been reported by Cartwright (1970)
and Sorey (1971). In Figure 9.8, Cartwright analyzes the
temperature data from wells in the Illinois Basin and
obtained a ground-water flow rate of 4.9 X lo-' cm/s.
The important limitations to this type of analysis reside
T,,, T,, T,, Measured temperatures

I): Leakage rate

Figure 9.5 Diagrammatic sketch of typical leaky aquifer

(from Bredehoeft and Papadopulos, Water Resources Res., v.
1, p. 325-328, 1965. Copyright by Amer. Geophys. Union).

applies. In this diagram, To is an uppermost temperature

measurement at z = 0, and TLis a lower-most temperatiire
measurement at z = L. These are the boundary conditions
for the one-dimensional steady-state problem described
by

(9.26)

The solution to this problem is (Bredehoeft and Papa-

dopulos, 1965)

where T, is the temperature at any depth over the thick-

ness L (Figure 9.5) and

which is the familiar Peclet number for heat transport.

A graphical solution to this problem is given by the
type curves of Figure 9.6 with the Peclet number taken
as the solution parameter. The curves are convex upward
or downward. A suggested method of analyses by Brede-
hoeft and Papadopulos (1965) requires a field plot of
(T, - To)/(TL- To) against the depth factor z / L at the
same scale as the type curves and superimposing this [exp(Ni,,P) - 11
plot on the type curve set. The value of NPEdetermined lexp(Ni,,] - 11
from this match is then used to calculate the velocity q Ffgure9.6 Type curves for solution o f one-dimensional
from Eq. 9.28. This assumes information on K,. conductive advective equation (from Ikedehoeft and Papado-
Sorey (1971) reports some data by Kunii and Smith pulos, Water Resources Rcs., v. I , p. 325-328. 1065. Copy-
(1961) that have been employed to verify the mathemati- right by Amer. Geophys. IJnion).
 9.2 Forced Convectfon 133

in the one-dimensional assuniptions for both the tempera-

ture field and the velocity field.

Heat Transport in Regional Ground-Water

Flow
Let us now continue the "experiment" started in the
previous section concerning the thermal outcome of vari-
ous orientations of streamlines of fluid flow and isotherms
of pure conduction. A conductive field is shown in the
two-dimensional region of Figure 9.9n. A two-dirnen-
sional flow field as shown in Figure 9.96 is superiniposecl
on this conductive field, with Figure 9 . 9 demonstciting
~
the alteration of the temperature field. The greatest alter-
ation occurs where the streamlines of flow are norni;il
[exp(N,,,) - 11
or nearly so to the conductive isotherms, that is, the
recharge and discharge areas. and the least alterdtion
Figure 9.7 Type ctirve match for temperature profile in takes place in the region of lateral flow. Because of this
glass beads (from Sorey, Water Resources Res., v. 7 , p. 963-
970, 1971. Copyright by Amer. Geophys. Union).

2=0' I
.s = 0 .v = 1.12 .v = I.

.v = 0 .v = I.
(l.1

Ffgure 9.9 Iliagrams illustrating (a)conductive heat triins-

lexp(x,, 4) - 11
port in a twodimensional region, ( b ) gener;tlized flow sys-
IexpW,+I - 1 I
tem for the twcxtimensional region, and (c) alteration o f the
Ffgure 9.8 Type curve match for Illinois Rasin (from Cart- isotherms (modified from Domenico and Palciauskas. 1973.
Wright, Water Resources Res., v. 6,p. 912-918, 1970. Copy- Geol. Soc. Amer. Pub., from Geol. Soc. Bull.. v. 84. p. 3803-
right by Amer. Geophys. Union). 38 14).
200 Chapter 9 / Heat Transport in Ground-Water Ij'low

convective alteration, the geothermal gradient will in- whereas for the one-dimensional problem of Bredehoeft
crease with increasing depth in the recharge area and and Papadopulos (1965), the characteristic length was
decrease with depth in the discharge area, the latter determined to be the length of the vertical region over
associated with a greater amount of heat flow at the which perturbations were measured. The expression for
surface. This problem was solved by Domenico and f ( z l , / L )is cumbersome and will not be given here. How-
Palciauskas (1 973) for the following conditions: ever, as the basin length-to-depth ratio becomes large,
the flow pattern is largely horizontal and f(zl,/L) ap-
1. The fluid properties are not affected by temperature proaches zero so that there is a minimal interference
and the medium is isotropic and homogeneous with with the conductive gradient, that is, v grad T = 0.
respect to both fluid flow and heat conduction. Small length-to-depth ratios produce flow systems that
2. The twodimensional flow problem is as described contain vertical components of flow throughout and are
in Example 5,1, where the water table is unspecified associated with large perturbations.
but the lower and lateral boundaries are of the no- For an approximate measure of the potential for con-
flow variety. vective alterations of the geothermal gradient, a modified
3. The temperature problem is described for the two- Peclet number may be given as
dimensional region of Figure 9.9 where the lateral
boundaries are nonconductive (dT/ilx = 0). The up-
per boundary is taken at some constant temperature
T, and the lower boundary is described in terms of
a constant temperature gradient T,',. Values less than one are associated with temperature
distributions that are not far from the limiting case of
Although the analytical solution to this problem was pure conduction. Other geometric configurations of flow
developed for any water table configuration, a special such as those describing local, intermediate, and regional
case was examined for the water table given in Figure flow (Figure 5.7) are also examined in terms of the inter-
9.9b for which an analytical solution is presented in Ex- pretation of the characteristic length associated with
ample 5.1. For this special case, the solution to the two- Peclet numbers for twodimensional flow systems (Do-
dimensional transport equation is of the form menico and Palciauskas, 1973).Van der Kamp and Bachu
(1985) describe the use of dimensional analysis for study-
ing thermal effects in various hydrogeologic regimes.
The study described was a theoretical one where the
(9.29) main interest was in determining the effect of fluid flow
on temperatures within the framework of the classic
where Ti is the temperature gradient across the bottom regional flow model introduced in Chapter 5. Morgan
boundary, NpEis the dimensionless Peclet number, and and others (1981) have used this model to investigate
all other terms are described in Figure 9.9. In this form, the threshold hydraulic conductivity where temperature
the temperature distribution as described in Eq. 9.29 is perturbations become significant in basins of fixed geom-
controlled by three essential features. The first bracketed etry within the Rio Grande Rift. Betcher (1977) solved
quantity describes the temperature distribution for pure the problem numerically with a finite element model and
conduction, that is, describes the conductive isotherms conducted a sensitivity analysis to ascertain the role of
shown in Figure 9 . 9 ~The~ . other quantities obviously the various parameters in controlling the temperature
describe the perturbation of the purely conductive field distribution. Parsons (1 970) used a numerical model to
and consist of a dimensionlessPeclet number and a group investigate the effect of hydraulic conductivity contrasts
of trigonometric and hyperbolic functions that deal and included a case where the upper boundary was taken
exclusivelywith the ratio of basin depth z0to basin length as a variable temperature boundary. Some results of his
[f(z,,/L)].For this case the Peclet number is described study are shown in Figure 9.10.Note in particular that the
threeorder-of-magnitude hydraulic conductivity contrast
(9.30) associated with the lower-permeability region more or
less preserves the pure conduction isotherms, as does
where K is the hydraulic conductivity and B is the mean the rather large area of lateral flow where v * grad T 0.
water table elevation as measured from the elevation of Smith and Chapman (1983) include the effects of tem-
the discharge area (Figure 9.9). Thus if the hydraulic perature on fluid density and viscosity in heterogeneous
conductivity or the mean water table elevation B gets anisotropic media. Figure 9.1la gives some of their re-
large with respect to the thermal conductivity (large sults for homogeneous material, showing the convective
Peclet number), forced convection dominates over con- influence on heat flow at the surface, which is most
duction and the conductive field is altered significantly. pronounced for higher values of permeability. For homo-
Note that B is the characteristic length in this problem, geneous but anisotropic material, the heat flow at the
 Next Page

202 Cbapter 9 / Heat Transport in Ground-WaterFlow

g:
a €
2
-- 60
100
.2 140

Distance (kilometers)
160.
Homogeneous; isotropic k
linear water table
120

80

0' I I I I I I I J
5 10 15 20 25 30 35 40
Distance (kilometers)

240

200

160

Y
120
I
I 80
100
40

O' 5 110 115 o:

Distance (km)
$5 o; 315 4b
Figure 9.11 (a)Thermal effects of ground-water flow i n a
basin of homogeneous, isotropic permeability with a linear
water table. Surface heat flow cxpressed in mWm-*. (0)Ther-
mal effects o f ground-water flow in a basin of anisotropic ho-
mogeneous permeability. Horizontal permeability is 8 X
10- m2, anisotropic ratio is 1, 10. and 100 for the three sim-
ulations shown. In all cases, ground-water flow is from left
to right on the diagnnis (from Smith and Chapman, J. Geo-
phys. Res., v. 88,p. 593-608, 1983. Copyright by Amer. Geo-
phys. IJnion).

surface increases with decreasing anisotropic ratios K,/ ore deposits by topographically driven flow. Some results
Kz0;igure 9.11b). from Garven and Freeze (1984a, 1984b) are shown in
Woodbury and Smith (1985) present a solution for a Figure 9.12. Figures 9 . 1 2 ~and b depict temperature dis-
three-dimensional flow field. Their analysis was con- tributions at low and high Peclet numbers, respectively,
cerned with the variations and areal distribution in heat due to differences in hydraulic conductivity. The advec-
flow at the surface as affected by relief on the water table tive perturbations are readily seen for the high conductiv-
and permeability variations. ity condition. Figures 9 . 1 2 ~and d depict temperature
The studies described are concerned with temperature distributions at high and low Peclet numbers, respec-
distributions in modern environments. Temperature dis- tively, due to differences in the thermal conductivity. As
tributions have also been ascertained for paleoflow sys- noted, the higher the thermal conductivity the closer the
tems that may have been important in the formation of temperature distribution apprwdches that of pure con-
Previous Page

9.2 Forced Convectfon 203

4 r
T = 20°C
.-
$ 3 -

g 2:
z 1 60
40

v. I , 1 I 1 I I

(If)
4 r
7 = 20°C
$ 3
a
.
K , = 100 rnlyr

0. I l l l l l t l l l l l l l l l l l
I
l

f 3c T=2O"C -
-0u , I I I I I I I I I

Kilometers
(g)
Ffgure 9.22 Tenipenture distributions at ( a ) low and (6)high Peclet numbers due to dif-
ferences in hydraulic conductivity; (c) high and ( E l ) low Peclet numbers due to differences
in thermal conductivity and for different water table configurations (e,f,g). In all cases,
ground-water flow is from left to right on the diagrams (from Garven and Freeze, 1984.
Amer. J. Sci., v. 284, p. 1085- 1124. Reprinted by permission of the Amer. J. Sci.).

duction. Figures 9.12e,$ and g depict temperature distri- presented in Geophysical Monograph d7 (Beck, Garren.
butions for different water table configurations, where and Stegena, 1989).
the effects of recharge and discharge along the water
table are reflected in the various distributions. Heat Transport in Active Depositional
The field observationsthat support the general interfer-
Environments
ence between isotherms and streamlinesdiscussed in the
theoretical work are provided in studies by Van Orstmnd The pressure evolution in active depositional environ-
(1934) and Schneider (1964). Several case studies are ments was seen in Chapter 8 to be a transient phenome-
204 Chapter 9 / Heat Transport in Ground-WaterFlow

non. As expected, the transient velocity field is responsi- t

ble for the convective transport of heat. Examining this
process from the perspective of the energy transport
equations provides some information on the pressure-
temperature history of a region. Although it is not possi-
ble to reconstruct all the factors that play a role in the
pressure-temperature history of basins, mathematical
models can be used to gain insights into certain processes
and for establishing some limiting or threshold condi-
tions.
In modeling studies, the conductive gradient is nor-
mally depicted as originating due to a heat flux across
the lower boundary of an accumulating sedimentary pile
(Figure 9.13). Most studies to date have assumed the
condition of forced convection although some free con-
vection cannot be ruled out. In addition, it is frequently
assumed that the surface temperature remains constant
during continued deposition of the sediment. Frequently,
a moving coordinate system is used to study this problem.
The results of such studies demonstrate that with uniform Temperature
sedimentation-in the absence of fluid flow-the ther- Ffgure9.24 Temperature patterns with moving boundaries
mal gradient will constantly decrease, and for uniform and convection.
erosion it will constantly increase. These facts are shown
in Figure 9.14 by profiles 1 and 2 when the distance axis
represents the depth from the sediment-water interface Sharp and Domenico (1976) developed an energy
to the impermeable basement rock. Profiles 1 and 2 transport model that incorporated onedimensional fluid
would also be expected from steady onedimensional flow and onedimensional heat transport. When applied
flow upward and downward, respectively, where the to Gulf Coast conditions for the loading history of Figure
sedimentary layer remains of uniform thickness (Brede- 9.15, they concluded that abnormal pressures were low
hoeft and Papadopulos, 1965). Profile 3 represents the for much of the geologic record and were increased
combined effect of onedimensiondl forced convection during the Eocene, the Oligocene, and most markedly
upward and the separation of the boundaries by the during the Late Pliocene-Quaternary period in response
accumulation of sediment. In this case, a zero or near- to rapid sedimentation (Figure 9 . 1 6 ~ )The
. period of low
zero velocity at the lower boundary and a maximum abnormal pressure was accompanied by pressure-depth
velocity at the growing sediment-water interface per- profiles very close to the conductive geothermal gradient
turbs profile 1 into the reverse "S" pressure-depth pro- (assumed at 30"C/km). For the Late Pliocene-Quaternary
file 4. period, the temperatures were uniformly lower than in
the previous period below depths of 6000 m (Figure
9.160). The reverse "S" profile dominates during this
period of rapid sedimentation, mainly because of the
moving boundary effect. They concluded that sediments
at present in the near offshore are now at their maximum
pore fluid pressure and minimum temperature at any
depth. Chia (1979) extended this study by incorporating
a three-dimensional loading or mean stress formulation
within the confines of the vertical transport of fluid and
heat. He also included the thermal expansion of fluids
as a pressure-producing mechanism, and a fluid-pres-
sure-fracture criterion to examine the conditions leading
to fracture initiation.
Cathles and Smith (1983) developed a two-dimensional
Bow model in combination with energy transport to ex-
amine episodic basin dewatering and the genesis of rnin-
era1 deposits. Bethke (1985) also has applied twodimen-
Pfgure 9.23 Schematic diagram of accumulating sediment sional flow and energy transport models to accumulating
with heat flux across the lower boundary. sedimentary piles. His calculations indicate a tendency
 7
Sandy clay
-
IS
ulation (
-
siana

-0
--I

..-
-1

-.-.- III
-.- -I P
M
-J
Q

I
I
9.2 Forced Convection

The nature of hydrologic disturbances in mountainous

terrain has recently been investigated by Forster and
Smith (1989). Their conceptual model is a two-dimen-
sional region not unlike Figure 9.9, with the bedrock
forming the upper surface and a conductive heat flux
across the bottom boundary. As a continuation of their
previous work described in Chapter 5 (Forster and Smith,
1988), the water table elevation is determined by using
a free surface approach where an infiltration rate must
be specified. Some of their results are shown in Figure
9.17a, 6,and c for a fixed infiltration rate and .a three-
order-of-magnitude increase in permeability going from
2 3

Figure 9.17~2to 9 .1 7 ~In

. these figures, the isotherms are
depicted with solid lines (%I, the heat lines are dashed
K Cretaceous
P Paleocene and show the heat transport to the surface by both con-
E Eocene duction and convection, and the dotted lines indicate
0 Oligocene
M Miocene
P Pliocene
Q Quaternary

20 0
Millions of Years Before Present
Figure 9.15 Estimated sediment accumulation rates for the
offshore Texas and Louisiana coasts (from Sharp and Domen-
ico, 1976, Geol. SOC.h e r . Pub., from Geol. Soc. Anier.
Bull., V. 87, p. 390-400).

for fluids to migrate laterally toward the edge of the

basins. Vertical flow occurs in the deeper parts of the
basin where the ratio of vertical to lateral permeability
exceeds the ratio of the 1engths.of vertical to lateral
pathways to the surface. He demonstrated further that
for small sedimentation rates the temperature gradient
Excess head (meters)
is not far removed from that expected from pure conduc-
((0
tion, as would be expected for situations approximating
normal compaction.

Heat Transport in Mountainous Terrain

2 Cretaceous-Eocene
The relief associated with mountainous terrain may pro-
mote an advective disturbance of the thermal regime that
is somewhat more severe than discussed earlier in the
chapter and may contribute to regional scale variations
in surface heat flow. In the southern Great Basin, for
example, Sass and others (1976) recognize a regional
anomaly of low heat flux, termed the “Eureka low,”
which includes the Nevada Test Site and Yucca Mountain
described earlier (Chapter 5). As the regional heat flow
50 100 150 200 250 300 350 400 450
within the crust beneath the “Eureka low” is likely of Temperature (;C)
the same intensiry as occurs in the Great Basin in general, (0)
a hydrologic disturbance was postulated where “ground Ffgure 9.26 Calculated excess head and temperature distri-
water is carrying much of the Earth’s heat in the upper bution in Gulf of Mexico sediment at the end of each of the
3 km and is delivering it elsewhere” (Sass and Lachen- four specified time periods (from Sharp and Domenico.
bruch, 1982). Wherever that “elsewhere” is, there is 1976, from Geol. SOC.Amer. Pub., from Geol. Soc. Amer.
likely to be yet another anomaly, this time a thermal high. Bull., V. 87, p, 390-400).
2B Cbapter 9 / Heat Transport in Ground-WaterFlow

fct

-
"c, Grwitd water p8tb iire
fk Heatfine
isothetrn t"C1
% Watertable
Regwn not penetrated
by basgf tws flw
~~~~~~~~~

wtlt
Figure 9.27 Patterns of ground-water flow and heat transport with a three-
order-of-magnitude increase in permeability when going from (a)to (c). Figure
9.17d has the infiltration rite decreased by a factor of 5 for the same permeabil-
ity of Figure 9 . 1 7 (from
~ Forster and Smith, J. Geophys. Res., v. 94, p. 9439-
9451, 1989. Copyright by h e r . Geophys. Union)

pathlines for fl.uid flow. Figure 9 . 1 7 shows

~ a conductive duces a rather weak thermal disturbance and a warmer
thermal regime, which becomes more disturbed as the thermal regime than for the higher-infiltration case of
permeability increases. In Figure 9.18b the disturbance Figure 9 . 1 7 ~ .
is minor, whereas in Figure 9 . 1 7 ~several changes are Figure 9.18 shows the effect of a high-permeability
noticeable. Such a permeable system is capable to trans- fracture on the flow system and thermal regime. A large
mit all the infiltration, and the water table falls below percentage of the flow and basal heat are captiired by
the bedrock surface. Further, almost a11 of the basal heat the fracture zone, resulting in an upwelling of isotliemis
flux is transported to the valley. This means that for in the fracture zone. Such features may produce thermal
regions above the valley bottom, temperatures are gov- springs at the fault outcrop. If the hydraulic conductivity
erned more by thermal conditions at the bedrock surface of the rock matrix is increased by one order of magnitude,
than by heat flow from below. a greater flux is captured by the fracture, the 75OC iso-
The flow system in Figure 9.17d has the same perme- therm of Figure 9.18 is reduced to 35"C, and the upwell-
ability as the system in Figure 9.17c, but the infiltration ing along the fault vanishes (Forster and Smith, 1989).
rate has been reduced by a factor of 5 , giving a fivefold In addition, the temperature in the upper regions ap-
decrease in the ground-water flux.The depth of the water proaches the temperature of the recharge because the
table is increased markedly, and the reduced flux pro- upper unit is not penetrated by the basal heat flux.
 9.3 Free Convection 207

where A has been introduced as the thermal diffusivity.

When compared with results based on the Rayleigh crite-
ria for pure fluids, this minimum temperature gradient
exceeds the required temperature gradient for the onset
of convection in fluids by a factor of (16/27)H'/n'~,,p,.
Thus, temperatures that establish free convection in flu-
ids are not sufficient to establish free convection in fluids
contained within porous solids, largely because of the
viscous drag of the solids on the fluids.
Lapwood (1948) demonstrated that the onset of free
Distance (Blornetersl convection in porous media occurs at Rayleigh numbers
Figure 9.28 Pettems of ground-water flow and heat trdns- on the order of 417'; that is,
port as influenced by a steeply dipping high-permeability
fracture zone (from Forster and Smith, J. Geophys. Res., v.
94, p. 9439-9451, 1989. Copyright by Amer. Geophys.
Union). This criterion is based on horizontal layers filled with a
nonmoving fluid and upper and lower boundaries that
92 Free Convection are both impermeable to flow and isothermal. Other ther-
mal and hydrologic boundary conditions for the hori-
Free convection has been defined as a flow driven by zontal layer case have been reported by Nield (196s).
density variations. As a buoyancy-driven flow, the visual- Pratts (1966) investigated the effect of horizontal flow
ization is of rising warm water, a cooling process, and on the onset of free convection and found that the con-
the sinking of tliese same waters. Thus, with free convec- vection parameter R, = 40 remains unchanged. The i n
tion we visualize ascending and descending currents that corporation of horizontal flow, however, implies that
form cells of various geometries. The transport of heat is both forced and free convection are occurring so it is
in the direction of the buoyant forces, and the isotherms expected that the nature of the convection rolls will be
depart markedly from their expected conductive distribu- controlled in part by the magnitude of the horizontal
tion. Figure 9.19 is a typical example of a thermal profile flow. This is demonstrated in the review of Combarnous
established by free convection. and Bories (1975),where for Nprless than 0.75, the axes
of convective cells appear to run perpendicular to the
The Onset of Free Convection mean flow direction (Figure 9.20b). For higher values of

Horton and Rogers (1 945) were among the first to address

the conditions for the onset offree convection in a porous
medium. Their theoretical results were based on an im-
posed vertical temperature gradient across a horizontal
layer, saturated with a nonmoving fluid, where the upper
and lower boundaries were taken as perfect heat conduc-
tors. According to this work, convection currents may
develop when the temperature gradient exceeds some
critical value
(b)
I
(9.32)
(0)

((.I
Figure 9.29 Free convection in a confined aquifer (from Ffgure 9.20 Forms of convective coils (from Combarnous
Donaldson, J.. Geophys. Res., v. 67. p. 3449-3459, 1962. and Bories, 1975. Reprinted with permission o f Academic
Copyright by Amer. <ieophys.IJnion). Press, from Advances in Hydroscience, v. 10. p. 2.31-507.
208 Cbapter 9 / Heat Transport in Ground-Water Flow

NpE,helicoidal rolls form, with their axes parallel to the

mean flow direction (Figure 9.20~).These can be com-
pared with the polyhedral cells associated with free con-
-
-------- Isotherms
Streamlines

vection operating in the absence of a bulk fluid motion

(Figure 9.204.

Example 9.1
The following is an example of the Rayleigh criteria as
demonstrated in Eq. 9.33. The parameters are as follows.
The coefficient of thermal expansion for the fluid is 2 X
OC-I, mass density is 1 g/cm3, acceleration of gravity
is 980 cm/s2, the permeability is 1 darcy or lo-' cm',
the sand thickness is 1000 ft or 3.048 X lo4 cm, the
Figure 9.22 Geometry of unicellular flow in an inclined
temperature gradient is 0.023 "C/ft or 7.55 X lo-' "C/ fluid layer (from Cornbarnous and Bories, 1975. Reprinted
cm so that the temperature change across 1000 ft of with permission of Academic Press, from Advances in Hydro-
sand is 23"C, the heat capacity is 1 cal/cm' "C, the fluid science, v. 10, p. 231 -307).
viscosity is 6 X lo-) g/cm s, and the thermal conductivity
is 4 x cal/cm s "C. and Rayleigh numbers above the critical value, the longi-
(980)(1)(1)(3.048 X 104)(10-n)(2X 10-4)(23) tudinal coils form.
NU =
(6 x 10+)(4 x lo+)
Geological Implications
= 57
In the study of sandstone diagenesis,it has become appar-
which exceeds the critical value of 40. ent that large amounts of mass must be transported to
account for certain diagenetic changes. Free convection
offers an attractive alternative as a mass transport mecha-
Sloping Layers nism as opposed to the "once-through" volumes of fluid
Combarnous and Bories (1975) discuss two important associated with compaction or even the thousands of
aspects of free convection in sloping layers. For such pore volumes of meteoric derived water through ancient
layers the required Rayleigh number is higher than the uplifted rock. Honon and Rogers (1945) were probably
criticalvalue of 40 so that for isothermal bounding planes the first to test this hypothesis for the Woodbine Sand
in Texas to account for the NaCl distribution. Their
Nm cos II/ > 4n2 (9.34) method was to calculate the minimum value of permeabil-
ity for which convection currents might occur (Eq. 9.33).
where 9 is the slope of the layer. For slopes of less Blanchard and Sharp (1985) took the same approach with
than 1 9 , the convective movement takes the form of the Rayleigh criterion (Eq. 9.33) in an attempt to account
polyhedral cells shown in Figure 9.20. For greater slopes,
the cells are in the form of longitudinal coils.
Gravitational potential surface,
A second point made by these authors is that convec-
tion will occur for all conditions where the isotherms 7-1
are not parallel to the geopotentials, that is, where the
temperature gradient is not collinear with the body
forces. It is thus recognized that there is always a convec-
tive movement in a sloping layer bounded by isothermal
planes of different temperature. The geometry of the roll
I

x-=' I I

is unicellular and two dimensional (Figure 9.21). A hot

current moves upslope along the bottom boundary and
then turns and moves downslope along the upper cold
boundary.
The three states of free convection just discussed may
Figure 9.22 Schematic diagram illustrating regions of fluid
be viewed as transitional. For small Rayleigh numbers stability in folded porous body. Boundaries of body assumed
below the critical value, the unicellular convective flow isothermal and impermeable. Reprinted with permission of
occurs in sloping layers. At Rayleigh numbers above the Geochim. Cosmochim. Acta, v. 46, Wood and Hewett, Fluid
critical value, the unicellular motion gives way to the convection and mass transfer in porous sandstone-a theo-
polyhedral cells for slopes less than 15".For steep slopes retical model. Copyright 1982. Pergamon Press PLC.
 9.4 Energy Resources 209

for the silica distribution in the Frio Sandstone in Texas. temperatures are under 650"C, and the stored energy is
Wood and Hewett (1982) examined the stability analysis in the several thousand times 10" calories range. Renner
in sloping layers and made some general statements on and others (1975) describe the hydrothermal convection
the large-scale diagenetic effects that might be attributed systems in the United States.
to large-scale (kilometers) unicellular currents (Figure A conductivedominated system generally relies on the
9.22). They conclude that although the currents are slow thermal blanket effect where a low-conductivity cap rock
(1 m per year), the mass flux associated with these cur- acts as an insulator for heat storage below. Yet another
rents operating over geologic time is sufficient to cause type of conductivedominated system is the abnormally
significant porosity changes. Cathles (1977) and Norton pressured reservoirs in the Gulf Coast. Fluid pressures
(1978) discuss convective fluid movements associated can reach as high as several thousand pounds per square
with intrusions and cooling plutons. inch, and temperatures can get above 300°C when one
considers depths of 15 - 20 km. The potential thermal
and mechanical energy is estimated to be between 46,000
9A Energy Resources and 190,000 megawatts for 20 years (Papadopulos and
others, 1975).
Geotbemtal Energy
It may be noticed that most of the references cited
Three types of geothermal systems are generally recog- here are mid-1970s in vintage. This is no coincidence and
nized: (1) hydrothernial convection, (2) hot igneous rock, reflects the effort to develop alternative energy sources in
and (3) conductive dominated. The hydrothermal con- response to the oil embargo of the early 1970s. With a
vective systems are either vapor dominated, which con- return to pre-1971 prices for oil and gas, the interest
sist of dry or wet steam, or hot water dominated. Dry in geothermal resources has diminished considerably.
steam fields are normally of low permeability so that little Geothermal systems are discussed rather extensively in
recharge occurs. Temperatures on the order of 250°C proceedings of a United Nations conference (1976) and
are not uncommon. The wet steam field is characterized in a bibliography by ERDA (1976).
by a greater amount of recharge and consists of steam- The development of geothermal resources often en-
water mixtures. The hot water systems may be classified counters problems not unlike those associated with the
as high (> 1 50"C), intermediate (90- 150°C), and low use of ground water as a water supply. Water injection
(< '9O'C) temperature systems. Both the vapor- and liq- i n hot aquifers is often considered as a viable management
uid-dominated systems require the following five charac- scheme to maintain the supply (Goguel, 1976). In other
teristics: (1) a heat source, (2) sufficient permeability, cases, the exploitation results in both land subsidence
(3) a source of replenishment, (4) a low-permeability and induced tectonic activity (Herrin and Goforth, 1975).
bottom boundary, and ( 5 ) a low-permeability cap rock. In response to a daily production of 130 million gallons
The temperature of the fluids in the foregoing classifi- of hot water from a reservoir located in Texas, four earth-
cation dictates their use as an energy source (Nathenson quakes of magnitudes 3.3 to 4.4 occurred within one
and MuMer, 1975). The high-temperature systems are month. Hundreds of microearthquakes were recorded
adequate for electrical generation. Nathenson and Muf- but they gradually diminished.
fler (1975) estimate the heat stored in these systems in The modeling of geothermal reservoirs must address
the United States as 257 X 10" calories. They suggest the coupled movement of heat and ground water. Mercer
that this is five times the amount stored in undiscovered and others (1975) have solved such a problem for the
high-temperature reservoirs. To provide a perspective of Wairakei Field, New Zealand. Reviews on the modeling
this energy source, 10'"calories is equivalent to 690 mil- methods are given by Pinder (1979) and Gary and Kassoy
lion barrels of petroleum (White and Williams, 1975). (198 1). A numerical model for a hot-waterdominated
The intermediate temperature systems are adequate for system is given by Sorey (1978).
space and process heating. They have a stored heat con-
tent of about 345 X 10'' calories in the United States.
Energy Storage in Aquifers
For both the high- and intermediate-temperature systems,
only about 25%of the stored energy is available at the Hot water storage in aquifers has been suggested as an
surface. Low-temperature systems are useful only under attractive means for storing energy produced during peri-
locally favorable conditions, for example, the heating of ods of low consumption for use during peaking hours,
homes or buildings by a local circulation system. or even seasons. According to Sauty and others (1982a),
Hot igneous systems are of two types, molten magma the important parameters in an energy storage problem
and hot dry rock. The molten magma is currently not include the recovery factor, which is the ratio of the
capable of exploitation due to the pressures and tempera- quantity of energy recovered to that injected, and the
tures that would be encountered in drilling. Exploitation temperature of the recovered water. These factors de-
of the dry hot rock systems would require a circulation pend on the properties of the aquifer, the water con-
system for water to be injected and then recovered. The tained within the aquifer, and the operating conditions
-
210 Chapter 9 / Heat Transport in Ground-WaterFlow

such as injection rates and duration. This problem has Academy of Sciences released their report entitled “The
been investigated numerically by Tsang and others (1 981) disposal of radioactive wastes on land. Several options
”

and Sauty and others (1982a and b). Moltz and others were considered, including disposal in space, in sub-
(1981) have described the results of a long-term field seabeds, in ice sheets, in “very deep” holes, liquid injec-
test that attempts to assess the feasibility of an energy tion in wells, rock melt disposal, transmutation, and a no-
storage system. action alternative. The heat generation aspects associated
Some of the physics involved in energy storage have with these alternatives did not go unnoticed. On the
been discussed by Sauty and others (19821). The warmer positive side the ice sheet disposal would utilize the heat
body of water that develops during injection is normally for the waste to melt its way to the bottom. On the other
restricted within a cylinder of radius r where hand, the seabed program ran into technical objections
because of the uncertain thermal effects of the canisters
(9.35) on the sediments. With regard to the no-action alterna-
tive, a presidential message on February 2, 1980 stated
that “civilian waste management problems shall not be
where c,,,is the specific heat of the fluid, c, is the specific deferred to future generations” (Department of Energy,
heat of the solids, Q is the injection rate, and H is the 1980). The mined geologic repository has emerged as
aquifer thickness. The lower the thermal conductivity, the most viable alternative.
the smaller the heat losses, and the higher the efficiency
of extraction. These features are demonstrated through
the use of dimensionless parameters. The Rock Types
The scientists in the early phases of the disposal program
were far more concerned with “geologic media” than
9.5 Heat Transport and Geologic with specific hydrogeologic environments. Among the
Repositories for Nuchar Waste Storage first two rock types proposed were salt and granite,
largely because the thermal conductivity of these sub-
Tbe Nuclear Waste Program stances is quite high. In fact, salt has one of the highest
For the past few decades, one of the largest applied earth thermal conductivities of all natural rock types (Table
science programs ever attempted throughout much of 9.4). T ~ L I from
S , the early beginnings there was concern
the world has been the investigations for safe disposal with the long-term production of heat due to decaying
of nuclear waste. Many countries have such a program, radionuclides. Salt formations, both domed and layered,
including Sweden, Belgium, France, Italy, Switzerland, have been selected as possible candidate sites in Den-
Canada, and the United States. The purpose of most of mark, West Germany, the Netherlands, and Spain. Crystal-
these programs is the disposal of nuclear waste associated line rocks have been selected in Canada, France, Japan,
with commercial nuclear power plants. The wistes can Switzerland, and the Scandinavian countries. On the
be classified into three major categories. The first is the other hand, Belgium and Italy are considering a clay-
fission products resulting from the splitting of uranium type rock. Some early potential candidates for the first
and other fissile nuclei used as fuel. These radioactive repository for commercial wastes in the United States
elements emit both beta and gamma radiation and pro- included the Michigan Basin (salt), the Gulf Coast (salt),
duce most of the heat in high-level waste. They are both the Permian basin in Texas (salt), and Vermont-New
short lived and long lived, the latter including ‘”I and Hampshire (granite). In addition, an active program for
pTc. The second category includes the transuranic ele- the disposal of defense nuclear wastes is being pursued
ments such as plutonium and neptunium, many of which in bedded Salt deposits near Carlsbad, New Mexico. This
are associated with extremely long half-lives. Another program is referred to as the Waste Isolation Pilot Plant
category includes neutron activation products, for exam- (WIPP). Some 9 miles or so of tunnels are already in
ple, “Ni. These products are generated in the fuel clad- place, and testing has already begun.
dings and structural materials in the core of the reactor. In the United States, rock types and locations other
To provide some idea of the magnitude of the problem than those cited earlier have been taken under consider-
caused by these wastes, the Office of Technology Assess- ation, including basalts at the Hanford Site in the state
ment (1982) has estimated that there will be something of Washington and volcanic tuff at the Nevada Test Site
on the order of 72,000 metric tons of spent fuel by the in southern Nevada. The Hanford Site has long been
year 2000. With reprocessing, these wastes would oc- assoc?dted with subsurfdce storage and disposal of de-
cupy a volume the size of a football field, one story high. fense I l l i C k d t wastes, and the Nevada Test Site has been
Without reprocessing, the volume of an entire stadium the location of subsurface nuclear detonations for de-
could be filled. cades. Hence, both are polluted already and both have
The problem of nuclear waste disposal was first ad- the advantage of possessing vast tracts of land that have
dressed in the United States in 1957 when the National been removed from the public domain. Further, some
 9.5 Heat Transport and Geologic Repositoriesf o r Nuclear Waste Storage 22

Table 9.4 Thermal Conductivity of Potential Repository Rocks

Temperature Thermal Conductivity
Material (“0 (W/m”C)
Quartz monzonite 2.1 -3.4
Silt (Louisiana) 2.2
Salt (Louisiana) 4.0
Tuff 2.4
Tuff 1.55
Salt 2.08-6.1 1
Granite 1.99- 2.85
Shale 1.47- 1 .bH
Tuff (welded) 1.2-1.9
Tuff (unwelded) 0.4-0.8
I3asAlt 1.16- 1.56

preliminary planning is underway for a second repository lected in the 1990s as the nation’s first repository for
sometime after the turn of the century. A “granite” pro- commercial waste. The three candidates selected in-
gram was formed, but now disbanded, and a sedimentary cluded a salt site in Texas and the two reservations just
rock (shale) program is more or less inactive. Because cited. In December 1987, Congress terminated site char-
of the political implications of nuclear waste disposal, acterization plans at the Hanford site and the Texas site.
whatever programs are active today may be abandoned At this time, only the Nevada Test Site remains as a candi-
in the future, or new ones added, but it is reasonably date for the nation’s first repository for commercial
certain that the disposal problem will be with us for waste.
many decades. Figure 9.24 shows a cross section of the Nevada test
Figure 9.23 shows the regions in the United States that site where the design repository is located in the unsatu-
were given consideration for the geologic disposal of rated zone. Figure 9.25 shows the projected temperature
nuclear waste. Nine sites were eventually selected for an within the repository rocks. The heat pulse produced
environmental assessment, seven in salt, one in basalt, will propagate through the host rock by conduction and
and one in tuff. The environmental assessments were convection. Other than establishing buoyancy-driven
published by the Department of Energy in 1986. From flows, the heat generated is expected to alter the preem-
these assessments, three were picked for extensive site placement condition of the host rock, and this is one
characterizations, and one of these three was to be se- of the main concerns. These concerns have resulted in

Figure 9.23 Regions considered for geologic disposal of radioactive waste.

212 Chapter 9 / Heat Transport in Ground-Water Ftbw

Solitario Canyon Fault

0
t
0

1 2
2 Miles

3 Kilometers
- -y- - Water table
Figure 9.24 Cross section of the Nevada Test Site showing the location of proposed reposi-
tory (from Department of Energy, 1988).

research in coupled phenomena, that is, the thermal- work on all potential repository rocks. Thus, for salt,
mechanical-hydrochemical processes associated with a we are not surprised to learn that the main thermal-
nuclear repository (for example, Lawrence Berkeley Lab- mechanical coupling is the deformation or creep (flow)
oratory, 1984). This research has not been limited to the of the salt in response to heating. In addition, brine pock-
unsaturated zone at the Nevada Test Site but includes ets have been known to migrate up the thermal gradient,
that is, toward the heat source. For saturated rocks, one
of the main concerns is the potential for fracturing by
thermally induced fluid pressures. For unsaturated zones
in general, the problems are complex and include the
potential for upward gas transport of radionuclides due
to the buoyant rise of hot air along with temperature-
induced vapor phase transport. For ail rocks in general,
there is concern for the change in mechanical strength
of the rock body and stress corrosion of the canisters.
A literature review of coupled thermal, hydrologic, and
mechanicalchemical processes pertinent to the pro-
Outer surface of canister posed nuclear waste repository at Yucca Mountain has
been prepared by the Southwest Research Institute
(1993).

Tbermobydrochenzical EBects
-
100’
0
I I I I I
100 200 300 400 500 600 700 800
I I I I
900 1000
From the study of hot springs and ore deposits, it has
long been recognized that fluids in hydrothermal systems
Time After Emplacement (years) react and subsequently alter the rocks involved in hydro-
Ffgure 9.25 Temperature versus time after nuclear waste thermal circulation (Browne, 1978). As the repository
emplacement (from Department of Energy, 1988). and adjoining rock will in effect be a “hydrothermal”
 Problems -
213

system for a few to several hundred years after emplace- heating process. The potentially large volume of pore
ment, there is concern over the nature of these alter- fluid released and enhanced permeability could allow
ations. Of special concern is the effect on the sorbing migration of the pore fluid in response to the fluid pres-
minerals or those that undergo exchange, which can sures produced during the heating episode.
provide a barrier to the migration of radionuclides. Many
of these minerals are unstable at elevated temperatures
and may be altered by reversible dehydration or mineral
Thennomechanical Eflects
reactions. For example, certain zeolite niinerals (hydrous If heating takes place in fractured rock that has an appre-
aluminum silicates) are abundant in the volcanic tuffs at ciable permeability, the thermal expansion of the fluid
the Nevada Test Site. With few exceptions, natural zeo- is easily accommodated by fluid movement, and the re-
lites are found in rocks that have formed at temperatures sponse may take place very close to the condition of
less than 200°C. According to Smyth (1 982) reactions constant fluid pressure (Chapter 8). However, these ;ire
that may occur in these rocks below 200°C are of two not the types of rocks that are likely to be considered
types, simple reversible dehydration and mineralogical for a repository in that the permeability is already too
phase inversions. The first of these involves a loss of high. A more likely selection might have discontinuous
bound water with contraction of the unit cell by as much pores or fractures of various orientation saturated with
as 10%with the development and propagation of frac- water and of such a low permeability so that no fluid
tures and the loss of mechanical strength. Smyth (1982) expulsion takes place during the heating episode. This
concludes that this constraint may limit to 50°C the maxi- is the extreme case of heating at constant fluid mass.
mum temperature rise in zeolite horizons that occur 30 This case was examined theoretically for rock properties
to 50 m from the emplacement horizons. at the Hanford Reservation (Palciauskas and Domenico.
Another important mineral frequently associated with 1982). The sample calculations indicate that fracturing
the retardation of radionuclide migration is montmoril- is inevitable.
lonite. As noted in Chapter 8, the alteration of a montmo- Yet a third type of environment, and one that is pre-
rillonite to a mixed clay in the Gulf Coast begins at tem- sumably ideal from the perspective of nuclide migration,
peratures on the order of 60-70°C. Based on first-order is the rock type that contains a few isolated pores or
kinetics of an Arrhenius form, Domenico and Palciauskas fractures and of such a low permeability that it contains
(1988) have examined the half-life (the time required for essentially no water or at least is in a state of low fluid
half of the expanded montmorillonite layer to transform saturation. Heating in this case involves only the solid
to illite) as a function of temperature. At 40"C, the trans- components and their thermal expansion. Significint po-
formation requires 1.6 X 10" years; at 120°C, it requires rosity changes are anticipated only if the original porosity
2400 years; and at 160"C, it requires 230 years. Thus, is very low, that is, on the order of lo-' (Palciauskas and
one might conclude that the process is too slow in com- Domenico, 1982).
parison to the time scales that are involved in the thermal Other potential problems dealing with thermomechan-
evolution of the host rock. However, the kinetics ad- ical effects include the triggering of latent seismicity,
dressed the irreversible transformation only. Like zeo- spalling, and loss of mechanical strength. From the per-
lites, a reversible transformation can occur when the spective of contamination, the safe disposal of nlldedr
temperature reaches the boiling point of water. The en- waste is a very difficult problem; in the presence of heat
ergy required to break the bonds of the hydration enve- generation from decaying nuclides, the problem is se-
lope is significantly lower, so that the kinetics of the verely compounded.
process involves time scales shorter than those of the

Problems
1. Calculate upper and lower bounds for the thermal a thermal conductivity of 5 X lo-' cal s-' cm-' OC-'.
conductivity of saturated sandstone consisting largely of
4. From the data of Example 9.1
quartz with a porosity of 20%)using Table 9.1. Repeat the
calculations for a dry sandstone with the same porosity. a. Calculate the minimum value of permeability to initi-
2. Figure 9.6 applies to both downward and upward ate free convection.
moving ground water. Which half of the family of curves b. Repeat this calculation for a sand thickness of 500 ft.
applies to upward moving ground water, and what is the c. For a Rayleigh number of 57, would free convection
significance of the curve where Npc = O? occur in a rock body with a slope of 30"?
3. From the type curve match of Figure 9.8, calculate 5. Compare the Fourier number of Eq. 9.22 with the
the velocity q through a layer 10 m in thickness. Assume Fourier number for fluid flow (Eq. 4.41). State the analo-
-
214 Cbapter 9 / Heat Transport in Ground-Water Flow

gous quantities and provide an interpretation of T* for rated zone in an attempt to freeze the water and the
heat flow. For a repository rock, what are the advantages contaminant in the unsaturated zone. Based on what you
of having a high thermal conductivity and a low heat ca- have iearned in this chapter and in previous chapters, is
pacity? it possible to make some approximate calculation of the
6. Consider onedimensional steady-state conduction effectiveness of one bore hole in contributing to this feat
across a layer of thickness L where the temperature at by the process of conduction alone? What equation or
the bottom (x = 0) and at the top (x = L ) is known. physical concepts may be modified to describe this prob-
Solve this problem for the steadystate temperature as a lem, and what is the role of the time constant in consider-
function of x, L , To, and T,. (Hint: see Example 4.3). ations of the rate at which freezing can be accomplished?
(Hint: If you are confused, consult Carslaw and Yeager,
7. A soluble toxic chemical is moving downward
1959, the chapter on the use of sources and sinks in
through a thick unsaturated zone in sands and gravels
cases of variable temperature, the section on the continu-
threatening an underlying water supply. An experimental
ous line source.) Is this section (or equation) analogous
program is being devised whereby refrigerants will be
to anything you have learned thus Par?
circulated through bore holes penetrating the unsatu-
 Chapter 10

Solute Transport

10.I Advection
10.2 Basic Concepts of Dispersion
10.3 Cbaracter of the Dispersion Coencient
10.4 A Fichian Model of Dispersion
10.5 Mixing in Fractzrred Media
10.6 A Geostatistical Model of Dispersion
10.7 Tracers and Tracer Tests

Mass occurring in water as ions, molecules, or solid parti- driven flow systems (Chapter 5 ) , the factors that influ-
cles undergoes not only transport but also reactions. The ence ground-water flow patterns-water-table contipu-
reactions redistribute mass among ions or between liquid ration, pattern of geologic layering, size of the ground-
and solid phases. Figure 10.1 lists the important mass- water basin, pumping or injection-control the direction
transport and mass-transfer reactions in ground water. ;md cite of mass transport. Thus, the background from
Over the next four chapters, we will study these pro- the earlier treatment of flow allows us to consider advec-
cesses, establishing the background necessary to under- tion in just a few pages.
stand the classical problems involved with ground-water The two flow nets in Figure 10.2 illustrate basic fea-
geochemistry and contaminant hydrogeology. Chapter tures about advection. First when only advection is op-
10 begins with the mass transport processes, advection erating, mass added to one or more streani tubes will
and dispersion, which spread dissolved mass in ground remain in those stream tubes. Other processes (for exmi-
water. Advection is mass transport due simply to the flow ple, dispersion) move mass between streani tubes. Sec-
of water in which the mass is dissolved. The direction ond, the direction of mass spreading in steady-state sys-
and rate of transport coincide with the ground water. tems is defined by pathlines even in relatively complex
Dispersion is a process of fluid mixing that causes a zone flow systems. Because advection is the dominant trims-
of mixing to develop between a fluid of one coniposition port process, knowledge of ground-water flow patterns
that is adjacent to or being displaced by a fluid with ;I guides us in interpreting patterns of cont;imin;uit mi-
different composition. gration.
The discussion above assumes that mass transport does
not influence ground-water flow. When this assumption
tails, for example, with contaminated ground water h;w-
-
10.1 Advection ing a density that is significantly different from the snibi-
The intimate relationship between ground-water flow ent ground water, the flow of water and mass may di-
and the process of advection means that knowledge verge. However, for most practical problems. ground
about flow systems is directly transferable to llflderstdnd- water and the dissolved mass will move at the s;im cite
ing advection. For example, in the case of topographically (in the absence of other processes) and in the s m e
-
216 Chapter 10 / Solute Transport

portant in fractured rocks, where the effective porosity

--- - - - -- - __ y- - - can be much less than the total porosity, often as low as
I X or 1 X lo+.
Mass transport processes
There are situations where the mean ground-water
Advection velocity is different from the advective velocity of the
Dispersion / mass. Studies by Corey and others (1963) and Krupp and
Chemical, nuclear, and biological processes
Acid-base reactions / others (1972) have shown that negatively charged ions
Soltution, volatilization, and precipitation can move faster than the water in which they are dis-
Complexation
Solution reactions solved. Electrical charges due to the presence of clay
Oxidation-reduction reactions
Hydrolysis reactions minerals can force anions to remain in the center of
Isotropic reactions pores, the location of the maximum microscopic veloc-
Figure 10.1 Conceptualization of mass transport in a ity. The water itself may flow through the lower velocity
ground-water flow system. regimes within a pore. Alternatively,a reduction in advec-
tive velocity is possible when the geologic medium takes
direction. Accordingly, the velocity of advective trans- on properties of a semipermeable membrane. Solutes
port is described by the Darcy equation developed in avoid entering the membrane because of electrokinetic
Chapter 3 or in some cases size constraints. The process of mem-
brane filtration has been implicated in the origin of brines
K ah in sedimentary basins.
v=--- (10.1)
n, a1 There are other occasions where the linear velocity of
where v is the linear ground-water velocity, K is the advective transport is apparently less than that of the
hydraulic conductivity, n, is the effective porosity, and ground water. An example of retardation is provided by
ah/dl is the hydraulic gradient. As discussed in Chapter sorption, which will be taken up in Chapter 12.
3, the linear ground-water velocity and, therefore, the
velocity of advective transport increases with decreasing
effective porosity. This relationship is particularly im- -
10.2 Basic Concepts of Dispersion
Thinking again about flow tubes in a steady-stateground-
water system is a good place to begin examining the
concept of dispersion. Dispersion spreads mass beyond
the region it normally would occupy due to advection
alone. This idea can be illustrated with a simple column
apparatus (Figure 1 0 . 3 ~ )The
. apparatus is similar to a
Darcy column with additional plumbing to permit the
controlled and continuous addition of a tracer. Initially,
a steady flow of water is set up through the column. The
test begins with a dissolved tracer at a relative concentra-
tion ClC,, = 1 added across the entire cross section of
the column on a continuous basis. Monitoring the out-
flow from the column establishes the relative concentra-
tion of the tracer as a function of time.
The relative concentration of tracer varies in water
going into and coming out of the column (Figure 1 0 . 3 ~ ) .
Such relative concentration-versus-timeplots are known
as the source loading curve and the breakthrough curve,
respectively. The breakthrough curve does not have the
same “step” shape as the source-loadingfunction. Disper-
sion creates a zone of mixing between the displacing
fluid and the fluid being displaced. Some of the mass
leaves the column in advance of the advective front,
which is defined as the product of the seepage velocity
and time since displacement first started. The position
5om of the advective front at breakthrough corresponds to a
(61 C/C,, value of 0.5 (Figure 1 0 . 3 ~ )In
. the absence of disper-
Ftgure 10.2 Mass spreading by advection alone in (a) a sion or other processes, the shapes of the loading func-
shallow, unconfined aquifer and (b) a local flow system. tion and the breakthrough curve are identical.
 10.2 m s i c Concepts ofmspersion 217

Loading Function Breakthrough Curve

,Test begins Position of advective front
\

Time - Time -

(CO

Figure 10.3(u) Experimental apparatus to illustrate dispersion in a column.

The test begins with a continuous input of tracer ClC,, = 1 at the inflow end.
The relative concentration versus time fiinction at the outflow characterizes
dispersion in the column.

A zone of mixing gradually develops around the advec- There are a few other distinctions between Figures
tive front (Figure lO.3b). Dispersion moves some tracer 10.315and 10.4. At some distance behind the advective
from behind to in front of the advective front. The size front for the case of onedimensional dispersion, the origi-
of the zone of mixing increases as the advective front nal source concentration C,, is encountered. The longer
moves further from the source. transport continues, the greater the length of column
This column experiment is an example of onedimen- occupied by water with a tracer concentration C 0 . In
sional transport involving advection and dispersion. Simi- this zone, the concentration is at steady state. This result
lar mixing will occur in both two and three dimensions has been demonstrated hundreds of times in column
(Figure 10.4). Again dispersion spreads some of the mass experiments where transverse spreading is eliminated.
beyond the region it would occupy due to advection When transverse spreading is included with longitudi-
alone. There is spreading both ahead of the advective nal dispersion (Figure 10.4), the concentration distribu-
front in the same flow tube and laterally into the adjacent tion changes significantly. There is a zone of dispersion
flow tubes. This dispersion is referred to as longitudinal beyond the advective front caused exclusively by longitu-
and transverse dispersion, respectively. Dispersion in dinal dispersion, and a resulting zone of mass depletion
three dimensions involves spreading in transverse and
vertical directions as well as longitudinally.

------ Equrpotential fine

Zorte of dlrperslan a - Fiow line
Concentration
(fJ1 Mass spread due ta
Figure 10.3(6) Schematic representation of dispersion Source of mass CmgU adwsction alone
within the porous medium at three different times. A pro- Ffgure 10.4 Comparison of mass distributions in a shdllow
gressively larger zone of mixing forms between the two flu- unconfined aquifer due to advection alone and advection-dis-
ids (C/C,, = 1 and C/C,, = 0)displacing one another persion.
-
218 Chapter 10 / Solute Transport

behind the advective front. At some distance behind the

advective front, depending on the extent of dispersion, Eddy or turbulent diffusion
the system is at steady state with respect to concentra-
tion. This part of the tracer plume is sufficiently Far be-
hind the zone of longitudinal dispersion that it does not
contribute mass to the frontal portions of the plume. The
steady-state concentrations are, however, less than the
original concentration because mass also spreads trans- I Eddy or turbulent diffusion
versely and occupies an area (taken in the vertical plane) / Vertical thermocline,
and deeper regions in
that is greater than the area where the mass was injected. lakes and ocean
Transverse dispersion will reduce concentrdtions every-
where behind the advective front, while longitudinal dis-
persion, will do so only at the front of a plume. H
,eat
I in HO
,
Dispersion occurs in a porous medium because of two
processes, diffusion and mixing due to velocity varia- Molecular diffusion
tions. This latter process is called mechanical dispersion. , 11 Salts and gases in HO
,
The following two sections provide an introduction to I Proteins in H,O
these very important processes.
Ionic solutes

Di@bsion / Salts in H,O

In Chapter 3, Darcy’s equation was introduced as a linear

law relating fluid flow q to the gradient in hydraulic head.
For mass transport, there is a similar law that describes Diffusion coefficients characteristic of various
how chemical mass flux is proportional to the gradient environments. Reprinted with permission from Lerman, A.,
in concentration. This law is known as Fick’s law and is in Non-equilibrium Systems in Water Chemistry; Hem, J. D.,
expressed for a simple aqueous nonporous system as: ed.; ACS Advances in Chemistry Series lob: American Chemi-
cal Society: Washington DC, 1971, p. 32. Copyright 1971
J = -Dd grad (C) (10.2) American Chemical Society.

In this formulation, J is a chemical mass flux with the

negative sign indicating that the transport is in the direc-
tion of decreasing concentration. The porportionality where V is the averaging volume, n is porosity, and T is
constant Dd is termed the diffusion coefficient in a fluid a tortuosity vector to account for the hindering of free
environment. If the amount of diffusing material, J, and diffusion by collision with the pore walls. The form of
the concentration, C, in Eq. 10.2 are expressed in terms Eq. 10.3 has prompted the definition of a bulk diffusion
of the same quantity, the diffusion coefficient Dd has the coefficient D; to account for the effects of tortuosity,
units of L2/T. In other words, if J is expressed in units giving the statement most commonly referred to as Fick’s
of moles/L2T, and the concentration is in moles/L’, the law for diffusion in sediments
proportionality constant has the units LZ/T in that the
n grad (C)
J = -0; (10.4)
gradient operator grad( ) has the units L-’.
Molecular diffusion is mixing caused by random molec- In the absence of an evaluation for D;, an effective diffu-
ular motions due to the thermal kinetic energy of the sion coefficient DA = D;n is generally employed so that
solute. Because of molecular spacing, the coefficient de-
scribing this scattering is larger in gases than in liquids J = -0; grad (C) (10.5)
and larger in liquids than in solids. The diffusion coeffi-
Several different empirical approaches are useful in de-
cient in a porous medium is smaller than in pure liquids
fining an effective diffusion coefficient (DA). The follow-
primarily because collision with the solids of the medium
ing equation relates DA for the fluid in a porous medium
hinders diffusion. Figure 10.5 provides a range in values
to the diffusion coefficient in a liquid Dd (Domenico,
for diffusion coefficients for various fluid environments.
1977):
In a porous medium, diffusion takes place in the liquid
phase enclosed by a porous solid. Averaging techniques 0;= D8Dd (10.6)
provide the following rigorous statement for Fick’s law
in the fluid phase of a porous sediment (Whitaker, 1967): where D* is some constant less than one that accounts
for the structure of the porous medium. In most cases
D’is a function of both porosity and tortuosity. According
to Hellferich (1966), the effective diffusion coefficients
 10.2 &sic Concepts ofmspersion 213

in exchange columns fall in a range Table 10.2 Geological Features Contributing to

Nonidealities in a Porous Medium
~~~~ ~~~~ ~ ~~~ ~~~

A. Microscopic heterogeneity: pore to pore

1. Pore size distribution
where the factor D' depends only upon porosity n.Other 2. Pore geometry
ways of expressing this relationship are (Greenkorn, 3. Dead-end pore space
B. Macroscopic heterogeneity: well to well or intraforma-
1983) tional
Di = u nD,, (10.8) 1. Stratification characteristics
a. Nonuniform stratification
where u is a number less than one, and (Greenkorn and b. Stratification contrasts
Kessler, 1972) c. Stratification continuity
d. Insulation to cross-flow
2. Permeability characteristics
(10.9)
a. Nonuniform permeability
b. PemeabiIity trends
where r is the tortuosity. In this latter formulation, the c. Directional permeability
tortuosity r is defined as the ratio of the length of a flow C. Megascopic heterogeneity: formational (either fieldwidc
channel for a fluid particle (LJ to the length of a porous or regional)
medium sample (0.The value of ( L J L ) is always greater 1 . Reservoir geometry
than one except for perfect capillary type passages. Bear a. Overall structural framework: faults, dipping strata,
(1972) defines tortuosity as (L/L,) resulting in values less etc.
than one, and ranging between 0.56 and 0.8 in granular b. Overall stratigraphic framework: bar. blanket, chan-
nel fill, etc.
media. He states further that the effective diffusion coef-
2. Hyperpermeability-oriented natural fracture systems
ficient is the product of tonuosity and the diffusion coef-
ficient in the bulk fluid. Whichever of these equations is From Alpay (1972), a practical approach to defining reservoir hcterogc-
employed, effective diffusion coefficients in porous me- neity. Copyright 1972, Society of Petroleum Engineers Inc.. JFT (July
dia increase with increasing porosity and decrease with 1972).
increasing path length to sample length ratios.
Table 10.1 cites diffusion coefficients in water for com-
mon cations and anions. With the exception of H' and cm'/s with the smallest values associated with ions hav-
OH-, values of DL, range from 5 X to 20 X ing the greatest charge.

Mechanical Llispersion
Table 10.1 Diffusion Coefficients in Water for Some
Ions at 25°C Mechanical dispersion is mixing caused by local varia-
tions in velocity around some mean velocity of flow.
Cation D,j Cm*/S) hiOn Dd Cmz/S) Thus, mechanical dispersion is an advective process and
H' 93. I OH- 52.7 not a chemical one. With time, mass occupying some
Na' 13.3 F- 14.6 volume becomes gradually more dispersed as different
K' 19.6 cr- 20.3 fractions of mass are transported in these varying velocity
Rb' 20.6 Br- 20.1 regimes. Variability in the direction and rate of transport
cs+ 20.7 HS- 17.3 is caused by nonidealities in the porous medium. The
HCO; 11.8 most important variable in this respect is hydraulic con-
Mg2+ 7.05 ductivity. Table 10.2 lists some of the geologic features
Ca2+ 7.93 c0;- 9.55 that produce nonidealities. Nonidealities exist at a variety
S8+ 7.94 so:- 10.7 of scales ranging from microscopic to megascopic, to an
Ba" 8.48
even larger scale (not included in the table) involving
Raz+ 8.89
MnL' 6.88 groups of formations.
Fe" 7.19 Nonidealities at all of these scales are responsible for
CP' 5.94 mechanical dispersion. For example, variability in veloc-
Fe3 + 6.07 ity at the microscopic scale develops because of differing
flow regimes across individual pore throats or variability
From Li and Gregory (1974). in the tortuosity of the flow channels (Figure 10.6~).
Reprinted with permission from Geochirn.Cosrnochirn.Acta, 38, Y.-H.
ti and S. Gregory, Diffusion of ions in sea water and in deep-sea sedi- The heterogeneities shown in Figures 10.60 and 10.6~
ments. p. 703-714. Copyright 1974, with kind permission from Elsevier produce dispersion by creating variabilic). in flow at a
Science Ltd, The Boulevard, Langford Lane, Kidlington OX5 lGB, U.K. macroscopic scale. The first case (Figure 10.6b) is an
-
220 Chapter 10 / Solute Transport

(a)Microscale Nonidealities (12) Macroscale Heterogeneity-Simple Layering

219.14
7 218.89
218.64
218.39
c
218.14
2 217.89
d 217.64
217 39
Distance along section (meters)
-4.0- -In(K)
‘f Location of core
CZI K < 10-3cmls
[c) Macroscale Heterogeneity-Complex Spatial Structure (from Sudicky, 1986)
Fggure 20.6 Examples of nonideilities at different scales giving rise to mechanical disper-
sion. Panel (c) is from Sudicky, E. A., Water Resources Res., v. 22, p. 2069-2082, 1986. Copy-
right by American Geophysical Union.

example of nonidealities created by layering. Contrasts in Pfannkuch (1962) explored these relationships by taking
hydraulic conductivity among the various layers produce existing experimental data and casting .hem in the form
differing velocities. More complicated macroscopic het- of a series of dimensionless numbers DJDd and z ~ d , , , / D ~ ~ ,
erogeneity is shown in Figure 1 0 . 6 ~This. example, where u is the linear ground-water velocity, d,,, is the
adapted from Sudicky (1986), represents the plotted re- mean grdin size, DI.is the longitudinal dispersion coeffi-
sults of a large number of hydraulic conductivity measure- cient, and Dtl is the diffusion coefficient. The first of
ments for the shallow unconfined aquifer at Canadian these numbers normalizes the observed dispersion in the
Forces Base Borden. column by dividing it by the coefficient of diffusion of
the tracer in water. The second number is the Peclet
number (Np,$), a ratio expressing advective to diffusive
transport.
10.3 Character of the LMspersion Figure 10.7 represents relationships derived by Pfann-
Coencient kuch (1962) with four classes of mixing related to the
Peclet number. The first class (Class 1) is for small values
Studies at the Microscopic Scale of the Peclet number (NpEC 0.01). For mixing in Class
The coefficient of hydrodynamic dispersion (D) is the 1, DI,/D,,does not change as a function of N,,, which
sum of the coefficientsof bulk diffusion (D;) and mechan- indicates that diffusion is the main cause of mixing. The
ical dispersion (D’). Values of the bulk diffusion coeffi- ratio DL/Dtrhas a value of approximately 0.67 (Fried and
cient can be estimated reasonably well within an order Combarnous, 1971) because at this velocity D,. is equal
of magnitude for granular media. However, questions to the bulk diffusion coefficient D:, which is less than
remain as to what is a realistic range of values for the the diffusion coefficient (Dd)in water.
mechanical component D‘ and what is its relative contri- With increasing values of Np, (Class 2, NpE0.1-4),
bution to hydrodynamic dispersion. Fortunately, results mixing is influenced not only by diffusion but by mechan-
from laboratory column experiments can address these ical mixing. DJD, increases as a function of NPE(Figure
issues. 10.7). With a further increase in velocity (Class 3, NpE
The fluid flow velocity and grain size are the main 4- lo4),mechanical dispersion dominates the mixing pro-
controls on the longitudinal dispersion in a column. cess except at the low end of the range of Npli.Values
 10.3 Character of the LMspersion Coemcient 22

tions are
D, = a,v and DT = ( ~ ~ 1 ' (10.10)

61s"
1051
1
104

103
where a, and ( Y are
~ the longitudinat and transverse dis-
persivities of the medium. Dispersivities have units of
length and, like hydraulic conductivity, are characteristic
properties of a medium. In practice, they quantiQ me-
C
chanical dispersion in a system. There are actually two
transverse dispersivities measured at 90" to each other.
For example, with horizontal flow, one component of
transverse spreading will occur in the horizontal plane.
and the second will occur in the vertical plane.
I I I I
10'' 1 lo1 lo3 lo4 lo5 lo6
Studies at Macroscopic and Larger Scales
Much of what is known about dispersion is based on
Figure 10.7 Behavior of U,./U,,and DJD,, as a function of
the Peclet number (NPE).The classes identify regimes where
early theoretical studies and column studies. However,
mixing results from the same processes (after Pfannkuch, important contributions have come from field studies at
1962 and Perkins and Johnston, 1963, A review of diffusion macroscopic and larger scales. The main emphasis in this
and dispersion in porous media). Copyright 1963, Society of research has been to estimate dispersivity values from
Petroleum Engineers Inc., SPEJ (March 1963). field experiments. Gelhar and others (1992) undertook
a critical review of field experiments at 59 sites around
the world. Tests yielded some 106 values of longitudinal
of DLover this range are approximately proportional to dispersivity ranging from 0.01 m to 5500 m at scales of
d.*. Greenkorn (1983) explains this exponent of 1.2 as 0.75 m to 100 km. It appears that longitudinal dispersivi-
the outcome of several different mixing processes. With ties increase indefinitely with scale. However, the de-
increasing NpE,velocity profiles develop within the tailed review of these experiments (Figure 10.8) deter-
pores. This effect produces mixing analogous to disper- mined that only 14 values were highly reliable. A further
-
sion in a capillary tube with DL v'. Other forms of 31 values were of intermediate reliability. Looking at the
mixing produce a velocity dependence with DL ti. A - most reliable results, however, the trend of increasing
case in point is mixing due to the tortuosity of the flow
paths due to the presence of grains. For values of NpEin
the range from lo4-lo6(Class 4), the effects of molecular
-
diffusion are negligible. DL u and the experimental
data plot as a straight line with a slope of 45".
One important result is the determination that longitu-
dinal dispersion is proportional to velocity. This relation-
ship has been generalized to describe dispersion at mac-
roscopic and megascopic scales. Column experiments
have also provided important information about trans-
verse dispersion. Shown in Figure 10.7 is the ratio
Dr/D<,plotted versus NPE (from Perkins and Johnston,
1963). The two curves in Figure 10.7 are similar in shape,
indicating that the process of transverse mixing is much
the same as longitudinal mixing. However, longitudinal
dispersion is greater than transverse dispersion at a given Reliability
NpE.For example, when NpE> 100, values of D,. are
approximately 10 times greater than D T .This tendency 0 Intermediate
for 0,to exceed DT applies to dispersion not only at
microscopic scaIes but at larger scales as well.

Scale (in)
Dispersivity as a Medium Property
Figure la8 Longitudinal dispersivity versus scale with
Column experiments have been useful in establishing data classified by reliability (from Gelhar, and others, Witcr
relationships between the coefficients of mechanical dis- Resources Res., v. 28, p. 1955-197-1. 1992. Copyright by
persion and the mean ground-water velocity. The equa- Amer. Geophys. Union).
-
222 Cbapter I 0 / Solute Transport

longitudinal dispersivity as a function of distance was of the method of interpretation. For example, characteriz-
more apparent than real (Gelhar and others, 1992). The ing the two-or three-dimensional spread of a tracer with a
most reliable dispersivity values are all at the low end of one-dimensional model requires an apparent dispersivity
the range. that increases as a function of distance (Domenico and
From such tests, a consistent view ;tbout macroscopic Robbins, 1984). However, the interpretation from a di-
dispersion has emerged. Heterogeneity at the macro- mensionally correct model yields a constant dispersivity.
scopic scale contributes significantly to dispersion be- There are results to suggest that scale-dependent dispersi-
cause it creates local-scale variability in velocity. Values vities exist.
of macroscopic dispersivity are in general two or more How can these nonconstant values be explained? One
orders of magnitude larger than those from column ex- way is to consider them simply to be a result of inconi-
periments. A typical range of values in longitudinal disper- plete spatial averaging. Consider a dispersion experiment
sivity from column experiments is from to 1 cm. In run in a heterogeneous unit with relatively large lenses.
field experiments, values range from approximately 0.1 The spread of tracer from a source to a nearby obsewa-
to 2 m over relatively short transport distances. At any tion point in the same lens should yield a relatively small
given scale, longitudinal dispersivities can range over 2 dispersivity that is comparable to a column value. The
to 3 orders of magnitude depending on the variability in tracer must move a substantial distance before it is able
hydraulic conductivity. Although no reliable, large-scale to fully interact with the heterogeneity to the extent
studies have been carried out, it is probable that longitudi- necessary to produce macroscale mixing. Thus, dispersiv-
nal dispersivity values in excess of 10 m exist. Values, ity values increase away from a source before becoming
however, may not be as large as estimates determined constant. Gelhar and others (1979) refers to this constant
using contaminant plumes and environmental isotopes, macroscale dispersivity as the asymptotic dispersivity.
which are considered less reliable (Gelhar and others, The approach to an asymptotic dispersivity might involve
1992). tracer spreading over 10s or 100s of meters.
Many dispersion studies have suffered from errors re- Another way to explain spatially varying dispersivity
lated to the collection and interpretation of field data. values is in relation to some representative elemental
These errors arise from (1) poor definition of the plume volume (REV) for dispersivity. With the spreading of a
due to a small number of monitoring wells and nonpoint tracer, eventirally a transition will occur from the micro-
sampling, (2) failure to account for temporal variations scale to macroscale. The implicit assumption is that for
in the advective flow regime, which can lead to lateral each scale a REV exists (Figure 10.9).
mass spreading, (3) incomplete information about the In the zone of transition between various scales, it may
source loading, (4) inherent limitations in some test pro- be difficult to define a continuum value for dispersivity.
cedures, and ( 5 ) oversimplified techniques for interpre- Spreading may not yet have encompassed a volume that
ting the test results, for example, assuming that a source is equivalent to a REV. As a consequence, small changes
of finite size can be approximated as a point source. in volume could result in significant variability in the
These problems typically force estimates of dispersivity spreading. Also, mnss may not be normally distributed as
values to be larger than actual values. Thus, many of classical theory would predict, and as a result, dispersivity
the published estimates of dispersivity are calibration o r values may not be definable. An example of this behavior
curve-fitting parameters (Anderson, 1984). is the concentration distribution from a tracer experi-
Results from field experiments also support the general
finding that horizontal transverse dispersivities are at least
an order of magnitude smaller than longitudinal values REV
(Gelhar and others, 1992). Further, vertical transverse megascopic
-
scale
dispersivities are typically some 1 to 2 orders of magni-
tude smaller than horizontal transverse dispersivities
(Gelhar and others, 1992). In media with pronounced
Transition
horizontal stratification, values of vertical dispersivity zone
may be similar to the bulk diffusion coefficient (Sud-
icky, 1986).
Accompanying the increased understanding about
large-scale dispersion is the realization that the fundamen- 4 ' REV
microscopic
tal concepts may not be as straightforward as they appear. scale
One important question is whether constant values of
macroscopic dispersivity exist. Several different field ex-
periments have described dispersivities that increase as
a function of the travel distance of the plume. In some
cases, these variable dispersivities are simply an artifxt
 10.4 A Fickiati Model of Dispersion 223

ment shown in Figure 10.10 (Sudicky and others, 1983). relatively short distances may not be representative of
Because the tracer was transported through different ele- asymptotic values. Further, it simply may not be practical
ments of the heterogeneous conductivity field in a sub- to carry out dispersion experiments on a routine basis
REV system, the tracer plume separated close to the over distances large enough to yield an asymptotic value
source. Asymptotic values of dispersivity could conceiv- (Sudicky, 1986).
ably exist for this system. However, in the earfy stages Problems associated with concepts of macroscale dis-
of spreading, defining a dispersivity value may be difficult. persivities have not invalidated their use. They remain,
The tendency for dispersivity values to be scale depen- in spite of limitations, a useful conceptual model of ;I
dent has important implications for how macroscopic complicated process. However, one must be aware of
dispersivity values are estimated and used. For example, limitations in extending classical concepts of dispersion
longitudinal dispersivity values from tracer tests run over from microscale to macroscale and to larger systems.

... -
10.4 A Fickian Model of Dispersion
D
. Dispersion or more formally hydrodynamic dispersion

... . . . ... . .
tI = 8days
occurs as a consequence of two different processes: diffii-
sion and mechanical dispersion. These two contributions
to hydrodynamic dispersion are represented matheniati-

. .
D cally as
0 .

.. D = D ’-+ D: (10.1 1)

0 . . .. . ’\
where D is the coefficient of hydrodynamic dispersion.
D’ is the coefficient of mechanical dispersion. and U: is
the bulk diffusion coefficient.
\. In most modeling studies, use of a coefficient of hydro-
I
I. t2 = 15 days dynamic dispersion implies that the behavior is Fickian.
,/

. By Fickian, w e mean the transport of a quantity in ;I

.. 0 . . . . direction down a concentration gradient at a rate that

is proportional to the gradient. While this relationship
describes molecular diffusion, it is not a mec1ianistic;rlly

.. correct representation of mechanical dispersion. The jus-

...
. . .. .
.
tification of why the diffusional model is indeed used for
mechanical dispersion is a practical one. The net effect

... . ..
./-1o--r
/ t3 = 29 days of both processes operating together is to cause the coii-

... ...
D ( 0 .
centration distribution to become normal within ;I short
\
rn I flow distance. In other words, although the microscopic
Fast zone
mechanisms for diffusion and mechanical dispersion dif-
fer completely, the macroscopic outcome for both pro-
. . . . .. . . cesses is similar, that is, a distribution that becomes nor-

-
mal. Given this outcome, we can represent dispersion

..
within the framework of a diffusion model.
Consider the normal distribution of Figure 1 0 . 1 1 ;Ind

0 2 4
Tracer injection wells
Meters -4W- CI- mg/L
Figure 10.10 Measured C1- concentrdtions from a tracer
experiment in the shallow aquifer at the Canadian Forces
Base. Borden. After a short time, the CI‘ plume separated
due t o different rates of advection in a “fast” and “slow” \
zone Cafter Sudicky and others, 1983). Reprinted from J. Hy-
drol., v. 63, E. A. Sudicky, J. A. Cherry, and E. 0. Frind, “Mi-
gration of contaminants in groundwater at a landfill: A case
study,” p. 81- 108, 1983 with kind permission from Elsevier
Science, NL, Sara Burgerhdrtstraat 25, 1055 KV Amsterdam,
The Netherlands. Figure 10.11 The normal or Gaussian distrihutioil.
-
224 Cbaptw 10 / Solute Transport

Table 10.3 Properties of the Normal Distribution to the dispersion coefficient as

Percentage of the Mass u = (2Dt)’I2 ( 0.

Values from Contained (Calculated as Two
the Normal Times the Value of the where t is time and D is the dispersion coefficient.
E The relative concentration of a pulse or tracer added
U Distribution Normal Distribution)
~ ~ ~~
at a point changes as a function of time with transport
0 0 0 in a constant-velocity flow system. After the distribution
0.1 0.0398 8 becomes Gaussian, the mean defines the position due
0.2 0.0793 15.8 to transport at the linear ground-water velocity and the
0.3 0.1197 24
variance describes the variation about the mean. Thus,
0.4 0.1554 31
0.1915
for a onedimensional column (Figure 10.12a),
0.5 38
0.6 0.2257 45
0.7 0.2580 52 (10.13)
0.8 0.2881 57.6
0.9 0.3159 63.2 where 0,.is the longitudinal dispersion coefficient. When
1.o 0.3413 68.3 the velocity is constant, Eq. 10.13 becomes
1.2 0.3849 77
1.4 0.4192 83.8
1.6 0.4452 89 (10.14)
1.8 0.4641 92.8
2.0 0.4773 95.4 where v is the linear ground-water velocity and x is the
2.5 0.4938 98.7 distance from the source. Both theoretical arguments and
3.0 0.4987 99.74 column experiments support these relationships.
4.0 0.49996 99.99 When mass spreads in two or three dimensions, the
infinity 0.5 100.00 distributions of mass sampled normal to the direction of
flow are also normally distributed, with variances that
increase in proportion to 2t. Thus, coefficients of trans-
verse dispersion are defined as
the accompanying properties given in Table 10.3. The
standard deviation u or the variance uzare measures of Dr=-
ff: (10.15)
2t
the spread in the data about the mean. Thus, from Table
10.3, a spread of 4 0 (x/u = 2 ) incorporates about 95.4% The two-dimensional spread of a tracer in a unidirec-
of the mass or area under the concentration distribution. tional flow field results in an elliptically shaped con-
One standard deviation contains about 68.3%of the total centration distribution (Figure 10.126) that is normally
mass. Statistics of the Gaussian distribution are related distributed in both the longitudinal and transverse direc-

0.4 I
\ Direction
of
V I \ transport /
F 0.2
2 0.0 rr
u
Distance
(n)

Source
0 __t

(6)
Ffgure 20.22 Variation in concentration of a tracer spreading in (a) one or (6)two dinien-
sions in a constant-velocityflow system.
 Next Page

10.4 A Fickian Model of Dispersion 22

tions. Typically, longitudinal dispersion is greater than shown in Figure 10.13a that is produced by a continuous
transverse dispersion (Figure 10.12b). Concentration dis- point source. At any point along the midline of the plume
tributions in three dimensions form ellipsoids of revolu- (for example, x, 0) a breakthrough curve (Figure 10.13b)
tion (football shapes) when the two components of trans- can be constructed by plotting the relative concentration
verse spreading are the same and longitudinal dispersion as a function of time (Robbins, 1983). The relative con-
is larger. When the vertical transverse dispersivity is centration is C/C,,, ,where C,,, is the highest concentra-
small, as is often the case, plumes take on a surfboard tion that will be observed at (x, 0). C,,, is always less
shape. than the source concentration C,.
The variances we have discussed differ in a subtle but This breakthrough curve is a cumulative normal distri-
important way. Those in Eqs. 10.13 and 10.15 are spatial bution with a 2u value that can be derived graphically
distributions, while the one calculated from a break- from Figure 10.1% as
through curve is the variance of a temporal distribution.
The following equation (Robbins, 1983) interchanges 2at = (tx, - ti61 (10.18)
space- time statistics relating to concentration distribu- where tg4and t l 6 are the breakthrough times correspond-
tions ing to relative concentrations of 0.84,and 0.16, respec-
ut = u*u: (10.16) tively, which accounts for about 68%of the mass (Table
10.3). With u known, the dispersion coefficient is calcu-
Thus, by recasting Eq. 10.3 in terms of r, lated with Eq. 10.17.
The same approach can establish values of dispersivity
( 10.17) in the transverse direction. This time the relative concen-
tration is plotted as a function of distance traveled trans-
the dispersion coefficient in the longitudinal direction verse to the direction of flow at t5,,(Robbins, 1983).
can be calculated for a breakthrough curve. The resulting concentration distribution is again normal
For example, consider the two-dimensional plume (Figure 10.1%). Once the variance is estimated, the dis-

Concentration contours ‘i

I
Y
( a )Spreading of a tracer from a point source

0
t16 ‘50 164

c _ Transverse Distance at t50

24
Time at (x. 01
( b ) Breakthrough curve at 0)
(s, (c) Variation in relative
concentration along (J,. )81-

Ffgure 10.23 Pattern of spreading of a tracer from a continuous point source in a two-
dimensional flow field. Two graphical procedures for estimating the standard deviation of a
normal distribution are shown in ( b ) and (c) (from Robbins, 1983).
Previous Page

-
226 Cbapter 10 / Solute Transport

persivity value is obtained from Eq. 10.14. The standard Mixing in a fracture system is the result of both me-
deviation in this case (Figure 10.13) is related to the half chanical mixing and diffusion. Concentrations in the frdc-
width of the distribution (r ) at a relative concentration tures will be affected by diffusion into the matrix, a chem-
of OSC,, (Robbins, 1983) ical process (Figure 10.14~2);variability within individual
fractures caused by asperities, an advective dispersive
u7.=-
r (10.19) process (Figure 10.146);fluid mixing at the fracture inter-
2.354 sections, a diluting or possibly diffusive process (Figure
This graphical approach can be used with any normal 10.14~);and variability in velocity caused by differing
distribution. As before, space-time statistics can be inter- scales of fracturing (Figure lO.14d) or by variations in
changed with fracture density (Figure 10.14e). Some of these processes
are coupled in a complex fashion and compete.
r., = r,v (10.20)
Diffusion into the matrix provides an important pro-
cess to attenuate the transport of contaminants. This
-
10.5 Mixing in Fractured Media
result contradicts experience with porous media, where
the contribution of diffusion to dispersion is generally
Many of the concepts of mixing from the previous sec- swamped by mechanical dispersion. Diffusion is more
tions apply to fractured media. There is nevertheless important in fractured media because localizing mass in
sufficient added complexity to warrant treating these fractures provides the opportunity for large concentra-
media separately. The discussion here will be focused tion gradients to develop. Theoretical studies by Grisak
on the only type of fractured medium that has been and Pickens (1980) and Tang and others (1981) explored
studied in detail. This is a case where fractures are the this process using numerical and analytical models of a
only permeable pathways through the rock. The hydrau- single fracture bounded by an infinite porous matrix.
lic conductivity of the unfractured part of the medium Dispersion in a single fracture is caused by the variability
is so small that fluid flow and hence advective transport in fracture aperture. This variability develops due to the
are negligible. However, mass can diffuse into the unfrac- roughness of the fracture walls and the precipitation of
tured rock matrix in response to concentration gradients. secondary minerals. At many locations, the fracture may

(a)Diffusion into the Matrix

“1

&bdvectlon and disprsian

I

Ctosed fracture
--- Fracture aperture (gum)
(6)Variability in velocity due to fracture roughness

7&
Source

(c) Mixing at fracture intersections ((0Variability in velocity due to

differing scales of fracturing
(4Variability in velocity due to
variable fracture density
Figure 10.14 A series of sketches illustrating the different causes of dispersion in fractured
rocks.
 10.5 Mixing in Fractured Media 23

Table 20.4 Tabulation of Dispersivity Values Based on Tracer Tests in Fractured Rocks

Migration Fissure
Distance Dispersivity Width
Experiments (4 (m) (mm)
Laboratory experiment with natural fissure 0.3 0.025 0.18
(Neretnieks and others, 1982)
Laboratory experiment with natural fissure 0.19 0.005 0.15
(Moreno and others, 1984) 0.27 0.01I 0. I4
Stripa natural fissure, two different channels 4.5 2.0 0.11
(Abelin and others, 1982) 4.5 0.62 0.14
Studsvik site 2 11.6 - -
(Landstriini and others, 1982)
Finnsjdn site 30 0.35/5 0.47
(Moreno and others, 1983) 0.98
Studsvik site I 22 6. t -
(Landstriini and others, 1978) 51 7.7
French site 11.8 0.8
(Lallemand-Barrks, 1978)
U.S. site 538 134
(Webster and others, 1970)
From Neretnieks (1 985)

be closed to flow and transport. Such aperture topology problems that confound the application of classical the-
(see Figure 10.14b)gives rise to channeling (Neretnieks, ory. If pure channeling develops, the dispersivity will
1985)where mass moves predominantly along networks increase continuously as a function of travel distance ;ind
of irregularly shaped pathways in the plane of a fracture. never reach an asymptotic value (Neretnieks, 1985). The
This channelization model has developed from tracer lack of spatial averaging in channelized fracture systems
tests carried out in fractured rocks. For example, studies may also inhibit the development of normal concentra-
at the Stripa mine (Neretnieks, 1985) showed how the tion distributions.
inflow of water to tunnels was extremely localized. Ap- Dispersion at the next larger scale occurs when the
proximately one third of the flow entered from approxi- geometry of the three-dimensional network begins to
mately 2% of the fractured rocks. These results are influence mass transport. A tracer moving along a fracture
strongly indicative of localized channel flow within indi- to an intersection (Figure 1 0 . 1 4 ~partitions
) into two or
vidual fractures. more fractures. Because the water is also partitioned and
Channels CAn be so poorly interconnected that they because concentration is the mass per unit volume of
may not interact with one another over appreciable dis- solution, partitioning by itself does not affect the concen-
tances (Neretnieks, 1985).This behavior is termed pirre trations. However, concentrations will be affected if
channeling. From this discussion, it is easy to understand there is mixing with water that does not contain the
why the smooth, parallel plate model used by some inves- tracer. This dilution depends directly on the type of mix-
tigators is really a simplified representation of a fracture. ing process at the intersection and quantity of water
Neretnieks (1985)has tabulated values of dispersivity flowing along the fractures.
from tracer experiments in fractured rocks (Table 10.4). The dispersion in Figures 10.14d and e is analogous to
Because of the relatively short transport distances, these that caused by heterogeneities in porous media. Differing
values reflect mixing caused by diffusion into the matrix scales of fracturing or spatial variability in fracture density
and mechanical dispersion within individual fractures. create variability in locar velocity. Even larger scale dis-
Thus, they are not true advective dispenivities. The val- persion could develop if individual discontinuities exist
ues, in general, increase as a function of travel distance on a regional scale or tlie variability in fracture density
as in porous media. The larger dispersion lengths for tlie includes several units.
last four tests, however, may be artifacts of the interpre- Many of the concerns about the applicability of cl;usi-
tive model, which did not account for matrix diffusion cal concepts of dispersion apply to fractured rocks. For
(Neretnieks, 1985). This example emphasizes how pub- example, is it meaningful or even possible to describe
lished dispersivity numbers need to be examined in rela- asymptotic dispersivities? How does dispersivity change
tion to the methods used for interpretation. as ;I function of distance? Is it even possible i n real net-
As is the case with porous media, there are conceptual works to define a REV! Presently, these questions remain
2 2 Chapter 10 / Solute Transport

unresolved. Looking into the crystal ball, it is possible

that many of the existing ideas about dispersion may y1

have to be extended or reworked for fractured rocks. y2

-
10.6 A Geostatistical Model
of Dfspersion
... y3

Exciting new contributions to the study of dispersion are

quantitative approaches that relate asymptotic dispersivi-
ties to geostatistical models of hydraulic conductivity.
Eventually, these approaches will form the basis for a
family of practical field techniques for characterizing dis-
.
.*

y,,
persivity. To begin, a few basic ideas need to be devel-
oped concerning how a geostatistical model can repre- Figure 10.15 A series of equdlly spaced In K values (from
sent heterogeneity in properties like hydraulic Freeze and others, 1989). Reprinted by permission of
conductivity, porosity, or cation exchange capacity. Ground Water. Copyright 0 1989. All rights reserved.
Medium properties in nature vary from place to place.
Consider the equally spaced hydraulic conductivity mea-
surements in Figure 10.15. Given such data, the goal of sediments are deposited, the correlation is usually aniso-
a geostatisticaldescription is to represent the heterogene- tropic. For example, spatial persistence in hydraulic con-
ity by a small number of statistical parameters. For the ductivity is greater in the direction of bedding than across
simplest case, three are sufficient, the mean, the variance, the bedding. Chapter 3 introduced the concept of statisti-
and the correlation length scales. This description implies cal homogeneity or heterogeneity as descriptors that de-
that the medium is stationary. In other words, there is scribe this characteristic of a medium.
no trend in the statistical character of the medium across
the field. Mean and Variance
For a series of data, the mean is the level around which
the parameter values fluctuate, while the variance de- The population statistics for a heterogeneous field are
scribes the spread about this level (Box and Jenkins, usually estimated from a finite number of samples. In
1976). When data vary over several orders of magnitude, terms of notation, the statistics of the actual population
they are often best described by a log-normal distribution are different than the sample estimates (Table 10.5).
(Freeze and others, 1989). In the case of hydraulic con- Thus, for a series of log hydraulic conductivity values,
ductivity, the following equation describes the distri- the mean estimated from the samples is
bution: - 1
Y = - c Y/ i = l . . .n (10.22)
(10.21) n
where Y , is the log hydraulic conductivity.The mean and where Bis the mean hydraulic conductivity and Y , are
variance for the population are p y and o:, respectively. n individual log hydraulic conductivity values. The sam-
The correlation length is a measure of the spatial persis- ple variance is
tence of zones of similar properties. For example, hydrau- 1
lic conductivity usually is correlated such that values s:.=-C(y,-
n T)Z i = 1 . . .n (10.23)
immediatelyadjacent to one another are similar and those
much further away are dissimilar. Because of the way where S:.is the variance in hydraulic conductivity.

Table 10.5 Summary of Notation for the Geostatistical Description of a Medium

~~~

Parameter Population Statistic Population Estimate Sample Estimate

Mean CLY P,. P
Variance u
: i?:. s:
Autocovariance 7,;Or Tv (h) &.?, or ?, (h) C1; or C,(h)
Autocorrelation PY, or Pi@) P, o',Pdb) rll.or rdb)
Correlation length A, A,
~

From Freeze and others (1989). Reprinted by permission of Ground Water. Copyright 0 1989. All
rights reserved.
 10.6 A Geostatistital Model of Dfspersion 2 3

0.37 (Figure 10.16)or declines to 0.37 of the covariance

at lag zero.
For example, with a finite number of vertical values
of log hydraulic conductivity, sample autocovariance (C,)
and autocorrelation (ry)functions can be calculated from
pairs of equally spaced data. The autocovariance function
at a separation h is

“0 1 2 3 4 5 6 7
Lag k Separation !J
Figure 20.26 Example of an autocorrelation function with where m is the number of pairs of sample points at
the correlation length defined as separation at which p = a given separation, yi is the spatial coordinate of the
0.37 (modified from Freeze and others, 1989). Reprinted by
measurement Yo,), and YOi + h) is the measurement
permission of Ground Water. Copyright 0 1989. All rights re-
h units away from Yo,) . autocorrelation function
The
served.
r,@) is formed by taking C,@) values and dividing by
C,-(O) or the variance Sf. of the sample. For unevenly
spaced data, values are grouped into a series of class
Autocovariance and Autocorrelation intervals with the separation distance given by the mid-
Functions point of each interval. The following example illustrates
A population can be described in terms of autocovariance the calculation of the mean, variance, and correlation
(7) or autocorrelation @) functions. These functions are length for a stationary medium.
related mathematically and describe how the correlation
between any two hydraulic conductivity values decays Example 10.1
with the separation or lag. A lag is simply some constant
Listed on Table 10.6 are values of hydraulic conductivity
separation interval. For example, with lag one, values
Y (where Y = In K) collected at 5-cm intervals along
being compared are a distance h apart, and with lag
a core. Calculate the mean, variance, and correlation
two,2h, etc. Although there is flexibility in defining an
length scale.
autocorrelation function, an exponential model is used
The solution is outlined on Table 10.6.The mean and
commonly (Freeze and others, 1989)
variance are 1 .O and 0.182, respectively. The autocovari-
p(h) = exp(-/hl/A) (1 0.24) ante coefficients at separations of 5 and 10 cm are 0.149
and 0.07, respectively, which when divided by the vari-
where p(h) is the autocorrelation of the population at a
ance give autocorrelation coefficients 0.81 and 0.38, re-
separation h. The correlation length scale (A) (or A,, A,.,
spectively. Because the autocorrelation coefficient at 10
A, for statistically anisotropic media) is the separation in
the given direction at which p takes on a value of e-’ or cm is nearly 0.37, A is about 10 cm.

Table 10.6 Work Sheet for Example 10.1

1 1.5 0.50 0.25 0.70 0.50 0.35 0.25

2 1.7 0.70 0.49 0.50 0.10 0.35 0.07
3 1.5 0.50 0.25 0.10 -0.2 0.05 -0.10
4 1.1 0.10 0.01 -0.20 -0.60 -0.02 -0.06
5 0.8 -0.20 0.04 -0.60 -0.40 0.12 0.08
6 0.4 -0.60 0.36 -0.40 -0.50 0.24 0..30
7 0.6 -0.40 0.16 -0.5 -0.10 0.20 0.04
8 0.5 -0.50 0.25 -0.1 0.0 0.05 0.0
9 0.9 -0.10 0.01 0.0 0.0
10 1.0
- 0.0 0.0
-
z 10 1.82 1.34 0.58
1
-c
n
1 0.182 0.149 0.07
230 Cbapter I 0 / Solute Transporl

Generation of Correlated Random Fields Estimation of Dispersdvdty

Some applications (for example, optimization of a pump Gelhar and Axness (1983) provide the following theoreti-
and treat system) depend on being able to generate syn- cal equation for predicting longitudinal dispersivity from
thetic hydraulic conductivity or permeability fields with a geostatistical description of the hydraulic conductiv-
specified geostatistical properties ( p Y ,A,, Ay , A=). Vari- ity field:
ous mathematical approaches can be used for this pur-
pose such as nearest-neighbor models (Smith and (10.26)
Schwartz, 1981) or turning bands methods (Delhomme,
1979; Mantoglou and Wilson, 1982). A new Fourier trans- where A/. is the asymptotic longitudinal dispersivity, cr$
form approach is capable of generating three-dimen- is the variance of the log-transformed hydraulic conduc-
sional, cross-correlated fields. The term cross-cowelrrte'cl tivity [i.e., Y = In K ] , A is the correlation length in the
means that in addition to generating, say, a hydraulic mean direction of flow, and y is a flow factor, which
conductivity field, the algorithm also generates a second Dagan (1982) considers equal to one. This equation
field like porosity, where porosity values are correlated assumes (1) unidirectional mean flow and (2) an expo-
to the hydraulic conductivity values in some specified nential covariance. The longitudinal asymptotic macro-
manner (Robin and others, 1993). dispersivity (A?)includes contributions from (1) the het-
The algorithm of Robin and others (1993) has created erogeneous structiire of the medium (A,,); (2) a local or
the example of a two-dimensional, correlated permeabil- pore-scale component (aL);and (3) diffusion
ity field in Figure 10.17.The field has a mean permeability
of 5.8 X 10-" m*, a variance in In k of 0.27, and correla- A; = A / . + LY/ + D;/v (10.27)
tion length scales A,v = 0.1 m, and A, = 0.02 m. The field where D: is the bulk diffusion coefficient and u is the
is statistically anisotropic because the correlation length linear ground-water velocity. The transverse asymptotic
in the x direction (or horizontal direction) differs from macrodispersivity is
the correlation length in the z direction (or vertical di-
rection). +
A:. = L Y ~ D;/v (10.28)

2m 23.0 24.Q 29.0

- In(k) m**2
Ffgure 10.17 Example of a correlated permeability field with dimensions 1.06 X 0.5 m having a rneiln k of 5.8 X lo-" rn', u'
of 0.27, A,r of 0.1 rn and A, of 0.02 m. Provided by R.A. Schincariol.
 10.7 Tracers arid Tracer Tests %
2

The transverse asymptotic dispersivity ( A /) is zero, im- methods. One limitation is the relatively large number
plying that the heterogeneous structure of the medium of hydraulic conductivity values that are required t o de-
does not create transverse dispersion. fine the spatial structure. The only practical way to ad-
Given the variance in In K and the correhtion length, dress this limitation is through indirect determination o f
one can estimate the overall contribution t o dispersion hydraulic conductivity using various kinds of borehole
that is related to the macroscale heterogeneity. The other logs or a borehole flowmeter.
two components, a/ and D;/z!, are relatively small and
can be estimated or in many cases neglected. The s u m
gives us A ; . Unfortunately, the characterization of the
statistical properties of a niediuni requires considerable - Tracers and Tracer Tests
10.7
hydraulic conductivity tl;ita, which practically are not Tracers and tracer tests are useful in understanding dis-
available. Borehole flow meters o r geophysical ap- persion and advection. The most important tracers in-
proaches are being developed t o provide ways of charac- clude (1) ions that occur naturally in a ground-water
terizing the statistical structure. system such as Br- or C1-, (2) environmental isotopes
An example in the estimation of longitudinal asymp- such as 'H,'H, or I8O,( 3 ) contaminants of ail kinds that
totic dispersivity is provided with data collected by Sud- enter a flow system, and ( 4 ) chemicals added t o a tlow
icky (1986). Shown in Figure 10.18 is the contoured system as part of an experiment. This latter group could
hydraulic conductivity field based on 1270 measurements include radioisotopes such as 'H. "'I, "Br; ions such as
along a single cross section. This section is oriented paral- CI-, Br-, I - ; and organic compounds such as rhodamine
lel to the direction of tracer migration at Canadian Forces WT, fluorescein, and uranine. Many of these ions or coni-
Borden (Freyberg, 1986; and Mackay and others, 1986). pounds do not react to any appreciable extent with other
The distribution in hydraulic conductivity when ex- ions in solution and the porous medium. They are consid-
pressed as In K is normally distributed with cri = 0.29, ered as "ideal tracers." Reactive tracers are used more
with horizontal and vertical correlation length scales (A, specifically to define the nature of reactions.
and A,) of 2.8 and 0.12 m, respectively (Sudicky, 1986). Given this variety of tracers and testing strateb'
lies, con-
These values confirm what is apparent from looking at fusion may arise as to which to use in site studies. In
the contoured field, namely that the correlation stnicture most cases, the choice is linked inexorably to the scale
is anisotropic. of the study or the presence of a contaminant plume.
The contribution of heterogeneity t o the asymptotic For example, tracing flow and dispersion in a unit of
longitudinal dispersivity can be calculated from Eq. regional extent (10s to 100s of kilometers) will involve
10.23 as either naturally occurring ions or environmental iso-
topes. Conducting a tracer experiment on such a large
scale is simply not feasible because of the long times
required for tracers to spread regionally. On a more local-
where the tlow Factor y is taken to be one. With the ized or site-specific basis (several kilometers), the pres-
contribution from diffusion essentially negligible and the ence of a contaminant plume that has spread over ;L long
local-scale dispersivity (a,)estimated to be 0.05 m, most time automatically makes it the tracer of choice. Agiiin.
of the longitudinal dispersion arises due to heterogeneity time is insufficient (except perhaps in karst) t o run an
in the hydraulic conductivity field. experiment at the scale of interest. Only for small systems
An attractive feature o f this method is that it does not (for example, some fraction of a kilometer) is there a
require that dissolved mass pass through the test volume. possibility of running a field tracer experiment. It is for
T~LIS it ,has the potential to be applied to large-scale these experiments when a selection must be made among
systems- too big to be tested using more conventional the various tracers.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Longitudinal Distance (meters)
Figure 10.18 Vari;ition in hydr-.ulic conductivity (as -In K ) along the cross section A-A' ;it
<:anadinn Forces Hnse Borden (modified from Sutlicky. \Water Resources Res., v. 12. p. 2060-
2082, 1986. Copyright by Americ:iri Geophysiail llnion).
-
232 Cbapter 10 / Solute Transport

Field Tracer Experiments Keeping the quantity of tracer small minimizes the distur-
bance of flow conditions. The resulting concentration
Tracers are transported in two kinds of flow environ- distributions provide the data necessary to determine
ments. In a natural gradient system, tracers move due to advective velocities, dispersivities,and occasionally equi-
the natural flow of ground water. In a system stressed librium and kinetic parameters.
by injection and/or pumping, transport occurs in re- Such an experiment typically requires a dense network
sponse to hydraulic gradients typically much larger than of sampling points. This chapter concludes with a brief
those in natural systems. A smallscale tracer experiment description of some of the important tracer tests of this
can involve either of these flow conditions. type run over the last decade.
Successin these tests depends on adequately character-
izing concentration distributions in space and time. Typi- Sing& Well Pulse Test
cally, a large three-dimensional network of monitoring The single well pulse test (Figure 10.19~)involves first
points is necessary to define the plume accurately. Point injecting a tracer followed by water into an aquifer at
sampling is essential to avoid concentration averaging some constant rate. After an appropriate period of injec-
within the well bore. For example, estimated values of tion, the aquifer is pumped at the same rate. The concen-
dispersivitytend to be larger when the number of individ- tration of tracer in water being withdrawn is monitored
ual sampling points is small or samples are collected over as a function of time or total volume of water pumped.
relatively large vertical intervals. This excess dispersion These concentration/time data provide a basis for esti-
represents error due to inadequacies in the sampling mating the longitudinal dispersivity and chemical param-
network. eters like Kdwithin a few meters of the well (for example,
In theory, mass-transport parameters can be estimated Pickens et al., 1981). Fried (1975) describes a more re-
from almost any test where a tracer is added in a con- fined version of this test where concentration/time data
trolled way. However, a few more or less standard tests are collected at different positions in the well with a
are preferred because there are simple procedures for down-hole probe.
interpreting the results. Unfortunately, many of these While historically of some interest, this test has limited
tests are subject to error because when run in their basic applicability in estimating values of dispersivity. It is gen-
configuration there is often an insufficient number of erally not possible to scale up dispersivity measurements
monitoring points. made at such a small scale to the larger scales of interest
in most problems. In addition, the lack of detailed obser-
Natural Gradient Test vations on how the tracer is spreading in the vicinity of
The natural gradient test involves monitoring a small the well makes any dispersivity estimate quite crude.
volume of tracer as it moves down the flow system. This test holds more promise for evaluating geochemical

(MI Single-Weti Pulse Test Cb) T w ~ W e lTracer

l Ted

Diveryeit1Radial Flow Canvergent Radial Flow

UbssrJirtion Tracer

Ambient t~aw
. .-
(c)Two-Well Tracer Test in Fractured Rock ( d ) Single-Well with Multiple Observation Wells
Rgure 10.17 Examples of field tracer tests.
 10.7 Tracers and Tracer Tests 233

processes. Because reactions operate on a local scale, well. The radially convergent test involves adding the
there is no real scaling problem. tracer at one of the observation wells and monitoring as
it moves toward the pumped well. Parameters can be
Two-Well Tracer Test estimated at scales of practical interest with accuracy,
In the two-well test, water is pumped from one well and given a reasonable number of observation wells. Between
injected into another at the same rate to create a steady- these two tests, the divergent flow test is preferable
state flow regime (Figure 10.190). The tracer is added (Gelhar and others, 1985). The converging flow field
continuously at a constant concentration at the injection (Figure 10.194) counteracts spreading due to dispersion
well and monitored in the withdrawal well. The simplest and is thought to be less useful.
way of running this test is by recirculating the pumped
water back to the injection well, as is shown on the
Estimates from Contaminant Plumes and
figure. However, the tracer concentration will begin to
increase at the injection well once breakthrough occurs
Environmental Tracers
at the pumped well. The test can also be run without Estimates of dispersivities at scales greater than several
recirculation by providing the water for the injection well hundreds of meters are based on mixing patterns caused
from an alternative source. The resulting concentration by the presence of a dissolved contaminant plume or
versus time data are interpreted in terms of processes and temporal variations in the chemistry of natural recharge.
parameters. These tests can be conducted over several These latter changes can be caused by the atmospheric
hundred meters in highly permeable systems. However, testing of nuclear weapons, which impacts tritium and
with only a single monitoring point, the test provides at chlorine-36 levels in rainfall, or by longer term changes
best only a crude estimate of dispersivity. Estimates can in climate, which impact oxygen-18 and deuterium (for
be improved by adding more observation wells between example, Pleistocene to present). The detailed examina-
the pumping-injection doublet. tion and interpretation of data from contaminant plumes
Modified versions of a two-well test have been con- and environmental isotopes will be reserved for several
ducted in fractured rocks (Raven and Novakowski, 1984). later chapters.
Packing off a small section of the borehole (Figure Estimates of advection and dispersion based on data
10.19~)isolates a single fracture plane between pumping from plumes or environmental measurements are less
and injection wells. This test is technically more de- reliable than field tracer tests. There is uncertainty in
manding because it requires working with packers. Nev- defining the loading function for the constituent of inter-
ertheless, it involves the same interpretive techniques est and, often, there is an inadequate number of sampling
and is subject to the same limitations that we just dis- points (Gelhar and others, 1992).
cussed.
Interpretive techniques for two-well tracer tests are
quite sophisticated. Grove and Reetem (1971) present a
Massively Instrumented Field Tracer Tests
graphical, curve-matching procedure that is appropriate Over the past decade, there have been several large-
for idealized flow conditions that develop in homoge- scale natural gradient tracer tests (Table 10.7) run in field
neous and isotropic units. A computer code for this pur- settings. Three tests, Borden, Cape Cod, and Columbus,
pose is provided by Grove (1971). Guven and others are unique because between five to ten thousand moni-
(1986) have developed an alternative approach to evalu- toring points provided precise three-dimensional pic-
ating two-well tracer tests. Their procedure interprets tures of the evolving tracer plumes over travel times of
variation in breakthrough at the pumped well due to approximately two to three years. These tests were run
variability in the velocity field caused by geologic lay- to learn more about the manifestations of physical ancl
ering. Essentially this analysis assumes that mixing due chemical transport processes at large scale in shallow
to local hydrodynamic dispersion is small relative to that unconfined aquifers, and to validate the geostatistical
caused by the heterogeneity of the system. The evalua- concepts of dispersion. The study of chemical processes
tion involves varying porosity and the hydraulic conduc- involved the use of reactive tracers in addition to conser-
tivity of the different units to match the simulated and vative tracers. The test at Borden used chloride and bro-
observed breakthrough curves. Other approaches can mide as the conservative (that, is nonreactive) tr;icers
account for the effects of dispersion (for example, Huya- and five organic compounds as reactive tracers (Table
korn and others, 1986). 10.7). The test at Cape Cod used bromide as the nonreac-
tive tracer, and lithium (Li-) and molybdate (MOO:-);IS
Single Well Injection or Withdrawal with Multiple reactive tracers. The test at Columbus used only consewa-
Observation We& tive tracers. The following two sections summarize some
These tests create a transient radial flow field by injection results from the Borden test and examine how well theo-
or withdrawal. The radially divergent test (Fig. 10.194) ries work that predict longitudinal dispersivity from statis-
involves monitoring the tracer as it moves away from the tical properties of the hydraulic conductivity field.
-
234 Chapter 10 / Solute Transport

Table 20.7 Information on Massively Instrumented Tracer Tests

Test Scale
Site Aquifer Material (m) References
Canadian Forces Glaciofliivial sand 90 Mackay and others, 1986
Base Borden, Ontario Freyberg, 1986
Sudicky, 1986
Cape Cod, Sand and gravel 250 LeBlanc and others, 1991
Massachusetts Garabedian and LeBlanc, 1991
Hess and others, 1992
Columbus, Mississippi Sandy gravel/gravelly 280 Bog@ and others, 1992
sand A d a m and Gelhar, 1992
Rehfeldt and others, 1992

Borden Tracer Experiment kg of CI- provided a mean concentration of 892 mg/l

The natural gradient tracer test at Canadian Forces Base (Freyberg, 1986). Figure 10.20 illustrates the vertically
Borden illustrated how ground-water velocities and the averaged CI- distribution at four times. Advection moved
extent of dispersion could be interpreted from “snap the tracer plume along the shallow flow system, and
shots” of the conservative tracer distributions. The CI- longitudinal dispersion elongated the plume in the direc-
and Br- concentrations were monitored on 14 occasions
over 1038 days at approximately 5000 observation
points. Here, we examine CI- data. The addition of 10.7
X

60

Slope = 2 4 = 0.066
50 I ) = 0.091 m/d
U, 0.033
(J = - = -
I I? 0.091
= 0.361~1
4c
0 200 400 600 800 1000
-E Time (days)

-
-E 30 a” I
X

c1 -
%

x Bromide
15 OChloride Transverse
20
Linear fit
-E
85 days

1 day (outline)

-10
-10
I I
0
I
10 20
I I
30
I
200
1
400
I
600
I
800
I
1000
y (meters) Time (days)
Figure 10.20 Vertically averaged concentration distribu- Figure 10.21 Fit of the Borden data to a constant dispersiv-
tion of CI- at various times after injection (from Mackay and ity model (from Freyberg, Water Resources Res., v. 22, p.
others, Water Resources Res., v. 22, p. 2017-2029, 1986. 203 1 -2046, 1986. Copyright by American Geophysical
Copyright by American Geophysical Union). Union).
 10.7 Tracers and Tracer Tests 2 3

Tub& 20.8 Estimates of Mass in Solution, Location of the Center of Mass, and Spatial Covariance
for the Chloride Plume
Center of Mass' Spatial Covariance2
Elapsed Mass in
Time Solution x, Yr 2,
Date (days) (k@ (m) (m) (m)
Cbloride
Aug. 24, 1982 1 6.7 0.2 0.1 2.78 2.1 2.4 0.5
Sept. 1, 1982 9 9.2 0.7 0.4 3.02 1.7 2.4 0.7
Sept. 8, 1982 16 9.2 1.6 0.7 3.06 2.3 2.8 0.8
Sept. 21, 1982 29 11.5 2.9 0.9 3.27 2.5 2.6 0.9
Oct. 5, 1982 43 11.3 4.1 1.6 3.34 4.4 2.7 1.2
Oct. 25, 1982 63 9.0 5.7 2.0 3.50 4.4 2.4 1.1
Nov. 16, 1982 85 11.2 7.7 3.2 3.75 5.7 3.3 0.8
May 9, 1983 259 11.5 22.7 11.6 4.52 17.8 4.4 3.7
Sept. 8, 1983 381 9.6 32.3 15.3 5.18 20.6 4.4 3.9
Oct. 26, 1983 429 9.2 35.9 17.2 5.25 24.3 6.0 3.2
Nov. 28, 1983 462 8.2 38.2 17.4 5.33 27.8 5.5 2.1
May 31, 1984 647 9.1 53.1 23.9 5.55 51.5 5.5 3.0
' Given in the field coordinate system.
'Given in rotated coordinates: x' is parallel t o linear horizontal trajectory, y' is perpendicular to linear horizontal trdjecton
From Freyherg, Water Resource Res., v. 22, p. 20.5 1-2046. 1986. Copyright by American Geophysical LJnion.

tion of flow. Horizontal-transverse spreading was, how- through ui and a: versus time data (Figure 10.21) pro-
ever, less marked (Figure 10.20). vided an estimate of D].and Dj. From these values, Frey-
Freyberg (1986) calculated the center of mass of the berg (1986) calcuiates LY/ and LY to be 0.36 and 0.039
CI- plume in three dimensions (Table 10.8)as well as the m, respectively. At later times, the straight-line approxi-
variance in concentration distributions in the longitudinal mation does not fit the data too well (Figure 10.21) indi-
and horizontal-transverse directions. Transverse-vertical cating that dispersivities may not be constant with time
spreading was negligible. or travel distance.
The data in Table 10.8 indicate that the center of mass
traveled 58.2 m in 647 days, which equates to a linear Validationof the Stochastic Model of Dispersion
ground-water velocity of 0.09 m/day. The simplest calcu- Section 10.6 introduced the idea of how asymptotic dis-
lation of the dispersivity is based on the relationships persivities could be estimated from parameters of the
between variance, time, and dispersivity (Eq. 10.121, as- heterogeneous hydraulic conductivity field. The theory
suming that dispersivities are constants. Fitting a line is based on stochastic continuum theories developed by
Dagan (I 982, 1984) and Gelhar and Axness (1 983). This
theoretical approach was tested at each of the three sites
Tubk 10.9 Geostatistical Description of Aquifers at the by comparing the longitudinal dispersivity value obtained
Three Sites and Comparison of Dispersivity Values in the tracer test with that estimated from geostatistical
Estimated from Field Tracer Tests and Stochastic information on the hydraulic conductivity fields. These
Theory statistical data came from a large numer of measurements
made using permeameters and borehole flow meters.
Longitudinal
The summary statistics for these sites (Table 10.9)
Aquifer Properties Dispersivity (m)
show that the aquifers at both Borden and Cape Cod are
Site & A, Sj Theory' Test relatively homogeneous compared to that at Columbia.
These differences are reflected in variances in the natural
Borden 2.8 0.12 0.29 0.45-0.6 0.36
logarithm of hydraulic conductivity. which for Borden
Cape Cod 5.1 0.26 0.26 0.35-0.78 0.96
Columbus 5.3 0.7 2.8 1.5 5- 10 and Cape Cod are 0.29 and 0.26. respectively, versus 2.7
for Columbus (Rehfeldt and others. 1992). The corrt!l;i-
' These calculations involve a variety of stochastic approaches and the tion Iength scales in the horizontal and vertical directions
values are not the same as those which would come from the simple are all quite similar (Table 10.9). The greater variance in
equation w e presented.
A,, = correlation length in the horizontal direction
In K at Columbus results in a larger estimated asymptotic
A,, = cornlation length in the vertical direction dispersivity at this site. This result generally is confirmed
Sf = sample estimate of the variance in In K by the results from the actual tracer tests, although :tf
-
236 Cbapter 10 / Solute Transport

Columbus, the dispersivity estimated from the tracer test field. Further, borehole flow meters provide a practical
is somewhat larger than would be predicted by stochastic approach to collecting the large number of hydraulic
theory (Adams and Gelhar, 1992). These results suggest conductivity measurements required for such an ap-
that it will be feasible to estimate dispersivity values from proach.
statistical information about the hydraulic conductivity

Problems
1. Consider two different media, one fractured and the be evaluated to provide an estimate of the variance in
other unfractured. The hydraulic conductivityis the same concentration as a function of time ( ~ 3 .
for each (1 X lo-' m/s) but the effective porosities are a. Modify Eq. 10.17 to a form in which the longitudinal
different (3 X and 0.30, respectively). Under an dispersivity aLis expressed as a function of the length
imposed hydraulicgradient of 0.01,determine how much of the column (L), time (t),and u:.
time is required for a tracer to be advected 30 m in both b. A breakthrough curve for a column experiment pro-
the fractured and porous media. vides a 0: value of 5.5 min2at a time of 82 min (i.e.,
2. The hydrogeologic cross section on the figure illus- time at which the advective front arrives). Knowing
trates the pattern of groundwater flow along a local flow that the column is 56 cm long, calculate the longitu-
system. Assume that a source of contamination develops dinal dispersivity of the medium (aL).
with advection as the only operative transport process. 7. A contaminant is added as a point source to ground
Describe the pathway for contaminant migration and water flowing with a constant velocity of 4 X m/s.
estimate at what time in the future the plume will reach Assuming longitudinal and two transverse dispersivities
the stream. (y and z directions) of 1.0, 0.1, and 0.01 m, determine
3. The approximate coefficient of diffusion for C1- in the spatial standard deviations (u& in the plume size
water is 2 X cm*/s. Estimate its value in a saturated after 400 m of transport.
porous medium having a porosity of 0.25 and a tortuosity 8. Following is a series of 15 hydraulic conductivity
of 1.25. measurements (K)measured on cores collected at a spac-
4. A useful rule of thumb is that mass has to spread ing of 0.2 m. Estimate the sample mean, variance, and
several tens of correlation lengths before the asymptotic vertical correlation length (using two lags). Assume an
dispersivity is reached. Explain why. exponential autocorrelation function.
5. Ground water flows through the left face of a cube Hydraulic Conductivity x lo-* cm/s
of sandstone (1 m on a side) and out of the right face
with a linear ground-water velocity of lo-' m/s. The 1 . . . .4 8 . . . . 36
porosity of the sandstone is 0.10, and the effective diffu- 2 . . . .13 9 . . . .12
sion coefficient is m2/s. Assume that a tracer has 3 . . . .48 10.. . .13
concentrations of 120, 100, and 80 mg/l (1.2, 1.0, and 4 . . . .64 11.. . .l8
0.8 X lo5 mg/m3) at the inflow face, the middle of the 5 . . . .62 12.. . .36
block, and the outflow face, respectively. Calculate the 6 . . . .58 1 3 . . . .48
mass flux through the central plane due to advection and 7 . . . .38 14.. . .52
due to diffusion. 15. . . .64
6. The essence of a column experiment for measuring 9. A series of hydraulic conductivity measurements for
dispersivity is to obtain a breakthrough curve, which can an unconfined aquifer provide a mean hydraulic conduc-

,,Contaminant source

I
I I
I I Unit 2 I
I I I

lOOm
K1 = 8.5x mlday
------_Equipotential line K2 = 8.5 x 1
0
'' mlday
meters above sea level n1.2 = 0.30

Problem 2
tivity ( 0 of 0.004 m/s, where Y = In K , a variance in
the log-transformed hydraulic conductivity (at)of 1.O,
and a correlation length in the direction of the mean
flow ( A , ) of 10.0 m. Estimate the asymptotic macroscale
dispersivity for the aquifer.
10. Following is an enlargement of the C1- plume mea-
sured after 462 days during the tracer experiment at
Canadian Forces Base Borden and in Figure 10.20.
a, Determine a:from the field data using the graphical
procedure and a conventional statistical calculation.
b. Assuming that the longitudinal dispersivity and ve-
locity are constant and that the plume originates as
a point source at the coordinates ( 0 . 0 , O . O ) in Figure
10.20, estimate a L .
c. Explain in this case why estimates of will not yield
estimates of the transverse dispersivity. Problem 10
Hint: Examine Figure 10.20
d Examine the character of transverse dispersion in
Figure 10.20. What qualitative assessment can you
make about the importance of transverse dispersion?
 Chapter 11

Principles of Aqueous Geochemistry

11.1 Introduction to Aqueous Systems

11.2 Structure of Water and the Occurrence
of Mass in Water
11.3 Equilibrium Versus Kinetic Descriptions of Reactions
11.4 Equilibrium Models of Reaction
11.5 Deviations from Equilibrium
11.6 Kinetic Reactions
11.7 Organic Compounds
11.8 Ground-Water Composition
11.9 Describing Chemical Data
Chapters 11 and 12 discuss the basic principles of aque- Ions, molecules, or solid particles in water undergo
ous geochemistry, and the theory and processes of mass not only transport (Figure 11. l ) but also reactions, which
transfer. This infomation is a prerequisite to redistribute mass among various ion species, or between
discussions of the chemistry of ground water and the liquid and solid phases. Our list of reactions is not
contaminants that later. M~~~ specifically, Chapter exhaustive but it includes the most important ones affect-
11 examines hndamental ideas about - - (,f
ing the chemistry o f ground water. These chemical pro-
water, the way mass occurs in water, and equilibrium cesses operate in a system consisting of one or more
and kinetic concepts in aqueous systems. It also discusses solid ph;ises, an aqueous solution phase, and a gas phase
the chemistry of organic compounds, the common con- (Figure 11.1) and redistribute mass within a phase or
stituents in natural ground water, and the unique graphi- between phases.
cal and statistical techniques developed to describe Most inorganic substances are electrolytes, dissolving
their occurrence. in water to form ions. Positively charged species, such as
Ca” or K’, are called cations. Negatively charged species,
such as HCO; or CI-,are called anions. Complex ions
-
1 f. 1 Introduction to Aqueous Systems form by combining simpler cations and anions. Organic
substances can also dissolve to form organic cations or
Mass can exist in the subsurface as (1) separate gas or anions. However, organic liquids are usually nonelectro-
solid phases, for example C 0 2 gas in the soil zone or lytes and dissolve in water as nonionic molecules. In the
minerals that form the porous medium; (2) a separate aqueous phase, the distribution of species is described
component of the liquid phase, for example, crude oil in ternis of a mohr Concentration or an activity, which
and liquid organic contaminants; or (3) mass dissolved is an effective concentration accounting for chemical
in the water itself (that is, solutes), for example, Na+ or nonidealities in reactions. The notation adopted through-
C1-. While all of these occurrences are important in a out the book is to represent molar concentrations with
geologic context, this chapter will focus on (3). The round brackets, for example, (Gal'), and activities with
nonaqueous phase liquids, gases, and solids will also be square brackets, [Ca?’]. Gases are represented in terms
discussed because they are often the sources of solutes of partial pressures of constituent gases (for example,
in ground water or potential contaminants. k o 2 ) or in nonideal systems by fugacities f w 2 . Solid
 11.1 Introduclfon lo Aqueous Systems 233

Gas Phase an ion is equal to the number of nioles of an ion multiplied

Partial Pressure P,, f; ,..., P, by the absolute value of the charge. For example. with
Fugacity L3
J!...*J,
a singly charged species such as Na-. 1 ,M Na- =
1 eq/L, and with a doubly charged species such as Ca'-,
1 M Ca'+ = 2 eq/L. Equivalent concentrations can also
be represented as equivalent charges per unit mass of
solution with units such as eq/kg or equivalents per
million (epm).
Mass per unit mass concentrations define a scalc i n
terms of the mass of a species or element per total miss
of the system. Many older analyses have been reported
using this scale with concentrations in parts per million
Solid Phase 3
A,,.x;,. ,:1 (ppni) or parts per billion (ppb). More recently, these
units of concentration have given way to corresponding
concentrations in mg/kg or pg/kg.
Mass per unit volume concentration is the most com-
mon scale for concentration. It defines the mass o f ;I
solute dissolved in a unit volume of solution. Concentr;i-
tions ;ire reported in units such ;IS mg/L or pg/l.. Again.
Figure 11.1 Generalized model of a chemicil system for there is a close correspondence between these last two
ground water at some temperature and pressure (modified scales of concentration. For dilute solutions, 1 ppni =
from Stiimm and Morgan, 1981, Aquatic Chemistry). Copy 1 nig/kg = 1 nig/L. As the total salinitj. of a solution
right 0 1981 by John Wiley & Sons, Inc. Reprinted by per- begins t o exceed approximately 10' mg/L, the conwr-
mission of John Wiley & Sons, Inc.
sion equation

mg/L
mg/kg = (11.1)
phase compositions (Figure 1 1 . 1 ) are represented as solution density
mole fractions and activities.
is necessary to move between these concentration scales.
Concentration conversions involve a few simple equ;i-
Concentration Scales tions. The most common scale change takes the reported
Molar concentration (M) defines the number of moles values of chemical analyses in mg/L (also mg/kg or ppni)
of a species per liter of solution (niol/L). A mole is the and converts them to molar concentrations:
formula weight of a substance expressed in grams. For
mg/L x
example, a o n e liter solution containing 1.42 g of niolarity = (1 1.2)
N a S Q has a (NalS04) molarity of 1.42/(2 X 22.99 + formula weight
32.06 + 4 X 16.00) or 0.010 M . Because Na2S0, dissoci- Conversion from mg/L to nieq/L is sometimes neces-
ates completely in water according to the following re-
sary to present chemical data grxphically or to check
action chemical analyses. This conversion equation is
NilzSo~= 2Nd' + SO;-
mg/L
the molar concentrations of Na' and SO:- are 0.02 and meq/L = ( 1 1.3)
formula wt/charge
0.01 M, respectively. This reaction states that one mole
of Na2S0, dissolves to produce two moles of Na' and To understand the conversion of scales. consider the
one mole of SO:-. Because substances react in molar following example.
proportions, thermodynamic calculations use concentra-
tions expressed in moles.
Example I I.I
Molal concentration (m)defines the number of nioles
of a species per kilogram of solvent (mol/kg). This scale The concentration of SO:- in water is 85.0 nig/L. Express
for concentrations in dilute solutions is almost the same this concentration as molarity. and meq/L.
as molar concentrations because a one-liter solution has
a mass of approxiniately 1 kg. For more concentrated
solutions, the two scales become increasingly different.
Equivalent charge concentration is the number of 85
equivalent charges of an ion per liter of solution with meq/L = = 1.77
(32.06 + 6 X 16.0)/2
units such as eq/L or meq/L. The equivalent charge for
240 Cbapter 11 / Principles ofAqueous Geocbemistry

Gas and SoBd Phases static interaction between a hydrogen and an oxygen
atom. These interactions produce an irregular tetrahedral
In thermodynamic calculations, concentrations of gases network (Figure 11.2~). This molecular structure ex-
are given as partial pressures. In a mixture of gases at a plains some of the unique characteristics of water such
fixed temperature and in equilibrium with the aqueous as its freezing and boiling points.
phase, the partial pressure of a gas is the pressure the The polar nature of water also contributes to the rela-
gas would exert if it occupied the whole volume of the tive solubility of electrolytes and the insolubility of non-
mixture by itself. For example, COzgas in the atmosphere electrolytes in aqueous solution. Charged species (for
has a Pco2 of about 10-j’ atmospheres (atm). It is also example, ions or molecules) bind with the polar mole-
common to report gas concentrations using a scale such cules in the water structure, which is an important factor
as ppm or mg/L. contributing to solubility. In water, a charged species
Concentrations for solids are represented as a mole forces a local rearrangement in the structure (Figure
fraction, which is the ratio of the number of moles of 11.24. Water molecules immediately adjacent bind to
the phase of interest to the total number of moles in the the charged species. Those nearby reorient themselves
system. However, thermodynamic calculations often do because of the charge. The zone of influence immediately
not require the solid phase concentrations because by adjacent to the ion is known as the hydration sheath. A
convention the solid phase concentration or activity for so-called random region separates the hydration sheath
a pure solid is one. from the bulk solution. Within the random region, the
water molecules are less oriented than either in the bulk
-
11.2 Structure of Water and the
solution or with the hydration sheath.
In the case of an uncharged species, such as many
Occurrence of Mass in Water nonpolar organic molecules, there is much less interac-
The structure of water influences the solubility of solids tion with the water molecule, and the nonpolar molecule
and organic liquids. The water molecule is comprised of in effect has to find a space for itself in the water struc-
two hydrogen and one oxygen atom. Shared electrons ture. These effects contribute to the relative insolubility
form covalent bonds between atoms. Hydrogen atoms of nonpolar molecules in water. Moreover, many of the
are located asymmetrically with respect to the oxygen thermodynamic relationships governing the dissolution
atom (Figure 1 1 . 2 ~ )Because
. of this structure and the process have molecule size as a variable. Solubility turns
covalent bonding, the electrical center of the negative out to be inversely proportional to molecule size (see
charges (electrons) has a different location than that of Section 12.2).
the positive charges. This separation makes the molecule
polar in terms of electrical charge. When a molecule is
perfectly symmetrical (for example, carbon tetrachlo- 11.3 Equilibrium Versus Kinetic
ride, benzene, and other hydrocarbons), there is no Descriptions of Reactions
charge separation and the molecule is nonpolar.
The water molecules are joined by a series of hydrogen The state of chemical equilibrium for a closed system
bonds (Figure 11.2b). These bonds form from electro- describes a position of maximum thermodynamic stabil-
ity. At equilibrium, there is no chemical energy available
to alter the relative distribution of mass between the
reactants and products in a reaction. Away from equilib-
rium, energy is available to spontaneously drive a system
toward equilibrium by allowing the reaction to progress.
Theoretical approaches are available for modeling the
chemical composition of a solution at equilibrium. How-
ever, these calculations provide no information about
the time required to reach equilibrium or the reaction
pathways involved. A different approach, based on ki-
netic models of reaction, is required.
(a) (6) (C) (d)
From a chemical perspective, ground water is best
Figure 11.2 Essential features of the structure of water
thought of as a partial equilibrium system. Some reactions
and adjustments due to the presence of an ion (adapted
from Pytkowicz, 1983): (a) the position of the hydrogen and
are at equilibrium and are describable in terms of equilib-
oxygen atoms in the polar water molecule; (b) hydrogen rium concepts. However, other reactions not at equilib-
bonding between water molecules, giving rise to (c) the tet- rium are best represented by kinetic concepts. How a
rahedral structure of water; (4water molecules rearrange reaction is finally modeled depends on its rate. An equilib-
themselves around a cation due to the polar nature of the rium reaction is “fast” in relation to the mass transport
water. processes that are working to change concentrations. A
 11.4 Equilibrium Models of Reaction 241

kinetic reaction is “slow” in relation to the transport fractions of seconds to at most a few minutes. These
processes. Thus, in applying an equilibrium model to a reactions are examples of homogeneous reactions, reac-
reaction, we assume mass is transferred instantaneously tions that occur in a single phase. Of the list in Figure
between reactants and products to move the system to 10.1, the acid-base, and complexation reactions are ho-
equilibrium. If the system transfers mass in a reaction at mogeneous. The dissolution-precipitation reactions are
a rate that is slower than physical transport, it requires examples of heterogeneous reactions or reactions involv-
a kinetic description. ing more than one phase. These reactions have half-times
The degree of competition between the rates of reac- varying from days to lo6 years (Figure 11.3). Heteroge-
tion and the mass transport processes influences how neous solution or exsolution reactions involving a gas
one might model a system. For example, it may be appro- phase have relatively larger half-times than liquid- solid
priate to model ion exchange in a sluggish regional flow reactions. Reactions on surfaces, which Langmuir and
system as an equilibrium process. However, the same Mahoney (1984) term adsorption-desorption reactions,
exchange process operating in an aquifer with high flow have half-times ranging from seconds to days. The larger
velocities could require a kinetic description. half-timesdescribe surface reactions in the porous matrix
Not all of the reactions are of the equilibrium type. of rock fragments. Radioactive decay and isotopic frx-
Irreversible reactions proceed in the forward direction tionation reactions are not included on the figure because
until the reactants are used up. Examples include radioac- the range in half-times is extremely variable.
tive decay, some redox reactions, and some organic reac- Redox reactions are often relatively slow because they
tions. Such reactions require a kinetic description. are mediated by microorganisms (Morel and Hering,
The kinetic approach to modeling reactions is much 1993). Typical half-times for these reactions range from
more general than the equilibrium one. Nevertheless, hours to a few years based on data for marine systems
most geochemical work involves models based on equi- (Berner, 1980). Transformation reactions involving or-
librium concepts. The greatest limitation in the applica- ganic contaminants in ground water depend on the physi-
tion of kinetic models is the lack of basic data concerning cal and chemical properties of the particular compounds
the order and rate of the reactions. The strength of the (for example, hydrolysis), and/or the abundance of mi-
equilibrium approaches is the ability to apply concepts crobes (for example, biodegradation).
of chemical thermodynamics to estimate equilibrium and This summary of reaction rates illustrates why the equi-
other constants as a function of temperature (Nordstrom librium approaches have proven so successful in model-
and Munoz, 1986). ing the chemistry of natural ground-water systems. Al-
most all of the important reactions have rates that are
faster than the typical rates of ground-water flow.
Reaction Rates
There are sufficient data on reaction rates to provide
guidance in deciding whether to use an equilibrium or -
11.4 Equilibrium Models of Reaction
a kinetic approach. Figure 11.3 compares the half-time Equilibrium concepts are useful in modeling many homo-
or half-life of many of the reactions on our list in Figure geneous and heterogeneous reactions. Consider the fol-
10.1. As a generalization, the fastest reactions are the lowing reaction, where ions C and D react to produce
solute-solute or solute-water type with half-times of ions Y and Z
cC + dD = y Y + ZZ (1 1.4)

Solute-solute
Solute-water
- Gas-
Water
- with c, d, y , and z representing the number of moles of
these constituents. For a dilute solution, the law of mass
action describes the equilibrium distribution of mass be-
Hydrolysis of rnultivalent
ions (Dolvrnerization)
, ,
tween reactants and products as
Absorption
desorption c _ _
Mineral-water
equilibria
. II (1 1.5)

c _
Mineral
recrystallization - where Kis the equilibrium constant and ( Y ) , ( Z ) , ( C ) and
( D ) are the molal (or molar) concentrations for reactants
,
I
and products. Depending on the reaction. we refer to
Secs Mins Hrs Days
Reaction rates, - Mos Yrs 106Yrs
this equilibrium constant as (1) an acidity or dissociation
Figure 11.3 A comparison of rcaction half-times for constant in acid-base reactions, (2) a complexation con-
many of the common reactions in aqueous systems (from stant in complexation reactions, (3) a solubiliv constant
Iangmuir and Mahoney, 1984). Reprinted by permission First in solid dissolution reactions or a Henry‘s law constant
Cdnadian/AmeriCaII Conference on Hvdrogeology. Copyright in gas dissolution, or (4) an adsorption constant for sur-
0 1984. All rights reserved. face reactions.
242 Cbapter 11 / Principles of Aqueous Geochemistry

Standard geochemical texts, for example, Morel and where A (Table 11.1) is a constant that is a function of
Hering (1993) tabulate values for these constants. Equilib- temperature, z, is the ion charge, and I is the ionic
rium constants expressed as a function of temperature strength of the solution. Mathematically, the ionic
are also included in the data bases of computer codes strength in mol/L is
for modeling aqueous systems such as SOLMNEQ
(Kharaka and Barnes, 1973), EQ3/6 W'olery, 1979), and I = 0.5 Mi%: (11.8)
PHREEQE (Parkhurst and others, 1980).
with (Mi)as the molar concentration of species i having
a charge z. Ionic strength is a measure of the total concen-
Activity Models tration of ions that emphasizes the increased contribution
of species with charges greater than one to solution non-
Activity models for dissolved species represent the non-
ideality.
ideal behavior of components in nondilute solutions.
Use of the Debye-Huckel equation is limited to solu-
Nonidealities result from electrostatic interactions among
tions with ionic strengths less than 0.005 M, which is
ions in solution. Mass action equations account for non-
fresh, potable ground water. It is inappropriate for use
idealities by replacing molar concentrations by activities
with more saline water.
or thermodynamically effective concentrations. For the
The extended Debye- Huckel equation provides esti-
case of species Y, the following equation relates activity
mates of y, to a maximum ionic strength of about 0.1 M
to molar concentration:
or a total dissolved solids content of approximately 5000
(11.6) mg/L (Langmuir and Mahoney, 1984). This equation is

where y,, is the activity coefficient. Typically, y is close t o (1 1.9)

one in dilute solutions and decreases as salinity increases.
Activities of species with more than one charge usually
are smaller than those with a single charge. At relatively where A and B are temperaturedependent constants and
high salinities, the activity coefficients in many cases ai is the radius of the hydrated ion in centimeters
begin to increase and even exceed one. (Table 1 1.1).
We can calculate y for ions from one of several differ- Geochemical studies of basinal brines require other
ent activity models. The simplest is the Debye-HUckel activity models to deal with the higher salinity water.
equation The simplest model takes an equation like Eq. 11.9 and
adds correction or curve-fitting parameters. The Davies
log y, = -Az;'(I)" (1 1.7) equation is an example. Establishing activities in even

Table 11.1 Parameters Used in the Extended Debye-Hilckel Equation

at 1 Atmosphere Pressure
~~ ~ ~

Temperature (C") A B (X 10")

0 0.4883 0.3241 2.5

5 0.4921 0.3249 3.0
I0 0.4960 0.3258 3.5
15 0.5000 0.3262 4.0-4.5
20 0.5042 0.3273
25 0.5085 0.3281
30 0.5 130 0.3290 4.5
35 0.5175 0.3297 5.0
40 0.5221 0.3305
50 0.5319 0.3321 6
60 0.5425 0.3338

8
9

11

Reprintcd with prrniission from Mnnov, G. <;. and From Klotz, 1950 with permission.
othen,J.Am. Chem. Soc., v. 65, 1943, p. 1765-
1767. Copyright 0 1943 Americ;in Chemical Society
 11.5 Deviations f r o m Equilibrium 23

more saline water requires the use of ion interaction strengths, activity estimates are difficult. Possible ap-
models of a type proposed by Pitzer and Kim (1974). proaches include making activities a proportion of the
This sophisticated activity model is more difficult to use ionic strength or equivalent to the activity of CO?.
than the Debye-Huckel or Davies-type equations, but Typically for ground waters, activity coefficients are
provides accurate estimates of activity u p to 20 M (Lang- calculated using the extended Debye- Huckel equation
muir and Mahoney, 1984). (Eq. 1 1.9) or the Davies equation. The following example
On an activity versus ionic strength plot for CI- and illustrates the use of the extended Debye-Huckel
Ca", note the typical curve shapes for monovalent and equation.
divalent ions (Figure 11.4). We have shown activities
calculated using the Pitzer model, but they are not strictly Example 11.2
comparable to the other three. At low ionic strengths,
all four methods give approximately the same activity Water contains ions in the following molar concentra-
coefficients. However, as the ionic strength increases tions: Ca" 3.25 X lo-', Na* 0.96 X lo-'?HCO; 5.75 X
(Figure 11.4), estimates of activity from the Debye- lo-', and SO:- 0.89 X lo-'. Calculate activity coefficients
Hockel, extended Debye-Hiickel, and Davies equations for Na' and Ca" at 25OC using Eq. 11.9.
differ noticeably. First, calculate the ionic strength from Eq. 11.8 or
Because the extended Debye-Huckel equation or the I = 0.5 X (3.25 X 2' 0.96 X 1'+ + 5.75 X 1' + 0.89 X 2')
Davies equation are applicable to charged ionic species, x lo-' = 0.012M
they cannot determine y's for neutral species that exist
as ion pairs or other complexes. For solutions with ionic This value of I is at the upper end of the range of applica-
strengths less than about 0 . 1 M, activity coefficients are bility of the extended Debye-HUckel equation. Now,
assigned a value of one. In solutions with higher ionic determine the y values
-0.508 X 1'(.012)"'
'Ogy"- =1 + .328 X lo-" X 4.0 X 10"(.012)"'
= -0.049
0.8)-
y,J. = lo-"1') = 0.89
-
0.61 II II \
\ -0.508 X 2'(.012)"'
log yc =
dJ-
1 + .328 X lo-' X 6.0 X 10w(.012)"5
= -0.183

-
--- Debye-Huckel
Extended Debye-Huckel
-yCJ!?= 10-'8' = 0.66

Pitzer
Davies
I Nonidealites relate to both gas and solids.However, for
-.- most problems, gas phase nonidealities can be neglected.
0.001 0.01 0.1 1.o 10.0
Ionic Strength Nonideal behavior in minerals is due to elemental substi-
1.01 I I I I tution in a crystal lattice. Some applications may require
treatment of nonidealities. However, all of the problems
in this book assume that pure solids have an activity of
one. Readers interested in a discussion of solid activities
should refer to Stumm and Morgan (1981).

-
11.5 Deviations f r o m Equilibrium
Viewing ground water as a partial equilibrium system
implies that some reactions may not be at equilibrium
(for example, dissolution or precipitation reactions). The
--- Davies departure of a reaction from equilibrium is determined
0.0 I as the ratio of the ion activity product to the equilibrium
0.001 0.01 0.1 1.0 10.0
ionic Strength constant (IAP/K). The ion activity product is calculated
Figure 11.4 Comparison of activities of CI- and Ca" as a by substituting sample activity values in the mass law
function of ionic strength, calculated tising the Debye- expression for a reaction. For example, given a ground
Huckel, extended Debye-Huckel, Davies, and Pitzer equa- water with known activities of [Cl, [ D l . [ Y l , and lZ1,
tions (unpublished data from A. S. Crowe). the ion activity product (IAP) for the reaction in Eq.
244 Cbapter 11 / PrlndpZes of Aqueous Ceocbemktty

11.4 is Calculations are easily carried out using one of several

different computer codes such as EQ3 (Wolery, 1979) or
(1 1.10) WATEQF (Plumrner and others, 1976). The codes correct
equilibrium constants for temperature and account for
If the ZAP > K , the reaction is progressing from right to the reduced activities of free ions due to the formation
left, reducing [ yl and [a and increasing [C] and [ D ] .If of complexes.
IAP < K , the reaction is proceeding from left to right. The
UP at equilibrium is equal to the equilibrium constant.
This theory provides the saturation state of a ground
water with respect to one or more mineral phases. When
- Kinetic Reactions
11.6
U P / K < 1 , the ground water is undersaturated with Chemical kinetics provide a useful framework for study-
respect to the given mineral. When U P / K = 1, the ing reactions in relation to time and reaction pathways.
ground water is in chemical equilibrium with the mineral, A kinetic description is applicable to any reaction, but
and when U P / K > 1, the ground water is supersaturated. is necessary for irreversible reactions or reversible reac-
Undersaturation with respect to a mineral results in the tions that are slow in relation to mass transport. Consider
net dissolution provided the mineral is present. Supersat- this reaction
uration results in the net precipitation of the mineral
should suitable nuclei be present. aA + bB + - - (other reactants) = rR + sS
*
k2

The saturation state is often expressed in terms of kl (1 1 . 1 1 )

a saturation index (SI), defined as log(W/K). When a +- * * (other products)
mineral is in equilibrium with respect to a solution, the
SI is zero. Undersaturation is indicated by a negative SI where ki and k2 are the rate constants for the forward
and supersaturation by a positive SI. and reverse reactions, respectively. Each constituent in
Example 11.3 illustrates how to calculate the state of Eq. 11.11 has a reaction rate (for example, r A ) that de-
saturation with respect to a mineral. scribes the rate of change of concentration as a function
of time (&A)/&}. Because of the stoichiometry, these
Example 11.3 rate expressions are related

Given a ground water with the molar composition shown

below, calculate the saturation state with respect to
calcite and dolomite. The activity coefficients for Ca2+, where a,b, r, and s are the stoichiometric coefficients.
MgZt,and C0:- are 0.57, 0.59, and 0.56, respectively. Reaction rate laws describe the time rate of change of
The equilibrium constants defining the solubility of concentration as a function of rate constants and concen-
calcite and dolomite are 4.9 X and 2.7 X re- tration. The rate law for a component (say& in Eq. 1 1.1 1
spectively.
has this form:
(Gal') = 3.74 X lo-‘ (Mg2+)= 4.11 X
<H+) = 1 0 - 7 . 9 rA = * - * (other reactants)
-kl(A)“’(B)”2
(Nat) = 2.02 X lo-‘
(11.13)
(K+)= 6.14 x 10-5 + k2(R)”l(S)’’12
- (other products)
7

(HCO,) = 1.83 X (CO:-) = 5.50 X

(so:-) = 1.24 x 10-3 (Cl-) = 1.19 X lo-’
where n l , n2, . . . and m l , m2,. . , are empirical or
stoichiometric coefficients. This equation expresses the
IApcaI= [Cazt1 [CO:-] rate of change in (A) as the difference between the rate
= 0.57 X 3.74 X X 0.56 X 5.50 X
at which the component is being used in the forward
reaction and generated in the reverse rf >:tion.
= 6.56 X lo-‘ Although reactions are written as a single, overall ex-
pression, they usually occur as a series of elementary
reactions. In other words, a reaction like Eq. 11.1 1 repre-
sents what is really a set of reactions that involves inter-
The sample is slightly oversaturated with respect to mediate species not included in the overall reaction. Ele-
calcite. mentary reactions are defined as reacilons that take place
IAP,,,, = [Ca2’l [Mgz+][CO:-12 = 4.89 X lo-’’ in a single step, for example (Langmuir and Mahoney,
1984),
Ht + OH- = H 2 0
The sample is stronglyundersaturated with respect to do- COz(aq) + OH- = HCO;
lomite.
H4Si04= SiOz(quartz) + 2Hz0
 11.7 Organic Compounds 23

For an elementary reaction, the coefficients in Eq. The reverse reaction has the following rate law:
11.13 are the stoichiometric coefficients. Thus, rate laws
d(FeSOf )
for elementary reactions are easy to establish. Unfortu- = k2(FeSO; ) (1 1.22)
nately, most reactions are not of this type and the rate dt
laws contain empirical coefficients. Their values are es- The change in (Fe") with time for the overall reaction is
tablished through laboratory experiments.
Many of the rate laws involve simple forms of Eq. -- + k(FeS0;)
dCFe'f) - -k,(Fe'+)(SO~-)
11.13. For example, (1) the irreversible decay of ''C dt
(1 1.23)
I4C+ "N +e (11.14)
Rate laws are even more complex for reactions in parallel
has this rate law or series and which cannot be written as a single over-
d("C) all reaction.
-- - -k,("C) (11.15) Solutions to rate equations like Eq. 1 1.23 describe how
dt
the concentrations of various components change from
and (2) the oxidation of ferrous iron between pH 2.2 some initial condition. This exercise is useful mainly for
and 3.5 characterizing reactions in laboratory experiments, but
not in modeling chemical changes in ground-water sys-
Fe" + to2+ &HzO+ FeOH?' (1 1.16) tems. In practice, kinetic equations like Eq. 11.23 must be
has this rate law in Fez' (Langmuir and Mahoney, 1984) integrated into a complete equation for mass transport,
which includes all of the other processes.
-- - -k!,(Fe'+)(Po,) (11.17)
dt

Both Eq. 11.14 and Eq. 11.16 are examples of irreversible

-
11.7 Organic Compounds
reactions, which progress until the reactants are con- Organic compounds are classified in terms of functional
sumed by the reaction. In this case, the rate-law expres- classes. The classes are distinguished by particular func-
sions for 'C and Fez* in Eqs. 11.15 and 1 1. I 7 describe tional groups and the unique structural relationships be-
these components as a consequence of the overall re- tween the functional group and other atoms in the com-
action. pound. A functional group (for example, hydroxyl group,
Kinetic reactions are described according to the order OH; carbonyl group, C=O) is a simple combination of
of the reaction. For the reaction, two or more of the following atoms: C, H, 0, S, N, and
P, that gives the chemical compound unique chemical
2A+B-+C (11.18) and physical properties.
the rate law in terms of (A) is The large number of different organic compounds, the
complexity of their structure, and the specific require-
ments of particular studies have prompted the use of a
variety of classification schemes. Our version (Figure
11.5) is a condensed and slightly reordered version of
The coefficients n 1 and n 2 define the order of the reac- the Garrison and others (1977) classification. It consists
tion in A and B. The order of the overall reaction is the of 16 major classes with all subdivisions except the first
sum of all the n's. Thus, Eq. 11.19 is of order n 1 + rz2. defined on the basis of functional classes.
Applying this scheme to the previous examples, Eq. 1 1.14 The actual steps involved in classifying a compound
is a first-order reaction, and Eq. 11.16 is second order. are complicated because most can be included in more
As Eq. 1 1.13 illustrates, the rate law for an equilibrium than one class. Garrison and others (1977) proposed ;I
reaction depends not only on how fast a constituent is simple set of rules to guide classification. A hierarchical
being consumed in the forward reaction but how fast it search of classes begins with Class 1 and proceeds
is being created in the reverse reaction. For this reaction through Class 16. For example, any hydrocarbon, cleliiied
(Langmuir and Mahoney, 1984) as a compound containing only carbon and hydrogen,
containing a halogen atom (chlorine. fluorine, bromine,
or iodine) is placed in Class 2 except the few that fall in
Class 1. Any compound containing phosphorous is
rate laws for Fe5+and SO:- in the forward direction are placed in Class 4 and so on. Readers can refer to Garrison
of second order and written and others (1977) for a more complete discussion of
rules for classification and a summary of subclasses.
The brief description of the various major classes in-
cludes a sketch of the various functional classes where
-
246 Cbapter I 1 / Principles of Aqueous Ceocbemistvy

1. Miscellaneous Nonvolatile Compounds 10. Aldehydes

2. Halogenated Hydrocarbons Basic structure Formaldehyde

Aliphatic Aromatic 0 0
I1 II

&
R-C-H H-C-H
CI
\ 7' 11. Alcohols
H /c=c\ CI Basic structure Methanol
R-OH CH, -OH
Trichloroethylene Chlorobenzene
12. Esters
3. Amino Acids
Basic structure Vinyl acetate
Basic structure Aspartic acid 0 H,C=CH - 0 - C - CH,
NH2 0 NH, 0 II 1I
I II I I1 R -C - O R ' 0
C OH- C - CH, - CH - C- OH 13. Ethers
4. Phosphorous Compounds
Basic structure 1,4-Dioxane
Basic structure Malathion c-0-c
/
i 0 0
\ II
S
II - S - CH -C - OC,H5
0
f o>
P (CH,O),P 14. Polynuclear Aromatic Hydrocarbons k 0 )
/p
0 I Phenanthrene
CH2- C - OCZH5
II
0
5. Organometallic Compounds

Tetraethyllead 15. Aromatic Hydrocarbons

C2H5
Basic structure Toluene
I
CH
,, -Pb -C2H5
I
'ZH5

6. Carboxylic Acid

Basic structure Acetic acid

16. Alkane, Alkene, and Alkyne Hydrocarbons
0 0

R-C-OH
II CH,-
II
C - OH
Ethane Ethene Ethyne
H H
7. Phenols I I
H- c-i- H
Basic structure Cresol I t
H H H
7 ="\ H
H-CEC-H

Figure 11.5 (Continued.)

&J &OH appropriate, a brief discussion of typical compounds in

8. Amines each class and their origin, and a sketch illustrating how
Basic structure Dimethylamine the functional class is represented in the structure of
Aliphatic Aromatic various organic compounds (Figure 1 1.5).
N- NH2
I I 1. MisceUaneous Nonvolatile Compounds are a
CH,- N- H diverse class of compounds that are difficult to clas-
sify in terms of functional classes. Naturally occurring
cornpounds include fulvic acid, humic acid, chloro-
9. Ketones phyll, xanthophylls, and enzymes. Industrially pro-
duced compounds include tannic acid, most dyes,
Basic structure Acetone
0 0
and optical brighteners. Fulvic and humic acids are
II II -CH, major components of a complex set of compounds
R-C-R' CH,-C called aquatic humic substances. According to Thur-
Figure 11.5 Classification of organic compounds (contin- man (1985), 40 to 60%of the dissolved organic car-
ued on next column). bon occurring naturally in ground water is comprised
 11.7 Organic Conipounds 247

of aquatic humic substances. The natural plant pig- functional group. Carboxylic acids are distinguished
ments, chlorophylls and xanthophylls, generally by a hydroxyl group bonded to tlie carbonyl group
have a low solubility. They are minor constituents (Figure 11.5). In natural systems. carboxylic acids
in ground water, as are enzymes. Tannic acids, dyes ionize and, therefore, are quite soluble, contributing
and optical brighteners occur in ground water mainly 5 to 8% of the organic carbon in surface water (Thur-
as contaminants. man, 1985). Common compounds in this class in-
2. Halogenated Hydrocarbons are one of tlie clude acetic acid, benzoic acid, butyric acid, formic
largest and most important groups of contaminants acid, palmitic acid, propionic acid, stearic ;wid. ;ind
found in ground water. This class, divided into the valeric acid. Carothers and Kharaka (1978) report
aliphatic and aromatic subclasses, is characterized by concentrations of carboxylic acid in excess o f 5000
the presence of one or more halogen atoms (Figure mg/L in formation waters. Carboxyiic acids are ;ilso
11.5). The term alipbatic describes a structure important components of landfill leacliates due to
where carbon atoms are joined in open chains. An anaerobic fermentation reactions. Many of these
aromatic hydrocarbon has a moleciilar structure acids are produced for industrial uses. The most ini-
based on benzene, C,H,, characterized by six carbon portant products are phenoxy acetic pesticides such
;IS 2,4-D, silvex, and 2,4,5-T. These compounds ;ire
atoms arranged in a ring and radially bonded hydro-
gen atoms. Three of the carbons are doubly bonded also widely used in the food industry.
and three are singly bonded. Included in the aliphatic 7. Phenols are a small but important class of naturally
subclass are solvents such as metliylene chloride, occurring and manufactured compounds. They are
chloroform, carbon tetrachloride, 1,1,I-trichloroeth- characterized by an aromatic ring with an attaclied
ane, and trichloroethylene; pesticides such as aldrin hydroxyl group. In uncontaminated ground water,
and dieldrin; and industrial chemicals such as vinyl Thurman (1985) cites natural abundances of
chloride, methyl chloride, and methyl iodide. The 1 pg/L or less with compounds such as phenol.
aromatic subclass contains common pesticides, such cresol, and syringic, vanillic, and p-hydroxybenzoic
as DDD and DDE. and industrial chemicals such as acids. While phenols are commonly associated with
1,2,4-dichlorobenzene and the various PCBs. petroliferous rocks, they originate in ground wafer
mostly as contaminants from industri;il wastes or bio-
3. Amino Acids are ch;iracterized structurally by an
cides. The wastes include cresol, phenol, pyroc;ite-
amino group, NH2 on the carbon atom adjacent to
chol, and napthol, while pesticides include dinitro-
the carbonyl carbon, COOH (Figure 11.5). These
cresol and 2,4-dinitrophenol. Chlorinated phenolic
compounds represent a relatively small proportion
pesticides, included in Class 2, are important contani-
of the naturiilly occurring organic matter in water.
inan ts near woodpreserving facilities.
According to Thurman ( I 985), the dominant amino
acids in ground water (for example, glycine, serine, 8. Amines considered here are mainly a group o f in-
alanine, and aspartic acid) are present at concentra- dustrial contaminants. In this system of classification
tions o f a few hundred .yg/L.
- some of the naturally occurring amino acids contiin
the amine functional class. The aliphatic functional
4. Phosphorous Compounds are distinguished by
class can be considered as derivatives of ammonia
the presence of phosphorous in phosplioroiis-siilfiir.
(NH,) with one, two, or three of the hydrogens re-
phosphate, or phosphite bonds (Figure 11.5). The
placed by alkyl groups (CH,) (Figure 11.5). In the
most important group of compounds within this
aromatic amines, a hydrogen is replaced by ;in ;wo-
class is a large group of pesticides. The unhaloge-
matic ring(Figure 11.5). Sources of potential cont;ini-
nated compounds include diazanon, malathion, para-
ination in ground water are wastes from herbicide
thion, and tributyl phosph;ite and halogenated com-
and synthetic-rubber production. Some common
pounds include clorthion, dicapthon, and dursban.
contaminants include diethylamine. dinieth~*laniine,
5. Organometallic Compounds are a highly toxic ethylamine, methylamine. aniline. henzidine. nap-
group of organic contaminants. They have a broadly thylamine, and pyridine.
diverse structure, distinguished by the presence o f 9. Keytones are also distinguished by tlie presence
a metal atom. Examples of compounds in this class of the carbonyi group (Figure 1 1.5). In rhe keytones.
include tetraethyllead, diethylmercury, and copper the carbonyl group is bonded to two alkyl groups.
phthalocyanine. which may be alike or different. These compounds
6. Carboxylic Acids are an important class of natu- occur naturally but are mainly contaminants in intlus-
rally occurring compounds and contaminants. These trial effluent and municipal sewage. Acetone. fluor-
organic acids are part of a family of functional classes enone, fenchone, 2-butane, butyl propyl ketone. ;uid
that includes keytones, aldehydes, and esters, which methyl propenyl ketone are examples of these com-
are characterized by the presence of a carbonyl COO pounds.
248 Chapter 11 / principles of Aqueous Geochemistry

10. AIclehydes have a structure similar to the key- alkanes are methane, ethane, propane, butane, and
tones except that the carbonyl group is bonded to the various structural isomers (same chemical for-
one alkyl group and one or two hydrogens (Figure mula but different structure). The cycloalkanes
113. They occur naturally at relatively low concen- (CsH2n) are characterized by a ring structure that
trations and are occasionally contaminants in contains single carbon - carbon bonds. Cyclopro-
ground water. Formaldehydeand iso-butyraldehyde pane and cyclohexane are examples.
are examples of these compounds. The alkenes are aliphatic hydrocarbons with a
carbon-carbon double bond (Figure 1 I .5). Alkenes
11. Alcohols are some of the most soluble organic
compounds in water. They can be considered as have a general formula C,H2,, with members of the
derivatives of a hydrocarbon in which a hydroxyl series including ethene, propene, named similarly
group replaces a hydrogen (Figure 11.5). In natural to the alkane series.
systems, alcohols are present only in trace quanti- The alkynes are characterized by a carbon triple
ties because they are nontoxic and broken down bond and the general formula CnH2,,-* (Figure 11.5).
easily in the food chain. They occasionallyare found The alkyne series is named like the alkanes giving,
as contaminants in ground water. A few of the alco- for example, ethyne, propyne, butyne, etc. These
hols are methanol, glycerol, terpinol, ethylene- hydrocarbons are usually found in ground water as
glycol, and butanol. a result of crude oil or refined petroleum products.

12. Esters like the carboxylic acids, keytones, and

aldehydes contain the carbonyl functional group
and have the basic structure shown in Figure 11.5.
-
11.8 Ground-Water Composition
Esters are usually found in ground water as the Inorganic and organic solids, organic liquids, and gases
result of contamination. Some commonly occurring found in the subsurface dissolve in ground water to some
compounds include pesticides such as butyl mesityl extent. Thus, the variety of solutes in ground water (Table
oxide, dimethrin, dimethy carbate, dinobuton, 11.2) is not surprising. Inorganic constituents are classi-
omite, tabutrex, vinyl acetate, and warfarine plus fied as major constituents with concentrations greater
a variety of industrial chemicals. than 5 mg/L (Table 11.2), minor constituents with con-
13. Ethers occur in water both naturally and as con- centrations ranging from 0.01 to 10 mg/L, and trace
taminants. Structurally, the ether functional group elements with concentrations less than 0.01 mg/L (Davis
consists of an oxygen bonded between two carbons and Dewiest, 1966). Naturally occurring, dissolved or-
as shown in Figure 11.5. Examples of ethers found ganic compounds in ground water could number in the
as contaminants are tetrahydrofuran and 1,4di- hundreds. They are typically present in minor or trace
oxane. quantities. By far the most abundant organic compounds
in shallow ground water are the humic and fulvic acids.
14. Polynucleur Aromatic Hydrocarbons (PAH) However, other organic compounds are also present
form from a series of benzene rings (Figure 11.5). (Table 11.2). Deep ground water or formation wd-
These compounds are important components of ter can contain concentrations of organic matter up to
ancient sediments and crude oils. They occur in 2000 mg/L (Hull and others, 1984). Examples of organic
potable water in trace concentrations and probably acids are acetate and propionate, anions that contribute
originate from coal tar or creosote contamination. to titratable alkalinity. The important ground water gases
Thurman (1985) lists phenanthrene, fluoranthrene, include oxygen, carbon dioxide, hydrogen sulfide, and
anthracene, and benzopyrene as examples of PAHs methane.
found in rivers and rainwater.
15 Aromatic Hydrocarbons are compounds with
a molecular structure based on that of benzene, The Routine Water Analysis
C6H6(Figure 11.5). These hydrocarbons are a majorIt is neither feasible nor necessary to measure the concen-
constituent of petroleum and in ground water are tration of all constituents that conceivably might occur
usually indicative of contamination from a spill of
in water. A “routine” analysis involves measuring the
crude oil or a petroleum distillate. Some typicalconcentration of a standard set of constituents. Such tests
benzoid hydrocarbons are benzene, ethylbenzene, form the basis for assessing the suitability of water for
toluene, methylbenzene, and o-xylene. human consumption or various industrial and agricultural
I & Alkane, Alkene, and Alkyne Hydrocarbons uses. The routine analysis typically includes the major
are important components of oil and natural gas. constituents with the exception of silicon and carbonic
Alkanes are characterized by the CH3 functional acid, and the minor constituents with the exception of
class and conform to the general formula CnHL,r+2 boron and strontium. Laboratory results are reported as
(where n is the number of carbon atoms). Common concentrations in mg/kg or mg/L. The reported concen-
 11.8 Ground-Water Composition 249

Table 11.2 The Dissolved Constituents water sample is evaporated to dryness. Specific conduc-
in Potable Ground Water Classified tance is a measure of the ability of the sample to conduct
According to Relative Abundance electricity and provides a proxy measure of the total
Major constituents (greater than 5 nig/L)
quantity of ions in solution. The measurement is approxi-
Bicarbonate Silicon mate because the specific conductance of a fluid with a
Calcium Sodium given TDS content varies depending on the ions present.
Chloride Sulfate The routine analysis identifies nearly all of the mass
Magnesium Carbonic acid dissolved in a sample. Unanalyzed ions and organic com-
Nitrogen pounds are usually a negligible proportion. One simple
Minor constituents (0.01- 10.0 mg/L) check on the quality of a routine analysis is to compare
Boron Nitrate the sum of concentrations of cations and anions in milli-
Carbonate Potassium equivalent per liter. Given that water is electrically neu-
Fluoride Strontium tral, the ratio of the sums should be one. Table 11.4
Iron Bromide
illustrates this calculation. While not exactly one, the
Oxygen Carbon dioxide
Trace constituents (less than 0.1 mg/L)
value is within the If: 0.05 range of acceptability used by
Aluminum Nickel most laboratories. The cation/anion ratio provides one
Antimony Niobium check that concentration determinations are not grossly
Arsenic Phosphate in error.
Barium Platinum Large numbers of routine analyses collected over time
Beryllium Radium provide basic data for research studies in many areas.
Bismuth Rubidium These data must be used with care. Errors can occur
Cadmium Ruthenium because of ;I failure to measure rapidly changing parame-
Cerium Scandium ters in the field, to preserve the samples against deteriora-
Cesium Selenium tion due to long storage, and to assure the quality of the
Chromium Silver
laboratory determinations. These issues are discussed in
Cobalt Thallium
Copper Thorium
more detail in Chapter 17.
Gallium Tin
Germanium Titanium Specialized Analyses
Gold Tungsten
Indium Uranium Laboratories can carry out a routine analysis in a rapid
Iodide Vanadium and cost-effective manner. Occasionally, there is a need
Lanthanum Ytterbium for more specialized analyses, for example, trace metals,
Lead Yttrium radioisotopes, organic compounds, various nitrogen-
Lithium Zinc containing species, environmental isotopes or gases.
Manganese Zirconium Such work is often related to problems of ground-water
Molybdenum
contamination or research and regulatory needs. Com-
Organic compounds (shallow)
Humic acid Tannins
Fulvic acid Lignins
Carbohydrates Hydrocarbons Table 11.3 Example of a Routine Water Analysis
Amino acids Parameter mg/L Parameter mg/L
Organic conipounds (deep)
Acetate PH 7.7 Conductivity 2500
Propionate Calcium 1' Magnesium 1
Sodium 550 Potassium 3.i
Modified from Davis, S. N., and I<. J. M. Dewiest, 19(6. Hydrogeology.
Reprinted by permission.
Iron 8.7 NO?, NO,' 0. I '
Nitrite 0.1' Chloride ii
Sulfate 59 Fluoride 0.25
Bicarbonate 1315 Hardnesb. T H
Alkalinity, T 1078
trations for metals (for example, Ca, Mg, Na, and so TDS' 1321
on) is the total concentration of metals irrespective of Balance 1.01
whether they are complexed or not.
A routine analysis often includes a few other items in 'Indicates Concentration "less than." Conductivity rcported in I.($.
pH in pH units. All metal parameters expressed as totals. hlkilinity
addition to concentrations (Table 11.3). Of note are pH,
and hardness expressed as calcium carbonate. Nitrare. nitrite. ;ind
total dissolved solids (TDS) reported in mg/L, and specific ammonia expressed as N.
conductance reported in microsiemens per centimeter. 'NO2 = nitrite, NO, = nitrate.
The TDS content is the total quantity of solids when a 'Total dissolved solids.
250 Chapter 11 / Principles of Aqueous Geochemistty

Table 11.4 Evaluating the Electroneutrallty of the Example

Routine Analysis

Cation Concentration Anion Concentration

mg/L meq/L mg/L meq/L
Ca2+ 1.o 0.05 HCO; 1315 21.6
Mg2+ 1.o 0.08 so;- 59 1.22
Na' 550 23.9 c1- 45 1.27
K+ 3.5 0.09 F- 0.25 0.01
Fe 8.7 -0.31 Total 24.1
Total 24.4 cation/anion ratio = 1.01

mercial laboratories should be able to analyze for all but ity is part of many ground-water investigations. Over
the most exotic constituents. However, given the cost time, different graphical and statistical techniques have
of this more specialized work, serious attention has to been developed to assist with this task. Each technique
be paid to quality-assuranceissues. has particular advantages and disadvantages in represent-
ing features of the data. Thus, in working with a set of
analytical results, one should examine alternative ways
-
11.9 Describing Chemical Data
of presenting results to select the most appropriate.
The methods divide into two major groups. First is a
Describing the concentration or relative abundance of group of graphical approaches for describing abundance
major and minor constituents and the pattern of variabil- or relative abundance. Second are approaches that pre-

Sample 1 Sample 3
Na +

15
t - .1
Samale .
I
c1

HCQ,

(C)
-
0 1000 mg/L
Scale of diameters

10

* * A * o
Scale of diameters

5 ~

(10

Layout of Plotting Axes

15

HCO,
Fe COJoptional)

Sample 1 Sample 3

Ffgure 11.6 Four different ways of plotting major ion data (modified from Zaporozec, 1972). Panel (a) is a Collins diagram,
Panel (b) is a Stiff diagram, Panel (c) is a pie diagram, and Panel (d)is a Piper diagram. Reprinted by permission of Ground Wa-
ter. Copyright 0 1972. All rights reserved.
 11.9 Describing Cbemical Data 2 3

Tab& J I . 5 Sample Chemical Data Used to Demonstrate Various Techniques for Plotting Chemical Data
Sample 1: Tertiary, Sample 2: Upper Cretaceous, Sample 3: Upper Cambrian,
Czechoslovakia Czechoslovakia Wisconsin
Chemical
Analyses mg/L meq/L meq (O/o) mg/L meq/L meq(O/o) mg/L meq/L meq(O/o)
Cations
Na’ + K’ 266.2 10.68 68.24 1913.7 81.54 70.24 7.9 0.34 4.4
Mg” 21.9 1.80 11.4 132.8 10.95 9.4 1 43.0 3.54 45.6
Ca” 61.7 3.08 18.5 468.5 23.38 20.50 78.0 3.89 50. I
MnZ’ Traces - - 0.67 0.02 0.01 0.14 0.004 0.05
Fe 2.3 0.08 - 0.15 0.005 0.004 0.11 - -
Sum - 15.64 100.0 - 115.89 100.0 - 7.77 100.0
Anions
c1- 11.3 0.32 2.05 850.0 23.98 20.63 17.0 0.48 6.4
NOS 0.0 - - 0.0 - - 0.7 0.0 0.1
HCO; 906.1 14.85 95.1 2568.5 42.10 36.23 364.0 5.96 79.5
so:- 21.2 0.44 2.82 2406.5 50.10 43.11 50.0 1.04 13.9
Sum - 15.61 100.0 - 116.18 100.0 - 7.50 l(W.0

From Zapovozec (1972).

Reprinted by permission of Ground Water. Copyright 0 1972. All rights resewed.

Figure 11.7 Example of how concentration data for an ion can be represented on a cross section. Shown here are S o : - data
from Blackspring Ridge, Alberta, Canada (from Stein and Schwartz, 1990). Reprinted from J. Hydrol. v. 117 by R. Stein and F. W.
Schwdrtz, On the origin of saline soils at Blackspring Ridge, Alberta, Canada, p. 99-131, 1990, with kind permission from
Elsevier Science, NL, Sara Burgerhartstraat 25, 1055 KV Amsterdam, The Netherlands.
252 Cbapter 11 / Princtples of Aqueous Geocbemistry

100
Water from
Peripheral bedrock
alluvial fans and outer
basin fill
x Salt flats

Figure 21.8 Example of the use of a Piper diagram for classifying ground water and defin-
ing a pathway of chemical evolution. The diagram shows how samples from the Salt Basin
in west Texas can be classified according to setting for example, limestone (dots), alluvial
fans, and basin fills (squares), and the salt flats (crosses). The evolutionary pathway is de-
fined by ground water moving from the firnestone, through the alluvium to the salt flats
(from Boyd and Kreitler, 19%).

sent patterns of variability in addition to abundance. This only that concentrations be plotted as a bar segment, a
group subdivides further into graphical/illustrative and point on a line, or a percentage of the pie. The appro-
statistical approaches. priate fields are shaded and possibly labeled in the case
of the Collins and Piper diagram (Figure 11.6). The Stiff
diagrams can be plotted with or without the labeled axes.
Abundance or Relative Abudance Plotting data for a Piper diagram is complicated be-
Several different graphical approaches can depict the cause there are three separate diagrams (Figure l l .&).
abundance or relative abundance of ions in individual The relative abundance of cations with the %meq/L of
water samples. The most common are (1) the Collins Na+ + K+, Ca2+,and Mgz+ assumed to equal 100%is first
(1923) bar diagram, (2) the Stiff (1951) pattern diagram, plotted on the cation triangle. Similarly the anion triangle
(3) the pie diagram, and (4) the Piper (1944) diagram. displays the relative abundance of CI-, SO:-, and
Data in Table 11.5 will be used to illustrate these ap- HCO,; + CO:-. Straight lines projected from the two
proaches. For plotting chemical data, concentrations triangles into the quadrilateral field define a point on the
need to be expressed as meq/L or %meq/L. The Collins third field (Figure 11.6d). To provide some indication of
and Stiff diagrams require absolute concentrations the absolute quantity of dissolved mass in the sample,
(meq/L), while the pie and Piper diagrams require rela- the size of the data point is sometimes related to the
tive concentrations (%meq/L). salinity 0 s ) . One advantage of all four techniques is
Figures 11.6~2,b, c, and d illustrate the sample data that they present the major ion data for a sample on one
plotted in different ways. The Collins, Stiff, and pie dia- figure. However, with exception of the Piper diagram,
grams are relatively simple to construct. They require these approaches are useful only in displaying the results
 11.9 Describing Chemical Data 2 2

Cation facies Anion facies for a few analyses, which are often "type" waters from
an area. Presenting a large number of these didgrinis
together is confusing and not much more helpful than a
table of concentration vdues.

Abundance and Patterns of Change

Grapliical/illustrative type diagrams or statistics c;in de-
fine the pattern of spatial change among different getr
logic units, along a line of section, or along a pathline.
One simple way of representing spatial change in a single
geologic unit is to take the single sample diagranis (for
example, pie or Stiff) and place them on a map. Such
maps can convey a sense of how the pattern of ion
abundances change within a unit (for example, Swenson,
1968). Including all of the constituents on a single map
can be advantageous. However, making sense out of ;I
collection of geometric shapes is difficult.
When chemical data vary systematically in space, it is
often best to plot and contour concentrations (or other
Percentage of Constituents
data) on maps or cross sections (Figure 1 1.7). This presen-
Ffgure 12.9 Templates for classifying waters into facies for tation it obvious how vdq,
cations and anions (from Back, 1961).
Any measured or calculated chemical parameter can be

Explanation 0Calcium-magnesium faeces

Ffgure 11.10 Example of hydrochemical facies mapping. Shown on the fence diagram
are cation data from part of the northern Atlantic Coastal Plain (modified from Back,
1961).
2? Cbapter 11 / Prtnclpks of Aqueous Ceocbemistry

represented in this way. One problem is the large number sive change in cation chemistry from calcium-magnesium
of figures that could be required to describe the chemistry water in the upland part of the Atlantic Coastal Plain to
of an area completely. However, given the usefulness of sodium water in deeper units located at the downstream
these diagrams, this limitation is not serious. end of the flow system.
Another problem arises with "noisy" data. Simply con- In closing, we mention two common statistical ap-
touring a set of such data can produce a complex and proaches. Summary statistics (for example, mean and
cluttered figure. A Piper diagram is preferable to concen- standard deviation) for various subsets of a chemical data
tration maps when data are noisy. By classifying samples set provide a simple way of describing patterns of variabil-
on the Piper diagram, one can identify geologic units ity. A useful classification technique is cluster analysis
with chemically similar water, and define the evolution (Davis, 1986). This procedure involves sorting a set of
in water chemistry along a flow system (Figure 1I .8). individual samples into smaller groups. Presenting the
Noisy data can also be smoothed before plotting on a results on a two-dimensional diagram called a dendogram
map or cross section. The facies mapping approach enables one to represent the pattern of clustering graphi-
(Back, 1961) provides one way of smoothing chemical cally down to individual pairs of samples and the natural
data. Samples are classified according to facies with two breaks between major groups. One advantage with this
templates for the Piper diagram (Figure 11.9), one for approach is that water can be classified on the basis of
the cations and the other for the anions. The limited any desired combination of chemical data. Unlike the
number of possibilities for classifying the chemical data Piper diagram, one is not limited to the same group of
effectively eliminates local variability yet preserves broad major ions. For a complete discussion of the method,
trends. The example (Figure 11.10) shows the progres readers can refer to Davis (19%).

Problems
1. The routine analysis of a water sample provides the dissolution is 4.9 X What does the saturation index
following concentrations (as mg/L): CaL+ = 93.9; indicate about the state of saturation with respect to
Mg2+= 22.9; Na' = 19.1; HCO; = 334.; SO:- = 85.0; calcite?
CI- = 9.0 and a pH of 7.20. 4. Assess in a preliminary way the quality of the analyti-
a. Calculate the concentrations of Ca'+, Na+, HCO;, cal results in question (1) by determining the cation/
and C1- in terms of molarity and milliequivalents per anion balance.
liter (meq/L). 5. Determine the order of the following kinetic expres-
b. What is the ionic strength of the sample? sions.
c. Calculate the activity coefficients for CaL' and
a. degradation of organic matter (org): d(org)/dt =
HCO; using the extended Debye-Huckel equation.
-kdorg)
2. Write mass law expressions for the following equilib- b. oxidation of pyrite: d(Fe3+)/dt= -k,Z(Fe'+), where
rium reactions. Z is surface area of reacting mineral
a. CaCO, = Ca2++ COj- c. general reaction: &A)/& = -k,
b. CO,(g) = CO,(aq) d. general reaction: d ( O / d t = -k,(A)(B)?
c. MnLt + Cl- = MnCI' 6. Represent the analytical results in question (1) graph-
3. For the water sample in question (l), determine the ically using (a) a Collins diagram and @) a Piper diagram.
saturation index for calcite at 25OC given that [CO{-] = (c) Classify the water using the concept of hydrochemi-
0.34 X and that the equilibrium constant for calcite cal facies.
 Chapter 12

Cbemical Reactions

12.1 Acid-Base Reactions

12.2 Solution, Exsolution, Volatilization, and Precipitation
12.3 Complexation Reactions
12.4 Reactions on Surfaces
12.5 Oxidation-Reduction Reactions
12.6 Hydrolysis
12.7 Isotopic Processes

A variety of physical, chemical, and biological processes A base is a substance with a tendency to gain a proton.
operate in the subsurface to transfer mass among the Acids react with bases in what are called acid - base reac-
fluids, gases, and solids found there. In this chapter we tions. In general, because no free protons result from an
classify and describe them, breaking down this relatively acid-base reaction, there must be two acid-base sys-
large and complex topic into manageable chunks. Fur- tems involved:
ther, this chapter will develop a sense of why reactions
are important in problems of contamination and the geo-
Acid, + Base? = Acid? + Base, (12.1)
chemistry of natural waters. In the forward reaction, the proton lost by Acid, is gained
by Base,, and in the reverse reaction the proton lost
by Acidz is gained by Base,. The following equilibrium
-
12.1 Acid-Base Reactions reaction illustrates this point:
Acid-base reactions are important in ground water be- HCO; + H 2 0 = HIO' + C0:- (12.2)
cause of their intluence on pH and the ion chemistry.
pH is a master variable controlling chemical systems. It In Eq. 12.2 protons are transferred from HCO; to H1O,
is defined as the negative logarithm of the hydrogen ion which in this reaction functions as a base, and from the
activity and describes whether a solution is acidic (pH acid H30+ to the base C O < - .
C 7),neutral (pH = 7), or basic (pH > 7). Hydrogen ion Applying the law of mass action to Eq. 12.2 yields
activity is represented here as [H'] so that pH =
-log[H+]. In solution, hydrogen ion exists as a proton (12.3)
associated with a molecule of water (H30+). Thus, we
can express hydrogen ion activity as [H,O+]. Writing This equation simplifies by assuming that [H20] = 1 and
the hydrogen activity as H 3 0 + differentiates between a [H,O+] = [H'], which gives
proton (a hydrogen atom that has lost an electron: H+)
and a hydrogen ion in solution. (I 2.4)
An acid is a substance with a tendency to lose a proton.
-
2-54 Cbapter 12 / Cbemical Reactions

By making this substitution, the acid-base reaction in- reaction describes the dissociation of water into hydro-
volving H20-H300+is neglected.This simplificationis usu- gen and hydroxyl ions, where the constant K,,is the ion
ally made in writing acid-base reactions in water. Never- product for water. This reaction is a simplified form of
theless, thinking of acid-base reactions in terms of the acid-base reaction
reaction pairs helps to understand these reactions.
In the previous example, the solvent water functions HzO + H z 0 = H3O' + OH- (12.13)
as a base. However, bases such as ammonia (NHJ also Writing the reaction this way implies that protons will
ionize in water be transferred between water molecules and between
NH, + HzO = NHd + OH- (12.5) H30+and OH- even in pure water. The dissociation reac-
Base, Acid, Acid, Base, tion for water is fundamental in establishing the relation-
ship between (H') and (OH-) according to the law of
which implies that water also functions as an acid. The mass action.
concept of an acid or base, thus, can be quite abstract. Next in Table 12.1 are weak acid-base reactions that
The strength of an acid or a base refers to the extent to originate mainly from adding C 0 2 gas to water. The first
which protons are lost or gained, respectively. Strength is reaction describes how C02 gas dissolves to produce
often reported in qualitative terms of strong or weak. carbonic acid H2CO;. By convention, the concentrations
Consider the following generalized ionization reaction of C02(aq) and HzCOJare represented by the single spe-
for an acid HA in water (Glasstone and Lewis, 1960): cies HLC03or
HA + HzO = H3O+ + A- (12.6)
(HICOD = (C02 aq) + (H2C0,) (12.14)
where A- is an anion like CI-. The strength of an acid
or base depends on whether equilibrium in a reaction The next pair of reactions in Table 12.1 describes the
like Eq. 12.6 is established to the right or left side. For two-stage ionization of carbonic acid, which is controlled
example, a strong acid will ionize and establish an equilib- by the first and second dissociation constants K ,and K 2 ,
rium to the right because of the significant proton transfer respectively. The last group of reactions describes the
to H 2 0 in spite of the fact that H 2 0 is a weak base. When dissociation of silicic acid in water. This acid originates
the acid (HA) is strong, its conjugate base A- will be mainly from the dissolution of silicate minerals.
weak. If HA is a weak acid, then its conjugate base will
be strong. This relationship with just a few exceptions
C02-Water System
applies to acids and bases of all types (Glasstone and
Lewis, 1960). The reactions involving COzin Table 12.1 show that C 0 2
dissolved in water partitions among H,CO;, HCO, and
Cog-. When the pH of a solution is fixed, the mass law
Natural Weak Acid-Base Systems
equations in Table 12.1 let us determine the concentra-
The few weak acid-base reactions that are important in tions of the species. Following is an example illustrating
natural ground waters are listed in Table 12.1. The first this calculation.

Table 12.1 Important Weak Acid-Base Reactions in Natural Water

Systems

-@
Reaction Mass Law Equation Eq- K(25'C)
H20= H t 4- OH- K,,= (H+)(OH-) 12.7 14.0
(HKO3
CO&) + H20 = HZCO! K<ol=
P~ol(H20)
12.8 1.46
(HCO;)(H+) 12,9
H2CO; = HCO; + H+ KI =
(HKO;)
6.35
(CO:-)(H+)
HCO; = C0:- + Hi Ki =
(HCO;)
12.10 10.33
(HSiO;)(H+)
H2Si0, = HSiO; + Ht K=
(H2Si0,)
12.11 9.86
(SiO:-)(H+)
HSiO; = SO:- +Hi K=
(HSiO;)
12.12 13.1
 12.1 Acidd-BaseReactions 2 2

ExampCe 12.1 A Bjerrum diagram illustrates the dominance of particu-

lar species over some pH range. For example, below pH
Assume that COz is dissolved in water so that (COJ, = 5, H2CO; is the dominant carbonate species (Figure 12.1).
M and In the range from approximately pH 7 to 9, HCO; is the
= (H2C0.T) + (HCO;) + (COj-) (12.15) most abundant species. The carbonate ion (COf-) is domi-
nant when pH is above 10.
What is the concentration of the carbonate species at
a pH of 6.35? We chose this particular pH to simpllfy
the calculation. Alkalinity
Substitution of the known pH into Eq. 12.9 gives So far, we have discussed only cases where the pH deter-
mines the distribution of carbonate species. In natural
water systems, the situation is more complex. The pH
(HLCO;) and carbonate speciation are interdependent, a function
Simplification gives (HCO;) = (H,CO;). Substitution into of not only the ionization equilibria for the carbonate
Eq. 12.10 provides species and water but also strong bases added through
the dissolution of carbonate and silicate minerals.
Formulating these processes mathematically requires
an understanding of the concept of alkalinity. For our
purposes, alkalinity is defined as the net concentration
and shows that (HCO;) >> (COi-). Assuming for the
of strong base in excess of strong acid with a pure C 0 2 -
moment that (Cog-) is negligible, a solution to Eq. 12.15
water system as the point of reference. When C 0 2 is
gives (HCO;) = (HzCO:) = M. Substitution into
dissolved in water at a fixed Pco,, the charge balance is
En. 12.10 gives (COi-) = lo--'"M.
(H') = (OH-) + (HCO;) + 2(CO:-) (12.18)
Similar calculations over a broad pH range would show This charge balance will change, for example, by adding
how the concentrations of the carbonate species change. a strong acid such as HCI and a strong base such as NaOH
A useful way of illustrating this relationship is with a to the solution. Assuming that NaOH and HCI ionize fully,
Bjerrum plot. It is a plot of the logarithm of activities (or the charge balance is
concentrations) of various species versus pH. A log C-pH
diagram for the carbonate system is shown in Figure 12.1
(Na') + (H') = (OH-) + (HCO;) + 2(CO<-) + (CI')
(12.19)
for (CO& = The arrows on the figure depict the
crossover points, where at pH 6.35 (H,CO;) = (HCO;) Rearranging Eq. 12.19 to group together ions contributed
and at 10.33 (HCO;) = (CO:-). These are known as the from the strong acids and bases yields
pK, values for aqueous CO, (i.e., pK, = 6.35 and
pK2 = 10.33).
(Na') - (Cl-) = -(H+) + (OH-) + (HCO;) + 2(CO<-)
(12.20)
The left side of Eq. 12.20 is the net concentration of
strong base in excess of strong acid and defines alkalinity
when Nd' > C1-. A more general form of this eqiidtion is
Alk = -
(i+)sb 2(OSa (1 2.21)
= -(H+) + (OH-) + (HCO;) + 2(CO<-)
where ( f + ) and ( f - ) are the positively and negatively
charged species from the strong acids and bases. When
the alkalinity is zero, Eq. 12.21 becomes the charge bal-
ance equation for a pure C0,-water system.
In natural ground water, the generation of net positive
charges through the dissolution of carbonate and silicate
minerals is always greater than the contribution of net
PH
negative charges from the ionization of strong acids.
Ffgure 12.1 Log C-pH diagram for the carbonate system Thus, as a general rule, ground waters are alkaline. Strong
with (CO,), = lo-' M.From F. M. M. Morel and J. G. He- acids are rare in natural waters and when they do occur
ring, 1993. Principles and Applications of Aquatic Chemistry. are the result of contamination. In these cases, the charge
Copyright 0 1993 by John Wiley Sr Sons, Inc. Reprinted by contribution from the strong acid may exceed that due
permission of John Wiley Sr Sons, Inc. to the strong base and create mineral acidity.
258 Cbapter 12 / Cbemical Reactions

So far, the pure C0,-water system with zero alkalinity Example 12.2
is the point of reference in defining alkalinity. In practice,
it may be more convenient to use other reference points What is the concentration of H2CO; dissolved in a small
and to define carbonate or caustic alkalinity (Morel and drop of rain at 25°C falling through the atmosphere?
Hering, 1993). Further, establishing the relative contribu- Assume that the COzin the droplet is in equilibrium with
tion of noncarbonate alkalinity may be useful. the atmosphere with a Pcozof 10-3.5 atrn. The Henry’s
As Eq. 12.21 shows, increasing alkalinity increases the law constant is atm-l/mol.
net positive charge on the left side of the equation. This
Pcoz
increase is not simply balanced by an increase in one of -
(HzCO;) = (COz aq) = -
KH
the negative ions on the right side, because equilibrium
relationships in the solution must be maintained. Part of
the increase in the concentration of negatively charged
species comes from the ionization of H2CO; to HCO; to
C0:- and an increase in pH. Thus, increasing the alkalin-
ity with a strong base ultimately leads to an increase in
pH. This behavior is commonly observed as ground wa-
ters evolve by dissolving minerals along a flow path. This example shows that C 0 2 gas is related to the
carbonate system through H2CO;. Thus, a change in the
partial pressure of COz gas not only changes (H2COf)
-
12.2 Solution, Exsolution, Volatilization, according to Henry’s law but also (HCO;), (Cog-), (H’),
and (OH-), which are related through the carbonate equi-
and Precipitation
libria. The concentrations of some cations may also
Because water is an excellent solvent, it dissolves many change because of equilibria with solid phases.
of the gases, liquids, and solids in the subsurface. Dissolu- The addition or removal of gases from solution, thus,
tion more than any other process is responsible for the can play a major role in controlling the chemistry of
large solute loading to ground water. Other processes like ground water. For example, a significant step in the
gas exsolution, volatilization, and mineral precipitation chemical evolution of natural ground water involves the
remove mass from water. solution of C 0 2 gas in the soil zone. This process not
only increases the concentration of HCO; but also en-
Gas Solution and Exsolution hances the aggressiveness of the water in dissolving
solids.
Gas solution and exsolution can transfer significant quan-
tities of mass between soil gases and ground water. Com-
monly, we model these processes using equilibrium con- Solution of Organic Solutes in Water
cepts based on Henry’s law. This mass-law equation
relates the concentration of dissolved gas in solution to Liquids other than water in the subsurface include oil or
the partial pressure of the same gas in an atmosphere in organic contaminants. These nonaqueous components
contact with the solution. Henry’s law strictly speaking can migrate as a separate liquid phase or be dissolved in
does not apply to gases such as C 0 2 or NH3 that react in ground water. This latter process is of interest here.
solution. However, in the case of COz, so little of the Organic compounds differ widely in their overall solu-
COz(aq) reacts that Henry’s law approximates the distri- bilities (Mackay and others, 1985). Some solutes like
bution of gas between the two phases. Thus, the concen- methanol are extremely soluble, while others such as
tration of H2CO; is almost all C02(aq) with a negligibly PCBs or DDT are sparingly soluble or hydrophobic. As
small concentration of H2C03. a general rule, the most soluble organic compounds are
The Henry’s law equation for COz is charged species or those containing oxygen or nitrogen.
Examples of this latter group are alcohols or carboxylic
(1 2.22) acids, which hydrogen bond with the water and are easily
accomodated in solution. Thus, one important factor in-
where KH is the Henry’s Law constant with units like fluencing the solubility of organic compounds is the func-
atm-L/mol or atm-m3/mol, Pco2 is the partial pressure of tional groups that are part of the molecule. Even minor
C02 (atm) in the gas phase and (COz * aq) is the molar changes in the structure of an organic compound such
concentration of COzgas in the solution (mol/L or mol/ as the position of the CI atom on a benzene ring can have
m3). In applications, the concentration of CO,(aq) in Eq. a noticeable influence on solubility. Further, solubility of
12.22 is the same as (HzCO;). alcohols decreases as the size of the molecule increases.
Example 12.2 illustrates how Henry’s law determines Without hydrogen bonding, solubility is much re-
the concentration of C02 in a solution in equilibrium duced. The energetics of dissolution, related to the ther-
with an atmosphere of a given composition. modynamics accompanying the organic-water interac-
 12-2 Solution, Exsolution, Volutilizatiort, and Precipitation 259

Table 12.2 A Comparison of Molecular Mass Versus

Solubility for Various Aromatics
Molecular Mass Solubility
Compound (g/mol) Wm9
Benzene (GMd 78.0 1780
Toluene (C-H,) 92.0 515
o-Xylene (C,Hd 106.0 I75
Cumene (C,H,,) 120.0 50
Naphthalene (C,,,HH) 128.0 33
Biphenyl (C12Hl,3 154.0 7.48
Reprinted with permission from Mackay, I>.. and Leinonm, P.J., Envi-
ron. Sci. Technol.. v. 9, 1975, p. 1179. Copyright 0 1975 American
ChemicJI Society.

tions and the formation of the space in the water for the Verschueren (1983) tabulated solubility data for many
molecule, do not favor dissolution. Many of the terms organic compounds. When data are unavailable or unreli-
describing this dissolution process are size dependent. able, measured octanol/water partition coefficients pro-
The larger the organic molecule is, the larger the space vide a basis for estimating solubility. The octanol/water
that is required in the aqueous solution, and the less partition coefficient (Kr,lJis a dimensionless equilibrium
soluble the compound is (Schwanenbach and others, constant that characterizes partitioning of an organic sol-
1993). For example, with aromatic compounds (Table ute (org) between octanol (an organic liquid) and water.
12.2), solubility is inversely proportional to the molecular The mass law expression for this reaction is
mass or size of the molecules. Temperature and the pres-
ence of other organic liquids are factors that also influ- (12.23)
ence solubility.
where (org * oct) is the concentration of the solute in
octanol and (org aq) is the concentration in water.
Values of K,,,,. for many organic compounds have been
tabulated by Hansch and Leo (1979). Because the pani-
tioning of an organic solute between water and octanol
is conceptually not much different from the partitioning
of an organic compound between itself and water. a
correlation exists between K,,,,.and solubility. This rela-
tionship is illustrated for various classes of liquid and
solid organic compounds in Figure 12.2. The regression
equation for all compounds is
log K,,,,. = 7.30 - 0.747 log Sb ( 1 2.24)
where Sb is the solubility ( M ) of the compound. Chiou
and others (1982) developed equations of this form for
all the classes of compounds shown in Figure 12.2.
4 Ethers

6 Alkynes 2' Volatilization

7 Alkenes
8 Aromatics Volatilization is a process of liquid or solid phase evapora-
9 Alkanes
10 Organophosphates
11 PCBs and DDT
} So,ids tion that occurs when contaminants present either as
nonaqueous phase liquids or dissolved in water con-
-1' I I I I I I I I tacting a gas phase. This situation commonly arises with
-9 -8 -7 -6 -5 -4 -3 -2 -1 0
Solubility in Water, log S,, (moles per liter) organic contaminants in the saturated and unsatimtecl
F&U?V 12.2 Relationship between solubility in water and zones (Figure 12.3) or during the sampling and ;inalpsis
octanol- water partition coefficients for various organic com- of volatile compounds. The process itself is controlled
pounds. Reprinted with permission from C. T. Chiou, D. W. by the vapor pressure of the organic solute or solvent.
Schmedding, and M. Manes, Environ. Sci. Technol., 16, 1982, The vapor pressure of a liquid or solid is the pressure of
p. 6. Copyright 0 1982 American Chemical Society. the gas in equilibrium with respect t o the liquid or solid
260 Chapter 12 / Chemical Reactions

. *.I .; V-'.,
'. . .;
, vapM Unsaturated Zone

-. ---.
Mobile, liquid
- -- - --__-_
--
-- -- --__-- Contaminant tn drssolverj - 'I-,

v ,/'
--- *_
Flow 44), contaminant ,
Saturated Zone
--c
---. .
---._-..-_.- - - -. ,

Figure 123 Migration of volatiles in the soil atmosphere away from an organic contami-
nant spill (modified from Schwille, 1985). Reprinted by permission Second canndidn/Ameri-
can Conference on Hydrogeology. Copyright 0 1985. All rights reserved.

at a given temperature. Vapor pressure represents a com- also mol/L to make KI, a dimensionless Henry's law con-
pounds tendency to evaporate and is essentially the solu- stant. The two different Henry's law constants are related
bility of an organic solvent in a gas. Raoult's law describes to each other by the following equation:
the equilibrium partial pressure of a volatile organic in
KI, = K,,/RT (12.28)
the atmosphere above an ideal solvent like benzene
where R is the gas constant 0.0821 L-atm/mol-"K, and 7'
Po0 = X,,P& (12.25)
is absolute temperature in O K . Zero on the Celsius scale
where Pogis the partial pressure of the vapor in the gas is equal to 273.15"K.
phase, xog is the mole fraction of the organic solvent, On occasion, one may be interested in representing
and P& is the vapor pressure of the pure organic solvent. the partial pressure of an organic compound in the vapor
Volatilization of dissolved organic solutes from water phase as a concentration. The following expression from
is described by Henry's law. Commonly, such Henry's Johnson and others (1990), which is based on the ideal
law equations are written as gas law, serves this purpose:
Pop = KH(organic * as) (12.26)
Gr&=
X,,,P::,#fW - 1000 (1 2.29)
where Pogis the partial pressure of the organic com- RT
pound in the vapor phase (typically, atm), and (organic where C,,, is the approximate concentration of an or-
* aq) is the concentration in the aqueous phase (typically, ganic compound in mg/L, x,,,is mole fraction of the
moVm' or mol/L). With the equation written in this man- compound in the liquid phase, P& is the pure component
ner, KH has units atm-m3/mol or atm-L/mol. Thus, the vapor pressure in atmospheres, and fw is the formula
Henry's law constants for reactions with volatile organic weight of the compound in g/mol.
compounds are the same as those for gases in Eq. 12.22.
The Henry's law constant, expressed in units of atm-
Example 12.3
Umol, is approximately equal to F'in atmospheres di-
vided by the solubility of the compound in water (moV
L). Thus, the most common convention treats volatiliza- Toluene with a formula weight of 92.1 g/mol is found
tion of pure solvents (such as benzene) using Raoult's as a pure liquid phase in the subsurface. Given that the
law, and the volatilization of solutes (such as benzene vapor pressure of pure toluene is 0.029 atm, calculate
dissolved in water) using Henry's law. its partial pressure and approximate concentration in the
An alternative form of Eq. 12.26 can be written as vapor phase at 20°C or 293.15"K.
The partial pressure is determined by substituting
-
(organic . vapor) = KI, (organic as) (12.27)
known information into Eq. 12.25:
where the concentration of the organic compound in
Purl:= 1.0 X 0.029
the vapor phase, (organic * vapor), is expressed in units
of mom and the concentration in the aqueous phase is = 0.029 atm
 12.2 Solution, Exsolution, Volatilixation, and Precipitation 261
The approximate concentration is determined by sub- compounds such as dichloromethane or 1 , ldichloro-
stituting known information into Eq. 12.29or ethane to nonvolatile compounds such as pentachloro-
phenol or DDT.
1.0X 0.029 X 92.1 X 1000 Volatilization is important for problems of organic con-
C”, =
0.0821 X 293.15 tamination within or close to the unsaturated zone, and
= 1 1 1 mg/L for issues related to sample collection and analysis. Soil
gases containing volatiles, when allowed to accumulate
in the basements of structures, can cause fires, explo-
Table 12.3,from Jackson and others (l985),lists vapor sions, or health problems. Contamination can be so seri-
pressures for a selected group of organic contaminants. ous that venting of the soil may be required. Alternatively,
Even a limited set of data shows vapor pressures ranging the tendency for some organic contaminants to volatilize
over 9 or 10 orders of magnitude from the most volatile forms the basis of a useful technique for detecting and

Table 12.3 Vapor Pressure Data for a Selected Group of Organic Contaminants

Henry’s
La+
Compound Formula V.P.O Constant
Halogenated hydrocarbons
Dichloromethdne CHlC12 349 3 x 10-4
Trichloromethane CHCI, 160 4.8 X lo-’
Tetrachloromethane CCl, 90 2.3 X lo-’
Bromoform CHBr, 5.6 (25OC) 5.8 X lo-’ c
1,l-Dichloroethane CHCllCHi 180 4.3 x lo-’ c
1,j?-Dichloroethane CHZCICHZCI 61 9.1 X 10‘’ c
1,l-Dichloroethene H?C=CCIL 500 -
Trans-l,2-dichloroethene CHCI=CHCI 200 (14°C) 4.2 X 1 0 - l ~
1 ,l,l-Trichloroethdne CCI,CHI 100 1.8 X lo-’
Trichloroethene CIIC=CHCI 60 1 x lo-‘
1,1,2-Trichloroethane CHlClCHCIl 19 7.4 x lo-’ c
Tetrachloroethane C12C=CC12 14 8.3 X lo-’
Aromatic hydrocarbons
Benzene 76 5.5 x lo-’
Phenol 0.2 3.0 X lo-- c
Chlorobenzene 8.8 2.6 X lo-’ c
Ethylenebenzene 7 8.7 X lo-’
Toluene 22 5.7 x lo-’
o-Xylene 5 5.3 x lo-’
Other organic solvents
Acetone 89 (5°C) -
Diethyl ether 442 5.1 X lo-’ c
Tetrahydrofuran -
1,4-Dioxane 30 -
Blocides
Pentachlorophenol 1.1 x lo-’ 3.4 x lo-(’
DDT 1 x lo-- 3.8 x 10-5
Lindane 9.4 x 10-6 4.8 X lo--
~~~~~~ ~~ ~

” Vapor pressure in m m Hg at 20°C; 1 atm = 760 m m Hg.

atm-m’/moIe; c indicates values calculated from vapor pressure and solubility data.
Modified from Jackson and others. 1985. Contaminant Hydrogeology of Toxic Organic Chemicals at a
Disposal Site, Gloucester, Ontario. 1. Chemical Concepts and Site Assessment, IWD Scientific Series 141.
Environment Canada, 114 p. R e p n d u c d with the permission of Environment Canada.
262 Cbapter 12 / Cbemical Reactions

monitoring contamination in the subsurface and for in pH contaminant plume as dispersion and other processes
situ approaches for recovering contaminants like gasoline increase the pH of the ground water. These examples
from the subsurface. show how ground water proceeds toward chemical equi-
Volatilization also causes problems in sampling and librium with respect to various minerals from undersatu-
analysis. When samples have access to the atmosphere, ration in the case of evolving natural ground water and
volatilization reduces the concentration of the contami- from supersaturation in the case of metal transport.
nant exponentially. If exposed to the atmosphere long
enough, the concentration of the contaminant can fall
Solid Solubility
below detection limits.
The solubility of a solid reflects the extent to which
the reactant (solid) or products (ions and/or secondary
Dissolution and Precipitation of Solids
minerals) are favored in a dissolution-precipitation reac-
Of all the processes that influence solute transport, the tion. In many reactions where the activity of the reacting
dissolution and precipitation of solids are two of the solid is equal to one, a comparison of the relative size of
most important in terms of their control on ground-water the equilibrium constant provides an indication of the
chemistry. Extremely large quantities of mass can be solid solubiIity in pure water. For example, Table 12.4
transferred under some conditions between the water shows the chloride and sulfate salts to be the most soluble
and solid mineral phases. For example, recharging phase and the sulfide and hydroxide groups to be the
ground water derives almost its entire solute load through least soluble. Minerals in the carbonate and in the silicate
the dissolution of minerals along flow paths. Mineral pre- and aluminum silicate groups have a small but signifi-
cipitation removes much of the metal present in a low cant solubility.

Table 12.4 Some Common Mineral Dissolution Reactions and Associated Equilibrium
Constants
M i n e r a l or Solid Reaction log K(25"C) Source
Chlorides and sulfates
Halite NaCl = Na' + C1- 1.54
Sylvite KCI = K' + C1- 0.98
Gypsum CaSOl . 2H20 = Ca2++ SO:- + 2H20 -4.62
Carbonates
Magnesite -7.46
Aragonite -8.22
Calcite -8.35
Siderite - 10.7
Dolomite - 16.7
Hydroxides
Brucite Mg(OH)2 = Mg" + 2 0 H - -11.1
Ferrous hydroxide Fe(OH), = Fez+ + 20H- -15.1
Gibbsite AI(OHX = Al" + 30H- -33.5
Sulfides
Pyrrhotite +
FeS = Fez+ SL- - 18.1 1
Sphalerite +
ZnS = zn2+ S2- -23.9 3
Galena +
PbS = Pb2+ S2- -27.5 1
Silicates and
aluminum silicates
Quartz SiOz + HzO = HzSiO, -4.00 1
Na-Montmorillonite 3Na-Mont + 1 1 % H 2 0 =
3% Kaol + 4H4Si04+ Na' -9.1 2
Kaolini te Kaol + 5H2O = 2AI(OH), + 2H4Si04 -9.4 2
' Morel and Hering (1993).
' Stumm and Morgan (1981).
Matthess (1782).
 12.3 Complexation Reactions 263
When other ions are present in a solution, the solubility Stability of Complexes
of a solid can be different from its value in pure water. and Speciation Modeling
Solubility increases due to solution nonidealities and de-
creases due to the common-ion effect. The previous chap- Most inorganic reactions involving complexes are kinet-
ter explains this issue of nonidealities in relation to activ- ically fast. Thus, they can be examined quantitatively
ity models. Generally, the solubility of a solid increases using equilibrium concepts. For example, the mass-law
with increasing ionic strength, because other ions in equations for Eq. 12.30 and Eq. 12.31 are
solution reduce the activity of the ions involved in the
dissolution -precipitation reaction.
The common-ion effect occurs when a solution already and
contains the same ions that will be released when the
solid dissolves. The presence of the common ion means
that less solid is able to dissolve before the solution
reaches saturation with respect t o that mineral. Thus, [CWX-N 1
the solubility of a solid is less in a solution containing a KG-~oH,~
=
[Cr(OH)'*] [OH-]
common ion than it would be in water alone.
and so on.
Calculation of the distribution of metals among various
complexes involves the solution of a series of mass-law
-
12.3 Complexation Reactions equations. We write the reactions in terms of the formil-
With the explosion of knowledge in contaminant hydro- tion of the complex from some combination of the metal
geology, the prevailing view about the importance of and an appropriate number of ligands. For example, writ-
complexation reactions has changed. For example, com- ing Eq. 12.31 as a series of association reactions gives
plexation facilitates the transport of potentially toxic met- the series of reactions and mass-law equations in Table
als such as cadmium, chromium, copper, lead, uranium, 12.5. These socalled mononuclear complexation reac-
or plutonium. Such reactions also influence some types tions have this general form
of surface reactions. Thus, a process that was mainly of
academic importance now has taken on practical signifi-
cance in contamination problems.
A complex is an ion that forms by combining simpler where M is a metal, L is a ligand, and H is a hydrogen ion.
cations, anions, and sometimes molecules. The cation or The stability constants, p,, for the reaction determine
central atom is typically one of the large number of metals the strength of the complex. Large values of @, are associ-
making up the periodic table. The anions, often called ated with stronger or more stable complexes. Stability
ligands, include many of the common inorganic species constants are related to the equilibrium constants for the
found in ground water such as CI-, F-, Br-, SO:-, series of reactions. Morel and Hering (1993, p. 332) have
PO:-, and C0:- . The ligand might also comprise various tabulated stability constants for a host of metal-ligancl re-
organic molecules such as amino acids. actions.
The simplest complexation reaction involves the com-
bination of a metal and ligand such as Mn" and CI-
Table 12.5 Association Reactions and
as follows:
Mass-LawExpressions Describing the
Hydrolysis of Cr
Mn" + CI- = MnCI' (12.30)
Mms-Law
A more complicated manifestation of complexation is the Equation Expressions
reaction series that forms when complexes themselves
combine with ligands. An example is the hydrolysis reac-
tion of Cr'+

Cr" + OH- = Cr(OH)?'

Cr(<)H)L' + OH- = Cr(OH)-; (12.31)
Cr(OH)t + OH- = Cr(0H):

and so on. The metal is distributed among at least three

complexes. Such series involve not only hydrolysis reac-
tions but other ligands such as CI-, F-, and Br-.
2
x Cbapter 12 / Cbemical Reactions

In some cases, polynuclear complexes form. This com-

plex is characterized by more than one metal (for exam-
ple, Cr3(OH)7+ or Cu,(OH):'). However, because these
complexes are usually rare and difficult to include in
speciation models, they are not often treated.
The stability constants for a series of metal ligand reac-
tions provide the basic information necessary to deter-
mine how the total concentration of a metal in solution
(M), is distributed as a metal ion and various complexes.
The following example illustrates the speciation calcula-
tion when Cr hydrolyzes.
PH
Figure 22.4 Log C-pH diagram for chromium hydroxide
Example 12.4
complexes.
A solution with a pH of 5.0 contains a trace quantity of
Cr {(Cr), = M}. Determine the speciation for Cr
among the various hydroxy complexes and Cr3+.Stability
constants (as base 10 logarithms) are 10.0, 18.3,and 24.0
for CI-(OH)~+,Cr(OH):, and Cr(OH)t, respectively. As- In Example 12.4,the speciation of Cr is controlled by
sume that the pH does not change with the addition of the solution pH. The log C-pH pfot for (Cr)T = (Figure
metal, no solids form, and the solution behaves ideally. 12.4) shows that at a pH below 4, C P is the dominant
The mole balance equation for (Cr), is Cr species. As the pH increases, the various hydroxy
complexes dominate.
(Cr), = (Cr3') + (Cr(0H)") + (Cr(0H):) + (Cr(0H)a This use of stability constants to determine the specia-
Substitution of the appropriate mass-law equations for tion of metals can be extended to a mixed group of
the association reactions into this equation gives mononuclear complexes. For example, if trace quantities
of Pb are present in a ground water, a mixed group of
(Cr), = (CP"') + &(C$+)(OH-) + f12(C6")(OH-)2 chloride, hydroxy, and carbonate complexes can form.
+ P3(Cr3+)(OH-)j Speciation is determined by taking the appropriate mole-
balance equation
(Cr), = (Cr)5+{1 + &(OH-) + &(OH-)2 + &(OH-)'}
(Pb)f = (Pb") + (PbCla + (PbC1;)
Because pH is fixed, (OH-) = M and all the terms (1 2.33)
in the brackets are known. Substitution of the known + (PbOH+)+ (PbC04
values solving for (CP') gives and substituting the appropriate mass-law equations. This
10-5 approach requires that metal concentration in the ground
((3')
=
1 + iolyio-4 + ioyio-IH) + 1024(10-27) water be sufficiently small so as not to influence the
major ion chemistry or pH of the solution. At some point,
-
- 10-5 however, adding large quantities of metal affects the pH
1 + +
10 1003 + 10-5 and carbonate chemistry. The speciation becomes much
= 10-6'ZM more difficult to deterniine in this case because the sys-
tem of mole-balance and mass-law equations is larger
The concentration of the complexes is calculated by sub- and more complex. An example of this more complex
stituting known values of (CP') and (OH-) in the mass- situation is the case where the chemistry of a ground
law equations water changes when the most abundant ions, Ca", Mg2+,
Na', K', and H', complex with the common ligands
(Cr(0H)") = P,(Cr5+)(OH-) SO:-, Cl-, HCO; , and OH-.
= 1O'O."( 10-6 '2)( 10-9
= 10-512~ Major Ion Compbxation and
Equilibrium Calculations
(Cr(0H):) = M
So far, in calculating mineral saturations, we have not
(Cr(OH)% = M considered that the free ion or ligand concentration is
less than the total concentration when complexes form.
Thus, most of the Cr occurs as Cr(OH)2+and Cr(0H):.
Only about 7.5% of (Cr), is the free ion C P . The calculated U P / K ratios for some minerals can de-
crease depending on the extent of complex formation.
 12.3 Complexation Reactions 265
Table 12.6 Calculated M o l a r Concentrations of both phosphate and carbonate complexes are imponant.
Complexes Formed from Major Ion Species (Based on This extensive complexing, especially over the neutral
Data from Example 11.2) to alkaline pH range, increases the solubility of some
LIgand uranium minerals by several orders of magnitude. Such
increases in solubility apparently are necessary to form
Metal HCO; c0:- so:- uranium ore deposits.
c*1+ 5.46 x 10-5 8.03 x lo-" 2.16 x lo-'
MgL+ 4.98 X lo-' 1.46 X lo-- 2.78 X lo-- Organic Complexation
Na + 1.54 x lo-' 1.13 x lo-' 6 . M x lo-'
Complexation can also involve organic ligands present
naturally in ground water or added as contaminants. For
example, at a site near Ottawa, Canada, Killey and others
(1984) showed how up to 80%of the ""Co present as ;I
contaminant in ground water occurred as weakly anionic
complexes. These complexes formed with naturally oc-
Let us reexamine Example 11.2 to illustrate this point. curring organic compounds. This and similar work em-
Assume now that the molar concentrations for the ions phasizes the need to describe organic systems quantita-
in Example 1 1.2 are total Concentrations. Table 12.6pres- tively. The task is difficult due to the diversity of organic
ents the concentrations of the major complexes. Even materials existing in water and problems in identifying
though the concentrations of the complexes are rela- them in laboratory analyses. However, it may be essential.
tively small, the reduction in [Ca"] and [CO<-]due to Killey and others (1984) indicate that concentrations of
complexation reduces the IAP/K ratio for calcite from organic matter as low as 2 mg/L can increase contanii-
1.34 to 1.03. For dolomite the ratio decreases from 0.018 nant mobility.
to 0.01 1. This example shows that even in relatively fresh We will restrict the overview of organic complexation
ground water (ionic strength = 0.023 M), the error in of metals to specific ligand groups including humic sub-
determining mineral saturation can be substantial when stances, artificial complexing agents such as nitrilotria-
complexation is not considered. This error can become cetic acid (NTA) or ethylenediaminetetraacetic acid
even larger as the water becomes more saline. In seawa- (EDTA), and specific compounds such as amino or car-
ter, for example, only approximately 40% of (SO& exists boxylic acids. The term huitzic szr6stunce (Section 1 1.7)
as SO:-, while the remainder exists as NaSO; (37%) and refers to a group of organic acids, which constitute the
MgSO;' (19%)(Morel and Hering, 1993).

Enhancing the Mobility of Metals

In general, metals in ground water are most mobile in
water with a low pH where most of the mass occurs as
a charged metal ion. Ignoring for a moment the effects
of sorption, mobility begins to decline once pH increases
to the point where equilibrium is reached with respect
to a solid phase, usually a metal-hydroxide, metal-
carbonate, or metal-sulfide. At this point, equilibrium
with the solid determines that most of the metal will be
associated with the solid phase.
Over the p H range common to most natural ground
waters, the concentration of most metals is small, re-
flecting this control. Complexes in these circumstances 2 4 6 8 10
PH
can enhance the solubility of metals. The transport of
uranium in ground water is a good example. Figure 12.5 Figure 22.5 Distribution of uranyl complexes vs. pH for
shows the relative abundance of various uranyl com- some typical ligand concentrations in ground waters of the
Wind River Formation at 2%. Pco2 = 10'" atm. ZF = 0 . 3
plexes in ground water with a composition typical of
ppm, Xcl = 10 ppm, XSO, = 100 ppm. 9 0 , = 0 . 1 ppm,
the Wind River Formation in Wyoming (Langmuir, 1978). CSiOL= 30 ppm (from Langmuir, 1978). Reprinted from D.
Across the entire range of pH represented on the figure, a Langmuir, Uranium solution-mineral equilibria at low temper-
significant proportion of the total uranium is complexed. atures with applications to sedimentary ore deposits, Copy
Below a pH of about 4, a uranyl-fluoride complex is the right 1978, p. 547-569, with kind perniission from Elsevier
most dominant species (Figure 12.5). In the range of Science Ltd., The Boulevard, Langford Lane, Kidlington OX 5
typical ground water for this formation (pH 6.6-8.3), lGH, UK.
-
266 Cbapter I 2 / Cbemical Reactions

most abundant fraction of naturally occurring organic on the S versus C plot (Figure 12.6). By repeating the
matter dissolved in water. Complexation of humic sub- procedure at the same temperature (hence, isotherm)
stances is usually modeled as a simple mixture of one or with different values of Ci,the family of points forms a
two “type” ligands. These could include well-character- sorption isotherni (Figure 12.6). This experiment is
ized components such as fulvic acid and humic acid or known as a batch test.
ideal classes of ligands within the humic acid group. Real isotherms have no prescribed shape. They can be
NTA and EDTA are used in detergents and cleaning linear, concave, convex, or a complex combination of
agents. By virtue of their strong complexing capability, all these shapes. Sorption is modeled by fitting an experi-
these two organic compounds are useful in studying mentally derived isotherm to theoretical equations. Two
metal-organic complexation. It is unlikely that these li- of the most common relationships are the Freundlich isu-
gands would be encountered in ground water, except therm:
perhaps in problems related to waste-water disposal into
the subsurface. However, as Morel and Hering (1993) S = KC” (12.35)
point out, these ligands are important end-members ap- and the Langmuir isotherm:
proximating the behavior of the most powerful ligands
one would likely encounter in nature. S=- Q’KC
A variety of other naturally occurring organic com- (12.36)
1+KC
pounds such as amino acids and carboxylic acids have
significant complexing capabilities. However, the most where K is a partition coefficient reflecting the extent
common amino acids, serine, glycine, and aspartic acid, of sorption, n is another constant usually ranging be-
are not present in sufficient concentrations to influence tween 0.7 and 1.2, and Q” is the maximum sorptive
metal transport in ground waters. In deep ground-water capacity for the surface.
systems, the concentration of organic compounds can be Figure 12.7 depicts Freundlich isotherms calculated
relatively large. For example, carboxylic acids, forming as with a K value of 1.5 and values of n of 0.5, 1 .O, and 1.5,
organic matter undergo diagenesis in a geologic basin, and Langmuir isotherms with values of K of 0.5 and 1.5
can reach concentrations as high as 10,000 ppm (Surdam and a Q“value of 30 mg/g. These examples illustrate the
and others, 1984). The presence of these acids could be range in curve shapes that can be fitted with these two
important in enhancing the solubility of aluminum in equations. If the fit is not satisfactory, there are other
formation waters. equations.
A Freundlich isotherm with n = 1 is a special case
because this linear isotherm is easy to incorporate into
-
12.4 Reactdons on Surfaces mass-transport models. The following equation relates S
to c:
Reactions between solutes and the surfaces of solids play
an important role in controlling the chemistry of ground S = K,C (12.37)
water. In natural systems, these reactions can completely
where Kd with units like cm’/g is the distribution coeffi-
change the cation chemistry, particularly in the case of
cient (i.e., the slope of the linear sorption isotherm).
the water-softening reactions. When contaminants are
Increasing values of K,, are indicative of a greater ten-
being transported, they can retard the spread of some
dency for sorption.
constituents or virtually immobilize them.

Sorption Isotherms
When water containing a trace constituent with a con-
centration Ci is mixed with granular solids and allowed
to equilibrate, mass often partitions between the solution
and the solid. The following equation represents this
process:
(Ci - C)(solution volume)
S= (12.34)
sm
where the equilibrium concentration, C, and the starting
concentration, Ci,have units such as mg/L or pg/L; s m r
is the sediment mass (for example, g); and S is the quan- Equilibrium Concentration C
tity of mass sorbed on the surface (for example, mg/g or Figure 22.6 Exdmple of a simple sorption isotherm con-
pg/g). Such an experiment provides the single point a structed using data points from batch tests.
 12.4 Reactjotis 011 Surfaces 267

8olI
s = 1.5 C1.5 - K,,= slope
140 -
-
70 120 -
-
s = 1.5 c' 0 100 -
-
-h'* 80-

60
-
-
30 x 1.5C
-
s = ____
1 + 1.5C 40 -
-
20 -
r
20/ I I I I I I I I I I
10 I 0 5 10 15 20 25
C (jlg/L)
I I I
0
10 20 30 40
C (milligrams per liter) S is plotted versus C, as shown in the figure to establish
Ffgure 12.7 Example of Langrnuir and Freundlich iso the isotherm. K,, is the slope of the isotherni. The calcikt-
therms. tion illustrates a strong affinity for the solid by pyrene.

Hydropbobic Sorption of Several studies (for example, Karickhoff and' others,

Organic Compounds 1979; Schwarzenbach and Westall, 1981) have shown
The K,,-based approach for modeling sorption is most how the distribution coefficient can be expressed as the
commonly applied to hydrophobic (or water-hating) or- product of constants describing the contaminant and the
ganic molecules as they sorb onto solid particles. These porous medium, or
mainly nonpolar molecules tend to partition preferen-
tially into nonpolar environments provided by small
quantities of solid organic matter (for example, humic
where K,,, is the partition coefficient of a compound
substances and kerogen). Organic matter can occur as
between organic carbon and water with typical units
discrete solids, as films on individual grains, or as stringers
such as cm3/g, andf,. is the weight fraction of organic
of organic material in grains. Overall, the more hydropho-
carbon (dimensionless), defined as g,,./g, or grams of solid
bic a compound is, the greater is its tendency to partition
organic carbon to grams of total aquifer solids.
into a solid phase. The following example illustrates the
When values ofJK and K,. are known. the K , , value
results of batch experiments with pyrene.
determined from Eq. 12.38 can be used to predict the
extent of partitioning. Organic carbon contents can be
Example 12.5 measured in the laboratory on porous-medium samples.
Table E l 2.5 summarizes data collected from batch sorp- However, this parameter is not well characterized. The
tion experiments with pyrene using 25 mg of porous range of values reported for geologic materials is e x
media in 50 mL of solution. Construct a sorption isotherm tremely variable (Table 12.7). In estimating K,,,., it is fortu-
and calculate the distribution coefficient in an3&. nate that a good correlation exists between log K,,, and
log K,,,,., the octanol- water partition coefficient. This cor-
(Ci - C ) X 0.050 pg/L. L relation is expected because the partitioning of an or-
S= , units + ~

0.025 g ganic compound between water and organic carbon is

not much different than between water and octanol.
Regression equations in practice describe the relation-
Table E12.S ship between K,,,. and K ',,,. (Griffin and Ro!., 1985).

Karickhoff and others (1979): log K,,,

1 19.4 5.9 27.0
2 39.7 9.5 60.4
3 60.9 14.1 93.6
4 79.1 20.7 1 17.8 Schwarzenbach and Westall (1981): log K,,,
5 100.3 26.1 148.2
= 0.49 + 0.72 log K,,,, (12.40)
 Next Page

268 Chapter 12 / Chemical Reactions

Table 12.7 A Synthesis of Some Data on the Organic Carbon Content of Sediments

WPe of Organic Carbon

Site Name Deposit Texture Content

Borden, Ontario' Glaciofluvial Fine-medium sand 0.0002

Gloucester, Ontario' Glaciofluvial Sands and gravels 0.0006
North Bay, Ontario' Glaciofluvial Medium sand 0.0001 7
Woolwich, Ontario' Glaciofluvial Fine-medium sand 0.00023
Chalk River, Ontario' Glaciofluvial Fine sand 0.00026
Cambridge, Ontario' Glaciofluvial Medium sand 0.00065
Rodney, Ontario' Glaciofluvial Fine sand 0.00102
Wildwood, Ontario' Lacustrine Silt 0.00108
Palo Alto, Baylandd ? Silty sand 0.01
River Glatt, Switzerland' Glaciofluvial Sand, gravel <o.ooo-0.01
Oconee River, Georgia6 River sediment Sand, 0.0057
coarse silt, 0.029
medium silt, 0.02
fine silt 0.0226
~~ ~~ ~~~

I Mackay and others (1986).

Jackson (personal communication, 1989)).
'J. Baker, University of Waterloo (personal cornmunicarion, 1987).
Mackay and Vogel(1985).
' Schwarzenbach and Giger (1985).
Karickhoff (1981).

Hassett and others (1983): log K,,, pounds (for example, methanol). Such compounds have
= o'088 + 0'909 log Kf'li' ('2'41) large aqueous solubilities and relatively low values of log
K,,,,.(for example, -0.66 for methanol). In contrast to the
where K O ,is a tabulated constant (dimensionless), and nonpolar compounds, these are preferentially partitioned
K , has units of cm3/g.The similarity in these equations is to the aqueous phase with much less affinityfor organic
evident in Figure 12.8. Overall, they provide a preliminary carbon. Karickhoff (1984)recommends that the model
estimate of K , when other information is lacking. In field
studies, K, values estimated from these equations need
to be refined with field or laboratory experiments.
The following example illustrates the application of
the empirical equations in estimating a distribution coef-
ficient. 4 Hassett et al. (1983)

Example 12.6 Schwarzenbac h

and Westall (1981)
An aquifer has a n f x of 0.01.Estimate the Kd value charac- r' -
terizing the sorption of 1,Zdichloroethane having a log -3
K O ,= 1.48.
Starting with the basic equation Kd = f,K,* and taking 1

logs of both sides gives

log K d = log& + log K , , I
Substitution for log K , with the Schwanenbach and
Westall (1981)equation yields 1 2 3 4

log Kd = log 0.01 + 0.49 + 0.72(1.48)= -0.445 Flgure 12.8 Correlation between the log octdnol-wdter
or 0.36 cm3/g.
The Kd value is 10-".4'5 partition coefficient (K,,,,.)
and the log organic carbon/water
partition coefficient as determined by three different studies
(modified from Griffin and Roy, 1985). Reprinted by permis-
It is tempting to extend this model Of organic sorption sion of the Environmental Instit. for Waste Management Stud-
to more-polar and lesshydrophobic (hydrophilic) com- ies, The Univ. of Alabama.
Previous Page

12.4 Reactions on Surfaces 263

of hydrophobic sorption should only be used for com- have developed an empirical equation to estimate the
pounds with solubilities less than lo-' M. critical organic fraction with KO,.and surface area as
When organic molecules carry a charge, hydrophobic controlling parameters.
sorption represents only part of the total sorption. Elec- There are other limitations in the model of hydropho-
trostatic forces caused by interaction with charged sur- bic sorption. A linear isotherm does not describe all sorp-
faces are also important. Organic acids and bases, for tion especially at large concentrations. Commonly, non-
example, ionize to some extent depending on the pH of linear forms of the Freundlich or other isotherms best
the ground water. Above a specified pH, a significant describe experimental data, especially when the concen-
proportion of the compound will occur in ionic form. tration of the organic compounds is relatively high. An-
The pH range over which ionization is significant de- other problem relates to the question of desorption, the
pends on the equilibrium constant for the reaction (pK,, partitioning of organic compounds back into the aqueous
or pK3. For example, with the following acid and base phase. Several studies show that isotherms for desorption
dissociation reactions may be quite different from the comparable ones for
sorption. This failure for reversible processes to behave
HA = Ht + A- BOH = B+ 4- OH- in the same way is termed hysteresis. Often, the hysteresis
the organic acid dissociates to a negatively charged com- results in an isotherm that during desorption exhibits a
pound (A-), and the base to a positively charged com- greater affinity for partitioning to the solid than was the
pound (B+). The extent of acid or base dissociation is case during sorption. Other limitations of the model in-
defined in terms of the ionization fractions (a and p) clude problems of competition between compounds and
where kinetic effects in sorption -desorption reactions.

and
' = [B']
[B'l
+ [BOH] (12.42)
Kd-basedApproacbes for Modeling the
Sorption of Metals
By substitution of the mass action equations
KKbased models for sorption have also been applied to
characterize the sorption of metals on mineral surfaces.
One of the most concerted efforts in this respect has
Eq. 12.42 becomes been in the evaluation of various sites as potential host
rocks for the disposal of nuclear wastes. The extent to
which sorption retards the spread of radionuclides has
an extremely important bearing on the suitability of a
given site.
These expressions provide a convenient way to deter-
Moody (1982) discusses the myriad difficulties in ob-
mine the ionization fractions for the organic acid or base.
taining consistent K,, values from experiments. The prob-
For example, in the case of organic acids, HA is sorbed
lem is that the metal-sorption processes are far too com-
but A- moves through the system without retardation.
plex to be accurately represented by a simple one-
Thus, a large ionization fraction (say > 0.5) implies a
parameter model. Thus, in spite of appealing and appar-
deterioration in the ability for hydrophobic sorption to
ent simplicity, one should be cautious in using the simple
retard contaminant spread. In the case of organic bases,
Kd approach in describing the sorption of metals. These
BOH is sorbed and B+ can undergo cation exchange.
limitations have promoted more sophisticated ap-
Thus, the effect of ionization on transport is more compli-
proaches.
cated for a base than for an acid.
Another instance where the hydrophobic model of
sorption begins to break down is with low fw values.
Multiparameter Equilibrium Models
Organic compounds can sorb to a small but significant
extent on inorganic surfaces. This sorption is negligible A more rigorous framework for surface reactions exists
in media with a large fw. At the critical level of organic that can account for properties of the solution and the
matter (McCarty and others, l981), the sorption due to solid surfaces. To illustrate this approach, we will exam-
organic and inorganic solids is equal. The critical level ine classical cation exchange reactions on clay minerals
of organic matter is not constant but a function of KO". and trace metal sorption on variably charged surfaces.
and the surface area of inorganic solids. Thus, for organic Cation exchange is the most well known of the surface
compounds with a high affinity for organic carbon (high reactions. Driving the reaction is the electrostatic at-
KO,,), the critical level is smaller than it would be for less traction between charged cations in solution and the
readily sorbed compounds. Similarly, the critical level surface charge on clay minerals or other oxide surfaces.
will be greater for a medium with a larger surface area Clay minerals in particular have a significant fixed nega-
(i.e., a large clay content). McCarty and others (1981) tive surface charge because of substitutions of cations of
270 Cbapter 12 / Chemical Reactfons

a lower valence in the lattice, and to a lesser extent The general form of a cation exchange reaction is
because of broken bonds at the edges of the mineral tzMX + mN"+ = nM"'+ + mNX (12.43)
(Table 12.8). Kaolinite differs from the rest of the clay
minerals. It has a variable charge as hydroxylated edges where M and N are metal cations with charges m + and
change their charge as a function of pH. n + , respectively, and MX and NX represent the metals
Cations binding to the exchange sites balance the nega- sorbed on the solid phase. The mass-law expression for
tive surface charges of the mineral. These cations can this reaction is
exchange with other cations present in solution. The [M I J l + ] l I [M(]Ill
term cution exchange cupucity describes the quantity of K= ( I 2.44)
[N"']'"[MX]"
exchangeable cations sorbed on the surface or indirectly
the negative charge of the soil (Yong, 1985). Cation ex- where the bracketed quantities are the activities of spe-
change capacity (CEC) has units of milliequivalents per cies in solution and on the exchanger. As with ions in
100 g of sample, and for the fixed-charge clay minerals solution, activity models are required for ions on the
is an important descriptive parameter. Values of CEC exchanger. The most common model assumes that the
vary considerably depending on the type of clay mineral activities of adsorbed species are equal to their mole
(Table 12.8). Illite and chlorite fall at the lower end of fraction, for example [NX]"' = xYx, and [MX]" = xtX,
the range of observed values while the vermiculite and where the mole fraction of NX is
smectite groups are at the upper end of the range. Much
of the variability in CEC is a function of the surface area (N*)
XhX =
of the exchanger (Table 12.8). (N*> + (M")
Clay minerals exhibit a preference as to which ion with (N*) and (M') representing the molar concentration
occupies an exchange site. Attempts have been made of species sorbed on the solid. With an appropriate activ-
over the years to establish a selectivity sequence, where ity model, the mass-law expression is
equivalent amounts of cations are arranged according to
their relative affinity for an exchange site. Many variations (12.45)
of this sequence have been published, such as the follow-
ing one proposed by Yong (1985): where K,is now known as a selectivity coefficient. Once
Li+ < Na+ < H+ < K' < NH; < Mg" < CaL' < Alj' an activity model is assumed for the adsorbed species,
the equilibrium constant and the selectivity coefficient
In general, the greater the charge on a cation, the greater are no longer equal.
the affinity for an exchange site. Although the idea of an As an example of how this theory is applied, consider
affinity sequence provides a useful concept for ad- the following exchange reaction between an ion in solu-
dressing exchange preference, it is too simplified to be tion P' and an ion Q' sorbed on the surface as Q-clay:
applied rigorously to evaluate exchange processes (Sposi-
tio, 1984). P+ + Qclay = Q' + Pclay

Table 12.8 Properties of the Common Clay Minerals

Surface
Lattice CEC Area Source of Charge
Clay Mineral Description (meq/lOO g) (m*/g) Charge Characteristics

Kaolinites 1 : 1, strong 5-15 15 Edges, broken bonds Variable charge

H bonds (hydroxyhted edges)
lllites 1 : 2, strong 25 80 Isomorphous substitution, Mostly fixed charge
K bonds some broken bonds at
edges
Chlorites 2: 2, strong 10-40 80 Isomorphous substitution Fixed charge
bonds
Vermiculites 1 : 2, weak 100- 150 n.d. Isomorphous substitution Fixed charge
Mg bonds
Montmorillonites 1 : 2, very 80- 100 800 Isomorphous substitution, Mostly fixed charge
weak bonds some broken bonds at
edges
n.d.-not determined.
From Yong (1985). Reprinted by permission of Second <:anAdian/Americm Conference on Hydrogeology. Copyright 0 1985. All rights reserved.
 12.4 Reactions on Suvaces 271

The mass law expression assuming that concentration is tively charged at low pHs. However, at higher pHs, the
equal to activity is surface becomes negatively charged and functions as a
cation exchanger. There is a pH in between, known as
(12.46) the zero point of charge (pH,), where the surface has
zero net charge. The term isoelectric point defines the
where P' and Q' are molar concentrations and xPclr). and pH, when the binding and dissociation of protons (H')
xyr.lay
are mole fractions. are the only reactions affecting surface charge. These
We can manipulate Eq. 12.46 further to learn more surface-charge relationships, unique for given solids and
about the distribution coefficient, K d . Assume that solutions, usually must be determined experimentally.
(P') << (Q'). and that no other ions are present in Tabulated in Table 12.9 are estimates of the isoelectric
solution. Because both ions are singly charged, the molar points for various solids.
concentrations and equivalent concentrations of the The variable surface charge is explained by the pres-
sorbed species are the same so that CEC = (P') + (Q') ence on the surface of species such as XOH, XO-, and
-
and because (P') is small CEC (4') and 7 = (P') + XOH: , where XOH represents the hydroxylated surface
(Q') - (Q'). If P' and Q' are the only ions in solution, and XO- the ionized surface site (Morel and Hering,
= (P)/P+ Q'), andxwlay= (Q*)/(P' + Q'). Substi-
xpr.,ay 1993). The following set of equilibrium reactions de-
tution of the approximations into Eq. 12.46 gives scribes the concentrations of these surface species:
XOH = H' + XO- XOH + H' = XOH;
(12.47)
These equations illustrate how the species composing
the surface depend on the pH of the ground water. At
Given that Kd = (P')/(P+), Eq. 12.47 becomes
low pH, XOH; is the dominant surface species, while at
K,CEC high pH XO- is dominant. The surface species are free
K,, = - (12.48) to react with metals or metal complexes. For example,
7
the sorption of a metal ion (M") is described by an
Thus, Kd is a function of both the properties of the equilibrium reaction similar to the preceding one or
exchanger and the solution. Values are not unique con-
stants and must be determined for particular ions on the M" + XO- = XOM'
exchanger as a function of differing solution concentra- By solving the resulting system of equilibrium reactions,
tions. This is one of the main weaknesses in the K,/ ap one can calculate the concentration of the metal on the
proach for modeling sorption. surface (XOM'). As Morel and Hering (1993) show, this
One important example of cation exchange is the natu- or other surface species bind the metal or metal com-
ral water-softening reactions. Ground water with rela- plexes. Choosing the appropriate species to include usu-
tively high concentrations of Ca" and Mgz+(hard water) ally involves matching experimental data to theoretical
and other minor species such as Fe" changes its chemical models. This model for exchange is sufficiently general
composition as it moves into marine clay or shale with to account for the sorption of metal complexes.
Na' on the exchange sites. The following reaction de- Stollenwerk (1991) studied the sorption of molybdate
scribes this change (MOO:-) on ferrihydrite, which coats quartz grains of an
outwash aquifer. A variety of reactions involving the
CaL' Ca
Mg" + 2Na-clay = 2Na' + Mg-clay
Fez+ Fe
Table 12.9 Examples of the lsoelectric
Selectivity coefficients for these reactions are such that Points (pH) for Various Solids
in clay-rich sediments, the ground water is softened with
nearly all of the Ca2' and MgL+replaced by Na'. Iron, Solid pH Source
which is often associated with problems of taste and Quartz a-SiO, 2-3.5 I
staining, is also removed. Chapter 15 will present more Albite NaAISilOn 2.0 I
specific examples of how this process affects the compo- Kaolinite AI(Si40,,,XOH), <?-4.6 1
sition of natural ground waters. Montmorillonite 52.5 1
Another important type of sorption reaction involves Hematite Fe,03 5-9 2
solids whose surface charges change as a function of Magnetite Fe,O, 6.5 -9
ground-water composition. Examples of such solids in- Goethite FeOOH 6--l 2
Corundum A120, 9. I 2
clude kaolinite, metal oxides (for example, S O L ,Alz03), -9 9
Gibbsite AI(OH),
and metal oxyhydroxides [for example, AI(OH),,
Si(0H)J. ' Murrdy and Parks (1980).
Typically, the surface on one of these solids is posi- Drever (1988).
-
272 Chapter 12 / Chemical Reactions

100, I with MOO:- in sorption reactions. Figure 12.10 shows

how the increasing concentration of phosphate signifi-
cantly reduces the percentage of molybdate sorbed on
the surface. Sulfate, however, affects molybdate sorption
to a much smaller extent. As implied in the figure, the
sorptive behavior of the surface can be modeled well
with a series of surface complexation reactions.
It is feasible to model surface reactions in a realistic
way. However, the mathematical approaches are com-
plex and require considerable data. Such models are just
now being incorporated into state-of-the-artcontaminant
transport models and, as Stollenwerk's study shows, are
being evaluated in relation to actual field problems.

PH
Figure 12.9 Variation in molybdate sorption as a function
-
12.5 Oxidation-Reduction Reactions
of pH. The circles present data from actual measurements. Oxidation-reduction or redox reactions are unlike any
The line presents the results of model simuiations. The differ- of the other reactions so far because they are mediated
ence in the results is due to the increased competition by microorganisms. The microorganisms act as catalysts
caused by elevated SO:- and POi- present in the water from speeding up what otherwise are extremely sluggish reac-
the deeper well F347-46 (from Stollenwerk, 1991).
tions. Microorganisms, mainly bacteria, occur ubiqui-
tously in the subsurface. While they can be found in
ground water, they usually form colonies attached as
films on the porous medium or fracture surfaces. Microor-
{S}FeOHocomplexation site causes the sorptive proper- ganisms use redox reactions as a source of energy. In
ties of the aquifer to change as a function of pH. At pHs spite of the help from microorganisms, redox reactions
below about 7, the ferrihydrite coatings are capable of are slow in relation to other reactions and typically are
sorbing molybdate (Figure 12.9). In this pH range, below
treated from a kinetic viewpoint.
the pHZwfor ferrihydrite, the surface has a net positive
charge. At very lower pHs, almost 100% of the total
molybdate ends up sorbed on the surface. Once the pH Oxidation Numbers, Half-Reactions, Electron
climbs above the pH,, the surface has a net negative Activity, and Redox Potential
charge and loses its capability of sorbing molybdate.
Stollenwerk's experiments also showed how other Our overview begins with an explanation of the oxida-
anions like phosphate (Po$-) and sulfate (SO:-) compete tion number. To take into account bond polarity, charges
defining the oxidation number are assigned to a com-
pound. For example, in the case of C O : - , the oxidation
number for carbon is (+ IV) and for oxygen (- II), giving
1001 f the molecule a net negative charge of -2. By convention,

I m Sulfate Exoerimental
F347-46
DH = 6.2
Mo(VI) = 0.05millimoles
I1 oxidation numbers are written as roman numerals.
The term oxidution refers to the removal of electrons

t
t
-A Phosphate'ExperimentaI
Model Simulations
per liter
Ferrihydrite = 0.015% from an atom, forcing a change in the oxidation number
of an element. For example, the following reaction
Fezt = Fe4+ + e-

I
6o

describes the oxidation of Fez', where e- is an electron.

In this reaction, the oxidation number for iron changes
from (+ID to (+IID.
Reduction refers to the addition of an electron to lower
the oxidation number:

1.2 Fe3+ + e- = Fez'

Sulfate or Phosphate Concentration (millimoles per liter) All redox reactions transfer electrons and involve ele-
Figure 22.20 Measured and simulated variations in molyb- ments with more than one oxidation number. Listed in
date sorption as a function of competition from Table 12.10 are some important elements, their typical
SO:- and PO:- (from Stollenwerk, 1991). oxidation states, and a few of the ions and solids that
 12.5 Oxidatfon-ReductfonReactions 2 3

Table 12.10 Some Elements Found with Adding these two reactions together gives the original
More Than One Oxidation State and reaction Eq. 12.50. Half-reactions are unique in that elec-
Examples of Ions or Solids Formed from trons are reactants or products. Otherwise, these reac-
Those Elements tions are similar to other equilibrium reactions.
Element and Mass-law equations can be written in terms of the
Oxidate State in Brackets Example concentration or activities of reactants and products (in-
cluding the electrons) and an appropriate equilibrium
constant. For example, the masslaw expression for the
half-reaction

Ox + ne- = Red

is

(12.52)

Rearrangement of Eq. 12.52 gives the electron activity

[e-I for a half-reaction:

(12.53)

In the same way that pH defines [H'], pe defines electron

From Morel, F. M. M., and J. G. Hering, 1993. Princi- activity, [e-1. Rewriting Eq. 12.53by taking the logarithm
ples arid Applications of Aquatic Chemistry. Copy- of both sides gives
right @ 1993 by John Wiley & Sons. Inc. Reprinted
by permission of John Wiley & Sons, Inc.
pe = -1ogIe-1 = (:){log K - log -

When a half-reaction is written in terms of a single elec-

form. A host of trace metals not included on the rable tron or II = 1, the log K term is written as pe" so that
also have variable oxidation numbers.
There is a direct analogy between oxidation- reduction
pe = pe" - log- [Redl (12.54)
reactions and the acid- base reactions studied in Section [Ox1
12.1. Instead of a transfer of protons from an acid to a
base, there is a transfer of electrons from a reductant Morel and Hering (1993, p. 430) and Pytkowicz (1983,
(electron donor) to an oxidant (electron acceptor). N o p. 249) tabulate values of pe" or log K. Some of the
free electrons result from a redox reaction because each important half-reactions together with values of pe" are
complete reaction involves a pair of redox reactions presented in Table 12.11.
Another way of characterizing redox conditions is in
Ox, + Red, = Red, + Oxl (12.49) terms of EH, the redox potential. Values of EH have units
of volts, acknowledging that a redox reaction involves
where Ox and Red refer to oxidants and reductants, the transfer of electrons. EH is related to pe by the follow-
respectively. This equation is analogous to Eq. 12.1, ing equation
which applies to acid- base reactions.
Following is an example of a complete redox reaction:
(1 2.55)
O2 + 4Fe2' + 4H' = 2H10 i-
4Fe"
where F is the Faraday constant defined as the electrical
O(0) Fe( + 11) 0(-11) Fe(+III) (12.50)
charge of one mole of electrons (96,500 Coulombs) with
OX, Redl Red, 0x1 2.3RT/F equal to 0.059 V at 25°C. In practice, pe is taken
to be the calculated value of electron activity and El, the
consisting of two separate half-redox reactions Ox,-Red, measured electrode potential for an electrochemical cell.
and Red,-Ox,: In other words, pe is a calculated quantity and Ell is a
measured one. The following example illustrates how pe
O2 + 4H' + 4e- = 2H20 of a ground water is calculated assuming redox equi-
(12.51)
4FeL+= 4Fei' 4- 4e- librium.
-
274 Cbapter 12 / Cbemical Reactions

Table 12.11 S o m e Important Half-Redox Reactions

pe" = l o g K
Hydrogen
H+ + e- = AH2&) 0.0
Oxygen
&O,(g)+ H' +
e- = lor&)+ iH20 +35.1
iO,(g)+ H' + e- = iH2O + 20.75
&H20,+ H' + e- = HIO + 30.0
(Note also HOE + H' = H20; log K = 11.6)
Nltrogert
NO; +
2H' +
e- = 4N2O8(g)+ H 2 0 +13.6
(Note also N,O&) = 2NO2(g); log K = -0.47)
4NO; + H' + e- = hN0; + & H 2 0 +14.15
(Note also NO; + H' = HNO,; log K = 3.35)
4NO; + 4H' + e- = fNO(g) + qHzO +16.15
+NO; + qH' + e- = AN20(g) + %HIO + 18.9
&NO; + %Ht + e- = hN1(R) + % H 2 0 +21.05
BNO; + qH' + e- = QNH: + $ H 2 0 +14.9
Sulfur
+
iSO$- H' + e- = 4.50:- + $H,O - 1.65
(Note also SO:- + H' = HSO;; log K = 7)
+SO:- $Hi+ +
e- = &,O: + PH,O +4.85
+ +
&SO:- %H' e- = &S:(s. on.) + $HIO +6.03
(Note also S]l(s. ort.) = S: (5. col.); log K = -0.6)
+
&SO:- HH' + e- = AS:- + #HLO +5.40
&SO:- + +
$H+ e- = & - + BH,O +5.w
(Note also S:- + H' = HS; ; log K = 6.1 and
HS; + H' = HZS,; log K = 3.5)
&SO:- +
flHi + e- = &S; + &H,O +5.12
+
(Note also Sj- H+ = HS;; log K = 7.0 and
HS; +
H' = HIS,; log K = 3.8)
&SO:-+ $H' + e- = QH,S(aq) + i H 2 0 +5.13
(Note also HIS(@ = HzS(aq); log K,,= - 1.0, and other
acid- base, coordination, and precipitation reactions)
Trace Metals
Cr
4HCrO; + 4H' + e- = 8Cr" + IHzO +20.2
(Note also HCrO, = ?$2,O:- + iHzO, log K = - 1.5;
HCrO; = H' + CrOs-, log K = -6.5; and various
Cr(II1) precipitation and coordination reactions.)
Mn
bMnO; + IH' + e- = bMn2' + bH20 +25.5
+
4Mn02(s) 2H' + e- = 4Mn" + H,O + 20.8
Fe
Fe" + e = Fe?' +13.0
bFe2+ + e- = $Fe(s) -7.5
&Fe,04(s)+ 4H' + e- = $Fez++ 2 H 2 0 +16.6
co
Co(OH),(s) 3H' + +
e- = Co'" + 3H,O +29.5
iCo,O,(s) + 4H' + e- = &02* + 2 H z 0 +j1.4
cu
Cll?' + e- = CU' +2.6
KUi+ + f?- = tCU(S) +5.7
 12.5 Oxidation-ReductionReactions 2 3

Table 12.11 (Cwitinuerl)

pe" = log K

Trace Metals
Se
$SeO;- + 2H- + e- = $H,SeO, + hHIO + 19.4
tH,SeO, + H' + e- = fSe(s) + $ H 2 0 +12.5
&e(s) + H' + e- = AH& -6.7
(Note also H,Se = Ht + HSe-, log K = -3.9;
H,SeO, = H' +
HSeO;. log K = -2.4;
HSeO; = H' + SeOi-, log K = -7.9;
SeOS- + H' = HSeO; , log K = + 1.7)
43
AgCl(s) + e- = Ag(s) + C1- 13.76
Ag+ + e- = Ag(s) + 13.5
Hg
hHgL+ + e- = BHg(l) + 14.4
Hg2' +
e- = $Hg:+ i-15.4
Pb
dPbOI + 2H' + e- = &Pb'- + H,O +24.6
(Note many other reactions for Mn, Fe, Co, Cu, Se, Ag,
Hg, and Pb)
* From Morel, F. M. M. and J . G . Hcring. 1993. Principles and Applications r~JAyuatic
C/x'mistry. Copyright 0 1993 by John Wiley Sr Sons, Inc. Reprinted by permission of
John Wilry I Sons, Inc.

Example 12.7 small or some of the reactants are not metabolized easily
by microorganisms. Another feature of many redox reac-
A ground water has (Fe") = M and (Fe"') = lo-?') tions is that with extremely large equilibrium constants
M. Calculate the pe at 25°C assuming that the activities they are essentially irreversible. Take for example a reac-
of Fe-species are equal to their concentrations. What tion where an excess of dissolved oxygen oxidizes or-
should be the measured El, of this solution? ganic matter (as CH20):
From Table 12.1 1 , the half-reaction for the reduction
of Fe" to Fez+is CHZO +0 2 + COL + HZ0
Fe3+ + e- = Fez+ with pe" = 13.0 This reaction progresses to the right until all of the or-
ganic matter disappears. This example raises the question
Substitution into Eq. 12.54 gives of how redox phenomena apply to nonequilibrium sys-
tems. A constant value of pe or EHexists when the concen-
(Fe")
pe = pe" - log- tration of one of the couples, OL/H20 in this case, is
(FP) much greater than the other. Because utilization of all of
10-33 the organic matter will require a negligibly small propor-
pe = 13.0 - log- tion of the oxygen, the redox potential defined by the
10-59
OJHzO couple does not change as a function of the
pe = 13.0 - 2.6 = 10.4 reaction progress. Thus, the measured EH may be domi-
nated by one couple, although mixed potentials are the
E,, = -
2.3RT = 0.059 X 10.4 = 0.61 V rule (Stumm and Morgan, 1981).
F
The concept of a dominant couple is useful in under-
standing the redox chemistry of natural and contami-
nated systems. However, research has shown that redox
Kinetics and Dominant Couples
equilibrium among all of the couples does not occur
So far, redox phenomena have been modeled as equilib- (Lindberg and Runnells, 1984). Effectively, pe values
rium processes. The implied assumption is that some of computed from each couple might span a broad range
the half-redox reactions are in equilibrium. However, of values. Thus, while the concept of master pe is alluring,
redox reactions often are slow relative to physical pro- the practice is not recommended (Lindberg and Runnells,
cesses because the number of active microorganisms is 1984). Calculations of pe provide indications of the direc-
276 Chapter 12 / Chemical Reactions

tion in which a system is evolving (Stumm and Morgan,

1981). This general absence of internal equilibrium im-
plies that care must be taken in assessing pe values.
Further, the difficulty in interpreting EH values measured
with electrodes means that a quantitative description of
redox reactions remains elusive for ground water.

Control on the Mobility of Metals

Redox reactions influence the mobility of metal ions in
solution. One example is the behavior of metals in sulfur
systems involving the sulfate/sulfide couple. Assume that
the total sulfur in a system (S,) exists as four possible
species, SO:-, H2S, HS-, and S2-, where in the first sulfur
occurs as S(+VI) and in the last three as S(-II). The
relationship between H2S,HS-, and S2- is fixed by equilib-
rium relationships, much like the carbonate system. In
oxic environments, SO:- is the dominant species of the
couple with essentially none of the S(-II) species pres-
ent. Sulfate ion is reduced in an anoxic environment and
the other three species dominate.
As the sulfur species change with changing redox con-
ditions, so too do the solids that precipitate. When
SO:- dominates, metal concentrations can be relatively 1
high because there are no solubility constraints to force PH
the removal of solids from solution. Once the system Figure 12.11 Example of an EH-pH diagram showing the
becomes reducing, almost all of the metal and sulfur are stability relationships of iron oxides and sulfides (from Ma-
removed from the system as metal-sullides, which are cumber, 1984). Reprinted by permission First Canadian/
usually relatively insoluble. Thus, in one case, the total American Conference on Hydrogeology. Copyright 0 1984.
AII rights reserved.
concentration of the metal is found as a metal ion, and
in the other as a solid.
In a more complex system where pH can change, other
ions are present, and one of several solids could form, it microbially catalyzed. In some cases, oxygen acts as the
would be useful to have a way of presenting the essential electron acceptor:
features of the redox and solid phase chemistry. EH-pH
or pe-pH diagrams serve this purpose nicely, Shown in fCH20 + fO2<g) = fCO2(@ + fHzO
Figure 12.11 is an EH-pH diagram for iron. The upper However, when O2is unavailable, other species, for ex-
and lower lines on the field represent the oxidation of ample NO;, Fe(+III), Mn(+iV), SOf, and C02, accept
water to O2 and the reduction of water to H2. These electrons. A few of the most important of these biotrans-
two half-reactions define the upper and lower limits for formation reactions include
oxidation or reduction. The other lines are boundaries
for what are known as stability fields. These fields are Fe(II1) reduction: fCH20+ Fe(OH), + 2H+
labeled with the one dominant ion or solid for the speci- = fC02(g) + Fe2+ + VHLO
fied EH and pH. When an ion dominates (for example, denitrifkation: +
CH20 #NO.; + $H+
Fe2+), the element is mobile because a relatively small +
= C02(g) %N,+ a H 2 0
proportion of the metal occurs as a solid. In the stability sulfate reduction: CH20 + @O:- + iH+
fields for solid phases (for example, FeS2 and Fe203), = 4HS- + H 2 0 + CO&)
nearly all of the iron will exist as that phase, making the
methane formation: CH20+ &C02(g)
element essentially immobile.
= iCH4 + CO&)

where CH20is a “type” organic compound.

Biotransformation of Organic Compounds
Biotransformation reactions are important when the
A large number of half-reactions involve the oxidation of organic compound is a ground-water contaminant. If bio-
organic compounds into simpler inorganic forms such transformation occurs rapidly, the reaction can attenuate
as C 0 2and H20. These reactions are referred to as biodeg- contaminant concentrations and perhaps alter the major
radation or biotransformation reactions because they are ion chemistry. The following chapter examines the im-
 12.7 Isotopic Processes 277

portant features of microbial systems and how they con- to the concentration of RX. As is the case with many
trol the kinetics of redox reactions. kinetic descriptions of reactions, this kind of simple rate
law oversimplifies what is often a much more compli-
cated relationship (Neeley, 1985).
- Hydrolysis
12.6 Both biotic and abiotic transformation reactions can
operate on organic compounds. Examples of these sys-
Hydrolysis is another kind of transformation reaction that
tems are presented in the following chapter.
operates on organic compounds. Unlike biodegradation,
hydrolysis is not catalyzed by microorganisms. Hydrolysis
involves the reaction of an organic molecule with water
or a component ion of water, for example, -
12.7 Isotopic Processes

R-X - H,O
R-OH + X- + Hi
where R-X is an organic molecule with X representing
Isotopes are atoms of the same element that differ in
terms of their mass. For example, hydrogen with ;in
atomic number of 1 has three isotopes, IH, iH, and :H,
an attached halogen, sulfur, phosphorous or nitrogen. with mass numbers (superscripts) of 1, 2, and 3, respec-
According to Neeley (1985), the introduction of a hy- tively. The first of these isotopes is stable, while the last
droxyl group into the parent molecule makes the product one {H (usually written 5H) decays radioactively to :He.
more susceptible to biodegradation, and more soluble. Thus, radioactive decay is one of the important reactions
Not all organic compounds will undergo hydrolysis. that involves isotopes. In radioactive decay, atoms of ;I
Table 12.12 from Harris (1982) (reprinted by Neely, particular isotope change spontaneously to a new, more
1985) lists the susceptibility of various functional groups stable isotope. Isotopic concentrations change also clue
to hydrolysis. First-order rdte laws generally describe hy- to processes like evaporation, condensation, or water-
dolysis with the rate of disappearance of RX proportional rock interactions. Typically, these processes favor one
of the isotopes of a given element over others, produc-
ing fractionation.

Table 12.12 Summary of the Susceptibilityof Various

Functional Groups to Hydrolysis Radioactiae Decay
Radioactive decay occurs mainly by the emission of an
Resistant to Hydrolysis'
a particle (;He) or a fl particle (electron -ye). Often ;K-
Alkanes Aroniatic nitro compottnds companying the emission of these particles is y radiation,
Alkenes Aroniat ic aniines which is electromagnetic energy of short wavelength.
Alkynes Alcohols This radiation forms when nuclides produced in an ex-
Benzene Phenols cited state (noted by *) revert back to a so-called
Biphenyl Glycols groundstate. As the following examples illustrate. a-de-
Polycyclic aromatic Ethers
cay changes both the mass number and the atomic num-
hydrocarbons Aldehydes
ber, @-decay changes the atomic number only. and y-
Halogenated aromatics Ketones
Hnlogenated hydrocarbon Nitriles emission changes neither:
pesticides (DDT, etc.) Carboxylic acid
Sulfonic acid
a-decay: '&jTh + 'ZRa + :He
/3-decay: 'ZRa + '$Ac +
Susceptible to Hydrolysis'
y-emission: '$U* + '$$tJ + y
Alkyl halides Epoxides and Iactones
Amides" I'hosphoric acid esters The arrows in these equations indicate that radioactive
Carbamates" Sulfonic acid esters decay is irreversible. The quantity of the reacting or pw-
Carboxylic acid esters ent isotope continually decreases, while the product or
"Most aniides have a half-life grater than a yaar at pH 7 and 25°C.
daughter isotope increases. Reactions become more com-
Structures such as C12CHCONHLhave a half-life less than a year. plex when the daughter itself decays through ;I series of
" All carbamates having only alkyl o r aromatic or N and 0 are persistent other products until a stable form is finally created. We
(i.e.. half-life > 1 year). refer to these reactions as decay chairis or clisiritrgr-cttiori
' Reprinted from Harris, J. C., Handbook of Chemical Property Estinia- srries (for example, ?$,Th,'45LJ. and 'ZU series).
tion Methods, W. J. Lyman, W. F. Reehf, and D. H. Rtisenblart, d s .
The decay of radioactive isotopes is independent of
Originally McGraw-Hill. 1982, suhscquently published hy Anieric;ui
Chemical Society, 1990. Copyright 0 1990 American Chemical Socirty. temperature and follows the first-order rate law:
' R~pfintedwith permission froni Neely, W. B. 1985, Hydrolysis. I n
Environment Exposure from Chemicals, eds. W. B. Nerly and G . E.
( 1 2.57)
Rlau, CRC Press,p. 157- 173. Copyright CRC Press. Inc.Uoca Raton. FL.
278 Chapter 12 / ChemZcaC Reactions

where N is the number of atoms of radioactive material, Table 12.14 Isotopes of EnvironmentalSignificance
t is time, and k is the rate constant for radioactive decay. and Their Relative Abundances
Integration of Eq. 12.57yields an expression for the quan-
Average Abundance 010
tity of parent remaining at any time following its for- Element Isotopes of Stable Isotopes
mation:
Hydrogen IH 99.984
N(t) = N&+' (12.58) :H" 0.015
fH" radioactive 10-14 t o 10-16
where Nu is the initial number of atoms of parent isotope.
99.76
The decrease in the activity of a radioactive substance 0.037
is commonly expressed in terms of a radioactive half-life 0.10
(tIl2), which is the time required to reduce the number
of parent atoms by one-half. The half-life relates to the Carbon 'ZC 98.89
IZ
C 1.11
rate constant for decay as follows: radioactive -1()-"'

0.693 95.02
tlll = 7 (12.59) 0.75
4.21
Listed in Table 12.13 are radioactive half-lives for many 0.02
radionuclides of interest in ground-water investigations. " Deuterium, often referred to by D.
Radioactive decay is important for two reasons. When " Tritium, oftenreferred to by T.
radioactive isotopes occur as contaminants, decay Modified from Fritz and Fontes (1980). Reprinted from Handbook of
through several half-lives can reduce the hazard when Environmental Isotope Geochemistry. v. 1, by P. Fritz and J. C. Fontes,
'Introduction," p. 1-19, 1980 with kind permission from Elsrvier
the residence time in a flow system is much greater
Science, NL, Sam Hugerhanstraat 25, 1055 KV Amsterdam, The Nether-
than the half-life for decay. This capacity to attenuate lands.
radioactivity provides the rationale to support the subsur-
face disposal of some radioactive contaminants. Radioac-
tive decay also forms the basis for various techniques
used to age-date ground water. The most important radio- Isotopic Reactions
active isotopes used for this purpose are .'H(tritium), %, The isotopes of hydrogen, oxygen, carbon, and sulfur
"'si, and %CI. are useful in studying chemical processes. Related to
these elements are seven isotopes of environmental sig-
nificance (Table 12.14). Commonly, ground-water stud-
ies involve *Hor D, "0, %, and %.
Because one isotope dominates the rest in terms of
Table 12.13 Radioactive Half-Lives for Elements of relative abundances (Table 12.14) and the changes due to
Interest in Age Dating and Contaminant Studies fractionation are too small to measure accurately, isotopic
abundances are reported as positive o r negative devia-
tions of isotope ratios away from a standard. This conven-
12.3 3 x 10' tion is represented in the following general equation
5 x 102 432 (Fritz and Fontes, 1980):
3.1 X 10' 7 x 10'
x - Rsim~;md
Rsamp~r
100 6.5 10' s= x 1000 (1 2.60)
29 3.5 x 10' R*lUhd
1.5 X 10" 2 x 10'
where 8, reported as permil (%a), represents the deviation
2 x 10' 2 x 10'
7 x 10" 1 x 10' from the standard, and R is the particular isotopic ratio
2 x 10- 90 (for example, '"O/"O) for the sample and the standard.
3 x 106 1.3 X 10" For example, we would express sulfur-isotope ratios as
30 1 .0
8.2 7 x 10"
22 4.5 x lo9
1.6 X 10.' 2 x 106 A 6% value of - 2 0 % ~means that the sample is depleted
22
in j4S by 20% or 2% relative to the standard.
8 x 104
This way of expressing isotopic compositions takes
1.4 X 10"'
advantage of the ability of mass spectrometersto measure
Modified from Moody (1982). isotopic ratios accurately. The errors involved in de-
 12.7 Isotopic Processes 273

termining 6 values is typically a small proportion of the increasingly more pronounced, the depletion becomes
possible range of values. more marked as air masses begin to move across cooler,
continental land masses, or are forced to higher and
colder elevations over mountain ranges. The temperature
Deuterium and Oxygen-I8 control on fractionation also explains why snow is more
This section discusses the occurrence of D and '"0in depleted in D and "0.
water, gases (for example, CH,), and ions (for example, If we assume that differences in the isotopic composi-
HCO; and SO:-). Deuterium and oxygen-18 composi- tion of rainwater reflects the increasing effects of "ciin-
tions of water are usually measured with respect to the out" from the vapor, then a relationship could exist
SMOW (Standard Mean Ocean Water) standard (Fritz and among samples. When the SD and SIHOcontent of rainwa-
Fontes, 1980). This choice of standard is particularly ap- ter from sampling sites around the world are plotted
propriate because precipitation that recharges ground together, they lie along a straight line (Figure 12.12).
water originated from the evaporation of ocean water. known as the meteoric water line (Craig, 1961). The
However, the isotopic composition of rain or snow in equation for this line is approximately = 86"O +
most areas is not the same as ocean water. Evaporation 1o'%,o.
and subsequent cycles of condensation significantly The meteoric water line provides an important key to
change the isotopic composition of water vapor in the at- the interpretation of deuterium and oxygen-18 data. Wa-
mosphere. ter with an isotopic cornposition falling on the meteoric
Water vapor in equilibrium with water is depleted in water line is assumed to have originated from the atnio-
6D and 6''O by 80%0and 10x0,respectively (Wallick and sphere and to be unaffected by other isotopic processes.
others, 1984). Thus, for ocean water (& and IS1"O - Deviations from the meteoric water line result from other
isotopic processes. In most cases, these processes affect
OXO), a vapor in equilibrium should have = -80%~
and 8'0 = -10%0.The first few drops of rain falling the relationship between SD and 8'0 in such a unique
from this vapor should theoretically have an isotopic way that the position of the data points can help to
composition that is the same as the starting water. This identifi a process. Figure 12.12 illustrates the direction
change in isotopic ratios (for example, 1'0/160) because away from the meteoric water line in which various pro-
of a chemical reaction (for example, water = water va- cesses push the composition of water.
por) is termed fructtonation. This process is described Two of the more commonly observed processes, evap
mathematically by a fractionation factor (a) (Fritz and oration from open water and exchange with rock miner-
Fontes, 1980), for example als (Figure 12.12), exhibit this deviation (Fontes, 1980).
When water in a pond or lake evaporates, there is an
enrichment in the heavier isotopes. However, because
of the dynamics of the process, the isotopic composition
of the pond follows an "evaporation line" with a slope
or in terms of del notation
1000 + SIHow,l,,
a= (1 2.61)
1000 + SI"OVaplr
-10
The isotopic ratios of water vapor in air masses associ-
ated with oceans, are however, not in equilibrium with -20 - Meteoric water line

the water. The vapor is more depleted than expected.

The extent of depletion is correlated mainly with temper- -30 -
ature (Gat, 1980). Rainfall sampled in coastal areas near
Evaporation from
the equator has an isotopic composition most like seiwa-
ter, but it is still slightly depleted. Moving away from the
equator to higher latitudes, the rainfall becomes more
and more depleted relative to seawater. For example,
along the western coast of the United States and Canada, Exchange with rock minerals
values of SD range from approximately -22 to - 101'X)o
and "0 from - 4 to -14%~)(Gat, 1980). This so-called
latitude effect is related to progressive temperaturecon-
trolled depletion of heavier isotopes in the vapor as they -80 I I I I I I I I I
-11 -10 -9 -8 -7 -6 -5 -4 -3
are preferentially removed during precipitation events. 6 IS 001,
This continuous removal of a condensate enriched in Figure 12.22 Ikviations in isotopic coniposirions ;iw;r!'
heavier isotopes is known as a Rayleigh process. from the meteoric water line as a consequence of viirious
Because fractionation effects at lower temperatures are processes (from IAEA Report No. 228, 1983).
280 Chapter 12 / Chemical Reactions

ranging from one to five depending on the local rates of Exchange between meteoric water and minerals con-
evaporation. The point of intersection of the evaporation taining oxygen results in the kind of deviation depicted
line with the meteoric water line usually is taken as the in Figure 12.12. Because only ‘’0 is involved in exchange,
isotopic composition of unaltered precipitation. In gen- the samples typically fall along a horizontal line, reflecting
eral, the farther along the evaporation line the data points an “oxygen shift” away from the meteoric water line.
lie, the greater the evaporation. Truesdell and Hulston (1980) indicate that the size of
Isotopic exchange between minerals and ground water the oxygen shift is proportional to the difference in origi-
is important in deep, basinal flow systems or in geother- nal P O between the water and rock, the temperature,
mal systems.The relatively high temperatures enable oxy- and the time of contact, and inversely proportional to the
gen and hydrogen to exchange between phases and to waterhock ratio. Yellowstone Park, Steamboat Springs,
achieve an equilibrium distribution. This equilibrium is Wairaki, and Salton Sea are well-known examples. Oxy-
described by the fractionation factor for the mineral. gen and hydrogen isotopes are extremely valuable in
Savin (1980) has summarized fractionation factors for a hydrogeologic studies. The most important applications
variety of different minerals and described how they include flow system tracing and indirect age determina-
change as a function of temperature. tions, which are discussed later.

Problems
1. Often (CO:-) is not reported in water-qualityanalyses mL of a solution containing various concentrations of Cd.
because the concentration is too low to measure directly.
Equilibrium (Cd)
Develop an expression from the appropriate m a s laws
Initial (Cd) Concentration
to calculate the concentration of ( C a - ) given pH and
Concentration (mg/mL) (mg/mL) x
(Hcoa.
0.0005 0.040
2. A water sample has the following ionic composition
0.001 0.093
at 25OC:
0.010 1.77
Ionic Concentration (M)x 0.050 39.3
Ca Mg Na HCO9 SO4 C1 pH 0.096 205.2
4.17 1.32 11.7 12.0 5.49 0.08 7.60 a. Find the equation for the Freundlich isotherm that
Given that the activity coefficients for Ca2+,Mgzt, and best fits these data.
HCO; are 0.52, 0.55, and 0.85, respectively, answer the Hint Transform the basic equation and data in terms
following questions. of the logarithm of concentration.
a. What is the partial pressure of C 0 2 in equilibrium
with the sample? b. Assuming that only the lowest concentrations are of
b. Assuming no complexation, what is the saturation interest, fit the same set of data with a linear Freund-
state of the sample with respect to calcite? lich isotherm (i.e., y1 = 1) to obtain a distribution
C. Assume that the concentration reported for calcite
coefficient (KJ.Be sure to indicate the units for K d .
represents (Ca), and that three significant ion pairs 5. At a site, the following contaminants are found (i)
form CaCO! , CaHCO,; , and CaSO:‘. Develop an equa- parathion (log K,,,, 3.801, (i) chlorobenzene (log K,,,,
tion to calculate the concentration of free Ca2+in 2.71), and (iii) DDT (log K,,,,,
6.19).
the sample, assuming the solution is ideal. a. Using the equation of Hassett and others (1983) in
3a. Given that the concentration of benzene (molecular Eq. 12.41, estimate the partition coefficient between
weight 78.1) dissolved in ground water at the water solid organic carbon and water ( K J .
table is 280 pg/L, calculate the equilibrium partial b. With anfx of 0.01, estimate the distribution coeffi-
pressure in the adjacent soil gas. The Henry’s Law cients (&) for these compounds.
constant for the partitioning reaction is 5.5 X c. On the basis of sorption alone, which of these com-
atm-m’/mol. pounds is the most mobile and which is the least
b. If benzene was present as the pure liquid having mobile?
a vapor pressure of 0.1 atm, what would be the 6. Compare the range in variability of the distribution
partial pressure? coefficients for carbon tetrachloride (log KO,,2.83)
4. Following are the results of a batch sorption experi- estimated using the three different regression equations
ment in which 2 g of porous media were mixed with 10 (Eqs. 12.39, 12.40, and 12.41) and an fM of 0.01.
 Problems -
281

7. The properties of six organic compounds are couple, producing a pe of 12.0. By assuming redox equi-
listed below. librium, determine how (Fe), = M would be distrib-
uted as Fez+and Fe".
Henry's Law Constant
Compound log K,,,, (atm-m3/moI) 10. An organic compound has a tendency to hydrolyze
1 2.04 5 x 10-5 with t,,? = 3.5 yr. What would an initial concentration
2 0.89 5 x 10-J of 100 pg/L be reduced to after 10 years?
3 0.46 7 x lo-' 11. Explain what is meant by the SMOW standard and
4 - .27 3 x lo-+ the convention for reporting isotopic results.
5 2.83 5 x 10-5
12. These isotopic measurements were made on sev-
6 I .48 4 x 10-5
eral prairie lakes and potholes.
a. List the organics in order of decreasing solubility.
b. List the organics in order of decreasing mobility. Sample P O (%o) SD (%o) (SMOW)
1 -6.5 -98.0
8. Consider the following half-redox reaction involving
2 -8.5 - 102.0
NO; with a concentration of Mand ammonium with
a concentration of lo-.' in ground water with a pH of 8 . 3 -2.6 -83.0
4 -4.0 -86.0
iN0.i + $Ht+ e- = BNH: + gH20
What is the best explanation for the deviation in these
What are the pe and EH of the system at 25°C assuming results from the meteoric water line? What is the esti-
redox equilibrium? mated concentration of unaltered meteoric water in
7. The couple Or/H20 is found to be the dominant this area?
 Chapter 13

Colloids and Microorganisms

13.1 A Conceptual Model of Colloidal Transport

13.2 CouOidal Transport in Ground Water
13.3 Microbiological Systems
13.4 Microbial Processes
13.5 Biotransformation of Common
Contaminants

~~ ~~ - ~~ ~~ ~ ~ ~ ~- ~~ ~~

This chapter is concerned with the occurrence and trans- which makes them efficient contaminant collectors. In
port of colloids and microorganisms. A colloid is a particle cases where some contaminant has a low mobility in
ranging in size from 1 nm to 1000 nm. Particles can solution, facilitated transport on particles may create un-
include bacteria, viruses, large macromolecules of dis- expectedly high contaminant fluxes.
solved organic carbon (for example, humic substances), Second, particle transport is important because of the
small droplets of nonaqueous phase liquids, inorganic related surface processes. For example, the original colo-
rock, or mineral fragments (for example, clay minerals, nization of some porous media by bacteria may have
ferric oxyhydroxides). Bacteria are small single-celled depended on their transport from above or laterally. The
organisms that occur ubiquitously in ground-water and biofilm created by the attached bacteria and other organ-
surface-water systems. A virus is considered to be a ge- isms can have an enormous influence on the transport
nome (that is, genetic information stored o n DNA or of nutrients and organic compounds dissolved in water
RNA) within some kind of protein coating (Chapelle, and often on the chemistry of ground water.
1993).
Particle transport is important in ground water for at
least two reasons. First, the migration of particles pro-
vides another way for mass to spread in the subsurface.
-
13.1 A Conceptual Model
This process may play a role in the genesis of soils or
of Colloidal Transport
ore deposits (McDowell-Boyer and others, 1986), but is Figure 13.1 is a conceptual model for colloid transport
of particular interest in relation to contaminant migration. developed by McCarthy and Zachara (1989). The trans-
The particles themselves may be contaminants, as is the port of colloids requires (1) a source for colloids in the
case with bacteria and viruses, o r the contaminants can medium, (2) conditions that promote the suspension of
be trace metals, radionuclides, or organic compounds particles or stability of colloids, and (3) advective trans-
sorbed onto clay or organic particles. Because of their port without significant size filtration o r surface inter-
small size, particles have relatively large surfwe areas, action.
 13.1 A Conceptual Model of CoUofdalTranspori 283

1 2 3 Stabilization
Genesisl Stabilization Transport/
The simple presence of colloids in ground water does not

-
mobilization/ aggregation/
suspension filtration
assure that they are transported. As Figure 13.1 implies,
Aggregation/ transport requires that suspended colloids be stable. In
straining filtration
Translocation a suspension of colloids, stability is determined as the
(from vadose
No outcome of competition between van der Waals forces
colloids
Suspended
that cause similar particles to aggregate and electrostatic
colloids in Physiccchemical forces that cause particles to separate. Van der Waals
ground water collection
forces are weak electrical forces that cause attraction
V No between atoms and molecules. These forces are created
colloids by oscillations in the charge distribution involved with
n a positively charged nucleus and negatively charged elec-
tron swarm. Electrostatic forces are produced by the
strong electrical attraction or repulsion between point
charges of differing or similar charges, respectively. Mc-
GS watertiow Carthy and Zachara (1989) discuss the complexities of
surface chemistry, particle mineralogy, and other factors
Ffgure 13.1 A conceptual model for the origin and trans- that determine stability. While these processes are com-
port of colloids in ground water (from McCarthy and Za- plex, certain generalizations are possible. For example,
chara, 1989). Reprinted with permission from J. F. McCarthy stability is promoted by a decreased ion concentration
and J. M. Zachara, Environ. Sci. Technol., v. 23, no. 5 , p. in the water, a decreased abundance of divalent ions (for
496-502. Copyright 0 1989 American Chemical Society.
example, Ca2+)relative to monovalent ions (for example,
Na'), and a pH that is different from the pH of the zero
point of charge (pH,) for the colloid surface.
Puls and Powell (1992) present experimental results
Occurrence of Colloidal Material that illustrate the concept of stabilization. The colloids
they studied were 150-nm Fe203particles. As we e x
Colloidal transport begins with a source of particles of
plained in Chapter 12, the surface charges of an iron
the correct size. One possibility is the translocation of
oxide changes as a function of pH. At the pH,, the net
colloids from other locations. Examples could include
surface charge is zero. Figure 13.2 is a plot of colloid
humic substances flushed from the vadose zone, viruses
diameter versus pH. Over the pH ranges of 2.0 to 6.5
and bacteria present in waste waters from septic tanks,
and 9.7 to 1 1 .O, the measured colloid diameters were
or organic macromolecules generated in landfill leachates
about 150 nm-the diameter of the starting colloids. In
(Gounaris and others, 1993). Colloids also can form from
the pH range from 6.5 to 7.6 (close to the measured pH,
precipitation of solids from solution. The cause is super-
of 7.0),the diameter of the colloids increased, indicating
saturation of a mineral phase stemming from abrupt
changes in the water chemistry due to varying pH, ion
concentrations, or redox potential (McCarthy and Za-
chara, 1989). For example, Gschwend and Reynolds 1000(
(1987) describe the formation of iron phosphate colloids 900 -
in the sewage plume at the Otis Air Force Base due to
the reaction of phosphate, present as a dissolved contami- -E 800 -
C 700-
nant in solution, and ferrous iron released from aquifer 5
solids. Other colloids formed by precipitation are small Z 600-
metal solids or radionuclides that can form in hydrolysis E
0 500-
reactions. Figure 13.1 also illustrates schematically how 0
- 400-
-0
colloids can form when small particles attached to the 3 300 -
surface of solids are remobilized.
200 -
There are other possibilities for generating colloids
i I 1 I I I I
besides those shown in Figure 13.1. For example, it is 100
2 3 4 5 6 7 8 9 10 11 I
well known in the petroleum industry that injection of PH
fresh water into saline formations can lead to the electro- Ffgure 152 Stability of 150-nm FelO, particles 3s P func-
static mobilization of clay coatings on grains. The solubili- tion of pH in 0.005 M NaC10, (from Puls and Powell, 1992).
zation of organic matter such as kerogen or bitumen, Reprinted with permission from R. W. Puls and R. M. Pow-
or the breakdown of biofilms, can also form colloids ell, Environ. Sci. Technol., v. 26, no. 3, p. 614-621. Copy-
(McCarthy and Zachara, 1989). right @ 1992 American Chemical Society.
284 Cbapter 13 / CoUofdsand Mfcroorganfsms

aggregation and suspension instability. Near the pH, the negatively charged above pH 7. As expected, in the range
particles are uncharged and van der Waals forces cause of pH 3 to 7, experiments showed there to be no colloid
them to aggregate. migration because the surfaces are positively charged
and the colloids are negatively charged. At pHs above
about 7.6, the colloids are mobile because the colloid
Transport and Filtration
and porous-medium surfaces are both negatively charged.
Stable colloids in solution are potentially transportable
as long as they do not interact substantially with the
porous medium. One possible mode of interaction is
through physical straining (also known as straining filtra-
-
13.2 Colloidal Transport
in Ground Water
tion), where the particles are too large to migrate through
small pore throats. However, colloids and most microor- Colloids can be mobile in ground water. One common
ganisms are commonly much smaller than the grain size health issue is related to the transport of viruses in ground
(and pores) of gravels, sands, and silts (Figure 13.3). water. A dominant source of viral contamination is raw
Physico-chemical collection on solid surfaces (McCarthy sewage disposed in the subsurface through septic tanks.
and Zachara, 1989) is a more important process of filtra- Other public health problems are related to waste-water
tion. The term physico-chemical collection applies to a renovation schemes that involve the recharge of second-
variety of different surface-active attractive forces. One ary sewage effluents. Keswick and Gerba (1980) summa-
example is electrostatic interaction, where a positive par- rize case studies that indicate the possibility for viruses
ticle is attracted to a negative surface. In general, nega- to be transported several hundred meters in ground wa-
tively charged colloids would be most mobile, given the ter. In exceptional cases, travel distances of close to one
tendency for most aquifer solids to be negatively charged kilometer are possible. Other literature has examined the
(McCarthy and Zachara, 1989). potential for radionuclide transport in colloidal form. For
Puls and Powell (1992) provide an interesting example example, at the Nevada Test Site, Buddemeier and Hunt
of how Fe203colloids interact with a porous medium (1988) describe the possibility of facilitated transport of
in column experiments. The medium was a quartz-rich radionuclides away from a nuclear detonation cavity. At
material with a relatively low pH,, because of the abun- Chalk River in Canada, significant radionuclide transport
dance of quartz (pH,, 2 2), and manganese oxides and was found associated with particulates (Champ and oth-
layered silicates (pH,,, < 4). Thus, when the pH is less ers, 1984).
than about 3, surfaces of the porous medium will be
positively charged. Above pH 3, surfaces exhibit a net
Sampling and Measuring
negative charge. Recall that the surfaces of the Fe20,
colloids would be positively charged below pH 7 and The colloid characterization process is complicated be-
cause most of the steps involved in sampling and analy-
sis-drilling a well, pumping out a sample, handling a
sample at the ground surface or in the laboratory-can
create particles not found in the original sample. For
100-10 m
example, McCarthy and Zdchdra (1989) explain how dril-
protozoa l n g i ling can add small particles when drilling mud is used.
-
5 p m-0.2 pm
bacteria
500-200 nrn
Pumping often creates unnaturally high flow velocities
that remobilize surface-bound colloids. Oxygenation of
rickettsiae chlamydiae
the samples at the surface or during preparations for
-
3%-20 nm
viruses
10-1 nm
laboratory analyses could change the redox conditions,
leading to the precipitation of Fe(I1I) oxides.
mac-ules
1 nm Good sampling practice involves pumping wells for
me-s several hours to eliminate fines near the well screen.

I I I I I I I I
-
Atoms
I I
Pumping rates are normally kept low, on the order of
100 mL/min, to avoid detaching colloids (Gounaris and
l O m m 1 rnm 0 . 1 rnm 0.001 mm others, 1993). A gasdriven, positive displacement pump
lOOpm 1Opm l p m
lOOOnmlOOnrnlOnm l n m 100 pm 10 pm avoids contact of the sample with oxygen. At the surface,
Ffgure 23.3 Comparison of the sizes of common microor-
samples are typically collected in nitrogen-flushed con-
ganisms with the grain sizes of sediments, molecules, and tainers or filtered for analysis in field glove boxes con-
atoms (modified from Matchess and Pekdeger, 1985, Ground- taining nitrogen.
water PoUution Microbiology, eds. C. H.Ward, W. Ciger, and Several different approaches are available to establish
P. L. McCarty). Copyright 0 1985 by John Wiley & Sons, Inc. the concentration and size distribution of colloids in sam-
Reprinted by permission of John Wiley & Sons, Inc. ples. Ultrafiltration systems process the samples through
 13.3 Microbiological Systems 285

a succession of smaller and smaller nanometer-sized fil- 7 1000'

ters. Knowing the volume of water passed through the - -- -- L- -- J
73"OO'
1-
72000'
system and the mass of colloid trapped on each of the
filters, one can determine colloid concentrations (Goun- 42'30'4 MASSACHUSETTS Boston
aris and others, 1993). The mineralogy of the particles
I
I
trapped on the various filters can be determined using c -- - - - -- _ _ _ _ 70'00'
energydispersive scanning-electron microscopes and x- '42"OO' 3
ray diffraction. Gschwend and Reynolds (1987) used a 0 20 40 60miles \ Cod
laser Iight-scattering approach, which can be interpreted I-+-++
to provide information on colloid size distributions and 0 20 4 0 60 kilometers ,/..
colloid concentrations. A variety of other light-scattering . 0
c
I

approaches can be used for colloid determinations as

well as centrifugation.
Otis Air Base

Studies at Cape Cod

Throughout this book, we make use of field studies that
illustrate both theoretical and practical concepts. One of
these important sites is the Cape Cod Toxic-Substances
Hydrology Research site, located near Falmouth Massa-
chusetts (Figure 13.4). The U.S. Geological Survey (USGS)
has operated this site since the early 1980s. It has been
the focal point for studies concerned with the fate and
transport of contaminants and the chemical and biologi-
cal behavior of dissolved and colloidal material in ground
water. Garabedian and LeBlanc (1 991) provide a detailed
overview of studies in relation to a large sewage plume
from the Otis Air Base and smaller scale process studies
conducted at an abandoned gravel pit, which is shown --.--
on Figure 13.4 as the tracer-test site. Our interest here
is with the behavior of bacteria and other colloids. Later,
we will examine other interesting features related to
contamination at the site.
-* 9
------ ---..-
w
0

0
2000feet

600 meters
EXPLANATION
At Otis Air Base, approximately 1740 m'/day of treated
Tracer-test site
sewage has been disposed through infiltration beds into
-100- Line of equal boron concentration. 1978-79 -
the subsurface since 1936. The disposal unit is an uncon-
fined aquifer comprised of sand and gravel outwash ap- - - -10-
Concentrations in micrograms per liter
Water-table contour, November 1979 -
Altitude of water table in meters Datum IS sea level
proximately 35 m thick. 7%esewage plume is about 915
m wide, 23 m thick, and 3.35 km long (Figure 13.4).The Ffgure 13.4 Maps showing the location of the Cipe Cod
treated sewage contains above-background concentra- Toxic-Substances Hydrology Research Site. the tracer-test
tions of sodium, chloride, ammonium, nitrate, phos- site, and the extent of the sewage plume from Otis Air Rise
phate, detergents, and several different volatile organic (from Garabedian and LeBlanc, 1991).
compounds.
Gschwend and Reynolds (1987) looked at features of
colloid transport. They found that samples of sewage and
that these colloids formed by precipitation from aqueous
ground water up to approximately 360 m down-gradient
solution. Reducing conditions in the plume promoted
from the infiltration beds contained colloids. There are
the dissolution of iron coatings from grains. while the
two distinctly different types of colloids present in the
required phosphate is a contaminant in the plume. These
ground water. Colloids in the actual sewage are organic
colloids are observed up to 300 m from the intiltration
macromolecules with a diameter of 6 nm. However, they
beds.
are not found in the nearest down-gradient well, approxi-
mately GO m away. Thus, the abundant colloids in the
sewage are not mobile (Gschwend and Reynolds, 1987).
Colloids in the ground water are approximately 100 nm
153 Microbiological Systems
in diameter and have a vivianite-type mineralogy (Fe, Microorganisms occur ubiquitously in the subsurface.
(POr)z-8H20). Gschwend and Reynolds ( 1 987) suggest Bacteria predominate, but protozoa and fungi are also
 n
286 Cbapter 1.3 / CoUoids and Microorganisms

common. Bacteria are small single-celled organisms. To

live, they metabolize dissolved organic matter present Bromide
naturally or due to contamination. Section 12.5 provided
examples of redox reactions that are bacterially cata- 0.3-
lyzed. Protozoa are somewhat larger, singlecelled organ-
isms that can feed on bacteria. Fungi live by degrading
9 0.2-
other organic matter and effectively recycle plant and 0

animal debris (Chapelle, 1993).

Bacteria occur in the subsurface in different ways. 0.1 -
Some are adapted to move freely in the water due the
presence of flagella that can propel them. The direction
of motion can be either random or directed in response 0.0
10 20 30 40 10 20 30 40
to substrate concentrations (chemotactic). This property Days after Injection
of motility can be important for the local redistribution
Figure 1.3.6 Breakthrough curves for Hr- and stained bacte-
of bacteria, and possibly influences larger scale transport. ria in a sampling device located 6.8 m down-gradient from
Most commonly, bacteria are immotile, bound to the the injection well. The solid line illustrates the fit of a theo-
surfaces of the aquifer solids. The attached population retical model to the observed data (from Harvey and Garabe-
forms what is called a biofilm, which consists of bacteria dian, 1991). Reprinted with permission from I<. W. Harvey
held together and to the particles by extracellular poly- and S . P. Garabedian, Environ. Sci. Technol., v. 25, no. 1 , p.
mers (McCarty and others, 1984). 178- 185. Copyright 0 1991 American Chemical Society.
In pristine ground-water systems, the bulk of the bacte-
rial population is attached to the particle surface (Ch'
1 iorse

and Wilson, 1988). Chapelle (1993) draws on theoretical have produced systematic observations on the transport
work by Kelly and others (1988) to suggest that when behavior of unattached bacteria at several scales. Harvey
substrate concentrations are low, immotile bacteria have and Garabedian (1991) describe the results of a natural
a competitive advantage over motile forms. There is some gradient tracer test conducted using Rr- and bacteria
indication that the proportion of motile bacteria may be (0.2-1.4 p m long). The bacteria were stained for identi-
larger in contaminated systems containing metabolizable fication and injected slowly into the aquifer (Figure 13.5).
organic compounds (Ghiorse and Wilson, 1988). With Samples were collected with time at a multilevel sampling
increasing substrate concentrations, motile bacteria may device located approximately 6.8 m down-gradient. The
become more competitive (Chapelle, 1993). breakthrough curves for Br- and the stained bacteria
Studies by Harvey and his co-workers on Cape Cod (expressed as dimensionless concentrations) are shown
in Figure 13.6 at the sampling device. The timing of the
breakthrough was similar. However, the peak abundance
was much less for the bacteria than for Br- and the
curve exhibited tailing. The reduced peak abundance
was attributed to the loss of bacteria by sorption on the
aquifer over the 6.8m-long travel path. However, these
bacteria behaved much like colloids because motility was
apparently not a factor in the transport.

6
w
4 Water table
The simplest conclusion from these experiments is
that unattached bacteria will not be particularly mobile
in the outwash aquifer over much more that several tens
of meters. However, other field observations point to the
transport of unattached bacteria over a substantial length
-
-1 m
t -
+
+
Ground water
flow
of the contaminant plume (Harvey and George, 1986;
Harvey and Garabedian, 1991). The main determinant
for transport appears to be whether the bacteria come
1 (-0.33m/d)
from an indigenous population that is actually living in
Injectate
cloud the aquifer or a nonindigenow population that is not
Ffgure 13.5 Schematic figure showing the design of the
living in the aquifer but is added with the sewage. Indige-
natural-gradient tracer experiment with Br- and stained bac- nous bacteria are potentially much more transportable
teria (from Harvey and Garabedian, 1991). Reprinted with because bacteria being lost by predation or due to sorp-
permission from R. W. Harvey and S. P. GArabedkdn, Environ. tion to the surfaces of p a in s in the aquifer are being
Sci. Technol., v. 25, no. 1, p. 178-185. Copyright 0 I991 repkdced by rapidly growing bacterial populations (Har-
American Chemical Society. vey and others, 1989). Organic contaminants within the
 13.3 Microbiological Systems 287

Diffusion layer Mallard (1984) discuss these sampling techniques in de-

tail and provide useful guidance on how to avoid sample
contamination.
Detailed microbiological studies require the collection
of soil cores because most of the microbial population
is surface attached. Beyond the difficulties in actually
collecting the core is the problem of how to avoid micro-
bial contamination during the drilling and sampling pro-
Figure 13.7 Conceptualization of a biofilm at a pore scale. cess. Boreholes should be drilled using techniques (for
example, hollow-stem augering) that do not require dril-
ling fluids (McNabb and Mallard, 1984). Convention;il
coring equipment-solid tube samplers, split-tube skim-
plume provide a source of organic matter necessary for plers, Shelby tube samples, and so on-can be sterilized
the growth of the bacteria. and used to collect the sample.
Although indigenous bacteria were used for the small- Because of the near impossibility in avoiding microbial
scale tracer test, growth rates were negligible during contamination on the outside surface of a core, its surfxce
trdnsport. Thus, the reduction in bacterial abundances is pared away as it is extruded. Figure 13.80 illustrdtes
were dramatic. Harvey and Garabedian (1 99 1) suggest an anaerobic glove box that can be used in the field. The
that staining the bacteria to observe them may have inhib- glove box is flushed with nitrogen for about 30 minutes
ited their growth, or that conditions for growth were at a rate of about 2500 L/hr (Armstrong and others, 1988).
not appropriate in the part of the aquifer where the The sample tube, sealed temporarily after collection, is
test was conducted. In summary, bacterial transport over
more than 1000 meters is possible in plumes containing
organic contaminants (Harvey and others, 1989).
Flow regulalar
and indiratn:
Biofilms
Biofilms are the site for transformation reactions that
are important in reducing the concentrations of some
organic contaminants. They are commonly conceptual-
ized as a simple layered structure wherein molecular
diffusion operates to bring metabolizable organic com-
pounds from the bulk pore fluid to clusters of bacteria
cells in the biofilm (Figure 13.7). Real biofilms are more
complex. For example, in cases where organic substrate
concentrations are relatively low, patchy or discontinu-
ous biofilms form (Rittmann, 1993). Detailed microscopic
studies of model biofilms demonstrated fluid flow within
biofilm channels (Stoodley and others, 1994). These
channels provide a delivery system for nutrients into the
biofilm that is more efficient than diffusive transport.

Sampling and
Enumerating Microbial Populations
A variety of techniques have been developed to enumer-
!h>
ate microorganisms in aquatic systems. The first critical
step is collecting the sample. Depending on the purpose Figure 23.8 Panel (a)A field-samplingglow box for pro-
of the study. one could consider either water samples or cessing a sediment core for miocrobial analysis. ( b ) The s i n i -
piing tube is inserted into the field glove box and the core
samples of soil material. Water samples can be collected
paring tool is attached t o the end of the tube. The cores ex-
from conventional monitoring wells. For example, at truded from the sample tube are collected in sterile cont:ain-
Cape Cod, Harvey and others (1984) collected water ers (from Armstrong and others, 1988). Reprinted with per-
samples from conventional monitoring wells using a sub- mission of NWWA/API Conference Petroleum Hydroc:arbons
mersible pump connected to the surface by Teflon tub- and Organic Chemicals in Ground Water: Prevention, Lktrc-
ing. After purging three to five casing volumes, samples tion and Restoration: National Ground Water Assoc.. Copy
were collected in sterile sample bottles. NcNabb and right 0 1988. All rights reserved.
288 Cbapter 13 / Colloids and Microorganisms

inserted into the box through an iris port and a sterile pacity of the system and the potential for contaminant-
paring tool (Figure 13.8b) is screwed on to the sample specific toxicity.
tube. The core is extruded using a rig-mounted extruder Plate counts provide information on the number of
and appropriate subsamples are placed in sterile contain- colony-forming units, but not the actual number of cells.
ers (Armstrong and others, 1988). A procedure that provides this information is the most-
Once the sample of water or solid is collected, tech- probable-number (MPN) approach. The approach creates
niques are available to characterize the microbial popula- many incubations that result from plating a large number
tion. In the case of a soil sample, however, an additional of successively diluted samples and appropriate repli-
step is required to remove bacteria from the solid materi- cates of each. The basic idea is to find out how much a
als. The most common procedure involves shaking a sediment slurry or water sample can be diluted before
sediment-distilled water slurry. Both chemical additives no growth takes place. The dilution information provides
(Chapelle, 1993) and glass beads (Kolbel-Boelke and oth- a basis for estimating the number of cells per volume (or
ers, 1988) are sometimes added to assist. mass) of the original sample. While this procedure is
somewhat different from a standard plate count, it can be
Plate Counts adapted as the standard plate count method to consider
A plate count involves smearing a sample (water or sedi- specific contaminants. It has most of the same limitations
ment-water slurry) containing the bacteria on a growth as well (NRC, 1993).
medium, incubating the plate for some time, and count-
ing the bacterial colonies. The growth medium can vary Direct Counting Procedures
but typically contains nutrients and a carbon “food Direct counting procedures use a microscope to count
source” for the bacteria, which are solidified by agar to the bacteria. Stains like acridine orange are used in con-
form a gel (NRC, 1993). The approach requires serial junction with epifluorescence microscopy to facilitate
dilutions because often a sample will create so many counting (NKC, 1993). Following this procedure, cells
colonies that they cannot be counted. By diluting the are separated from the solids as before in a sediment-
starting water sample or sediment slurry more and more water slurry. The fluid fraction containing the cells is
with sterile water, the number of bacteria will also be separated, stained with the fluorescent dye, and filtered
reduced until the number of colonies can be counted. so that bacterial cells are collected on a membrane filter
Accounting for the dilutions, the number of bacteria can (Chapelle, 1993). Under ultraviolet light provided with
be expressed, for example, as colony-forming units per the microscope, the fluorescing cells can be counted.
milliliter (CFU/mL) in the case of water samples, or CFU/ This procedure provides an estimate of the number of
(g dry weight of sediment). cells per gram of material.
This procedure has limitations because not all bacteria
in a sample grow in the environmental conditions af- Biochemical Techniques
forded by the agar medium. One accommodation in the Several tools are able to identify bacteria (NRC, 1993).
testing procedure involves using sterilized soil extract DNA probes identify genetically similar materials in a
that is stabilized by agar as a component in the growth given cell. The degree of match helps to identify the cell.
medium. This organic-poor mixture can increase the One, however, needs relatively detailed information on
number of colony-formingunits because growing condi- the genetic sequences that are likely present in the un-
tions more closely resemble the natural environment. known cell to create a useful probe. Another useful tool
Similarly, when enumerating anaerobic populations, an- is fatty acid analysis. This procedure identifies bacteria
aerobic conditions must be maintained with appropriate on the basis of the unique fatty acids found in the cell
electron acceptors such as nitrate or sulfate. membranes.
We noted that not all bacteria present in a sample
grow. Further, as the samples are slurried, a significant
Rates of Microbial Reactions
proportion of the bacteria will remain on the sediment
surface and not be counted (Chapelle, 1993).Thus, plate Beyond knowing the numbers of bacteria present or
counts provide a minimum estimate of bacterial abun- those able to degrade specific organic compounds, one
dance. often may be interested in the rates of biotransformation
In site cleanups, a concern is not the overall size of reactions. This information is particularly useful in the
the bacterial population, but the size of the population design and evaluation of bioremediation systems. Micro-
of microorganisms capable of biotransforming a specific cosms are one tool to examine the rates of microbial
organic contaminant. Plate counts can provide this infor- processes. The microcosm provides a laboratory setting
mation. The growth medium is modified to utilize the for isolating and studying subsurface environments (Wil-
organic compound of interest as the only carbon source. son and Noonan, 1984).
Thus, by reporting the abundance of contaminant-degrad- In its simplest form, a microcosm consists of a porous
ing organisms, one can examine the biodegradation ca- medium, water, and a functioning population of microor-
 13.3 Microbiological Systems 283

ganisms isolated in a container. The particular design The mixture of aquifer solids and ground water is spiked
is a trade-off between the convenience of operation in with a mixture of contaminants and incubated in the
maintaining a given environment, cost, and the faithful- dark. (Nielsen and Christensen, 1994a).Periodically,sam-
ness with which given processes are simulated (Wilson ples are collected for analysis.The large volume of sample
and Noonan, 1984). With the flow-through systems (Fig- in the container assures that the effect of removing the
ure 13.9a), fluids are moved continuously through the sampies is minimal.
microcosm in a column-style experiment. Examples of degradation curves for four aromatic hy-
Figure 13.96 illustrates a static microcosm. These ex- drocarbons, which were developed with the laboratory
periments typically involve a set of identical microcosms, batch microcosm, are presented in Figure 13.10. The
including three replicates, that contain the sample of experiment was performed with sediment and aerobic
aquifer material spiked with a known concentration of ground water from a shallow glaciofluvial aquifer in Den-
the organic compound or contaminant of interest. These mark (Nielsen and Christensen, 1994a). The microcosms
are incubated in the dark at a fixed temperature. At spe- were incubated in the dark for 149 days. From Figure
cific time intervals, three replicated microcosms in the 13.10, considerable variability in the rate of degradation
set are sacrificed to provide a measurement of the con- is apparent. The rate for biphenyl is approximately 50
centration of the organic compound. A set of sterilized pg L-' day-' while that for p-dichlorobenzene is 1.6 pg
microcosms is also monitored as a control to establish L-' day-'.
that no other processes or sample-handling problems A much more difficult procedure for estimating biodeg-
cause a loss of the organic tracer. These concentrations, radation rates involves tracer tests run with biodegrdd-
plotted as a function of time, describe the rate of loss able organic compounds. Such a test was conducted by
of the organic compound from the microcosm, which Barker and others (1987) at Canadian Forces Base Borden.
usually can be represented by a kinetic model. About 1800 L of ground water were collected from the
Figure 1 3 . 9 illustrates
~ a laboratory batch microcosm. shallow aquifer and spiked with 2360 pg/L benzene,
This microcosm is constructed from a large glass bottle. 1750 pg/L toluene, 1080 pg/Lp-xylene, 1090 pg/L in-

Microcosm Sampling
apparatus

Dilution-water
n
Teflon-lined
screw caD

50-mL test tube

(6)Static Microcosm (c) Laboratory Batch Microcosm

(from Wilson and others, 1985) (from Nielsen and Christensen. 1994)
Figure 13.9 Examples of (a) flow-through (6) static microcosms, in Ground Water Quality, eds. C. H. Ward. W. Giger. and
P. L. McCarty. Copyright 0 1985 by John Wiley & Sons, Inc. Reprinted by permission of John Wdey & Sons, Inc., and (c) a lab-
oratory batch microcosm reprinted from J. Contam. Hydrol., v. 15, by P. H. Nielsen and T. H. Christensen,Variability of hiologi-
cal degradation of aromatic hydrocarbons in an aerobic aquifer determined by laboratory batch experiments, p. 305-320. 1994
with kind permission from Elsevier Science, NL, Sara Burgerhartstraat 25, 1055 KV Amsterdam, The Netherlands.
290 Cbapter 13 / CoUoids and Microorganisms

1.o

0.8b'

.r
0.8

0.21 .
,
b

'c
0.
*r

I
**. 0 .

1
. .
r:%.
I
I
...
..
I
. . I I - -- - +
0
Days Days Days Days
Figure 13.20 Normalized concentrations of benzene, o-xylene,pdichlorobenzene, and biphenyl as a function of time in labo-
ratory batch microcosms for two replicate sediment samples (from Nielsen and Christensen, 1994). Reprinted from J. Contam.
Hydrol., v. 15, by P. H.Nielsen and T. H. Christensen, "Variability of biological degradation of aromatic hydrocarbons in an aero-
bic aquifer determined by laboratory batch experiments, p. 305-320, 1994 with kind permission from Elsevier Science, NL, Sara
Burgerhartstraat 25, 1055 KV Amsterdam, The Netherlands.

xylene, 1290 pg/L o-xylene, and 1280 mg/L CI-. This a pristine ground-water system is relatively meager. Thus,
water was injected into the aquifer just below the water microbial populations appear adapted for growth and
table. The spreading of the various organic compounds survival in nutrient-poor (oligotrophic) conditions (Balk-
with time was monitored using a dense network of multi- will and others, 1989). Theoretically, in a system closed
level piezometers. Figure 13.11 compares the vertically to organic carbon, the concentration of organic matter
integrated concentrations at three different times. Both should decline until it is no longer possible to support
benzene and toluene are biodegrading. The zeroorder the microbial population (Ghiorse *md Wilson, 1988).
biodegradation rates were -30 mg day-' for benzene, However, this end point has not been discovered in large
-37 mg day-' for toluene, and -33, -47, and -55 mg flow systems.
day-' for 0-, m-,and p-xylene, respectively. Again, rates Results from several studies show that even down to
are quite variable depending o n the organic compound several hundred meters, microbial abundance does not
(Barker and others, 1987). diminish with depth. However, there can be considerable
spatial variability both vertically and laterally. For exam-
ple, in holes drilled as part of the Deep Probe Project of
Microbial Ecology of tbe Subsurface
the U.S. Department of Energy, Levine and Ghiorse
The subsurface is home to a diverse set of microorgm (1990) found a highly significant correlation between
isms. In pristine ground-water systems, population densi- bacterial abundance and hydraulic conductivity. Bacteria
ties range between lo5and lo7cells per gram dry weight were more abundant in sandy sediments and much less
(gdw) (Ghiorse and Wilson, 1988). Generally, these pop- abundant in clayey sediments. Similar distributions are
ulation densities are small relative to those found in the observed in shallow environments as well. This pattern
soil zone and the unsaturated zone. This difference in may reflect the much greater difficulty in originally colo-
the size of microbial populations is caused by a decrease nizing the fine-grained sediments because of straining
in the concentration of organic matter as recharging filtration of migrating organisms. In addition, the larger
ground water moves through the soil. fluxes of water through the more permeable units would
The distribution of microorganisms depends in a com- likely make significantly more dissolved organic matter
plex way on the residence time of water in the system available. Fredrickson and others (1990) found significant
and the geochemistry of organic matter trdnsported along diversity in the ability of bacteria to utilize various types
the flow system. As Ghiorse and Wilson (1988) suggest, of organic compounds even when nearby samples were
the quantity of organic matter and nutrients passed along compared. Their data indicate a great diversity in bacteria
 13.4 Microbial Processes 231

0 0 0 0 0 0 0 0 0 0

Chloride Benzene Toluene

0 0 0 0 0 0 0 0

n
0 0 0 0 0 0 0 0

0 0 0

0 0 0 Day 108

0 0

0 0

0
lo 0
0

0
0

0
Day 53

0 0 lo 0 0 Day 3

1: 0
0

l o 0 1 0
I
0 0
0

injection well
0 0 0 I Injection well
0 0 0
Injection well
0 0 0

I I 1 ff I I 1 fi I I I I I I
-4 -2 0 JJ -4 -2 0 JJ -4 -2 0 2 4 6
ycoordinate frn)
Figure 1511 Vertically integrated concentmtions of chloride, benzene, and toluene. Concentrdtion units are nig . m/L for chlo-
ride and pg . m / L for benzene and toluene (from Barker and others, 1987). Reprinted with permission of Ground \Varer Monitor-
ing Review, p. 64-71. Copyright @ 1987. All rights reserved.

at the population and organism levels even within the of providing sufficient energy to maintain the bacterial
same geologic unit. population.
A variety of organic compounds can serve as the pd-
miry substrate for microbial metabolism (NRC, 1993).
-
13.4 Microbial Processes Table 13.1 lists examples of primary substrates. Generic
examples of these primary utilization reactions were pre-
There are two different modes in which organic sub-
sented in Section 12.5 wherein the organic compound
strates participate in biodegradation reactions. With pri-
(the electron donor) reacts with various possible electron
mary utilization, a dissolved organic compound provides
acceptors (for example, oxygen, nitrogen. sulfate). Fol-
the main source of energy and carbon for the microorgan-
lowing are examples of overrill reactions involving these
isms. In effect, the microorganisms produce energy b y
common organic compounds:
the electron transfer involved with redox reactions. In
other cases, an organic compound or substrate of interest
is cometabolized along with some other primary sub- benzene: CGHr) + 7.501 “OL + 3H?0
strate. The secondary substrate effectively is not capable Iiexane: Cr,H,, -I-9.502+ K O : + 7H:O

Tab& 13.1 Examples of Primary Substrates

Aerobic and anaerobic Glucose, acetone, isopropanol, acetate, benzoate, phcnol
Aerobic primarily Alkanes, benzene, toluene, xylene, vinyl chloride, methane. propane
292 Cbapter 13 / Colloids and Microorganisms

Writing the reactions in this way, however, hides the pounds biodegrade in ground-water settings. Table 13.2
detailed biochemistry of the reaction. For example, ben- summarizes several of the most important factors in
zene degradation involves several intermediate com- this respect.
pounds that are related to various enzyme systems. En-
zymes react with the organic substrates to make BioJlm Kinetics
compounds that are able to react at rates appropriate to Biotransformation reactions take place within the bio-
maintain life (Chapelle, 1993). film. There are two coupled processes of interest, the
Some chlorinated solvents, like trichloroethene (TCE), transport of substrates or organic compounds from the
dichloroethene (DCE), dichloroethane (DCA), and vinyl bulk ground water into the biofilm, and the growth (or
chloride (VC), are relatively oxidized compounds and decline) in the mass of the biofilm. The processes are
resist biotransformation because they do not provide coupled because transport of the substrate depends in
much energy. They, however, can be cometabolizedwith part on how much biofilm there is and the mass of the
other organic compounds. For example, Roberts and oth- biofilm depends in part on the availability of the substrate
ers (1989) demonstrate how methanogenic bacteria are as a source of energy (McCarty and others, 1984).
stimulated through the addition of methane and oxygen. Ritman and McCarty (1980a) developed a model to
The oxidation of methane provides the energy and car- simulate biofilm processes where the concentration of a
bon necessary for bacterial growth and leads to the pro- single substrate is sufficiently small to be rate limiting.
duction of enzymes required to degrade the secondary Later, we will examine a case where the availability of
substrates like TCE and VC. Chemically, both TCE and the electron acceptor (for example, 0 3 limits the reac-
VC end up being mineralized to C02.This possibility for tion rate.
cometabolizing chlorinated solvents holds promise as a The organic compound moves by diffusion through
remedial approach for aquifers contaminated by TCE. the diffusion layer and into the biofilm and ultimately is
Other organic solvents like 1,1,1-trichloroethane lost by bacterial utilization. The following empirical ki-
(TCA),tetrachloroethene (PCE), TCE, DCE, DCA, and CT netic law describes the rate of loss (Bouwer and
undergo secondary utilization in reduction reactions. The McCarty, 1984):
organic compound in this case is reduced and another
compound (for example, a transition metal) is oxidized.
However, unlike the oxidation-type reactions that lead to (13.1)
the mineralization of the compound, reduction reactions
typically produce a less chlorinated compound. where S, is the concentration of the rate limiting sub-
strate, X, is the active cell density in the biofilm (mass of
bacteria L-3), k is the maximum specific rate of substrate
Issues in Biodegradation utilization (mass of substrate/mass of bacteria per unit
Before leaving this discussion on microbial reactions, we time), and K, is the Monod half-maximum-rateconcentra-
will mention briefly some of the factors that come into tion (mass of substrate L-'). K, is t h e concentration at
play in determining whether and how organic com- which the specific rate of utilization is one-half of k.

Table 23.2 Summary of Complicating Factors Affecting the Microbial Degradation of Organic Compounds

Factor Explanation
Toxicity Relatively large concentrations of organic substrates can prevent or slow metabolic reactions (NRC,
1993). How this factor comes into play depends on the particular substrate and its concentration, as
well as features of the microbial population.
Acclimation Often a time lag of days to months occurs before the microbial population begins to degrade the or-
ganic compound. Explanations for the lag include time necessary to expand an initially small popula-
tion, adaptation to a change in electron acceptor, and possibly time for genetic changes to occur
(Chapelle, 1993).
Low concentrations Microorganisms often cannot reduce substrate concentrations below some minimum level. Rittman
and McCarty (1980a, 1980b) define the minimum substrate concentration as S,,,,. During primary utili-
zation, substrate concentrations below S,,,,, cannot support the growth of the biofilm.
Sequestration Although organic contaminants may be abundant in the subsurface, they may be unavailable to micro-
organisms. This situation can develop when contaminants occur as nonaqiieous phase liquids or are
sorbed to aquifer solids. Biodegradation often requires that substrates enter the cell, a difficult process
in these cases (Alexander, 1994). Enzymes and/or inorganic nutrients are subject to sorption, thereby
interfering with the biodegradation process.
Mixtures Often with a mixture of organic contaminants, some compounds are selectively utilized, while others
are ignored until the first compound is completely utilized.
 13.5 Biotransfomation of Common Contaminants 2
x

Substrate utilization is a function of two constants (k biodegradation. The reaction rate increases when addi-
and K J , the substrate concentration, and the active cell tional oxygen is provided.
density X,. McCarty and others (1984) present expres- The mathematical description of material transport to
sions for bacterial growth, in essence describing how X, the biofilm in this case would involve two transport equa-
changes with time as the difference between growth tions, one for the substrate and one for the electron
rates (that is, substrate conversion) and loss rates. By acceptor. Again, both reactants move by diffusion in the
solving the coupled mathematical problem of substrate diffusion layer and film and disappear through utilization
transport and bacterial growth, it is possible to model a by bacteria. The two governing transport equations are
steady-state biofilm system. coupled through the reaction term because the rate of
For a secondary substrate with S, much less than K,, substrate utilization (following Bailey and Ollis, 1986) is
Eq. 13.1 simplies to the following first-order rate law a function of both the substrate and electron acceptor.
(Bouwer and McCarty, 1984): This rate expression is a modified version of Eq. 13.1.
The kinetic equation for the utilization of the electron
(13.2) acceptor depends on both concentrations as well.
More sophisticated biofilm models can account for
Equation 13.2 states that substrate utilization within the contributions of both the substrate and electron ac-
biofilm is proportional to the concentration of bacteria ceptors to cell growth. Borden and Bedient (1986) and
and the ratio k / K , . Under some conditions, bacterial utili- Molz and others (1986) have embedded biofilm processes
zation, as described by Eq. 13.2, can be the rate limiting in a contaminant transport model where both substrate
process controlling transport of the secondary substrate and oxygen transport are modeled together with chang-
within the biofilm. In other words, diffusion is relatively ing cell mass.
fast through the biofilm and boundary layer relative to
this loss rate. In this situation, we can model the biofilm
at a macroscopic scale by a simple kinetic expression
(that is, Eq. 13.2). The reaction is characterized by a half-
-
13.5 Biotransformation
of Common Contaminants
life of substrate decomposition (Bouwer and McCarty,
1984): Chapter 17 examines families of organic compounds that
are prevalent in problems of ground-water contamination
(13.3) (Table 13.3). Here, we will briefly survey biotransforma-
tion processes in relation to these families. Commonly,
where Eq. 13.3 comes from the integration of Eq. 13.2. biotransformation processes are examined in relation to
This model is appropriate when the concentration of two prototypical settings. In aerobic settings, oxygen is
active organisms is small and ground-water movement is the major electron acceptor for the various biotransfor-
slow (Bouwer and McCarty, 1984). mation reactions. In anaerobic settings, oxygen is absent,
Conditions exist when the biodegradation of a sub- and other compounds, like nitrate, sulfate, and carbon
strate is limited by the availabilityof an electron acceptor. dioxide, function as electron acceptors.
For example, the aerobic biodegradation of an organic
liquid in water can utilize the small quantities of oxygen
Hydrocarbons and Derivatives
naturally dissolved in the water. In spite of an abundance
of organic substrate, the lack of sufficient oxygen, as Hydrocarbons and related compounds constitute one of
dictated by the stoichiometry of the redox reaction, limits the most important contaminant families in ground wa-

Tab& 23.3 Important F a d e s of Organic Contaminants Found In Ground

Water
Chemical Family Examples of Compounds
Hydrocarbons and derivatives
fuels benzene, toluene, o-xylene, butane, phenol
PAHs anthracene, phenanthrene
alcohols methanol, glycerol
creosote mcresol, o-cresol
ketones acetone
Halogenated aliphatics tetrachloroethene,trichloroethene, dichlorornethane
Halogenated aromatics chlorobenzene, dichlorobenzene
Polychlorinated biphenyls 2,4'-PCR, 4,4'-PCB
-
294 Chapter 1.3 / CoUoids and Microorganisms

ter. In aerobic ground-water systems, aromatic hydrocar- CCI, = CCI,

bons with up to two benzene rings are mineralized rela-
tively rapidly with minimal lag times. For example, the 1
CHCl = CCI,
microcosm studies of Nielsen and Christensen (1 9 9 4 ~ )
found that benzene, toluene, and o-xylene degraded with
short lag times (about 4 days). Initial concentrations of
1
CHCl = CHCl
approximately 150 pg/L of benzene, and toluene de-
creased to <2 pg/L in less than one month. The degrada- 1
CH, = CHCl
tion of o-xylene required about three months and was
substantially less complete. These laboratory findings of Figure 1312 Reductive dehalogenation from PCE to TCE
to DCE and VC.
relatively rapid degradation rates are also corroborated
by field tests. For example, the rapid biodegradation of
BTX compounds in field tests at Canadian Forces Base anogenic conditions. Smolenski and Sulflita (1987) found
Borden was discussed previously. that p-, m-,and o-cresol also degraded under anaerobic
Hydrocarbons degrade in anaerobic systems as well. conditions. The remaining hydrocarbons in Table 13.3,
Convincing field data come from a study of an oil spill at alcohols and ketones, are readily degradable in ground-
Bimidji, Minnesota (Cozzarelli and others, 1988). Detailed water.
chemical sampling indicated that under anaerobic condi-
tions concentrations of benzene, toluene, and 0-, m-,and
Halogenated Aliphatic Compounds
p-xylene decreased in ground water over a distance of
approximately 10 m. Ethylbenzene, however, was resis- Halogenated aliphatic compounds include the common
tant to biodegradation. Lyngkilde and Christensen (1 992) industrial solvents like carbon tetrachloride (CT), tetra-
followed the transport of a variety of organic compounds chloroethylene (PCE), trichloroethylene (TCE), trichloro-
within a landfill-leachate plume. Along the first 100 m ethane (TCA),vinyl chloride (VC), and dichloroethylene
of the plume, a redox sequence from methanogenic, (DCE). These compounds are formed from carbon atoms
sulfidogenic, to ferrogenic conditions prevailed. Over the joined together in chains with attached atoms. In aerobic
first 20 m (methanogenic-sulfidogenic environment), systems, they resist degradation and are extremely persis
there was some indication that various isomers of ben- tent. There are no known bacteria that can oxidize these
zene as well as toluene and xylenes were able to degrade. compounds as a primary substrate because the reaction
Within the ferrogenic zone, there was obvious degriida- yields so little energy (Chapelle, 1993). Thus, a major
tion of these compounds as well as ethylbenzene. activity for contaminant hydrogeologists is in cleaning
Other components of fuels (or coal-tar derivatives) are up solvent spills.
polyaromatic hydrocarbons (PAHs) such as naphthalene, An important exception to this general result is when
anthracene, or pyrene. These compounds are comprised the aerobic system contains methane either naturally or
of a series of benzene rings. In general, PAHs are biode- applied in a bioremediation scheme. Compounds like
gradable in aerobic settings, with the rate decreasing as TCE may be metabolized as a secondary substrate when
the molecular weight of the compound increases. Durant methane is present as a primary substrate. Under anaero-
and others (1W5) found through microcosm studies that bic conditions (methanogenic), halogenated compounds
bacteria were capable of aerobically mineralizing naph- are commonly biotransformed. The reactions involve a
thalene and phenanthrene at the site of a former manufac- sequential loss of chloride atoms in a processes termed
tured gas plant. In anaerobic systems, PAHs may be some- reductive dehalogenation. Figure 13.12 illustrates reduc-
what more resistant to degradation. However, Lyngkilde tive dechlorination reactions beginning with PCE. The
and Christensen (1992) report the reduction of naphtha- last step from vinyl chloride to COz is quite slow relative
lene concentrations under ferrogenic conditions. to the other dechlorination reactions, so that vinyl chlo-
Phenolic compounds are also found as components in ride often tends to accumulate.
fuels or as industrial contaminants. As discussed in Sec-
tion 1 1.7, phenols are characterized by an aromatic ring
Halogenated Aromatic Compounds
with an attached hydroxyl group. Examples of these com-
pounds include phenol, nitrobenze, o-cresol, and o-ni- Halogenated aromatic compounds have a structure char-
trophenol. Microcosm studies by Nielsen and Chris- acterized by a benzene ring and attached halogens. Exam-
tensen (19946) show that these compounds are ples of these compounds include chlorobenzene, dichlor-
mineralized in aerobic systems. Indications are that they obenzene, and trichlorobenzene. In general, these
biodegrade relatively rapidly with broad variation in the compounds are degradable under aerobic conditions
rates. At a former wood treatment facility in Pensacola, (Chapelle, 1993). However, the ability to biodegrade is
Florida, Bekins and others (1993) reported the degrada- reduced as the molecule becomes more highly chlori-
tion of phenol and 2-, 3-,and 4-methylphenol under nieth- nated (NRC, 1993). In anaerobic systems, these com-
 Problems 2 3

pounds degrade via reductive dechlorination reactions,

although there often may be a time lag (Chapelle, 1993).
Nielsen and others (1995) examined the fate of 0-and
p-dichlorobenzene in the leachate plume of the Vejen
landfill. Under anaerobic conditions ranging from metha-
CH,CCI, CH,CHCI,
nogenic to iron(II1) reducing, these compounds were
not transformed.
A

Polychlorinated Bipbenyb (PCB)

CH,CHCI CH,CH,CI I
PCBs were used historically as a dielectric compound in
transformers. Although the manufacture and use of these
compounds has ceased, they are still occasionally discov-
ered at sites. The biphenyl structure involves two linked
benzene rings with between one and 10 attached chlo-
rine atoms per molecule. Generally, only less chlorinated
PCBs oxidize under aerobic conditions. The more chlori-
B
CH,CH20H

rfj
CH,COOH

Acetic

nated PCBs are generally very resistant to oxidation under c*2 co2 COZ
natural conditions. Under anaerobic conditions, highly
Figure 13.13 Probable fate of TCA under methanogenic
chlorinated PCBs transform via reductive dehalogena- conditions: biotic transformation pathways are denoted by
tion reactions. lines marked B and abiotic transformation pathways by lines
marked A . Reprinted with permission from Vogel. T. M..and
McCarty, P. L., Environ. Sci. Technol., v. 21, 198'. p. 1209.
Compkx Transformation Pathways
Copyright 0 1987 American Chemical Society.
Both biotic and abiotic transformation reactions can oper-
ate on organic compounds in a complicated way. Con-
sider the common industrial solvent l , l ,l-trichloroeth- middle pathway (Figure 13.13) by reductive dehalogcna-
ane (TCA), which is often a contaminant in ground water tion to 1,l-dichloroethane (1,l-DCA) and chloroetliane
(Vogel and McCarty, 1987). Under abiotic conditions, (CA). The next reaction along this pathway, the hydroly-
TCA transforms to 1,l-dichloroethylene (1,l-DCE) along sis of CA to ethanol, has a pseudo-first-order rate constant
one reaction pathway and to acetic acid along another of about 0.37 yr-I, giving CA a half-life of about 1.9 yr
(Figure 13.13). The pseudo-first-order rate constants for (Vogel and McCarty, 1987). In biologically active sam-
the formation of 1,I-DCE and acetic acid from TCA are ples, ethanol would be transformed to C 0 2 .
about 0.04 yr-' and 0.20 yr-', respectively (Vogel and Exactly which of these pathways will be dominant and
McCarty, 1987). These intermediates will transform bio- which intermediates of TCA are found at a site depend
logically under methanogenic conditions to vinyl chlo- on the potential for methanogenic biotransformation. Ir-
ride WC) and C02, respectively (Figure 13.13). In the respective of the pathway, however, the intermediates
presence of a highly active population of methanogenic should be present because of the rate limitations imposed
bacteria, most of the TCA would biotransform along the by the various reactions.

Problems

1. For an aerobic reaction, k / K , is estimated to be 5.0 4 84.5

L/mg cells per day. Given a concentration of microorgan- 8 58.9
isms of 0.01 mg cells/L or lo4bacteria/mL, estimate the 15 30.5
half-life of substrate utilization.
Estimate the decay rate constant and the half-life for this
2. In a laboratory experiment, the concentration of an particular compound.
organic compound is found to decline over a period of
15 days. 3. Assume TCA occurs in an aquifer with minimal bio-
logical activity at a concentration of 100 pg/L. (a) After
Elapsed Time Concentration five years, what other compounds besides TCA would
(days) W.1 be present? (b) Use the transformation rate data provided
0 123.0 in this chapter to estimate which of the various com-
1 111.7 pounds would be most abundant.
2 102.2
 CbaDter 14

The Equations of M a s s Transport

14.1 Mass Transport Equations

14.2 Mass Transport with Reaction
14.3 Boundary and Initial Conditions

Hydrogeologists traditionally have worked to extend It remains to describe the nature of the flux J, which
their understanding of ground-water flow through the could be diffusive transport or advective transport com-
precise language of mathematics. Not surprisingly, this bined with diffusion.
same approach has been applied to problems of mass The right-hand side of the conservation statement as-
transport. The equations developed in this chapter pro- sumes that the gains or losses in mass inside the volume
vide a comprehensive framework for quantitatively de- are proportional to changes in concentration in mass
scribing mass transport with and without accompanying inside the unit volume. Thus, a proportionality constant
chemical reactions. These equations are at the heart of is required that expresses the mass lost or gained when
approaches used to model the chemical evolution of the concentration changes a unit amount. In this case,
natural ground water and contaminant migration. the proportionality constant is unity because mass occu-
pies the pores, where concentration is mass per unit
volume of fluid (rn/V,) and porosity n for a fully satu-
14.1 Mass Transport Equations rated medium is the fluid volume per unit total volume
(VJV,). Thus, the product Cn is mass per unit total
As with flow, the starting point for developing the mass volume and the conservation statement becomes
transport equations is a conservation equation. This equa-
tion in words is
(14.1)
mass inflow rate - mass outflow rate =
change in mass storage with time This equation states that the net mass outflow rate per
unit volume equals the time rate of change of mass within
in units of mass per unit time. This statement applies to
the unit volume.
a domain of any size. For a representative volume with
inflow through three sides, the left-hand side of this word
equation is the divergence of the flux of interest (see The mflusion Equation
Chapter 4), that is,
For a diffusive system defined by Fick’s law of the form
- div J = net mass outflow rate per unit volume in Eq. 10.9, Eq. 14.1 becomes
236
 14.1 Mass Transport Equations 237

a (Cn)
div [D;n grad C] = - (14.2)
Expanding the advection terms and assuming const;int
iJ t D; and n give

With the further assumption that porosity does not vary

spatially or temporally, and that the diffusion coefficient
D;V'C - V -VC + CV * v = -
i Jt
'" (14.8)
is constant spatially, we can write
For steady ground-water flow, V v = 0 and
& div [grad C] = -
ac dC
at D;fV'C - V - V C = - ( 1 4.9)
i)t
or
When the velocity goes to zero, that is, no advection.
we recover the diffusion equation for a stationary fluid,
Eq. 14.4. When the concentration profile is steady. Eq.
14.0 becomes
This diffusion equation is analogous to the unsteady flow
equation introduced in Chapter 4 and the unsteady con- D g l C - V'VC = 0 (14.10)
ductive heat transport equation in Chapter 9. The solu-
tion to this equation determines how the concentration Equations 14.9- 14.10 are called the advection-
at any point in a three-dimensional field is changing with diffusion equations. They describe the manner in which
time. If the net outflow is zero so that the concentration mass is moved from one point to another by advection
is not changing in time as modified by diffusion. The one-dimensional form of
Eq. 14.9 is

(14.11)
Equation 14.4 is Laplace's equation, as were the steady-
state fluid flow equation and the steady-stateheat conduc- The first term on the left describes mass transport by
tion equation. The solution of this equation describes diffusion, and the second describes mass transport by ad-
the value of the concentration at any point in a three- vection.
dimensional field. Before leaving the advection-diffusion equation, we
would like to consider the concept of dimensional analp-
sis as applied to mass transport. In Chapter 4. the method
The Advection-Diflusion Equation of dimensional analysis was introduced and applied to the
For systems with fluid motion, mass transport can be due unsteady flow equation. The result was a dimensionless
to both advection and diffusion. This combined flux can group known as the Fourier number. Applying these
be described mathematically by combining the advective same ideas to the advection-diffusion equation lets us
flux in Eq. 1 0 . 1 and the diffusive flux in Eq. 10.4, or define a dimensionless concentration C- = C/C,., where
C , is some characteristic concentration and the super-
ac
= -nD: - + v,,cn
1,. (14.5a)
script + indicates a dimensionless quantity. The diffu-
dX sion-advection equation in dimensionless form is
I1C
J,. = -nD: 7 + V,.Cn (14.5b)
dY
ilC
J= = -n& - + u,Cn (1 4.5c) where the dimensionless quantity ilL/D,; is the Peclet
8.z number for mass transport (NpE)with L as some ch;ir;icter-
where v,Cn is the advective flux. istic length. The Peclet number expresses the transport
by bulk fluid motion or advection to the mass transport
For isotropic conditions, by diffusion. The dimensionless quantity L'/t,.D: is the
Fourier number for mass transport.
J = -nD$ grad C + vCn (1 4.6)

where v, the linear ground-water velocity, is a vector with The Advection-Dispersion Equation
components v,, z!,., v,. Substitution of the rate expression
(Eq. 14.6) into Eq. 14.2 and carrying out the necessary Diffusion alone does not fully account for mixing during
operations gives mass transport. Scaledependent mechanical dispersion
in many systems is usually more important. Logically. if
dispersion is to be incorporated in the advection -
V.(nD,;VC) - V (vCn)
* =-
a(nc) ( 1 4.7)
at diffiision equation, it should be reflected in the velocity
238 Cbapter 14 / Tbe Equations of Mass Transport

term. Unfortunately, it has not been possible to do this law. This law, for example, can be expressed in terms
in a simple way. Instead, the coefficient of hydrodynamic of the rate of decrease of a reactant, or the rate of increase
dispersion incorporates the combined effects of diffusion of a product, depending on which constituent is being
and mechanical dispersion. If the dispersion coefficient described by the transport equation. It may be necessary
is constant, this equation is to track both a reactant and a product using two coupled
transport equations. One example is particular kinds of
DV'C - V *VC i- C v * v = -
ilc (14.13) biotransformation reactions where both the reactant and
at the products are contaminants.
Prior to examining some rate laws, let us consider the
where D = D' + 0;.The justification for treating disper-
concept of dimensional analysis as applied to transport
sion in this manner is purely a practical one and stems
with reaction. In order to do so we replace the quantity
from the fact that the macroscopic outcome is the same
r / n in Eq. 14.14 by the product of a reaction rate coeffi-
for both diffusion and mechanical dispersion. The actual
cient k and concentration C,that is, kC where k is the
physical processes, however, are entirely different. By
volume reacted per unit volume per unit time. Applying
replacing the diffusion coefficient by a coefficient of hy- the dimensionalizing procedures to the steady-state form
drodynamic dispersion, the mixing process is Fickian. of Eq. 14.14,we arrive at two dimensionless groups
Thus, in cases where mechanical dispersion is small com-
pared to diffusion, Eq. 14.13 reverts to the advection-
diffusion equation. In the absence of ground-water flow, k?L (Damkohler number I)
2,
Eq. 14.13 reduces to the diffusion equation.
The advection-dispersion equation is the workhorse kLL
D
(Damkohler number II)
of modeling studies in ground-water contamination. We
will have much to say about its solution and application
later in the book. where the Damkohler numbers may be taken as a mea-
sure for the tendency for reaction to the tendency for
transport. Damkohler number I is important at high
-
14.2 M a s s Transport with Reaction
Peclet numbers, and Damkohler number I1 is important
at low Peclet numbers. Note the ratio of number I1 to
The equations presented so far describe the processes number I reduces to the Peclet number.
of advection, mechanical dispersion, and diffusion. With Let us now examine some rate laws and show how
reactions, the statement of mass conservation expands they are included in mass transport equations. Before
to incorporate sources of material within the porous proceeding, however, we will review the terminology
volume. The equation is modified by source or sink terms used to organize the discussion. Reactions are classified
depending on whether a constituent is being added or as homogeneous, operating within a single phase, or
removed by chemical processes. The appropriate state- heterogeneous, operating between two phases. In addi-
ment of mask conservation when reactions are consid- tion, reactions can be described from an equilibrium or
ered is kinetic viewpoint depending on the rate of the reaction
relative to the mass transport process.
mass inflow rate - mass outflow rate 2
mass production rate = change in mass storage with time First-Order Kinetic Reactions
The plus or minus term designates either a source or a One example of a simple kinetic reaction is the first-
sink and takes on different forms for different reactions. order decay of a constituent due to radioactive decay,
At this point, it is convenient to represent such sources biodegradation, or hydrolysis.
or sinks in a symbolic form. For example, in the case of
onedimensional transport, the word equation above be- (1 4.15)
comes

4,-a2c- q-
aC r ac
2 -=- (14.14)
where A is the decay constant for radioactive decay or
a2 ax n at some reaction rate coefficient for biodegradation or hy-
drolysis. In all cases, A has units of time-' and is the
where r is taken symbolically as the mass produced or volume reacted (or disintegrated) per unit volume per
consumed per unit volume per unit time, moles/L"T.This unit time. With this formulation, the onedimensional
equation applies only to a single constituent. Involving transport equation becomes
several dissolved species in the transport requires a sys-
tem of such equations, one for each constituent.
Source terms are usually specified in terms of a rate
 14.2 Mass Transport with Reaction 2 3

Equilibrium Sorption Reactions where K,, is the distribution coefficient. Combining Eqs.
14.21, 14.22, and 14.23 gives
An example of a heterogeneous equilibrium reaction is
the sorption of mass from solution. The rate law is
(14.24)
r = -dC* ( 14.17)
at which is a function of the fluid concentration alone.
Substitution of Eq. 14.24 into the general transport equ;i-
where C is the concentration of the solute on the solid
tion and rearranging terms give
phase. The one-dimensional transport equation that in-
corporates this reaction is

The bracketed quantity o n the right-hand side is ;I con-

The rate law incorporated in Eq. 14.18 is a general one, stant termed the retardation fctctor, R, , or
appropriate for both equilibirum and kinetic sorption
reactions. A solution to this equation requires that a spe-
cific rate law be incorporated. For kinetic nonequilibrium
sorption reactions, Equation 14.25 becomes
ac*
-= s i c , C*) ( 14. I 9) (14.27)
at
or, in words, the rate of sorption is a function of both Thus, the retardation factor merely serves to decreise
the concentration of mass in solution and the mass sorbed the values of the transport parameters D and 1 7 . As local
on the solid. For equilibrium sorption reactions, equilibrium implies that the net rate of reaction is zero,
we anticipate similar effects for other equilibrium-con-
C’ =f ( C ) (14.20) trolled isotherms of the formf ( C )although the niathemat-
Because the concentration of sorbed mass is a function ics do not work out so simply. Aris and Aniundson ( 1973)
of mass in solution, the &?/at term in Eq. 14.19 can be cite several other forms of sorption isotherms. which ;ill
expressed in a more tractable dC/r?t term. This substitu- have this property. Equilibrium-controlled reactions form
tion produces a single differential equation containing a basis for much of the modeling conducted in contami-
one dependent variable, which is solvable by analytical nant hydrogeology. Similar assumptions have also been
methods. employed by geochemists in models describing the din-
The following derivation explains how the transport genesis of marine sediments (Berner, 1974). by chemical
equation for the simple case of linear sorption is obtained. engineers interested in the effects of diffusion and longi-
The rate of mass sorption per unit volume of porous tudinal dispersion in ion exchange and chromatogr;iphic
medium is columns (Lapidus and Amundson, 1952; Kasten and oth-
ers, 1952), and by soil scientists interested in the effects
(1 4.21) of dispersion and ion exchange in porous niedi;i (Lii ;ind
Jurinak, 1971; Rubin and James, 1973).
where S is the quantity of mass sorbed on the surface
and ph is the bulk density. The bulk density can also be Heterogeneous Kinetic Reactions
defined as
The general case of a reaction between a solid and ;I
Ph = PS(1 - n) (1 4.22) solution involves several steps in series. with the slowest
being rate controlling. The rate-determining step can he
where ps is the mass density of the minerals making up
transport controlled or surface controlled. depending o n
the rock or soil, normally 2.65 g/cm2 for most sandy
the relative magnitude between the rate at which the
soils. The quantity p,(l - n) is the total mass of solids
mass moves to or reacts on the surface.
per unit volume of porous medium.
The starting point for discussing transport-controlle~l
What now is required is an expression for an equilih-
reactions is a model of the reaction process. Assume that
rium sorption isotherm. Although several are available.
a thin stationary layer occurs at the interfilce between ;I
we will use the simplest (and likely most useful) one,
solid and the bulk fluid and that diffusion is slow relative
the linear Freundlich isotherm. By taking this isotherm
to actual reactions at the surface so that the coming
and differentiating with respect to time, w e obtain
and going of atoms is rate controlling. If the process o f
transport across the thin layer is diffusion. there must be
( 14.23)
a concentration gradient. This gradient can be assunietl
-
300 Cbapter 14 / Tbe Equations of Mass Transport

linear and represented as (C; - C ) / u where C:,, is the controlled reactions, the combined parameter D d / u is
saturation concentration at the solid surface, C is the referred to as the mass transfer coefficient LT-' and is
concentration in the bulk fluid, and u is the thickness of equivalent to k , in Eq. 14.33. The combined parameter
the stationary layer separating the two.The fluxJ across DS*/u has the units T-'.
the surface in moles vol-' time-' is The use of the reversible equation for sorption is wide-
spread in transport problems of an academic nature.
Transport equations incorporating this term have been
(14.28)
solved first for the condition of advection only (Amund-
son, 1950), and later for advection and dispersion (Lapi-
whereA is the surface area of the material per unit volume dus and Amundson, 1952). Deju (1971), supported by
of water and Dd is the diffusion coefficient. The ratio ud/ earlier work (Deju and Bhappu, 1970), used this form
u is the mass transfer coefficient k,,and has the units of of adsorption to describe the chemical weathering of
velocity. In this form, the rate of dissolution of the solid sediments. More recent theoretical studies dealing with
is the product of a driving force, which is the departure sorption in relation to contamination problems includes
from saturation (Cz,,- C), a mass transfer coefficient, and the work of Relyea (1982), Melnyk and others (1983),
a surface area of the material per unit volume. Diffusion- Valocchi (1985) and Bahr and Rubin (1987). However,
controlled reactions of the form represented in Eq. 14.28 kinetic models for sorption are rare in field applications
can be expressed as because of the difficulty in determining the rate expres-
sions.
(14.29)

where S* is the specific surface of the porous medium

(Domenico, 1977). It is possible to include Eq. 14.29 in
-
14.3 Boundary and Initial Conditions
a mass transport model. Solving timedependent differential equations requires
Now let us consider the surface-controlled reactions. boundary and initial conditions. The task of the boundary
The best examples of these reactions are the sorption conditions is to account for the effects of the system
and desorption reactions. According to Klotz (1946), one outside of the region of interest on the system being
of the earliest formulations assumes that the irreversible modeled. Once the boundary conditions are specified,
local rate of sorption is the outside effects are accounted for and need not be
directly modeled. The boundary conditions discussed ear-
act - k,C(C:,, - C*) (14.30) lier for flow have their counterparts in relation to mass
n at transport. The first boundary condition is a specified
concentration along a given boundary. An important spe-
where k , is a constant and c;S,is the saturation concentra-
tion on the solid. Thomas (1944) considered the form cial case for this boundary condition is where the speci-
fied concentration is constant, termed a constant concen-
i a c t - k,C(C:, - C*)- kZC* tration boundmy. A second type of boundary is where
(14.31) the normal component of the concentration gradient is
n at
specified. Given a normal component of the concentra-
where kz is a constant. For a porous medium, the follow- tion gradient along a boundary, there must be a diffusive
ing equation describes a reversible sorption process: flux across the boundary. Thus, we use the term mass
fluboundary where mass enters or leaves a region
act - S*(k,C - bzC*) (14.32) across a boundary. One special type of mass flux bound-
n at
ary is a no-flux boundary, that is, a region across which
where k l is a forward (sorption) rate constant and k2 no mass is permitted to leave. Expressed mathematically
is a backward (desorption) rate constant with units of
velocity. This equation states that sorption and desorp- _
aC -
-0 (14.34)
tion occur simultaneously and at different rates. This aN
equation can be rewritten as
where N is the normal to the boundary. This means there
is no concentration gradient normal to the boundaries
= S*kl(C - C:,,) (14.33) so mass cannot leave the region. A second special type
n at
of mass flux boundary is a variable-flux boundary.
for the case where (kl/k2)C*= C;. This equation is of The onedimensional problem in Figure 14.la, involv-
the same form as the diffusioncontrolled reaction dis- ing contaminant migration in an aquifer away from a
cussed earlier. Here the rate constant bl has the units long narrow ditch, provides an example of boundary
LT-' and the specific surface has the units L-I so that conditions. The contaminants enter the system at an u p
the combined term has the units of T-I. With diffusion- per boundary at x = 0. A lower boundary not shown in
 14.3 Boundaty and Initial Conditions 301

stant concentration or constant mass flux boundary

overspecifies what the final result will look like so that
a variable-flux boundary is called for and is of the form
D ac
vc- - (14.35)
f3X

This equation states that at the exit the rate at which

Upper boundary No-f low boundary
some contaminant leaves the system must be equal to
(n)
the combined rate of advection and dispersion by which
the species is transported through the system.
,D'tc River In many mass transport problems (e.g., Figures 14.la,
b), contaminants enter the system through one or several
boundaries. This loading can be specified using either a
fixed concentration or a variable-flux boundary condi-
tion. Equation 14.35 is an example of a variable-fluxcon-
dition. A constant-concentration condition for a one-di-
t t mensional case is
Upper boundary Lower boundary
(6) C ( 0 , t ) = g(t) (14.36)
Figure 14.1 Contaminant migration in a onedimensional This equation indicates that at the boundary (x = 0) the
system that is (a) semi-infinite or ( b ) finite in length.
concentration is given for all time as a function of time.
Examples of some commonly used boundary condi-
tions for onedimensional systems are listed on Table
14.1. The second condition shown on the table is an
the figure is located far down the system at x = 00. In this
example of pulse loading. The boundary condition
figure, the source (ditch) can be depicted as a constant
changes after some time (to)has elapsed. Thus, it is neces-
concentration boundary and the lower impermeable bot-
sary to provide two boundary conditions and the time
tom as a no-flux boundary. A right-hand boundary need
for which each applies.
not be specified as the boundary is too far removed from
An initial condition describes the distribution of a con-
the interior to have any effect. The location of this bound-
taminant within the domain when the simulation begins
ary makes the system semi-infinite. These same condi-
(time = 0).For a onedimensional problem, a general
tions for the ditch and impermeable bottom boundary
form of an initial condition is written
apply in Figure 14.lb, but now something must be said
about the right-hand side boundary at the river. A con- C(x, 0) = f ( x ) (t = 0) (14.37)

Table 14.I Examples of Some Boundary Conditions That Are Used to Add Contaminants to a One.
Dimensional System
Name -me Form
Constant concentration Fixed concentration C(0, t) = C,,

Pulse-type loading with constant concentrations

Exponential decay with source concentration 0
Fixed concentration
Fixed concentration
C(0, t ) =
I
c,,, 0 < t 4 t,,
0,t > t,,
C(0, t) = C,,e-"'
Exponential decay with source concentration + C, Fixed concentration C(0, t ) = C, + C,,e-"'

Constant flux with constant input concentration Variable flux 2+

( -
-D VC = tC,.

Pulse-type loading with constant input fluxes Variable flux (-D$ + vC) = IIC,,,0 <t t,,

C,, ,C., Cb= various constant concentrations.

(Y = decay constant.

to = time at which concentration changes due to pulse loading.

-
302 Cbapter 14 / Tbe Equations of Mass Transport

where@) is a function describing the onedimensional tions and boundary conditions is the first step in modeling
variation in concentration in the x direction. Some com- mass transport. The step of solving the resulting equation
mon initial conditions are C(0,x) = 0,C(0,x) = C,,which using various modeling approaches will be taken up in
provide for a constant concentration along the system. Chapter 18.
Formulating the transport problem in terms of equa-

Problems
1. Consider onedimensional steady-state diffusion name two other parameters or groups of parameters that
across a layer of thickness L where the concentration C have these same units, one treating fluid flow and the
at the bottom (x = 0) and at the top (x = L) is known. other for the transport of heat by conduction?
Solve this problem for the steady-state concentration as 4. Compare the Fourier number of Eq. 14.12 with the
a function of x, L , C , , and C,. (Hint: See Example 4.3.) Fourier number for fluid flow (Eq. 4.83) and the Fourier
2. Provide a dimensional analysis of the steady-stiate number for heat conduction (Eq. 9.22). State the analo
version of Eq. 14.14 where v / n is replaced by kC,and gous quantities and provide an interpretation of T' for
obtain two Damkohler numbers. mass transport by diffusion.
3. The diffusion coefficient has units of L2/T. Can you
 Chapter 15

Mass Transport i~ Natural

Ground-Water Systems

15.1 Mixing as an Agent for Chemical Change

15.2 Chemical Reactions in the Unsaturated Zone
15.3 Chemical Reactions in the Saturated Zone
15.4 A Case Study of the Milk Riuer Aquifer
15.5 Age Dating of Ground Water

The mass transport processes and chemical concepts composition. Mechanical dispersion and diffusion crt.ate
presented in the last four chapters provide a framework a zone of mixing between the two waters. Accordingly.
for studying the chemically related problems in hydroge- phenomena thought of most often in relation to labora-
ology. This general framework applies equally well to a tory columns operate on a much larger scale and strongly
variety of core hydrogeological problems like the chemi- influence the chemistry of natural ground waters.
cal evolution of ground water along flow systems or
ground-water contamination. Moreover, this process ap-
proach lets us address other important problems like the
The Mixing of Meteoric
diagenesis of carbonate rocks, karst formation, the origin and Original Formation Waters
of some kinds of ore deposits, and the secondary migra- There are several reasons why ground-water mixing c;in
tion of petroteurn, which historically have resided in occur. A common situation is related to large-sc;ile struc-
other areas of geology. tural changes in a sedimentary basin when. following
This chapter examines the processes that control the uplift, meteoric water has the opportunity to tlusli unit5
chemical evolution of ground water and some manifesta- that previously contained connate water o r formation
tions of these processes in relation to the geologic water. Meinzer (1923) defined connate water ;is w;iter
“work” of ground water. This chapter also examines that is trapped in the pores of a sediment during deposi-
approaches used in the dating of ground water. tion. For marine deposits, this suggests forni;ition water
of the composition of seawater. Whether o r not this
composition is retained up to the beginning of burial
-
15.1 Mixing as an Agent depends strongly on the depositional environment ;IS
well as numerous other geologic factors. For ex;iniple.
for Chemical Change
Land and Prezbindowski (1981) point out that the Ed-
Solutes and stable isotopes are affected by advection and wards Formation was massively altered by meteoric dia-
dispersion. These processes become apparent along a genesis prior to burial, where the original co11IIiItc w t e r
flow system when a water of one chemical composition was first mixed with several pore volumes of meteoric
displaces or simply ends up next to water with a different water.
304 Cbapter 15 / Mass Transport in Natural Ground-WaterSystems

Whatever its chemical composition at the time of

burial, the displacement of formation water will generally
begin when the following two conditions are satisfied:
(1) the permeable unit is uplifted so that its outcrop
occupies a position that permits recharge by meteoric
water, and (2) the downdip portions of the unit have
outlets through which the formation water can be dis-
placed. When permeable beds pinch out with depth,
definite outlets are not readily available and the water
must escape through overlying less permeable forma-
tions.
The typical pattern is for fresh, unaltered meteoric
water to be found in and near regions of outcrop with
progressive increases in salinity down the flow system.
Fresh water or “noses” of relatively fresh water grading
into saline water have been noted quite frequently. Some
examples include the Madison Formation west and south
of the Williston Basin (Downey, 19&1), the Inyan Kara
Formation in southwestern North Dakota (Butler, 1984), Figure 25.2 Dissolved solids concentration for the Mission
and the Dakota-Newcastle Formation of the Dakota Sand- Canyon Formation, in parts per thousand (after Mitsdarffer,
stone in western South Dakota (Peter, 1984). These for- 1985).
mations have limited recharge areas that coincide with
the uplifted portions of the aquifers.
Two field examples of the development of freshwater
noses are shown in Figures 15.1 and 1 5.2. Figure 15.1 forced upward in a compactiondriven flow system. An
from Downey (1984) demonstrates how freshwatcr is example of this mixing occurs in the Gulf of Mexico
beginning to invade regions around areas of limited out- basin (Bethke and others, 1988), where abnormal fluid
crop in western United States. Figure 15.2from Mitsdarf- pressure is driving pore fluids out of the basin. In the
fer (1985) shows a welldeveloped freshwater nose in last two million years, when the effects of overpressuring
the Mission Canyon Formation in the Williston basin with have been most pronounced, brine migration has appar-
the source of recharge probably the Black Hills of South ently pushed back an apparently much more extensive
Dakota. An interesting question is whether the entire freshwater incursion into the basin (Bethke and others,
freshwater portions of the Dakota Sandstone represent 1988).
an immense freshwater nose surrounded by more saline The implication of this process in terms of ground-
waters in the neighboring basins. water chemistry is that brine can move into parts of a
Another type of mixing occurs whenever brines are basin where it otherwise may not have formed. An exam-
ple is the Na-Ca-C1brine found in Lower Cretaceous shelf
carbonates of the Edwards Group in south-central Texas
(Land and Prezbindowski, 1981). This brine may have
formed in the overpressured portion of the Gulf of Mex-
ico basin from evaporites known to occur there, and
subsequently migrated into shallower, updip parts of the
Edwards Group.
Not all mixing in ground water is related to the large-
scale tectonic deformation of sedimentary basins. In units
having a low hydraulic conductivity, the porewater can
often be relatively old and chemicallydifferent from mod-
ern-dayrecharge. Examples of such materials are the clay-

I
----___
WYOMING 1\-1
NEBRASKA
_ _ _ ----
-
rich tills found in the northern part of the United States
and in Canada (Desaulniers and others, 1981; Bradbury,
1984). These deposits often have hydraulic conductivi-
0 200 mi ties of lo-* cm/s and lower. Because of the relatively
Rocks exposed l -1 -10,000- Line of ?qua1 short time since deposition, they can contain water incor-
at land surface 0 200 km dissolved
solids mg/L porated when the till was deposited. Even where this is
Figure 25.2 Dissolved solids concentration in water from not the case, the low permeability of till means that any
the Cambrian-Ordovician aquifer (after Downey, 1984). long-term changes in the chemistry of recharge will be
 Z5.1 Mixing as an Agent f o r Cbemical Change 305

reflected in the chemical patterns along the flow system Water depth-960 ft
C f 01,"""
because of the age differences of the water. Section 15.5
0 40 80 120 160
describes a site in Wisconsin where changing values of 6Tl
and 8 '0 document significant differences in the isotopic ft. rn.
composition of modernday recharge and original pore c 100-
water, possibly present when the glacial till was depos- 2
4-
E -50
ited about 10,000 years ago (Bradbury, 1984). I
w
g 200-
c

mflusion in Deep Sedimentary Environments <2 300-

v) -100
Diffusion has been described as the transport of mass in
response to concentration gradients. As the diffusive flux 400
- Salt penetrated
is proportional to the concentration gradient, this pro- I I I
20 30 40
cess can cause significant mass redistribution whenever % HO
,
large concentration gradients can occur and can be main- (n)
tained over geologic time. Thus, this process should be
most effective where zones of hypersalinity underlie Water depth-2702 ft
brackish water. Diffusion acts to eliminate the concentra- CI- Olnn

tion gradient and, thus, becomes less significant as this

I '
I I 1-

equalization occurs. ft. rn. Hole 70 I

I
The occurrence of brines in sedimentary basins has 100 - I
been attributed to three causes: dissolution of original
evaporites, retention of original formation waters buried
s
.- -50
p 200
c
with the sediment, or membrane filtration. Whatever the
mechanism, the resulting concentration gradients can
give rise to significant mass redistribution over geologic E
m

( 500
A \ I
I I
b, I
600
20 30 40
% n,o
1 (6)
Figure 15.4 Plots of depth versus chlorinity and water con-
-
!k
tent in shallow sediments of the Gulf of Mexico. (a)Rock
c salt is encountered at 433 feet. The dashed line represents
g 2 the distribution of C1- at steady state due to diffusion be-
-
.-
t tween two boundaries of constant C1-. (6) Away from dia-
5
0
piric structures there is no significant variation in CI- concen-
: tration with depth (from Manheim and Bischoff. 1969).
3 Reprinted from Chem. Geol., v. 4, by F. T. Manheim and J. F.
Bischoff, Geochemistry of pore waters from Shell Oil Com-
pany drill holes on the continental slope of the northern
Gulf of Mexico, p. 63-82, 1969, with kind permission from
Elsevier Science, NL, Sara Burgerhanstraat 25, 1055 KV Am-
4 sterdam, The Netherlands.
I I I I
100 200 300 400 5
Salinity (o/oo)
time. The common pattern that emerges is a consistent
Figure 15.3 Variation in maximum porewater salinity as a increase in porewater salinity (mainly Na- and Cl-) with
function of depth in the Alberta, Illinois, and Michigan basins
depth (Figure 15.3). The data from north Louisiana, in
as well as the Louisiana Gulf Coast (from Ranganathan and
particular, show a linear increase in NaCl over 2700 m
Hanor, 1987). Reprinted from J. Hydrol., v. 92, by V. Ranga-
nathan and J. S. Hanor, A numerical model for the forniation to a depth at which salt is known to occur.
of saline waters due to diffusion of dissolved NaCI in subsid- One interpretation as to how these chemical patterns
ing sedimentary basins with eviporites, p. 97- 120, 1987, have evolved comes from theoretical studies by Rangana-
with kind permission from Elsevier Science, NL, Sara Burger- than and Hanor (1987) and field studies by Manheim and
hartstraat 25. 1055 KV Amsterdam, The Netherlands. Bischoff (1969) and Manheim and Paul1 (1981). Rangana-
3 2 Cbapter 15 / Mass Transport f n Natural Ground-Water Systems

than and Hanor (1987) solved one-dimensional forms (5) cation exchange, for example,
of the diffusion and diffusion-advection equation with
realistic sets of transport parameters and concluded that
Ca” + 2Na-X = 2Na+ + Ca-X
salinity-depth relationships develop over several tens of Cazt + Mg-X = MgZt + Cd-X
millions of years that are similar to those observed in
Mg2++ 2Na-X = 2Nat + Mg-X
a variety of sedimentary basins. Manheim and Uischoff
(1969) examined the porewater composition from six where Na-X is Na adsorbed onto a clay mineral.
holes on the continental slope of the Gulf of Mexico (6) organic reactions
and showed how the presence or absence of salt diapirs
affected porewater salinity. Samples from holes drilled This list is generally the conceptual model for unsatu-
near salt plugs showed systematic increases in salinity rated zone processes proposed by Moran and others
with depth that was interpreted as being caused by salt (19781). While their model was not intended to be gen-
diffusion (Figure 1 5 . 4 ~ )The
. development of the diffu- eral, combinations of these processes explain the chemis-
sion profile in Figure 1 5 . 4 required
~ about 400,000 years. try of water in different settings.
Samples from holes drilled away from diapiric structures
showed little change in chemistry as compared to seawa- (11 Gas LHssolution and Redistribution
ter except for minor diagenetic changes and a loss of
SO:- (Figure 15.46). Similar studies along the southeast- The dissolution and redistribution of C02(g) are im-
ern Atlantic coast of the United States showed the same portant soil-zone processes. Rainwater or melted snow
result over a greater range in depth (Manheim and contains relatively small quantities of mass, is somewhat
Paull, 1981). acidic, and has a Pco2 of about atm. As this water
moves downward, it rapidly dissolves C 0 2; this process
occurs in soil at partial pressures larger than the atmo-
- Cbemdcal Reactions
15.2 spheric value. Elevated COz pressures are due primarily
to root and microbial respiration and to a lesser extent the
in tbe Unsaturated Zone
oxidation of organic matter (Palmer and Cherry, 1984).
In the unsaturated zone, it is mainly the mass transfer Values for soils range generally from atm to more
processes that ultimately exert the greatest control over than 500 times larger (Palmer and Cherry, 1984). C02
the concentrations of major and minor ions and organic dissolved in water is further redistributed among the
compounds in ground water. Almost all the reactions weak acids of the carbonate system. For a pH range of
that we discussed in Chapter 12 operate to some extent. 4.5 to 5.5, H2CO; is the dominant carbonate species with
Following is a summary of the important chemical and HCO; and Ht the most dominant anion and cation, re-
biological processes affecting the inorganic and or- spectively.
ganic constituents. One direct result of dissolving COz(g) in water is a
gas dissolution and redistribution rapid increase in the total carbonate content of the water
and a decrease in pH. Calculations show the extent to
CO&) + HZO = HZCO; which the Pco2 in the unsaturated zone influences the

H2COS = HCO; + Ht
HCO; = COi- + H+
weak acid-strong base reactions, for example,
calcite: CaCO,(s) + H+ = Ca” + HCO;
anorthite: CaAI,Si30,(s) + 2H+ + H2 0 = kaolinite
+ CaZ+
albite: 2NaAlSi30,(s) + 2H+ + 5H20 = kaolinite
+ 4H2Si0, + 2Na’
enstatite: MgSiO+(s)+ 2H+ = Mg2++ H,Si02
Ffgure 25.5 Idealization o f flow through a hypothetical car-
sulfide mineral oxidation, for example, bonate sequence. The circled letters are reference points for
4FeS2(s) + 150, + 14H20= 4Fe(OH),<+ 16H+ calculations discussed in the text (modified from Palmer and
Cherry, 1984). Reprinted from J. Hydrol., v. 75, by C. D.
+ 8SOS- Palmer and J. A. Cherry, Geochemical evolution of groundwa-
precipitation/dissolution of gypsum ter in sequences of sedimentary rocks, p. 27-65, 1984, with
kind permission from Elsevier Science, NL, Sara Burgerhart-
CaS04 * 2H20(s) = Ca2++ SOf- + 2HL0 straat 25, 1055 KV Amsterdam, The Netherlands.
 15.2 Chemical Reactions in the Unsaturated Zone 307

Table 15.1 Calculated pH and Total Carbonate COi- increases at the expense of H2CO:. Overall both
Concentration for Water in Equilibrium with Specified the alkalinity and cation concentrations increase.
Pco, Values Calculations again illustrate these effects (Palmer and
Partial Total Cherry, 1984). Referring to Figure 15.5. the point of
Pressure C 0 2(atm) PH Carbonate" (mM) interest is now B, where water encounters limestone
bedrock. The system is assumed open ( i t . . the Pi:02 re-
10-" 5.62 0.0I9 mains fixed) with sufficient residence time to be at s m m -
lo-" 5.12 0.178 tion with respect to calcite. Pc:02 again is a key variable
4.62 1.72 in determining the dissolved carbonate content, the tot;il
" (H,<:O:) + (HCO;) + KO:.) in niillimoles per liter; tempen- Ca content (mainly Ca"), and the total quantity of calcite
tiire = 10°<:. dissolved to reach equilibrium. As the Pco2 is increased
Modilied froni Palmer and < : h e m (1984). Reprinted from J. Hydrol.. froni 10-i5 to to lO-'.', pH decreases from 8.30 t o
v. 75, hy C. I). Rilmer and J. A. Chern. "Geochemical evolution of
7.64 to 6.99 (Table 15.2). The dissolved carbonate con-
groundwater in sequences of sedimentary rcrks." p. 27-65, I984 with
kind permission from Elsevier Science. NL. Sara Rurgerhanstr:i;it 25.
tent (mainly HCO;) and total Ca increase. The tendency
1055 KV Amsterdam, The Netherlands. for weak acids to ionize is reflected by the dominance
of HCO; .
This simple system involves the dissolution of only one
evolving porewater chemistry (Palmer and Cherry, 1984). mineral. Langmuir (1971) calculated how HCO; and pH
Consider the hypothetical system in Figure 15.5. In mov- change as the water approaches equilibriunl with respect
ing to position A (Figure 15.5), the only process is the to a second mineral (dolomite). For a system that is open
open system dissolution of CO1 (Pm2 remains fixed). As with respect to CO?, water evolves to equilibrium with
expected, increasing the Pco2 from lo-", to 10-" to calcite along lines of constant Pco2 (Figure 15.6). For
lo-'' atm increases the total amount of carbonate in example, moving along the Pco? line of lo-?' atni. the
solution by about two orders of magnitude and lowers pH of a solution in equilibrium with calcite is 7.65. The
the pH from 5.62 to 4.62 across the range of partial second solubility curve labeled "dolomite" illustrates the
pressures (Table 15.1). pH and HCO; concentration for soil water in equilibrium
Another important soil zone-atmospheric process is with dolomite alone or calcite and dolomite together. At
the dissolution of O&). The resulting levels of dissolved the temperature of calculation. the solubility curves for
oxygen are sufficiently large at least initially to control dolomite and dolomite plus calcite are the same (Ling-
the redox chemistry in shallow ground water. muir, 1071).
Another feature of this diagram is the series of paths
labeled "no CO? added." These lines describe how the
(2) Weak Acid-Strong Base Reactions
concentration of HCO,; and pH change under closed
CO,-charged ground water is effective in dissolving min- system conditions. In calculating these pathwa}s. the
erals. The most common reactions involve the weak acids P m l decreases as C 0 2 is depleted by the dissolution of
of the carbonate and silicate systems and strong bases carbonates. Whether the system is open or closed with
from the dissolution of carbonate, silicate, and alumino- respect to <:02 can have an important bearing on the
silicate minerals. As we showed in Section 12.2, this porewater chemistry. For example, an open system (Pc:02
process causes the weak acids to dissociate. In the car- = lo-' atm) has a HCO; concentration of about 260 ppni
bonate system, the relative abundance of HCOi and and pH of about 7.34 at calcite saturation (Figure 15.6).

Table 25.2 Calculated pH, Total Calcium Concentration, Total Carbonate Concentration and Amount of Calcite
Reacted in Water at Equilibrium with Respect to Calcite and with Specified Pco2Values

Partial Total Calcite"

Pressure CO, (atm) PH Ca (mM) Reacted (mM)
8.30 0.628 1.26 0.62~
7.64 1. i o 2.96 I .40
6.90 3.22 8.13 3.22
~~ ~ ~~ ~~ ~ _______

" (H1<:O:) + (HCO,) + (COi ) in millimoles per liter.

" Amount of calcitr dissolved in the unsaturated zone; temperatitre = 1O0C
Mcdilird froni Palmer and C h e w (198.4). Reprinted fromJ. Hydrol.. v. 75. by C. 1). Palmer andJ. A. Cherry. "Ceocheniical rvolittion ol'#round\vater
in scqiicncesof sedimentary rocks," p. 27-65. I984 with kind permission from Elsrvier Science. NL. Sara I3urgerhartstraat IS. I055 KV Ani.*tcr&im.
'The Nethrrkiiids.
308 Cbapter 15 / Mass Transport in Natural Ground-WaterSystems

Under closed system assumptions, these values are 170

ppm and 7.70, respectively.
A model involving just the two processes considered
so far, the dissolution of COzgas accompanied by calcite
dissolution,describes the chemistry of water in the unsat-
urated zone in carbonate terranes. Shown in Table 15.3
is a chemical analysis for water from a near-surfacespring
in a limestone in central Pennsylvania (Langmuir, 1971).
This analysis probably approximates the composition of
water from the unsaturated zone. The most abundant
ions are Caz+and HCO; , with relatively small concentra-
tions of Na+ and C1-. The calculated Pco2for this sample
is atm and like nearly all of the spring water is
undersaturated with respect to calcite and dolomite. Pre-
sumably, transport through fractures has been so rapid
that calcite equilibrium was not achieved. Farther along
the flow system, calcite equilibrium finally constrains
increases in Ca" and HCO; concentrations.
As our sample list of weak acid-strong base reactions
shows, silicate and alumino-silicateminerals will also re-
act to some extent when they are present. The relatively
low solubility of these minerals, however, means that 7.5 8.0
their contribution to the mass dissolved in porewater PH
will be relatively small when soluble minerals like calcite Ffgure 15.6 Possible approaches to equilibrium of ground
are present. In systems where there are no carbonates water in contact with calcite and (or) dolomite at 12°C. The
present, these reactions control the ion chemistry. solubility curves of dolomite and calcite plus dolomite are
Garrels and MacKenzie (1967) looked at a case where the same at this temperature. Reprinted from D. Langmuir,
waters rich in C02reacted with a suite of minerals found The Geochemistry of some carbonate ground waters in cen-
in igneous rock (Table 15.3). Again by using water from tral Pennsylvania, Copyright 0 1971, p. 1023-1045, with
ephemeral springs as a surrogate for soil water, we can kind permission from Elsevier Science Ltd, The Boulevard,
Langford Lane, Kidlington OX5 lGB, UK.
observe the behavior of a system with no carbonate min-

Table 15.3 Comparison of the Chemistry of Waters from Four Different Recharge Environments and the
Unsaturated Zone Processes Giving Rise to These Waters

Study Ca2+ Mg2+ Na+ K+ HCO; SO:- C1- pH log Pcoz SI, Sh
1. Langmuir (1971) 41. 4.4 6.5 2.1 134 - 12 7.13 -2.08 -0.78 -1.16
2. Garrels and MacKenzie
(1967)" 3.1 0.7 3.0 I .1 20 1.0 0.5 6.2 -1.8 - -
3. Kimball (1984) 57 21 ' 26 0.4 210 57 2.6 7.9 -2.57 0.26 0.14
4. Moranet al. (1978b)" 62 25 469 7.9 748 577 4.7 8.0 - - -
1. Processes 2. Processes 3. Processes 4. Processes
CO, dissolution and redis- COz dissolution and redis- COz dissolution and redis- C 0 2 dissolution and redis-
tribution tribution tribution trlbution
weak acid-strong base weak acid-strong base 0,dissolution 0,dissolution
calcite plagioclase weak acid-strong base weak acid-strong base
biotite calcite calcite
K-feldspar dolomite pyrite oxidation
albite
precipitation-dissolution
pyrite oxidation
of gypsum
cation exchange
Notes: Concentration in milligrams per liter.
Partial pressures in atmospheres.
SI, saturation indices for calcite and dolomite.
'' Mean values for several samples.
 15.2 Cbemical Reactions in the Unsaturated Zone 303

erals (Table 15.3). The concentration of all ions is low Melted snow
because only plagioclase feldspar, biotite, and K-feldspar
dissolve. In fact the concentration of most ions is so low
that the initial composition of precipitation had to be
considered when Garrels and MacKenzie worked with
the data. Dissolution apparently takes place under closed-
system conditions with about half of the CO?consumed. Parent material

63) Suwde Oxidation

Sulfide oxidation is one of the important redox reactions
within the unsaturated zone. Minerals like pyrite or mar-
casite are oxidized to produce Fe(OH),(s), SO:-, and H-.
In fact, pyrite oxidation is one of the most important Unweathered geologic deposit
n Water table
acid-producing reactions in geologic systems (Moran and
others, 1978a). In coal mining areas like the Appala-
chians, this reaction can be the cause of serious acid-
mine drainage problems (Mordti and others, 19784.
An example of how various cheniical reactions, includ-
ing pyrite oxidation, control recharge chemistry is a study Figure 15.7 A conceptual model of subsurface flow in the
by Kimball(1984) in the Piceance Creek basin of north- Plains region of the United States and Canada. (1) Annual po-
western Colorado. The initial acquisition of solutes in tential evapotranspirationgreatly exceeds annual precipita-
the recharge to the Uinta Formation is controlled by tion, and most infiltration is lost by evapotranspiration.(2)
three processes considered so far, namely: open-system Exceptional precipitation events produce recharge (from
Momn and others, 197%).
dissolution of C 0 2 and 02,weak acid-strong base reac-
tions, and sulfide mineral oxidation (Table 15.3). The
main difference as compared to the essentially niono-
mineralic system described by Langniuir (1971) is that
the dissolution of dolomite and albite, and the oxidation
change reactions are the water-softening reactions,
of pyrite, add Mg”, Na+, and SO: to the water (Table
where Ca” and Mg” in the water exchange with sorbed
15.3).
Na- as ground water moves through clayey material. Shal-
low ground water from a study area in North Dakota
( 4 ) Gypsum Precipitation and LMssolution (Moran and others, 1978a) provides an example of how
all of the processes we have considered so far work
Conditions exist when the solute load in soil waters can together. The chemical data on Table 15.3 represent
be much higher than we have considered so far. The mean values for 39 water samples collected from various
cyclical precipitation and dissolution of gypsum is proba- near-surface drift units. The abundance of Na‘ and
bly the most important process in this respect. Over SO;’-in the shallow ground water is attributed to gypsum
much of the Plains region of the United States and Canada, dissolution, sulfide oxidation, and cation exchange. The
annual potential evaporation exceeds annual precipita- cation exchange process is important here because it has
tion by a considerable amount. Thus, water that infiltrates effectively increased Na- concentrations at the expense
in normal precipitation years evaporates and deposits a of Ca” and Mg”.
small quantity of gypsum. With repeated rain or snow-
melt, gypsum accumulates in the upper part of the soil
horizon (Moran and others, 1978a). Exceptional recharge (6)Organic Reactions
can dissolve some of this soluble material and move it Work by Wallis and others (l98l), Thurman (1985). and
down and into the ground-water system (Figure 15.7). Hendry and others (1986) has identified a variety of im-
In some arid areas, recharge water could have SO:- con- portant organic reactions such as dissolution of organic
centrations in excess of 5000 mg/L (Hendry and oth- litter at the ground surface, complexation of Fe and Al.
ers, 1986). sorption of organic compounds, and oxidation of or-
ganic compounds.
The dissolution of organic litter at or close to the
(5) Cation Exchange ground surface is the major source of dissolved organic
The precipitation and dissolution a€gypsum in the Great carbon (DOC) in soil water and shallow ground water.
Plains region is also accompanied by cation exchange DOC concentrations typically fall in a range from 10 to
(Moran and others, 1978a). The most important ex- 50 mg/L in the upper soil horizons and less than 5 nig/
310 Cbapter 15 / Mass Transport in Natural Ground-WaterSystems

L deeper in the unsaturated zone (Thurman, 1985). Con- weak acid-strong base reactions, for example,
centrations are highest at their source and decline with
depth through sorption and oxidation. The most im- carbonate minerals + H+ = cations + HCO;
portant fraction of the DOC is a group of humic sub- silicate minerals + H+ = cations + H,SiO3
stances, consisting mainly of humic and hilvic acids. I'an-
nins and lignins, amino acids, and phenolic compounds alumino-silicate minerals + H+ = cations + H2Si0,
are often present in smaller concentrations (Wallis ;ind + secondary minerals (for example, clay minerals)
others, 1981). dissolution of soluble salts, for example,
The complexation of Fe and Al with organic matter is
an important process facilitating the transport of these halite: NaCl(s) = Na' + CI-
poorly soluble metals from the A horizon of the soil to anhydrite: CaSOXs) = Cal' + SO:-
the B horizon (Thurman, 1985). This is one key feature of
the soil-forming process called podzolization (Thurman, gypsum: CaSO, * 2HzO(s) = Cd" + SO:- + 2H20
1983). In terms of ground-water systems, this reaction is - -
carnalite: KCl MgCI2 6H20(s) = K' Mg2++ +
not particularly important because most of these com- 3CI- + 6H20
plexed metals sorb in the B horizon of the soil. kieserite: MgSO,-HIO(s) = Mg" + SO+- + HzO
Dissolved organic compounds originating in the upper
sylvite: KCI(s) = K f + CI-
part of the soil horizon are not particularly mobile due
to sorption. Many of the sorption models we discussed redox reactions, for example,
in Chapter 12 operate within the soil zone. Hydrophobic @&) + Hf + e- = &H?O
sorption occurs because of the relatively large quantities
of solid organic matter present in the upper part of many $Fe20,(s>+ 3Ht + e- = Fez++ $ H 2 0
soil horizons (Thurman, 1985). Similarly, the abundance hMn02(s) + 2HC+ e- = hMnZ++ H,O
of metal oxides, hydroxides, and clay minerals leads to
surface complexation reactions and electrostatic interac- QSOf- + tH+ + e- = BHS- + i H 2 0
tions. Again in terms of the chemistry of shallow ground
KO&) + HC + e- = KH,(g) 4- f H 2 0
water, this is one of the processes that keeps the quantity
of DOC in recharge at relatively low concentrations. %O&) + Hf + e- = %H,O + f H2 0
Oxidation reactions involving organic matter can i n cation exchange, for example,
fluence the chemistry of shallow ground water. For exam-
ple, the oxidation of dissolved organic matter (repre- Ca2+ Cd
sented as CH20) provides a source of C02gas within the Mgzt + 2Na-clay = 2Na' + Mg-clay
unsaturated zone, which is readily dissolved in soil water.
A second reaction, involving the oxidation of a sulfur- Fez+ Fe
containing compound (represented by the amino acid
cysteine), is thought to play a major role in the accumula-
( 2 ) Weak Acid-Strong Base Reactions
tion of gypsum in shallow soils (Hendry and others,
1986). This reaction is the organic counterpart to the If the ground water is not yet in equilibrium with carbon-
pyrite oxidation reaction. In arid areas, this process can ate, silicate, and alumino-silicate minerals, they will con-
also produce recharge with large SO:- concentrations tinue to dissolve in the saturated zone. Because of their
(Hendry and others, 1986). relative abundance, reasonably fast reaction rate with H",
and reasonable solubilities, these reactions increase the
cation concentrations, alkalinity, and pH.
-
15.3 Chemical Reactions The examples presented in the previous section are
again useful in illustrating these effects. Consider the
in the Saturated Zone
flow system in Figure 15.8, where water flows through
The chemistry of ground water depends on not only the a sequence of overburden, limestone, and then dolomite
chemistry of the recharge but also the reactions operating (Palmer and Cherry, 1984). Recall that as the water
along the flow system. Processes in the saturated zone moved to reference point A, it dissolved C 0 2 at a fixed
are more complex than in the unsaturated zone because partial pressure, and at point B it had become saturated
the mass transport processes play a more important role, with respect to cafcite. In the saturated zone at point C,
and because geologic, hydrogeologic, and geochemical ground water reaches equilibrium with respect to both
settings are much more diverse. However, most of the calcite and dolomite. In moving to C, the pH, the total Mg
same processes affecting ion concentrations in the iitisat- concentration, and the total CO, concentrations increase
urated zone are also operative in the saturated zone in- (Figure 15.8).The Pco, declines from atm to 1W2-'',
cluding: because once the system closes below the water table,
 15.3 Cbemical Reactions it1 the Saturated Zotre 3 3

Pigwe 15.8 Pattern of mass transfer as water proceeds Figure 15.9 Pattern of mass transfer as water proceed!,
through an idealized overburden-cirbonate rock sequence through :in idealized overburden-cirbon;tte rock sequelire
(modified from P i h e r and Cherry, 1984). Reprinted from (niodified from Palmer and Cherry. 1984). Reprinted from
J. Hydrol., v. 75, by C. I). Palmer and J. A. Cherry, Geochemi- J. Hydrol., v. 75, by C. D. Palmer ;md J. A. Cherry, Geochcnii-
cal evolution of groundwater in sequences of sedimentary cal evolution of groundwater in sequences of sedimentan,
rocks, p. 27-65, 1984,with kind perniission from Elsevier rocks. p. 27-65, 1084. with kind pcrmission h n i Elxevic-r
Science, NL, Sara 1lurgerhartstra;it 25. I055 KV Amsterdam. Science, NI., Sara Burgerliartstraat 25. 1055 KV hmstrrl;ini.
The Netherlands. Tlie Netherlands.

C 0 2depletes by the dissolution of dolomite. The total Ca depending on what carbonate mined: , 4 were eiicoiin
concentmion decreases because, in dissolving dolomite, tered and in what order.
the ground water became supersaturated with respect Looking further at the example presented by GiIrrels
to calcite. In moving from B to C, 1.59 mmol of calcite and MacKenzie (1 967). evolution through ;icicl-biise
precipitates and 0.76 mmol of dolomite dissolves (Fig- re;ictions increases the quantity of mass dissolved i n the
ure 15.8).
The chemistry of the ground water at point C depends
on not only the processes involved but also the order of
encounter (Freeze and Cherry, 1979). A slight change in
geology, placing the dolomite above the limestone, has
a significant impact (compare the data in Figure 15.8
with those in Figure 15.9). The pH in the second case
is lower, and 2 (>a, 2 Mg, 2 CO, , and Pco?are all higher.
Holding the system open while dolomite dissolves to
equilibrium enables more mass to dissolve. Further, in
moving from point B to point C, the chemical composi-
tion of the ground water does not change (Figure 15.9).
The example of evolving composition in a predonii-
nantly carbonate terrane (Langmuir, 1971) illustrates
how the composition simply proceeds toward equilib- t /-
rium with respect to those minerals available for dissolu- ;A , *'* -...L - .I."
tion (for example, calcite and dolomite). Once in the 1.5 -1 0 0.5
zone o f saturation, the ground water approaches satiira- Sf,,
tion with respect t o both calcite and dolomite because Figare 15.10 Saturation index for calcite versus the s;ititr;i-
tion index for dolomite. Spring wiiters are design;ited I>!
of the longer residence time (Figure 15.10; Table 15.4).
open synihols; well waters by solid s!nihols. Irimigles t l ~
Also, water from a predominantly limestone source rock
notc ;I limestone source rock; circles denote ;I dolonlitic
more closely approaches saturation with respect to source rock. Crosshatched area show's the limit!, o f unc'er.
calcite than dolomite. This Fact is illustrated in Figure taint!, in SI, and SI,,. Reprinted from I). Impiiiuir. *l'k
(ieo-
15.10 by the abundance of data points for limestone cheniistn of some carbonate ground waters in centr:il I'CIUI-
source rocks (triangles) lying above the dashed line sylvania, Copyright 0 1971. p. 1 0 3 - 1045. with kind
(SI, = SI,,). Even in this simple carbonate system, there permission from Elsevier Science Ltd, The Houle\.ard. Imp-
are differences in the pathways of evolution apparently ford I.nne. Kidlington OX5 1GB. U K .
-
312 Cbapter 15 / Mass Transport in Natural Ground-WaterSystems

Table 25.4 Averages of Various Chemical Parameters Wa_ter table

for Spring and Well Waters from a Carbonate Terrane

Parameter Spring Water Well Water

-
Specific conductance" 347 499 No vertical
47.5 55.0 or
Ca?' (ppm) horizontal
Mg" (ppm) 13.9 29.6 scale
HCO; (ppm) 183 265
PH 7.37 7.46
SI'P -0.41 -0. I5
Sb -0.63 -0.1Y Fresh water Brine and cavities Disrupted rock

'' Millinihos per centimeter. Figure 15.22 Schematic representation of the circulation
"Saturation index. of fresh water and brine in arras of salt dissolution in west-
Modified from Langniiiir (1971). Reprinted from D. Lang- ern Oklahoma. There is no scale, but the length of the sec-
muir, The Geochemistry of some carbonate ground waters in tion could range from 1 to 15 km and the thickness from 30
central Pennsylvania, Copyright 0 1971, p. 1023- 1045, with to 300 m (modified from Johnson, 1981). Reprinted from J.
kind permission from Elsevier Science Ltd, The Boulevard, Hydro]. Special Issue: Symposium on Geochemistry of
Langford Lane, Kidlington OX5 IGB, UK. Groundwater, v. 54, by K. S. Johnson, Dissolution of salt on
the east flmk of the Permian Basin in the southwestern
United States; p. 75-94, 1981 with kind permission from
ground water. The effects of increased residence time Elsevier Science, NL, Sam Rurgerhartstraat 25, 1055 KV Am-
can be evaluated by comparing the chemistry of ephem- sterdam, The Netherlands.
eral and perennial springs (Table 15.5). The system be-
haves as expected. The concentrations of cations in-
crease due to the continued hydrolysis of biotite and
plagioclase as does the alkalinity (reflected in the This process (Figure 15.11) leads to the formation of
HCO; concentration and pH). An important source of saline brines in the shallow ground water of western
Ca'+ is the dissolution of small quantities of carbonate Oklahoma and the Southeastern part of the Texas Panhan-
minerds. In the deeper parts of the system, montmoril- dle aohnson, 1981). Freshwater recharged through per-
lonite occurs as a weathering product of plagioclase in meable units moves downward until it encounters salts
addition to kaolinite. at depths ranging from 10 to 250 m. The dissolving salt
produces cavities at the updip limit or the top of the
salt uohnson, 1981). Periodically, the rocks overlying the
(2)Dissolution of Soluble Salts cavities collapse. The process is apparently self-perpetu-
Section 15.1 explained how the presence of evaporites ating because the collapse and fracturing of overlying
in a sequence could affect the water chemistry through units provide improved access to the salt for fresh water.
diffusion. There are other situations (for example, Figure The evaporites, mainly halite and gypsum/anhydrite,
15.11) where active ground-water flow directly encoun- are interbedded with a thick sequence of red-beds. Given
ters evaporites. Mineral salts are extremely soluble and the particular salts involved, high concentrations of Na'
dissolve to produce a brine whose composition depends and CI- are not surprising. Clearly when ground water
on the particular minerals present (for example, halite, encounters large quantities of soluble salts in the subsur-
anhydrite, gypsum, carnalite, kieserite, and sylvite). face, the impact on the chemistry is considerable.

Table 35.5 Chemical Composition of Ephemeral arid Perennial Springs of the Sierra Nevada

Sample
Source Ca2+ Mg'+ Na+ K+ HCO; so,'- c1- sioz PH
Ephemeral 3.11 0.70 3.03 1.09 20.0 1 .00 0.50 16.4 6.2
springs
Perennial 10.4 1.70 5.95 1.57 54.6 2.38 1.06 24.6 6.8
springs
"Concentntions in milligrams per liter.
Reprinted with permission from R. M. Garrels. and F. T.MacKenzie, in Equilibrium Concepts in Natural Water Systems; Gould, R. P. ed., American
Chemical Society: Washington, D.C., 1967, p. 224-225. Copyright 1c967 American Chemical Society.
 15.3 Cbemical Reactions in the Saturated Zone 313

CH20 + 4H’ + 2Mn02(s)= 2Mn2- + 3 H 2 0 + CO?

( 1 5.3)
Once the pe is sufficiently reduced, sulfide species appe;ir
from the reduction of SO:- or
2CH20 + SO:- f H- = HS- + 2HZ0+ 2COl ( 15.4)
I
400-300 mV 200-100 mV When the rate of sulfide production exceeds the dissolu-
Direction of Ground Water tion rate of iron and manganese oxides, Fez- ;ind Mil2
Flow or Distance from
Recharge Area concentrations fall (Figure 15.12) as metal sultides precip-
itate as the stable phase.
Figure 15.12 Different redox zones existing along a con-
fined aquifer. Reprinted from Jackson and Inch (1980). Hy- A study at Chalk River Nuclear Laboratories near O t -
drogeochemical processes affecting the migration of radionu- tawa, Canada, demonstrates how redox conditions can
clides in a fluvial sand aquifer at Chalk River Nuclear change along a flow system. By careful measurements of’
hboratories, NHRI Scientific Series, Paper 7, Environment pH, EH. dissolved oxygen (DO). and the total conceiitr;i-
Canada, 58 p. Reproduced with the permission of Environ- tion of sulfides (S+-) as well as other key chemical p;icinie-
ment Canada, 1997. ters, Jackson and Patterson (1982) defined three redox
zones. The main geologic units were two fluvial s;incl
aquifers separated by a thin layer of interbedded CliI!
(3) Redox Reactions about 1 m thick (Figure 15.13). Ground water flows
southward from an upland area toward Perch Like. In-
The redox conditions encountered along a flow system
flow to the aquifers is near the area labeled “disposA
are important in controlling the chemistry of metal ions
area” (Figure 15.13) with none further down-gradient.
and solids (for example, Fez+,MnL’, and Fe20,), species
The lower aquifer in the recharge area contains dis-
or solids containing sulfur (for example, SO:-, H2S, and
solved oxygen and has an EH of about 0.55 V (Figure
FeS,), and dissolved gases containing carbon (for exam-
15.14), which is quite close to the theoretical range of
ple, C02, CH,). It is possible in some flow systems to
0.71 to 0.83 V. Both Fez- and Mn’- concentrations ;ire
define redox zones. These zones are parts of an aquifer
low, and sulfide is undectable. The declining E,, along
in which pe is controlled by a dominant redox couple.
the deep flow system coincides with the reduction of
Field studies (for example, Champ and others, 1979;Jack-
oxygen. Increasing concentrations of iron and niangii-
son and Patterson, 1982) have shown that oxygen, iron-
nese are evident once the oxygen is depleted (Figure
manganese, and sulfides zones will often be present. The
15.14). Zone I1 is probably not extensive because sultide
probable half-reactions controlling pe in these zones are
is becoming relatively abundant at E,, values below 0.2
the reduction of oxygen to water, the reduction of iron
V. The downstream end of the flow system is a sultide
or manganese oxides, and sulfate reduction to HS- or
zone with EH values approaching the theoretical range
H2S. In a few cases, a methane zone can form from the
for the sulfate-sulfide couple (i.e., -0.21 to -0.32 V).
reduction of CO?.
Concentrations of Fe”, Mn’-, and SO:- decline in zone
Reduced sulfide and methane zones tend to develop
111 from the precipitation of ferrous sulfides (Jackson ;tnd
in contined flow systems containing excess oxidizable
Patterson, 1982).
DOC and a lack of recharge containing oxygen down-
The reactions that oxidize organic ni;itter (Section
gradient in the system. Thus, redox zones will be ob-
12.5) generate C02as a product. This C 0 2is redistributed
served in an extensive artesian aquifer that receives re-
among H2CO;, HCO; , and CO<-. In aquifers where <>I?.
charge from a limited area of outcrop or in an aquifer
;ind Mg” are exchanged onto clay minerals for Na.. the
with confining units.
possibility exists for carbonate dissolution iuitl even
Figure 15.12 summarizes the common changes i n re-
higher HCO; concentrations. With CO? generatecl 11). re-
dox chemistry from zone to zone uackson and Inch.
dox reactions, ion exchange, and carbonate dissolution.
1980). In zone I, oxygen initially present in reclyarge will
the water will evolve chemically to a sodiuni-bicarI,onatc
decline through reduction by organic carbon
type like that from the Atlantic coastal plain (Foster.
CHZO +0 2 = CO? + H j 0 (15.1) 1950), the Eocene aquifers of the east Texas basin (Fogg
The concentrations of ~ ~ and 2 M
- ~ ’ +in 11 increase and Kreitler, 1982), and the Milk River aquifer (Hentln’
because the Fe(1II) and Mn(1V) minerals, which are oxi- and Schwartz*l99O).
dized solids, are not stable in the more reducing environ- Chapelle and others (1987) investigated the source o f
ment. These reactions are C 0 2 in ground water from the Atlantic Coastal Plain i n
Maryland. The units of interest included the M;igotIi!.-
“20 + 8H+ + 4Fe(OH)5(s) = *Fe’+ + “H?O + ‘ 0 2
Upper Patapsco aquifer and the Lower Pat;ipsco ;tquift.r.
(1 5.1) These aquifers crop out between Washington and Ikilti-
314 Cbapter 15 / Mass Transport in Natural Gmund-Water Systems

,.. ' L .

Figure 15.13 Cross section through the lower Perch Lake basin aquifer at Chalk River Nuclear
Laboratories, Canada, showing equipotential lines, piezometer tips, and the position of the water
table. Hydraulic head measurements averaged over the period of 1973-1975 (from Jackson and
Patterson). Water Resources Res., v. 18, p. 1255-1268, 1982. Copyright 0 by American Geophysi-
cal Union.

0.61
HA1
, MA1
I
~
I
~ 4 oa
I
~ 5 K04,5
I
- Piezometers

$ 0.oL
.-
c > Lower Sands
- A - 0.12
k
6- Aquifer
/
/ \,Fe,(o) -
- 0.10 ,j$
E 5- A / \\ S:-C+) -
e
.-M 4- / \
\
- 0.08
.-
\ -0.06
e
E
\
\
\
- 0.04 2
.-
5 1
\
\
- 0.02 7E
'ye;

A 0.00
o loo 200 300 400 500 600 700 aoo goo 1000
Distance from HA (meters)
Figure 15.14 The pH, El,,O , , and total dissolved iron and sulfide values for the deep, con-
fined Lower Sands aquifer. The piezometer numbers are shown on the top abscissa (from
Jackson and Patterson). Water Resources Res., v. 18, p. 1255-1268, 1982. Copyright 0 by
American Geophysical Union.
 Next Page

15.3 Chemical Reactions in the Saturated Zone 3 x

( 4 ) Cation Exchange
The most important exchange reactions are the natural
water softening reactions, which take Ca'- and Mg" out
of water and replace them with Na'. The main require-
ment for this process is a large reservoir of exchangeable
38"45 Na+, which is most often provided by clay minerals de-
posited in a marine environment. One does not have t o
go far in the United States or Canada to find clays or
shales capable of ion exchange. A case in point is bedrock
in the Wabamun Lake area of central Alberta, Canada
(Schwartz and Gallup, 1978). Shale, sandstone, and coal
of the Edmonton Group are overlain by two younger
bedrock units that include the Pembina coals and sand-
38"15 stone of the Paskapoo Formation. Bedrock in turn is
mantled by sand and gravel. The shales of the Edmonton

-
Group and those interbedded with the Pembina coals
were deposited in a marine environment. Figure 15.16
area of the is a plot of Ca" versus Na' concentrations for samples
0Outcrop
Patuxent Formation \21
Potentiometric contour
(in meters above sea level of
the Lower Patapsco aquifer) from the drift, Pembina coal, and Edmonton Group. It
m Outcrop area of the
Arundel Formation Approximate direction of
shows how the pattern of cation dominance shifts from
Ca'" (and Mg", not shown) to Na' as water niovesdeeper
Outcrop area of the ground-water flow (in the
Patapsco Formation Lower Patapsco aquifer) into the marine shale.
Figure 15.15 Map showing the regional outcrop area for When ion exchange takes place, its effects on the
the Patuxent, Arundel, and Patapsco Formations on the Atlan- cation chemistry of water should be unmistakable. How-
tic coastal plain in Maryland. Ground-water flow in the ever, as the following case study shows, an equivalent
Lower Patapsco Formation is away from the outcrop area increase in Na' concentration that is matched by an in-
(from Chapelle and others). Water Resources Res., v. 23, p. creasing CI- concentration may mean that other pro-
1625- 1632, 1987. Copyright 0 by American Geophysical cesses are at work.
Union.

more (Figure 15.15) and dip toward the east. Recharge

occurs along the outcrop areas with flow toward the
Potomac River and under Chesapeake Bay (Figure 15.15). /---/-'
\<Edmonton Group
Although these units contain n o calcareous minerals,
\
HCO; concentrations increase significantly in the direc-
tion of flow. Where the Patapsco aquifer crops out,
HCO; concentrations range from 0 to 50 mg/L. Approxi- and sand
mately 30 km downgradient, they have increased to
about 150 to 200 mg/L, and finally near Cambridge, Mary-
land (Figure 15.15), HCO, concentrations range from
400 to 500 mg/L (Chapelle and others, 1907). This pat- Coal zones '
tern is similar to that observed by Foster (1950) farther
south in Virginia.
The potential sources of C 0 2 are the bacterially medi-
ated oxidation of solid organic matter or the abiotic decar-
boxylation of these materials. The presence of both sul-
fate-reducing and methanogenic bacteria in cores,
however, points to the generation of CO? in redox reac-
I I
tions. Although sulfate-reducing bacteria are present, the If 10 100 1( I0
relatively small quantities of SO:- present make the con- Ca2+ (rniiiigrams per liter)
tribution of COz from sulfate reduction small relative to Figure 15.16 Fields defined by plotting <;a'- versus N i
that from methanogenesis. The bacteria facilitating these data for three different geologic units in the W;tb;imiin Lake
reactions occur together because the theoretical pe of a arra o f central Alberta, Canada (from Schwnrtz mid (hiliip.
system is nearly the same with either -action. 1978).
Previous Page

3L6 Chapter 15 / Mass Transpod in Natural Ground-WaterSystems

-
15.4 Case Study of tbe An exciting feature of this aquifer system is the well-
Milk River Aquifer defined patterns of variability in the major and minor
ions and the stable or radiogenic isotopes. CI-, S'"0, and
We have singled out the Milk River aquifer for detailed SD change markedly along the flow system (Figure 15.18)
discussion because the interpretation of chemical pat- and appear to act as conservative tracers (Hendry and
terns requires the integration of concepts of mass trans- Scliwanz, 1988). C1- concentrations increase from less
port and mass transfer. Several studies have documented than 1 mM to more than 100 mM at the downstream end
the geologic and hydrogeologic setting including Mey- of the flow system. S"O and SD values increase from
boom (1960), Schwartz and Muehlenbachs (1979), and about -2O"/,,, and - 150°/,,, , respectively, in the recharge
Hendry and Schwartz (1988, 1990). The aquifer is part areas to about -8.5"/,, and -85"/,, respectively, at the
of a thick sequence of Cretaceous rocks (Figure 15.17). northern end of the aquifer.
Units of interest from oldest to youngest are the shale of Several different studies have been concerned with
the Colorado Group, the Milk River Formation whose explaining these patterns, such as Schwartz and Muehlen-
lower member is the aquifer, and shale of the Pakowki bachs (1979), Domenico and Robbins (1985a), and Phil-
Formation (Figure 15.17).The aquifer pinches out to the lips and others (1986). A reexamination of these ideas
north and east as the result of a facies change. prompted the development of one that involves disper-
Ground-waterrecharge occurs where the aquifer crops sion and advection in the aquifer and diffusion in the
out in northern Montana and along an area of outcrop- aquitard (Hendry and Schwartz, 1988). This model takes
subcrop in southern Alberta. Flow is mainly northward into account the effects of both mass transport processes
or downdip in the aquifer. Leakage from the aquifer and geologic change in controlling ground-wdter chem-
moves upward through the Pakowki Formation and over- istry.
lying units. Before extensive development of the ground- T6th and Corbet (1986) showed how the configuration
water resource, flowing wells were common. of the land surface changed with time. As recently as five

a3!

Ffgure 25.27 Study area for the Milk lUver aquifer investigation. The geological cross sec-
tion illustrates the most important drift and bedrock units in southern Alberta, Canada (from
Hendry and Schwartz, 1990. Reprinted by permission of Ground Water. Copyright 0 1990.
All rights reserved.
 15.4 A Case Study of the Milk River Aquifer 317

Rg. 15 Rg. 10

Tp. 10 Milk Rivw Formation

High Cf isotopically enriched
~

((I)

Tp. 5
Recharge
111 -------- ........7-

w
Tp. 1 Canada Figure 15.19 The effect of geologic changes on the c1ic.m-
Montana United States istry of ground water in the Milk River aquifer. In ( ( I ) the ail-
uifer is deeply buried. With time. erosion uncovers upstream
Rg. 15 Rg. 10
end of the aquifer, causing inflow of meteoric water (niotli-
fied from Hendry and Schwartz. 1990). Heprinied by prriiii~-
sion of Ground Water. Copyright 0 1990. All rights resen.c.tl.

million years ago. the land surface was probably 700 111
higher than now. Even in the early Pleistocene. the land
surface was about 200 ni higher than present (Ttith ;itid
Corbet, 1986). When the aquifer was deeply buried (Fig-
lire 15.19), ground water probably was enriched isotopi-
cally relative to present-day meteoric water ( P " 0 -
-
-2O/,,, and C93 - 150'/,,,,) and had ;I me;in (:I- concen-
tration of about 85 mill (Hendry and Schwartz. 1988).
Montana United States Continuing erosion eventually exposed a relativel!. large
recharge area that provided a new source of w;itw to the
Rg. 15 Rg. 10
aquifer. Proximity to the surface woiild result in recharge
with a low C1- concentration (Figure 15.196). This
change in the flow dynamics of the aquifer probably
occurred about 1 X 10" years ago. Somewhat 1;iter. the
stable isotope composition of the precipitation bec;iiiie
similar to present-day meteoric water due to ;I ch;ing-
ing climate.
This chemicallydifferent water with time has disp1;icetl
the original formation water. However. in so doing. it
has created diffusion gradients between the isotopicall!,
depleted, low chlorinity water in the aquifer ;ind tlie
original water remaining in the shale of the <:olor;iclo
Group. Thus water moving down the ;iqiiifer espcri-
enced an increase in C1- concentration ;inti cnriclinirnt
Montana United States
0 10 20 in 8'0 and 673 due to aquitard diffusion. This siiiiplc
model of advection and dispersion i n thc ;iquifer and
Kilometers diffusion in the aquitard was tested niatliem;itic;ilI!~I>!.
Figure 15.18 Areal variation in CI-, 8"O, and 6D in water Hendry and Schwartz (1 988).
from the Milk River aquifer (from Hendry and Schwartz). Wa- Na' and HCO; concentrations iiicreiise i n the direction
ter Resources Res., v. 24, p. 1747- 1764, 1988. Copyright 0 of flow (Figure 15.20). SO:- behaves oppositely \\.it11 tlic
by American Geophysical Union. highest concentrations in tlie recharge ;irci ant1 ;I s!'stcm-
atic decrease down-gradient in the confined pwt of the
aquifer (Figure 15.20). Not shown are maps for C;i2. ;ind
Mg". Except for a few areas where the Milk River crops
3 3 Cbapter 15 / Mass Transport in Natural Ground-Water Systems

its abundance in formation waters in the basin (Hendry

and Schwartz, 1990). If the Na+ distribution is in fact
related to CI-, a relationship should exist between Na-
and CI- concentrations. The lines in Figure 15.21 are
hypothetical pathways of chemical evolution generated
by adding equivalent amounts of Na' and CI- to water
from the recharge area. The lines fit the observed trends
in the data extremely well, suggesting that the process
controlling the Na' ion concentration is also the one
affecting CI-.
Sulfate reduction cannot explain the increase in
HCO; and decline in SO:-. There is a lack of measurable
sulfide in either ionic or gaseous forms. The most reason-
able explanation of the SO:- distribution is a geologic
one. Approximately 35,000 years ago, glaciation contin-
ued to open up the recharge area of the aquifer and
deposited till. The sulfate chemistry of the aquifer in
areas o f siihcrop simply reflects unique processes that
operate within the till to produce high SO;- (Hendry and
others, 1986). Thus, high SO:- water is limited t o the
vicinity of the subcrop because time has not been suffi-
cient for this chemically distinctive water to move fur-
ther. On average, the transport of a twcer through the
confined part of the aquifer requires about 3 X lo5yr
0 20km (Hendry and Schwartz, 1988).
u Mass transfer calculations down-gradient of the SO{-
bulge suggest that minor cation exchange and sulfate
reduction is occurring but not nearly to the extent sug-
Tp. 10 gested initially by the major ion distributions. The increas-
N ing HCO; concentrations and large concentrations of
methane gas in the down-dip end of the aquifer come
t I from the reduction of organic matter. It is not certain
whether the reaction is occurring in the aquifer or
Tp. 5 whether both CH, and HCO; are diffusing from the Coltr
raclo shale. Given that a major gas field occurs ;it the
down-dip end of the aquifer, it is most likely that the
shale unit is the source rock for the gas.
Tp. 1 lMore recently, the Milk River aquifer was the focus
Ffgure 15.20 Concentration distribution (mM) of NilA, for a large multidisciplinary study designed to test isoto-
HCO; + COf , and S O - in waters of the Milk River aquifer pic and geochemical techniques for testing old ground
(modified from Hendry and Schwartz. 1990). Reprinted by
permission of Ground Water. Copyright 0 19%). All rights re-
served.

out, concentrations of these ions are less than a few

milligrams per liter.
The classical interpretation of these data would point m 'Site 6 5
to sulfate reduction and ion exchange as the key pro- z I

cesses responsible for generating these patterns. How-

ever, other processes are probably more important. The
aquitard diffusion model, explaining the distribution of
CI-, also constrains the behavior of the cations. Diffusion
of only CI- would produce a charge imbalance. Thus,
counter ions must diffuse from the shale to maint:iin
electroneutrality. Na' would be the likely species given
 15.5 Age Datirig of Ground Water 319

water (Ivanovich and others, 1991). Beyond standard Br, and I) and their radioactive isotopes "'CI and ""I,
major ion and environmental isotope analysis (that is Fabryka-Martin and others (1991) suggest that tlie m;rjor
tritium, D, '"O),samples were measured for "C, .KCI, influence on the major ion chemistry is tlie diffusion of
various noble gases (He, Ne, Ar, Kr. Xe, and L2zRa),Ar modified seawater from low-permeability lenses or units
isotopes ("Ar, j''Ar), uranium and thorium isotopes, and within the Milk River aquifer. Thus, while diffusion re-
various isotopes of krypton. mains a dominant process, this model looks t o soiirces
An important result of this work is yet another model within the aquifer to provide dissolved m;iss ;ISopposed
to explain the origin of salinity patterns within the Milk to sources outside the aquifer (Hendry and Schw;trtz.
River aquifer. Based on the geochemistry of halides (CI, 1988).

- Age Dating of Ground Water

15.5 larly attractive with this radionuclide is the long hirlf-life
(cIl2 = 3.01 X 10' years) and the smaller number o f
There are two different approaches for dating ground
potential reactions (as compared to I I<;) that need to he
water. The direct approaches involve interpreting the
accounted for in calculating an age date (Bentley ;incl
concentration distribution of radioactive elements that
others, 1986). One limitation with this radioisotope is
occur as environmental tracers. The most common envi-
the need for tandem accelerator mass spectrometry for
ronmental tracers are tritium, carbon-14, and chlorine-
measurements, which at present limits the av;iil;ihility of
36. The indirect dating methods rely on knowledge of
the approach.
how particular isotopes or organic chemicals change as
a function of time in recharge. Examples are S"0, which
Tritium
has changed on a continent-wide basis as a consequence
Tritium concentrations are reported in terms of tritium
of deglaciation, and short-range tracers, the chlorofluo-
units (TU) with 1 TU corresponding to one atom o f 'H
rocarbons, which have been increasing in the atmo-
in 10'' atoms of 'H (Fontes, 1980). Tritium occurs natw
sphere since they began to be manufactured in tlie 1930s.
rally in tlie atmosphere, with concentrations in precipit;i-
tion usually less than 20 TU. However, tritiiinl gencrxtrcl
Direct Metbods by thermonuclear testing in the atmosphere between
;ibout 1952 and 1963 has swamped the natur;il produc-
The direct dating techniques interpret the distribution
tion of tritium. The long-term record of 'H in precipit;ltion
of a radioactive species in ternis of a first-order kinetic
at Ottawa, CatIada, shows that during the period o f 1111-
cite 1;rw for decay. The residence time of mass in the
clear testing concentrations were often greater t1i;in I 0 0 0
system or the ground-water age ( t )is described niathemat-
TU (Figure 15.22). Tritium levels declined once weapons
ically as
testing stopped in 1963 but present-day levels reniiiin
above the natural background. Details on the seasoixil
(15.5)
variation and latitude dependance o f 'H levels in prccipi-
tation are presented by Gat (1980) and Fontes (1980).
wliere t,,, is the half-life for decay, A,, is the activity assum-
Ideally, by knowing the concentm1ion of '14 in precipi-
ing no decay occurs, and Aoh is the observed or measured
activity of the sample.
Tritium ('H, tllL= 12.35 yr) and carbon-14 ("C, tl,?=
5730 yr) are conimonly used for age dating. However,
both suffer from limitations. The relatively short half-life
for 'H makes it is useful for dating water only less than 500t--fl-
about 40 years old. In addition, the cessation of nuclear
testing in the atmosphere has eliminated the global
source of new tritium. Within a few more decades the
tritium levels in precipitation will decline to an extent
that tritium will be less useful as a tracer. "C with a much
longer half-life has the potential to date water up to
I I I L
about 40,000 years old. However, interpreting 'C data 1J
1950 1960 1970 1980 1986
is difficult because o f the need to account for other pro- Year
cesses besides radioactive decay that influence the niea- Figztre 15.22 Tritium levels in precipit;ition ;it 0 t t ; i n x On-
sured I t activity. tario (modified from Robertson and Cherr)). \Vatu Kc-
Phillips and others (1986) have demonstrated the po- sources Res., v. 25, p. 1097- 1 109, 1989. Copyright 0 ly
tential of using chlorine-36 ('%I) in dating water. Particu- Anierican Geophysical IJnion.
- Cbapter 15 / Mass Transport in Natural Ground-WaterSystems
320

tation (the source) and its distribution in ground water, mately 100 pmc (Fritz and Fontes, 1980) and slightly
one should be able to date the water. In most cases, higher in the ions coming from the dissolution of CO:.
however, tritium cannot be used in such a quantitative The dating method works because once carbonate spe-
way. The main problems stem from the uncertainty and cies move below the water table I'C begins to decay and
complexity of atmospheric loading. For example, at most there are no additional sources.
places, there are usually insufficient data to establish the The main problem in applying this method is that some
historical pattern of 'H loading. Sometimes this limitation reactive minerals contain carbon, and carbon transfers
can be overcome by correlating partial local records to in and out of the ground water. These interactions can
stations like Ottawa, Canada, with long-term records. The reduce the "C activity in the water and, thus, need to
complex loading function also provides a problem in be accounted for in estimating the age. In terms of Eq.
interpreting 'H data. Without a great deal of information 15.5, the value of& (the IiC activity assuming no decay)
about tritium distributions in the ground water, it is diffi- would be lower than 100 pmc, reflecting the fact that
cult to determine whether a sample with 30 TU is a late other processes besides radioactive decay influence the
1950s water that has decayed through three half-lives or "C activity of the sample. Any age calculation is meaning-
a 1970s water originally with 75 TU that has decayed ful as long as A,, and A,J, differ only d l i ~the effects of
through one half-life. radioactive decay.
The main application of tritium is to differentiate pre- Mook (1980), Reardon and Fritz (1978), and Wigley
1952 water from younger water. Tritium, then, is often and others (1978) list the following processes that can
used as an indirect method that relies mainly on knowl- alter the "C activity of ground water:
edge of how it is loaded to ground-water systems. The
logic is that assuming pre-1952 water to have had an 1. The congruent dissolution ol carbonate minerals,
original 3H concentration of 5 TU, the concentration in which adds "dead carbon" or carbon without "C
1988 would be at maximum 0.6 TU,which is close to activity to the ground water. Overall, this process
the detection limit even using enrichment techniques of lowers the I'C activity measured for the sample.
analysis. Thus, any detectable tritium in a sample implies 2. The incongruent dissolution of carbonate or other
that the water contains some component of more recent Ca-containing minerals, accompanied by the precipi-
or post-1952 water (Fontes, 1980). tation of calcite. This process will remove "C as
Some of the limitations of tritium dating can be over- calcite precipitates, and if dolomite is the mineral
come by adding measurements of helium-3 ('He), the dissolving, additional dead carbon is added through
stable daughter of tritium decay. The isotopic decay reac- (1) above. This process could occur in the zone of
tion is written as saturation following the rapid solution of calcite to
equilibrium with subsequent precipitation as dolo-
'H + 'He* + 0- mite slowly dissolves.
where the * indicates a tritiogenic source of 'He (Solomon 3. The addition of dead carbon from other soiirces such
and others, 1995). Other sources of 'He exist that must as the oxidation of old organic matter, sulfate reduc-
be appropriately accounted for. When both 'H and 'He tion, and methanogenesis. These again reduce the
measurements are available, summing 'H (TU) and 'He IlC activity of the sample.
(as TU) eliminates the decay of 'H (Solomon and others, 4. Possible isotopic exchange involving COi- and car-
1995). Thus, one could determine, for example, the peak
bon;ite minerals, which could lower the IiC activity.
concentration coinciding with 1960's bomb testing. In
This process is generally considered to have a negligi-
addition, one could determine the 'H/'He age of the
ble effect at normal ground-water temperatures.
water. Solomon and others (1995) demonstrated this &it-
ing technique at a site on Cape Cod. Two approaches are available for estimating a ground-
water ;ige (Kimball, 1984). One way is to interpret ages
Carbon-14 on the basis of the ion and isotopic data for a single
Measurements of IiC are reported as percent modern "C sample without information from other samples. The
(pmc) determined as the ratio of the sample activity to second approach involves using the ion or isotopic data
that of the international standard expressed as a percent- from many samples in ill1 integrated way and mass bal-
age. "C originates naturally in the upper atmosphere ance modeling to sort out all the major inputs and outputs
through a reaction involving nitrogen and neutrons. Like of carbon.
jHH,weapons testing in the atmosphere has affected its The simplest way to establish Actis to account only for
concentration in recent years. However, except [or the most important process affecting activity, which
young waters, this increase does not affect the interpre- is the congruent dissolution of calcite. The reaction be-
tation. tween water containing C 0 2 and calcite is
"C in ground water comes from the solution of CO&)
in the soil zone. The activity of in C 0 2 gas is approxi- CO? + H 2 0 + CaCO,(s) = Caz+ + 2HCO; (15.6)
 1S.S Age Dating of Ground Water 3 2

At equilibrium according to this reaction, half the bicar- (Ingerson and Pearson, 1964). Again, a carbon-mixing
bonate would be generated from a source containing "C equation can be written (Mook, 1980)
(CO,) and the other half would be generated from a dead
source (calcite). Assuming that the activity of the COLis (a + b)S"C,. = (a + 0,56)8''G:,),+ 0.566'C, (15.1 1)
1 0 0 pmc and the calcite is 0 pmc, A" would be 50 pmc, where Sl3CCI.,8'?Cc-),3 and 6l3C, are the measured values
reflecting the equal contribution of carbon from both of 6I3C in the ground water (as total carbon). soil gas.
sources. and calcite, respectively, and the other terms ;ire molar
in many cases, it is unlikely that the reaction would concentrations. With some algebraic manipulation, we
be at equilibrium due to the lack of carbonate minerals can express Eq. 15.1 1 in terms of (a + 0.56)Mn + 6)or
or kinetic effects (Pearson and Hanshaw, 1970). The fol-
lowing form of the reaction describes this more realistic (U + 0.56) -- (S"C, - S'C,.)
situation (Mook, 1980) (a + 6) (8"G.oL- S1'C,.)
(a + 0.56) C 0 2 + 0.56 CaCO, + H1O where the left side of the equation is the unknown factor
= 0.56 CaLt + 6 HCO; + a C 0 2 (15.7) in Eq. 15.9 correcting the A,, value (Mook, 1980). The
6% content of C 0 2gas in the soil zone is known rcison-
Excess C02in this reaction results in a "C activity greater ably well (i.e., -27 +- 5 / , , , PDB in temperate regions ;incl
than 50 pmc and perhaps close to 85 pmc. - 13 +. 4"/,., in tropical regions; see Mook, 1980) ;is is
Two of the simpler techniques for correcting A,, values the value for marine carbonate (0 to +Z'/,,,).
are based on these ideas. For example, one empirical All of these approaches are useful to some extent.
approach simply assumes A,, has a value of 85 2 5 pmc. However, they do not account for many of the processes
Measurements of dissolved carbon in ground water of affecting the "C content of the water (Pearson and Han-
northwest Europe suggested that this value was in fact shdw, 1970). More comprehensive schemes exist that
representative of soil water and shallow ground water we will briefly mention. Fontes and Gamier (1970). for
in temperate climates (Vogel, 1967, 1970). Another less example, account for mineral dissolution and isotopic
empirical approach is based on Eq. 15.7 and provides an exchange reactions. The most comprehensive model to
A,, value that is the weighted contribution of C 0 2 and date is that of Wigley and others (1978). Their scheme,
calcite. Mathematically, this correction is a simple mixing which is complicated enough to require a computer
equation or code, takes into account an arbitrary number of carbon
(a + b)A,, = (a + 0.56)A,:,,,+ 0.56 A,. (15.8) sources (for example, dissolution of carbonate minerals
and oxidation of organic matter) and sinks (for example,
where A,,,, is the estimated activity of C 0 2 in the soil mineral precipitation, C 0 2 degassing, and methane pro-
zone (usu&y 100 prnc), A, is the estimated activity of duction), and the equilibrium fractionation between
calcite (usually 0 pmc), and (a + b) is the total moles phases.
of carbon in the water (C,). Rearranging Eq. 15.8 and One way in which the Wigley and others (1 978) model
assumingd, is 0 provides the desired correction equation can be used is in conjunction with the code BALANCE
for A,, (Kimball, 1984;Plummer, 1984).As a first step. BALANCE
estimates the transfer of carbon into and out of solution.
A, o2
A,, = (a + 0.56) ___ (15.9) These estimates provide input data for the model of Wig-
(a + 6) ley and others (1 978) to calculate fractionation fiictors.
the predicted S'C and the corrected value of A,,.
or written in terms of C1
Kimball(1984) examined several methods for interpre-
A( o2 ting "C ages of ground water in tlie Piceance Creek lxisin
A,, = (C, - 0.511) - (1 5.1 0) in Colorado. He compared four different techniques: the
c7 mass balance model coupled with the procedure of W i g
Equation 15.10is known as the Tamers equation (Tamers, ley and others (1978), the Tamers equation. tlie Ingrrson
1967, 1975). The first term on the right side of Eq. 15.10 and Pearson equation, and the Fontes nnd <;;irniereqii;i-
assumes that for water with a normal pH the tot;il carbon tion. The results in particular for waters from the lower
is found as COL(or H2CO: depending on conventions) aquifer illustrate some important features iibotIt ' Y: cliit-
and HCO;. C, is determined using the major ion data ing. First, there is a clear requirement t o ;iccount for the
together with ;I speciation program like WATEQF. The chemical processes in producing ;I date. I n all c;ises. tlic
molar concentration of HCO; (i.e., 6) is known from the corrected ages are less than the analptical ages ;ind in a
water analysis, and A, ,>1 is taken as 100 pmc. few cases more than an order of magnitude less. Second,
Another way of establishing how much dead carbon there is a marked variability in the dates clepending on
has been contributed from carbonate dissolution is to which correction technique is used. For ex;imple. the
use 8°C as a measure of the extent of carbonate reactions dates derived from mass balance modeling sometimes
%
3 Cbapter 15 / Mass Transport in Natural Ground-WaterSystems

were an order of magnitude less than those obtained that provide a basis for establishing times. This section
from the other methods (i.e., 1600 versus 18,000 yr). It will review the use of two promising approaches that
appears from Kimball's work that the Tamers equation, involve environmental isotopes (6'"Oand SD)and anthro-
which accounts only for mineral dissolution, gives the pogenic contaminants with a known history of use (chlo-
oldest dates. The other approaches yield yoringer dates. rot1 uorocarbons).
Beyond this, there is little consistency among the results.
These results raise the question as to what interpretive 6 ' 9 and SD
method yields the best date. Can we say that because The record of changing 8'0 and SD composition of pre-
the mass balance procedure accounts for more processes cipitation recharging ail aquifer can be preserved in
the date is better? Probably not. Maybe the reaction ground water from areally extensive aquifer systems or
model could be wrong, or perhaps important processes from units having an extremely low hydraulic conductiv-
were left out. In spite of a long history of development, ity. In either case, water recharging the aquifer may be
the 14Cmethod is at best a semiquantitative tool. Confi- several thousand or tens of thousands of years younger
dent predictions can be made only when the processes than the ground water.
affecting the carbon chemistry are absolutely defined - Desaulniers and others (1981) showed the shift in the
the exception rather than the rule. The hope of collecting isotopic composition as a function of depth in ground
a single sample of water and extracting a date seems to water collected from glacial till in southwestern Ontario,
have faded in light of the effort required for process iden- Canada. The 8'0 content of shallow ground water falls
tification. in a range of -9 to -lO"/,,, (SMOW), which is similar to
present-day precipitation. Water samples collected
Cblortne-36 deeper in the various till units are isotopically lighter
There is strong evidence (Bentley and others, 1986;Phil- (that is, - 14 to - 17"/,",). Corrected "C ages for these
lips and others, 1986) that *CI will emerge as a useful water samples are in excess of 8000 yr R.P. This age
radioisotope for dating water up to two million years indicates that the deeper water was either the original
old. Again, the most important source of this isotope is pore fluid in the till or was recharged soon thereafter.
fallout from the atmosphere, although small quantities The so-called isotopic shift that is recorded with depth
are produced in the subsurface. The results of analyses thus represents the gradual climatic warming following
are reported as the ratio of *CI/CI, with a typical value deglaciation.
for meteoric water lying in a range from 100 to 500 Rndbury (1984) found similar results in studies of tills
X Because there appear to be comparatively few along the south shore of Lake Superior. At a site near
processes that affect the MCl/Cl ratio, the dating equa- Duluth, he documented a lo"/,,, shift in P O over approxi-
tions are straightforward modifications of Eq. 15. l l . The mately 50 m of low-permeability till. Again the shallow
simple geochemistry of CI in ground water avoids the samples were isotopically enriched (- 10"/,,, 8'0) rela-
complexity inherent with 14C.About the only processes tive to the deepest samples (- 17''/c,,,8'0).
contributing dead CI to ground water are the dissolution By carefully documenting the isotopic character of
or mixing of older water containing CI-. Thus Far, 'CI samples from low-permeability sites in formerly glaciated
has been used to date ground water in the Great Artesian areas, it may be possible to date ground water indirectly.
basin of Australia (Bentley and others, 1986) and the Milk Water that is isotopically lighter than present-day mete-
River aquifer in Alberta, Canada (Phillips and others, oric water is likely relatively old, originating during colder
1986). glacial climates. This approach to indirect dating appears
Like tritium, levels of .W have been elevated up to to have promise in selecting sites for the long-term con-
two or three orders of magnitude as a consequence of tainment of hazardous wastes (Desaiilniers and others,
global fallout from high-yield nuclear weapons tests in 1981).
the 1950s. The presence of so-called bomkpulse %I in
a water sample provides a clear indication of the relatively Chlorofluorocarbons
young age of the sample. To date, it has been useful in Chlorofluorocarbons (CFCs or Freons) are a Family of
sorting out the patterns of complex unsaturated zone organic compounds w e d widely as propellants in aerosol
recharge (Fabryka-Martin and others, 1993). cans and refrigerants. In recent years, their use is being
phased out because of concerns about the destruction
of ozone in the Earth's atmosphere. Two of the (CFCs)
Indirect Methods in particular, dichlorodifluoroniethane (CCl2F2or CFC-
Indirect chemical methods depend on interpreting sys- 12) and trichlorofliiorometliane (CCI,F or CFC-11) ac-
tematic changes in the chemical composition of indicator count for the greatest use commercially (Busenberg and
species or isotopes along ground-water flow paths. Un- Plummer, 1992).CFC-12and CFC-11 have been manufx-
like the radiogenic isotopes, Some chemical pattern is tured since the 1930s and 1940s,respectively.
interpreted in light of events or other independent data The potential of environmental CFCs in dating ground
 15.5 Age Datiiig of Ground Water 323

pected CFC-12 and CFC-I 1 concentrations in the atiiio-

sphere. Figure 15.23, taken from Duiikle ;ind othcrs
(1093), shows how the atmospheric concentcition of
these compounds has changed ;IS a function of timc.
Busenberg and Pluninier (1992) discuss i n detail Iiow
700
these atmospheric concentrrition v ,I I lies were recon-
stnictecl .
p 600 The dating method ;issiinies that the concentr;ition o f
i
CFCs in precipitation recharging the ground water is in
equilibrium with the atmospheric gas-phasc concentr;~-
ti ons ttccording to Henry’s 1;i w . The eqiiili brium const:in t
describing the partitioning of g;tses into w;iter is tcnipcm-
t lire dependent. Thus,the ;tctu;il concentrations of (XX ;-
11 and CFC-12 found in the ground water clepenrls on
the ground-w:iter temperature. Figure 15.23 depicts the
expected concentration of the two main CFCs i n ground
water assuming equilibrium with the ;itmospheric con-
centrations ;it 9°C.
To employ CFC-11 and CFC-12 in dating, one would
collect ;isample of ground water, determine the concen-
trations of CFCs, and use the observed contentr;itions
with Figure 15.23 to obtain ;I dite of rechmge. All of
Year
these steps are complic;itrd. To asstire t1i;it s;iniplrs ;ire
Figure 15.23 Reconstniction of atmospheric concentra- completely isolated from tlie air, Busenberg and Iluninier
tions of CFC-I 1 and CFC-I 2 since 1940 in parts per trillion (1002) sealed samples in glass ampiiles by he;it fiising
volume per volume air (after Busenberg and others, 1993; immediately following their collection. The 1abor;itory
Elkins and others, 1993) and calculated corresponding solu-
measurement of CFCs is also difficult bec;iuse it is inipecr-
bilities of CFC-I 1 and CFC-12 in water at 9°C in picogrmis
rive to avoid contamination o f samples (especiall!, old
per kilogram witer. Reprinted from Dunkle and othcrs, Wn-
ter Resources Res., v. 29, p. 3837-3860, 1993. Copyright 0 samples) with modern air, which contains high CFC con-
by American Geophysical Union. centrations. Readers interested in a detailed discussion
of the analytical techniques can refer to Busenberg ;incl
Plummer (1 992).
water has been recognized since the mid-1970s (Thomp- An example in the application of CFC dating is the
son, 1976; Thompson and Hayes, 1979). These com- study of the shallow ground water of the Delni;irv;i Penin-
pounds have also been used in tracer experiments (Ran- sula. which is located along the eastern side of <:hrs;i-
dall and Schultz, 1976). Interest in CFC dating has been peake Hay (Dunkle and others. 19%). Table 15.6 sunini;i-
renewed due mainly to efforts of scientists at the U.S. rizes data for ;I nest of wells (Figure 15.24) along :I cross
Geological Survey (Busenberg and Plummer, 1992; Dun- section (F1 -FI’) completed in a shallow surfici;tl aquifcr
kle and others, 1993). n t w tlie town o f Fairmount. The concentrations of <:I:<:-
Using the history of production of these compounds 11 and CFC-12 (pg/kg) are used with the relationship
and direct atmospheric measurements since 1977, it is presented in Figure 15.23 to provide the estimated ;ige
possible to develop relatively accurate estimates of ex- in years. Usually, the oldest of the pair o f d;ites (CFC-I I

Table 15.6 Sample Data for Ground Water from the Dehnarva Peninsula
CFC-11 CFC-12

Depth to Screen Concentration Age Concentration Age

Well (Top) (m) (Pdkd (Year) (Pg/kS) (Year)

57 6.I 661.3 1985.0 2~1.9

58 12.2 582.5 1982.5 260.4
59 18.3 409. I 1976.0 216.0
(a 24.4 232.3 1971.5 152.4
61 29.6 207.1 1970.5 95.3
From Duiiklc and others (1993).
324 Cbapter 15 / Mass Transport in Natural Ground-WaterSystems

40

-60
-80 \
....
-'O0\etticai Scale Greatly Exaggerated 0 0.5 1 mile
Explanation 1....I....I
66 -Well number 0 0.5 1 kilometer

I 1984 -CFC-modeled recharge year

0 . 3 9 --Concentration of nitrate (mg/L as nitrogen) for June 1989.

-
Numbers in parentheses are nitrate for December 1989. January 1990.

General direction cf ground-water flow

-9-1
Water table

Figure 15.24 Cross section of the Fairmount network along F1 -F1' showing CFGmodeled recharge years, nitrate
concentrations (in milligrams per liter as N), and generalized ground-water flow paths. Reprinted from Dunkle and
others, Water Resources Res., v. 29, p. 3837-3860, 1993. Copyright 0 by American Geophysical IJnion.

and CFC-12) is selected as the best estimate of the age. are limitations. For example, CFC concentrations in an-
The distribution of ages (Figure 15.24) agrees with rela- aerobic environments may be reduced through microbial
tive ages based on hydrologic arguments (Dunkle and degradation. Processes like dispersion and sorption can
others, 1993). In other words, samples furthest along the also influence CFC Concentrations (Dunkle and others,
flow system are oldest. 1993). In spite of limitations, the approach has potential
As with many other of the age dating techniques, there for dating relatively young w;iters.

Problems
1. Explain why the isotopic composition of ground wa- 4. In carbonate rocks subject to recharge, the most
ter found in glacial till in Canada and parts of the United dramatic changes in major ion chemistry occur over a
States may be depleted in 6'"Oand 6D relative to precipi- rekitively short distance as infiltration first enters the unit.
tation sampled at the same locations. Later changes are often almost insignilicant by compari-
2. Over the Plains region of the northern United States son. Explain why we use arguments related to kinetics
and Canada, carbonate-rich till overlies marine shale ~ r l and mineral equilibrium.
sandstone. In nonarid areas, recharge from snowmelt
might take on the following chemical composition as it
moves downward through till and into shale bedrock Table 15.7 Chemical Composition of Shallow Ground
(see Table 15.7). Water
Interpret the chemical evolution of the water in terms
of the most likely mass transport processes. Concentmion (milligrams per liter)
Unit Ca" Mg" Na' HCO; SO:- CI- pH
3. Explain why the order in which ground water en- till 79.0 50.0 210.0 436.0 61.0 16.0 7.80
counters minerals can be important in determining how 5,0 o.5 450.0 1044.0 b.o 53,0 8 ,
the major ion chemistry evolves.
 Problems 325

5. Using the concept of redox zones, explain why H2S 8. Use the Tamers equation to determine the age o f
gas is rarely found in ground water close to the water a water sample having a total inorganic carbon coii-
table in recharge areas. tent (C,) of 16.7 mM and a HCO; concentration o f
6. Explain how tritium can be used in hydrogeologicai 997 mg/L. The measured carbon14 activity of the sample
investigations. is 2.7% modern carbon.
7. Isotopic measurements on a sample yield a measured 9. The concentration of CFC-11 in four ground-water
carbon-14 activity of 6.9%modern carbon, and a mea- samples was measured as 1.9, 76.9, 176.1. and 430.4
sured SI3CTas total carbon of d.@/,,,. Assuming a S'jC of pg/kg (data from Dunkle and others, 1993). Estimate the
for calcite and a S W of -27"/,,, for soil gas, estimate
8'/,,, year in which recharge occurred, assuming a recharge
the age of the ground water. temperature of 9°C.
 Cbalbter 16

Mass Transport in Ground-Water Flow:

Geologic Systems

14 1 Mass Transport in Carbonate Rocks

162 Economic Mineralization
16.3 Migration and Entrapment of Hydrocarbons
144 Self-organization in Hydrogeologic Systems

The ability of ground water to dissolve rocks and minerals in more detail some diagenetic changes that affect car-
and to redistribute large quantities of dissolved mass has bonate rocks. The process responsible for most of these
broad geologic implications. The so-called geologic work changes is mixing- the displacement of connate water
of ground water that is related to mass transport thus by recharging meteoric water, the same process we dis-
includes classical problems such as chemical diagenesis, cussed in Chapter 15.
the formation of some types of ore deposits, soil salinity, The environment of diagenesis for emerging carbon-
and evaporite formation. Our objective in treating these ates is demonstrated in Figure 16.1. This idea was ad-
topics here is to develop a sense of how mass transport vanced by Vernon (1969) and modified by Hanshaw and
processes in ground water work at various scales and to others (1971). This type of environment develops every-
describe specific physical and chemical factors that relate where on an emerging carbonate platform where fresh
to these problems. water drives out the marine fluids to some depth below
sea level. Three diagenetic settings are evident: (1) a
fresh-water zone, replenished by meteoric water; ( 2 ) a
-
141 Mass Transport mixing zone where fresh water mixes with seawater;
and (3) a deep marine zone, dominated by seawater.
in Carbonate Rocks Environments such as those depicted in Figure 16.1 do
Hydrogeologists for years have had a fascination with not represent small-scale phenomena as this type of envi-
carbonate rocks. Not only are they often productive aqui- ronment must, at some point in time, exist everywhere
fers, but with karst give rise to spectacular surfxe land- on an emerging carbonate bank. The slower the rate of
forms and cave systems. However, karst is but one mani- emergence, the longer each part of the bank becomes
festation of chemical diagenesis. The term chernicul subjected to meteoric water. Diagenesis progressively
diugenesis refers to any chemical change that occurs in advances in the direction of the retreat of the sea from
a sediment following deposition. Now, let us examine the area.
 1 6 1 Mass Transport in Carbonate Rocks 327

Air + fresh water Fresh water Mixed fresh t Salt water

in pores in pores salt water in pores
in pores

Figure 16.1 Diagrammatic cross section showing a mixing zone in

an emerging carbonate rock (after Longman, 19882, as modified from
Vernon, 1969). Reprinted by permission.

Tbe Approacb Toward Cbemical Equilibrium and 500 mg/L if the partial pressure of CO? is lo-’ bars.
in Carbonate Sediments again at a pH of 7. Whatever the solubility of calcite, the
solubility of aragonite is 1.38 times higher (D. Lmgniuir,
Modern carbonate sediment in marine environments con- 1987, personal communication). Thus, fresh water enter-
sists of the metastable phase aragonite (CaCO,), Iiigh- ing the emerged aragonite is undersaturated with respect
magnesium calcite, and a limited amount of low-magne- to both calcite and aragonite and starts to dissolve the
siiim calcite. Ancient carbonate rocks, on the other hand, aragonite. When a sufficient amount of CaCO, is in solu-
are virtually void of aragonite and high-magnesium tion, the solubility of calcite is achieved so that the solu-
calcites and consist chiefly of low-magnesium calcite and tion becomes saturated with respect to calcite but r e
dolomite. Hence, sometime after emergence from the mains undersaturated with respect to aragonite. With
marine environment, some major transformations take continued dissolution of aragonite, the waters beconic.
place, some accompanied by a gain in porosity and others
by a loss in pore space. Virtually hundreds of references
on this topic are available, with the broad review in
Bathurst (1971) being the most extensive. I I1
Sequence of dtagenetic stages
111 IV
- V
100 I I I
Inversion is the replacement of a mineral by its poly-
morph, for example, aragonite by calcite. Inversion oc-
curs by either a solid-state transformation, which makes
it very slow, or by a solution-reprecipitation process,
which is likely faster (Carlson, 1983). Land and others
(1967) view the process as occurring in five progressive
stages, each o f which is associated with a loss in porosity
(Figure 16.2): stage I is unconsolidated sediment con-
sisting of aragonite and high-magnesium calcite; stage V
is a stabilized limestone consisting of mostly calcite. a
a
.
Paramoruhic calcite from aragonite

If the transformation from aragonite to calcite is one Calcite (identifiable as former Mg calcite)
of solution and reprecipitation, it may be viewed within
a simple thermodynamic framework. Comparing the
Figure 162 Sequential stages in the calcitization of a n u -
phases calcite and aragonite, calcite is the less soluble. rine carbonate sand consisting of aragonite xnd Mg calcite
However, we cannot stipulate a solubility for calcite with- (from Geol. Soc. Amer. Bull., Land and others. 1967. Repro-
out further information for some of the environmental duced with permission of the publisher. The Grologic;il Soci-
parameters. For example, calcite has a solubility of about ety of America, Boulder, Colorado USA. Copyright 0 I%-.
100 mg/L if the partial pressure of CO?is about lo-’ bars The Geological Society of America. Inc.).
3 2 Cbapter 16 / Mass Transport in Ground-Water Flow: Geologic Systems

supersaturated with respect to calcite, initiating its pre-

cipitation. With only aragonite as our initial starting mate-
rial, and with calcite having a lower solubility than arago-
nite, it is patently impossible to achieve saturation with
respect to the high-solubilityphase, and aragonite is al-
ways in a state of dissolution. Hence, in the absence of
an infinite supply, it disappears from the rocks, with all
the CaC03going to calcite. With high-magnesiumcalcites
and aragonite in the starting material, the dissolution-
reprecipitation process continues down the solubility
gradient until the transformation to the lowest-solubility
phase (calcite) is complete. Thus, with both aragonite
and high-magnesium calcites in the initial starting mate-
rial, the waters can be supersaturated with respect to
the two lowest-solubilityphases but never supersaturated
with respect to all three. When the sediment becomes
depleted of the high-solubility phase, the saturation
concentration shifts downward, becoming lower than
the saturation concentration of the remaining high-
solubility phase. Thus the ultimate transformation to
calcite.
**
Distance
(bl
-
The foregoing arguments are based purely on thermo-
dynamic grounds. What cannot be ascertained from ther-
Ffgure 263 Spatial variation of CaCO, in solution where
modynamic reasoning is the distance over which the
the entering fluid has a concentration of C,,, the equilibrium
flow system must persist until saturation with respect to concentration for calcite is C,, the concentration C, denotes
the lowest-solubility phase (calcite) is achieved. If this the position where the net dissolution is zero, and C, indi-
state of saturation and, ultimately, supersaturation cannot cates the saturation Concentration of aragonite (from Geol.
be achieved, the whole aragonite deposit will be com- SOC.h e r . Bull., Palciauskas and Domenico, 1976. Repro-
pletely dissolved and removed from the geologic record duced with permission of the publisher, The Geological Soci-
with no coprecipitation of calcite. Because calcitic lime- ety of America, Boulder, Colorado USA. Copyright 0 1976,
stone is ubiquitous in the geologic record and absent The Geological Society of America, Inc.).
in modern sediments, supersaturation with respect to
calcite is undoubtedly achieved over distances that are
short relative to the size of the emerging carbonate body. -I
c, - co (16.1)
The determination of “how short” is a problem for trans- ’c, - c e ,
port theory. l - l l + - I
Consider Figure 16.3a,which shows the spatial distri-
bution of CaCO, in solution for the problem being ad-
dressed. Meteoric water enters the emerged aragonite where xsis the saturation distance; a, is the longitudinal
deposit undenaturated with respect to aragonite and dispersivity; u is the ground-water velocity; S’ is the spe-
starts to dissolve the CaCO, deposit. The concentration cific surface of the originalmaterial, defined as the surface
eventually builds up to the saturation concentration of area of the pores per unit bulk volume of sediment; k is
calcite. Continued dissolution of aragonite gives rise to a reaction rate coefficient, defined as the volume of the
a state of supersaturation with respect to calcite, and original deposit dissolved per unit surface area of rock
calcite starts to precipitate. As the concentration continu- per unit time; and the concentrations C are as defined
ally rises, aragonite is dissolving faster than calcite is in Figure 16.3
precipitating. Finally, where the concentration no longer The numerator on the right-hand side of Eq. 16.1 sug-
changes with distance in the system, the net dissolution gests that the saturation distance increases with increas-
is zero, that is, as much calcite is precipitating as aragonite ing values of longitudinaldispersivity.Within the denomi-
is dissolving. Note that this occurs at a solubility less than nator, the important dimensionless quantity is the
the solubility of aragonite. Eventually, aragonite will be Damktihler number
totally removed from the system with precipitation of
calcite. The distance to saturation with respect to calcite
shown on Figure 1 6 . 3 is ~ described as (Palciauskas and -
S’k
a, (16.2)
Domenico, 1976) V
 1 4 1 Mass Transport in Carbonate Rocks 323

The larger this quantity, the smaller the saturation dis- 1.01 1
I
tance. Hence, this interesting interplay of variables sug-
gests that if the dissolution rate of the original material
is fast with respect to the velocity at which the dissolution
products can be carried away by moving ground water,
the saturation distance is rapidly achieved, and the tfiins-
formation to calcite takes place within ;I relatively short
distance. If the rate of dissolution is slow compared to
the velocity at which the products can be carried away, 0

more of the sediment is dissolved with no coprecipitation

of calcite; that is, the solution remains undersaturated
with respect to calcite. If the reaction rate coefficient is
close to zero, the saturation distance occurs at infinity;
that is, the fluid remains undersaturated throughout the
flow domain. Thus, a highly dispersive system thins the
mass in solution, whereas a strongly advective one re-
duces the contact time between the dissolving agent
(water) and the solid surfaces making up the porous
medium, and both intrude upon the kinetics.
Equation 16.1 also applies to the approach to calcite
Length of pack 20 cm 20 cm
equilibrium in terrestrial limestones where meteoric wa-
ter enters the formations in an undersaturated state. I n
0.01
limestones, supersaturation may be achieved because of 10 20 30 40 50
Velocity-' ~ c m )
the dissolution of Ca- and C0,-bearing minerals other
than calcite (for example, gypsum), which are more solu- Figure 164 Departure from saturation versus velocity I
ble than calcite. The solution to this problem is given in from calcite-pack experiments (from data presented by Weyl,
graphical form in Figure 16.3b. As noted, the concentra- 1958. pp. 173- 174) (from Geol. Soc. Amer. Bull., Palci;iuskis
tion once more increases exponentially with distance and Domenico, 1976. Reproduced with permission of the
from C,)to Css,passing through the saturation concentra- publisher, The Geological Society of Americ;c. Boulder. Colo-
rado USA. Copyright 0 1976. The Geological Society o f
tion at some intermediate point. In this diagram r is
America, Inc.).
the production rate of the high-solubility phase, which
produces ions in solution contributing to the supersatura-
tion, and S'k is the dissolution rate constant for the low-
solubility phase. The quantity r has the units M/L'T, or against l/z) so that the slope of the line equals S * M .
mass produced per unit volume per unit time, and S*k As shown in the figure, the departure from saturation
has the units T-', or volume dissolved per unit volume increases (approaches one) with increasing velocity. For
per unit time. As the production Kite r becomes small a given velocity, the degree of saturation increases with
with respect to S'k, C,Gt approaches C,,, and the saturation increasing viilues of S*k. The dimensionless D;imkiihlrr
distance x, becomes larger. Conversely, if r is large rela- group S"kL/ir for experiment I is 0.7 13 and for the second
tive to S*k, the saturation distance decreases. In a experiment is only 0.206. The most likely re;ison for this
monomineralic terrain, that is r = 0, the terminal con- is that experiment I had finer particles and. consequently.
centration is the saturation concentration and supersatu- a higher specific surface.
ration is not possible.
The competition between kinetic and physical pro-
The Problem of Undersaturatioir
cesses has been observed at the laboratory scale for disso-
lution- precipitation reactions at large Peclet numbers Up to now our main interest has been in the r;ite ;it which
(advective systems). This is demonstfitted in Figure 16.4, states of saturation and supersaturation are achievecl in
which shows the departure from saturation in two calcite carbonate terrain. However, we recognize t1i;it the exis-
packs that differ only in particle diameter. In these experi- tence of karst is proof of the ability of w;iter to remain
ments, water that is undersaturated with respect to in some undersaturated state over relativel!. hrge flow
calcite is sent through a calcite pack at different veloci- distances. The mechanism for this unders;itur,ition was
ties, and the concentration at the end of the pack is understood only when a number of dissolution experi-
determined for each test. Given the exit concentration, ments on limestone were carried out (Berner ;ind Morse.
the degree of saturation with respect to calcite is then 1974; Plummer and Wigley, 1976). I t was found that
determined. The departure from saturation is plotted dissolution rates decrease sharply as s;ituCition is ;ip-
330 Cbapter 16 / Muss Transport in Ground-WuterFlow: Geologic Systems

proached. For instance, Plummer and Wigley (1 976)

found dissolution rates could be approximated by a sec-
o n d a d e r equation at low solute concentrations

F = K(C, - C)' (16.3)

where I; is the dissolution rate, K is a constant that w;is

found to vary by two orders of magnitude between speci-
mens, Cis the concentration of dissolved calcite, and the
subscript s denotes a saturation concentration. However,
closer to equilibrium, the reaction was found to vaiy
between a fourthorder process and an approximare
eighthorder process. Two specimens of Iceland Spiu
were used in these experiments, the second of which
was not as clear as the first. Dissolution rates of the
second specimen were markedly slower than the first,
and the authors suspected the calcite to have trace inhihi-
tors. Trace concentrations of lead, copper, and phos-
phate in particular have been found to significantly re-
duce dissolution rates close to saturation (Terjesen and Figure 16.5 Fields of occurrence of common natural wa-
others, 1961; Berner and Morse, 1974). Thus it is the ters plotted on a graph of salinity versus Mg/Ca ratios. As sa-
linity rises, higher Mg/Ca nitios are required to promote dolo-
asymptotic approach to saturation that allows slightly
mitization (from Folk and land, 1975). Reprinted by
undersaturated water to penetrate for considerable dis-
permission.
tance in carbonate rocks.

From the preceding discussion, the lower the salinity,

Dolomitiration the lower the required Mg/Ca ratio required for dolomiti-
ration. Three models have been proposed where ground
The mineral dolomite, CaMg(C03)2,frequently forms by
replacement of calcium carbonate, CaCO, . This obvi- water is the carrier fluid for the magnesium. Figure 1 6 . 6 ~
ously requires some input of Mg, normally considered is the reflux. model proposed by Zenger (1972), where
to be provided by a moving fluid. The transformation
hypersaline waters in the su6kba environment have a
ultimately results in a 13% reduction of the space occii- density greater than that of the lagoonal seawater, so
pied by the minerals. Land (1973) performed some calcu- that waters with a high Mg/Ca ratio seep downward,
dolomitizing the carbonates beneath the lagoon. A varia-
lations that focus on the number of pore volumes of fluid
tion of this model, also proposed by Zenger (1972), is
required to dolomitize completely 1 m' of sediment with
40%porosity without any loss of porosity. His calcula- the evaporative pumping model, where marine waters
tions are as follows: are pulled landward toward the sabkba due to evapora-
tion on the sabkha during dry spells (Figure 16.66).Yet
44 pore volumes of hypersaline brine that has precipi- a third conceptual model has been proposed for the
tated gypsum mixing zone concept of Figure 16.1 by Hanshaw and
others (1 97 1). In these environments, dolomite is able
807 pore volumes of normal seawater
t o form at low Mg/Ca ratios at progressively reduced
8070 pore volumes of normal seawater diluted by a Factor salinities. At salinities on the order of 90 to 95%fresh
of 10 by fresh water water and 10 to 5%) seawater, dolomite forms at a Mg/
Ca ratio of about 1 : 1 (Scoffin, 1987). Figure 1 6 . 6 ~is
From these calculations, dolomites may form in a variety
taken from Land (1973) to account for the dolomitization
of environments where the conditions just described
of Pleistocene limestones of Jamaica.
hold. Folk and Land (1975) focus on these environments
with a diagram that plots salinity against Mg/Ca ratios of
the aqueous environment in which the transformation
takes place (Figure 16.5). Thus, for dolomites to form i n
- Economic Mfneralization
16.2
high-salinity sahkhas (the Arab word for salt flats), the White (1968) has identified four factors responsible for
Mg/Ca ratio must be on the order of 5 : 1 to I 0 : I , whereas the formation of ore deposits by ground water:
for low-salinity lakes, the ratio can be less than 1 : 1. With
1. A source for the mineral constituents
freshwater aquifers with mixing zones similar to that
shown in Figure 16.1, the ratio must be on the order o f 2. Dissolution of the minerals in water
1 : 1 or 2 : 1 , depending on the salinity. 3. Migration o f the fluids
 162 Econonric Miiiernlizntioii 331

gene sulfides, to regional scale convection deep i n the

crust in the case of porphyry copper or lode gold clepos-
its. Space is not sufficient here to examine all these depos-
its in detail. However, a glance ;it the ktctors contributing
to precipitation clearly demonstrates the role o f the mix-
ing zone in the precipitation process. The dominant con-
tributing factors include abrupt chemic;il ch;inges
(changing EH-pH conditions, decomplexation). mixing
of oxidizing and reducing waters with precipit:ition ;it
the redox front, the mixing of meteoric and ni;igm;itic
fluids, and declining temperatures.

Roll-Front Uranium Deposits

A cl;issic study on the relationship between tlie occur-
rence of iiraniiim and ground-water flow is tlie work o f
_-c- Marine
Galloway (1978) on the uranium deposits o f the Ciit;i-
waters houla Formation o f the Texas Co:ist;rl Plain. This Oligcr
Ill) cene t o Miocene-aged fomiation mnges i n thickness
from 60 to 300 m and was deposited in ;I complex tlu\.i;iI
environment. The unniiini deposits in cross section ;ire
commonly crescent shaped and are formed by miner;il
precipitation ;it a redox bountlan. An import;int ;rspcct
of this study involved mapping the distribution o f tine-
and coarse-grained facies because the distribution of m i -
niuni coincided with the distribution and orient;ition o f
permeable sand belts.
GalIoway's conceptual model (Figure 16.7) proposed
that le;iching of volcanic ash layers provided the source
(cl of uranium. This process began shortl!. after the f1iivi;iI
Figure 16.6 (a)Seepage refliix and (0) evaporative pimip- sediments were deposited. Flow through semicontined
ing models for dolomitization (after Zenger, 1972). Figure (c) sands transported uranium through the most pernic;iblc
shows mixing zone dolomite on the north cwast of Jamaica parts ofthe flow system. The flow pattern w x s controlled
(from Land, 1073). Panels (a) and ( 0 ) reproduced with per- by the geometry of the channel sands ; i d tlic presence
mission of J. of Geological Education. Panel (c) reprinted
of down-dip faults (Figure I&-). The tcinsport continuecl
from Sedimentology v. 20, by L. S. land, Holocene meteoric
dolitimization of Pleistocene liniestone in North Jamaica, p. until urriniuni and other trace met:ils like iron precipi-
41 1-422, 1973, with kind permission from Elsevier Science. tated somewhere close to the sulfate-sultidr reclos
NL, Sara Hurgerhartstraat 25, 1055 KV Amsterdam, Nether- boundary. The redox front itself would have continued to
lantls. niorc down-gradient with existing miner;ils continu;ill!,
solubilized iind precipitated. Effectivel!.. the reclos front
provides a place for initiall!. dispersed tmce met;ils t o be
concentrated. The migration of tlie front must eventuall!.
4. Precipitation of the minerals in response to physical
stop because of postdeposition;il chnnges t o the rlow
and chemical changes in the fluid and/or the po-
system that reduce tlie overall perme;ibility ;ind tlic tlux
rous medium.
of water. These changes could have included ( 1 ) the
With some rewording, a similar set of conditions can be compaction and sealing of boundan. ;iquit;irds. (I) t he
applied to the migration and entrapment of hydro- continued displacements along fiiults. ;ind (2) the cliagc-
carbons. netic modification of porous. perme;ible units ((hllo-
way, 1978).
I n oxidizing environments. the ur;in\.l ( + VI) species
Origin of Ore Deposits
.
(for example, [JOi- IJ02CO'i.l!OISO','. m d 1 I0,OH ) ;we
Mass precipitating from ground water or left as ;I product mobile. Various uranyl complexes contribute to this mo-
of weathering can in some cases form valuable mineral bility in ;in important way (Langmuir. 1 9 3 ) . Thus. ur;t-
deposits. Examples of deposits formed in this way are nium would be transported at early st;iges \vIien tlie
listed in Table 16.1. These deposits form in a variety of system is oxidizing. Mobility woulcl be m;iint;iinccl ;is
different ground-water settings that range from local wa- El, declined through ox!.gen and iron m;ing;inesc reclos
ter table conditions in the case of nickel laterites or super- zones. Galloway (1978) suggested th;it rcduct;ints might
3
% Cbapter I 6 / Mass Transport in Ground-Water Flow: Geologic Systems

Tab& 161 Some of the Different Kinds of Ore Deposits Whose Origin Depends in Part on Flowing Ground Water
Factors Contributing to
Type of Deposit Example Type of Flow System Precipitation
Nickel laterite New Caledonia Shallow. water table Weithering and changing El,-pH
at the water table
Laterite bauxite Jamaica Shallow, water-table drainage Accumulation as residual deposit
helped by karst accompanying weathering
Supergene sulfide Chuquicamate, Chile Shallow, water table Weathering and changing E,,-pH
at the water table
Calcrete uranium Yeelirrie, Australia Discharge end of shallow Dissolution from source rock,
ground-water flow system transport, and precipitation due
to evaporation and decomplex-
ation
Roll-front uranium Texas Coastal Plain Shallow ground water Leaching of ash, transport, and
precipitation at redox front
Unconfonnity-related Athabasca district Saskatchewan, Deep ground-water flow re- Mixing of oxidizing uraniferous
uranium Canada lated to faulting and reducing waters
Mississippi Valley type Pine Point, Northwest Territories, Gravity or compaction flow Leaching from sedimentary source
lead zinc deposits Canada of brines from deep sedimen- rocks, transport, and deposition
tary basins due to declining temperatures
and possibly changing El,-pH
Porphyry copper Sm Manuel, Kalamazoo, Arizona Convection in response to in- Mixing of meteoric and magmatic
trusion of a stock or dike fIuids and cooling
Lode gold deposits Carlin, Nevada Fluid convection of meteoric Leaching of source rocks, trans-
water deep in the crust port and deposition in fractured
rocks due to declining teniper-
ature

include organic debris and sulfides in the sandstone or its. All occur primarily, though not exclusively, in pre-
gases migrating vertically. The highly insoluble solids ferred horizons in carbonate rocks and are sufficiently
such as coffinite (USiO,) or uraninite (UOJ would be the conformable to the bedding so as to be described as
stable phase as redox fell to a point close to where the stratiform or stratabound. All contain fluid inclusions
sulfate-sulfide or a similar couple became controlling. filled with brines that indicate a temperature of formation
The chemical behavior of uranium is shown graphically of 100°C to 150°C. Most have similar structural associa-
in Figure 16.8. The stability field for amorphous and tions, being located in regional highs such as the Ozarks,
crystalline forms of uraninite (not shown) lies in a similar Cincinnati Arch, and so on, where they are found near
position on the EH diagram as coffinite (Galloway and the top or along the flanks of the domal structures (Fig-
Hobday, 1983). Details concerning what minerals precip- ures 16.9 and 16.10).
itate and the origin of the strongly reducing conditions A postulated ground water origin for these deposits
in the sandstones are discussed by Galloway (1978) and was stated as early as 1854 by Whitney. By the late 1800s,
Galloway and Hobday (1 983). most investigators agreed on the meteoric origin. Their
The distribution of other metals besides uranium is differences, however, had to do with the source of the
controlled by these same redox processes. For example, meteoric water. One group Favored a downward percola-
paralleling the occurrence of uraniiim in the roll front is tion theory from overlying source beds whereas the other
a zone of iron disulfide mineralization. The EI,-pH dia- group favored lateral- or upward-moving waters. Some
gram for iron shows that the stability field for pyrite lies difficulties with both these theories came about with the
in a similar position to that of coffinite or uraninite. later introduction of fluid inclusion data, which suggested
a formation temperature of 100°C to 150°C. With the
current geothermal gradient, the lowest temperature
Mississippi VaUey-Type Lead-Zinc Deposits
cited requires a depth in excess of 5000 ft whereas the
Carbonate-hosted lead - zinc deposits are found in the
deposits known to date are above 2000 ft in depth.
midcontinent region of the United States and in Canada
and are often referred to as Mississippi Valley- type depos- Waterfrom Compaction Stoiber (1941) was one of
its. Ohle (1959), and later Anderson and Macqueen the first to propose that the fluids that formed the lead-
(1 982), summarized the similarities between these depos- zinc deposits in the Tristate region were closely related
 162 Economic Mineralization 333

Uranium leached
from surftcial ash
.r
! \%.

Uranium mwesinta 7

Uranium transpwted
down hydrodynamic
gradient in miconfined Uranium emsentrated
aquifer near matgin of iron
oxidation tongues

Explanation
a Reduced fluvial and c1em5efac1e.j
fron oxrdatiort tmgue (aspect of mwe general
alteratiw phenomenal
Scale: variable-In Catahoula may range from:
1 0 10

Figure I 6 7 Diagrammatic representation of the origin of roll-front-type uranium

deposits in the Catahoula fluvial systems of Texas (modified from Galloway, 1978).
Reproduced from Econ. Geol., 1978, v. 73, p. 1656- 1676.

to oil field brines and were derived from compaction. agreement. The conceptual model of Nobel (1963) is
Nobel (1963) advanced this idea in general to account shown in Figure 16.11. As noted, the ore-forming fluids
for all Mississippi Valley-type deposits. This idea was move out as hot brine through permeable zones to pre-
accepted and amplified by Jackson and Beals (19671, cipitate eventually on basin margins that currently act as
Dunham (1970), and Dozy (1970). Dozy (1970) also ground-water recharge areas.
noted that both oil and mineral deposits were common The hypothesis of Nobel (1963) and others has been
around the Ozark uplift area, indicating that the processes examined from the perspective of mathematical (numeri-
that control the migration of oil and ore should be in cal) models by Sharp (1978) and Cathels and Smith
334 Cbapter 16 / Mass Transport in Ground-Water Flow: Geologic Systems

(1083). Sharp (1978) used a onedimensional compac-

1.7 tional model coupled to heat transport with provisions
for faulting for the overpressured fluids of the Ouachita
10 basin to account for lead-zinc deposits near the Ozark
Dome. Faulting was presumed to rupture the overpres-
0.8 sured basin during the Late Pennsylvanian-Permian pe-
riod, providing hot pore fluids to the northern Arkansas-
0.6
southeastern Missouri area where lead - zinc deposits are
known to occiir. The main emphasis of Sharp's (1978)
-
w
e
0.4
work was to determine if critical temperatures could he
maintained along with peak flow rates required in the
5
w"
0.2
mineralization. Cathels and Smith (1983) suggested a
pulselike release of compactional fluids based on the
0
modeling of a basin similar to the Illinois Basin. The
0.2
consensus of these studies indicates that the postulated
compactional origin of Mississippi Valley- type lead-zinc
0.4
deposits is possible.
Cruvfty Flow Orfgfns The Pine Points district in the
-0.6 Northwest Territories of Canada contains carhonate-
hosted lead-zinc deposits similar in many respects to
0 2 4 6 8 1 0 1 2 1 4 those of the midcontinental United States. In this region,
PH Jackson and Beals (1967) postulated that the metals were
Figure 16.8 E,I-pH diagram showing the stability fields (if derived from an argillaceous sediment or from metal s d -
crystallie and amorphous coffinite (USiOJ in carbonate-rich
fides in associated black shales. A compaction model
ground water containing approximately 60 mg/L of H2Si0,,
200 mg/L total carbonate species, and 240 pg/L uranium.
was proposed to account for the Pine Point ore deposit.
The arrow depicts the possible variation in EH-pH along the Garven and Freeze (1984a,b) examined a gnvityilriven
flow system (modified from Galloway and Kaiser, 1980). flow system as a potential origin for these stratabound
ore deposits. Their conceptual model is shown in Figure
16.12, where fluid flow is directed across a thick shale
unit and focused into a basil carbonate unit. The ore-
forming constituents are leached from the shales and
are deposited at the discharge end of the basin. The

.-In
I
a
Barite-Galena

Southeast Missouri

Ore deposits

Northern Arkansas

Figure I69 Generalized sections of two lead-zinc mining districts showing structural and stnti-
graphic relations of the ore deposits (from Geol. Soc. Amer. Special Paper, v. 24, Bastin, 1939. Re-
produced with permission of the publisher, The Geological Society of America, Boulder, Colorado
USA. Copyright 0 1939, The Geological Society of America, Inc.).
 16.2 Ecorrorriic Mitiernfizntion 3 3

0 60
U
Kilometers

Figure 16.12 Conceptual model o f fluid Ho\v iii :I c:irIxiii-

ate unit that could be the locus of a strat:ihouncl orchotl!.
(from Amer. J. Sci., Garven and Freeze. I W 4 x Reprintctl I>!
permission of Anier. J. Sci.).

noted in the figure, the fluid in the margin sediments

may be forced into the adjacent continent. that is. ;ic-
cording to Oliver (1986), the thrust sheet wts like ;I
Ffgure 1 6 1 0 Map showing the relationship of the ore de-
posits to sedimentary basins and arches and uplifts (modified
squeegee driving fluids diead of it.
from Hey1 and others, 1970). As noted in Figure 16.13, metamorphism o f coal c1r.-
creases with distance from the orogenic belt ;inel pas
fields are closer to the orogenic belt th;in o i l tielcls. both
mathematical models included not only fluid and energy presumably due to higher teniper;itures. l'he Ic;itl-zinc
transport, but mass transport as well. Their simulation deposits of the type we have been discussing ;ire fiirthcst
results indicate that gravity-driven flow can provide favor- from the belt and presumably were carried by the
able flow rates, temperatures, and metal concentrations same fluids.
for the formation of an ore deposit in a relatively short It has been mentioned that Dozy (19-0) noticed the
geologic time. Betlike ( 1 986) also considered a topo- close association between leiid-zinc deposits o f the hlis-
graphic drive system to account for Mississippi Valley- sissippi Villey- type and liquid hy<lroc;irbons.'I'liis ;issoci-
type deposits from Illinois basin brines. ation is shown in Figure 16.14. The hy~~othesis o f Oliver
The Expulsion of Fluids f r o m Orogenic Belts and (1986) suggests that both nil and niinernls wc trnnsportecI
Continental Collisions In 1986, Oliver proposed ;I by the tectonic brines from the main orogcnic I>clts011
mineralization hypothesis of continental proportions, the the continent. Such a system m;iy be seeti to bc cotillxiti-
essence of which calls for the expulsion of fluids from ble with the Garven-Freeze ( 1 9 8 4 ; ~I>) gr:ivity tlow
buried and deformed continental margin sediments to model. Here the tectonic brines originating from the
foreland basins and the continental interior during over- precollision marginal sediments ma\. mix ivith mctcoric
thrusting associated with plate collisions. A simplified water driven by the hydraulic Iie;itl cstahiishctl b!, tcctoni-
history of the Appalachian orogen in response to tlie cally created topographic relief (Figure 10.15).
closing o f the ancestral Atlantic Ocean and collision of In recent years we haye seen p;irts o f the h! pothesis
North America and Africa is shown on Figure 16.IS. As by Oliver (1986) embraced I>!. mcnibers o f thc iiiodeliiig
community--at 1e:ist those parts th:u c;in be re:iilii!, pro-
grammed for a digital computer. l'hus. in 19x8. ( h - i x - i i
proposed that o i l migration in the A1bcrt;r Ikisin tlirotlgh-
out Tertiary time was the result o f :I forc1;mcl basin uplift.
The similarity between the App;ilachi;in orogen ;in11 thc
Cordilleran orogen in Alberta has been re;ison;ibl!~\\.ell
established. C;arven (1988) st;ites that thc tr;iiisport of'
oil waned in the h t e Tertiary ;is the region;il tlo\\~tlissi-
pated due t o further erosion of the landscapc. K o o n s
(1988) discusses the fluid flow regime clue t o ;I contincn-
tal collision in New Zealand and the occiirrence of;ictit.c
zones of mineralization. Bethke (1988) ;inel Ikthkc ;inel
Mirshak (1 990) conclude on the basis o f moclel studies
Ffgure 1611 Idealized section showing aquifer transmit- tlyat deep basin brines have migrated for Iiunclretls o f
ting water of compaction from source beds (from Noble, kilometers through tlie interior hisin o f Sorth Amcric+;r.
1963). Reproduced from Econ. Geol., 1963, v. 58, p. 1145- The brines redistributed hydrocarbons i n their present
1156. reservoirs and formed Mississippi Valley-t!.pc dqosits.
 Next Page

336 Cbapter 16 / Mass Transport in Ground-Water Flow: Geologic Systems

c--
Margin sediTents
\
Thrust sheet

.;.. . .. . ..
, .. ..
Ocean crust
Transition crust

Africa
c-- \
Thrusisheet

..
I. . .

Schematic
Subsidence due to load (not to
I -Tectonic brines] scale)
(4
Figure 161.3 Simplified history of Appalachian orogen. (a) Passive continental margin with
margin sediments. (6)Thrust sheet overrides margin sediments; load causes subsidence. (c) Con-
tinent-continent collision advances thrust sheet and terminates orogeny. ( d ) Block diagram of
orogen at time corresponding to panel (6).Heavy arrows schematically illustrate flow of the
brines expelled to buried sediments. Gas and anthracite deposits are closer to orogen than oil
and bituminous coal, respectively. Continentil crust is 35 km thick; horizontal dimension in dia-
gram is 500 km. (from Geology, Oliver, 1986. Reproduced with permission of the publisher,
The Geological Society of America, Boulder, Colorado USA. Copyright 0 1986, The Geological
Society of America, Inc.).
--------
I Missouri ' Illinois
IInq
I-
They mention further that t h e migration events coincide
both temporally and spatially with episodes of orogenic
deformation along t h e margins of the craton, which pro-
vided the topographic relief to establish the paleo flow
systems. Garven (1995) presents a good overview of con-
tinental scale ground-water flow and geologic processes.

I ' I J
A Tristate .$l:?J..oilpools
100 200 km 6 N. Arkansas Ffgure 26.15 Simplified illustration showing injection of
'w
50 "'mi'es
C
D
S.E. Missouri
IllinoislKentucky
,$f,/ Ore districts tectonic brines into hydrologic flow induced by tectonically
created topography (from Geology, Oliver, 1986. Repro-
Figure 1614 Distribution of oil and ore occurrences duced with permission of the publisher, The Geological Soci-
around Ozark uplift (from Dozy, 1970). Reprinted by permis- ety of America, Boulder, Colorado USA. Copyright 0 1986,
sion of the Institution of Mining and Metallurgy. The Geological Society of America, Inc.).
Previous Page

16.3 Migration and Entrapment of Hydrocarbons 337

Noncommercial Mineralization: Saline Soils and cides almost exactly with an area of ground-water dis-
Evapodtes charge. We can see the evidence in Figure 16.17 hy
Any tinle that mineralized ground water gets within a observing that salt accumulates in places where the head
few meters of the ground surface in an arid climate, the gradient between the shallow ground water i d a deeper
potential exists for saline soils or thick evaporite deposits artesian system is at a maximum.
to form. Evaporation concentrates whatever salts were The Texas example illustrates many of the same fea-
initially present in the water. The quantity of salts gener- tures, only on a slightly larger scale. Recharge originating
ated at or near the ground surface is a function of the in rocks and alluvial fans peripheral to the salt Hats is
mass flux and the time over which the flux is operative. eventually evaporated in areas where the water table is
Minerals precipitate in the inverse order of solubility less than 1 m from the ground surface (Boyd and Kreitler.
as the solution reaches saturation with respect to the 1986). Gypsum is the most commonly found mineral,
various solid phases. The particular mineral assemblage occurring in the form of gypsum mud and gypsum sand.
that forms depends very much on the composition of the However, dolomite, calcite, magnesite, halite, and native
starting water. Figure 16.16, which is a slightly modified sulfur are also found.
version of one presented by Hardie and Eugster (1970), Boyd and Kreitler (1986) have documented the pattern
describes the possible pathways along which the major of major ion evolution along the flow system. Ground-
ion chemistry of an evaporating water could proceed. water flow from the Permian limestone bedrock is ini-
As is apparent on the figure, there are critical points at tially quite fresh. Salinity increases progressively down
which the solution will evolve in one of two possible the flow system with a shift toward Na- and CI- donii-
ways. The direction of change depends on the relative nance as the effects of evaporation become evident. Ulti-
abundance of key cations and anions. Drever (1982) dis- mately, the ground water evolves to a Na-Mg-SO,-CI-type
cusses this model in detail along with a few niore re- brine with TDS values between 50,000 and 300,000 nig/
cent refinements. L (path IV in the Hardie-Eugster model) in the most
Here we will consider two examples. The first is a case evaporated parts of the salt flats. Halite, in fact, occurs
of saline soil development in east-central Saskatchewan. locally in areas where tbe brine concentration is close
Canada (Meyboom, 1966b). and the second is ;in evapo- to the maximum observed.
rite deposit in the northern Salt Basin in west Texas and Hardie (1991) presents a good overview on the forma-
New Mexico (Hoyd and Kreitler, 1986). Shown in Figure tion of evaporites.
16.17 are features of “the Great Salt Plain” that occupies
a broad depression between hills on the east and west
sides of the study area in Saskatchewan. The shading
-
16.3 Migration and Entrapment of
outlines an area with large patches of salt efflorescence Hydrocarbons
(mainly NdZSO,) on which only a few salt-tolerant plants Whenever two immiscible fluids like oil and water oc-
are able to survive. The area of salt accumulation coin- cupy the same porous medium, there will be a siniulta-
neous flow of each fluid, with each propelled by its own
driving force. The treatment of this topic is complex s o
we will use here the abrupt interface approximation.
This interface is assumed to segregate the fluids rigidly
but will be subject to adjustments with the flow of one
Path I Path II
2mca.2 >Alkalinity Alkalinity > 2mCa.2
or both of the fluids. As with the problem discussed
for the intrusion of salt water, this approximation will
generally be a good one if the mixing zone between
the fluids is relatively narrow compared to the region
occupied by the fluids.

Displacement and Entrapment

When two homogeneous fluids occupy adjacent regions
in space (Figure 16.18),each is characterized by its own
liydradic head (Hubbert, 1943)

Figure 1616 Conceptual model o f brine evolution and the

sequence of mineral precipitation (from Boyd and Kreitler. P
h,=Z+- (16.4b)
1986; modified from Hardie and Eugster, 1970). Mineral. SOC. PZg
Am. Spec. Pub., v. 3, p. 273-290. Reprinted with per- where p2is greater than pI . Because the pressure is contin-
mission. uous across the interface, the interface is shared by both
338 Chapter 16 / Mass Transport in Ground-WaterFlow: Geologic Systems

West of Second Meridian

Legend

./> Saline soil

0 12 Head difference, in feet, betwecn piezometric surface of artesian aquifer and water table
-25- Approximate line of equal head difference. in feet, between piezometric surface of artesian
aquifer and water table

Figure 1 6 17 Saline soils on the Grc:at Salt Plain, Saskitchewin, and their relation-
ship to the hydrogeology. (Reprinted from Meyboom, P., 1966b. Groundwater Studies
in the Assiniboine River Dninagc Rasin, Part 1: The Evaluition of a Flow System in
South-central Saskatchewin. <;eologic;d Survey o f Canatla, Bull. v. 139, 65 p.).

fluids. This makes it possible to solve for the pressure in

either of Eqs. 16.4 and substitute the result in the other (16.6)
equation. Solving for the less dense fluid

(;)
hi = - b 2 - [
-yq% (16.5)
where z is the elevation of the interface given in Fig-
ure 16.18.
A question now arises as to the slope of the interface
which is an expression for the head of a lighter h i d hi if one o r both of the fluids are in movement. The slope
in terms of the head of the more dense fluid. Solving this of the interface is given as the change in its elevation z
expression for z gives along the interfwe s
 16.3 Migration and Entrapment of Hydrocarbons 333

If the denser fluid is in motion and the light fluid static,

sin a is negative, and the slope of the interface will
increase downward in the direction of flow. The greater
the flow rate of the denser fluid q2,the greater the nega-
tive tilt. This is demonstrated in Figure 16.20 forground-
water movement in an anticlinal structure containing a
PZ nonmoving liquid hydrocarbon. The relationship be-
tween the slope of the oil-water interface and the hy-
Figure 1618 Slope of the interface between two immjsci-
draulic gradient for the fluid can be expressed (Hub-
ble fluids (from Hubbert, 1940). Reprinted by permission of
the j.of Geol. Univ. of <:hicago Press. Copyright 1940. bert, 1953)

where the subscript u' is for water and o is for oil. As

Substituting Darcy's law, where the gradient is taken as the hydraulic gradient decreases and approaches zero,
ah/&, gives the slope of the oil-water interface approaches zero
(horizontal). According to Hubbert (1 9531, stnictiires
with leewardclosing dips less than the tilt of the oil-
water interface cannot hold oil. Hence, in the presence
If the dense fluid qr is not in motion of low-velocity ground water (flat hydrdulic gradients),
even minor structures become more efficient in their
capability to hold the hydrocarbon. On the other hand,
for steep hydraulic gradients, the hydrocarbon may be
As pz > p l , sin a > 0 and the slope of the interface will flushed out of the trap.
increase upward in the direction of flow. The greater the
flow rate q,of the lighter fluid, the greater the slope. It Basin Migration Models
follows that a light fluid will always displace a denser fluid
if the latter is not in motion, provided the displacement is Tcith (1988) has summarized the various forces that give
done immiscibly. Consider, for example, that the heavy rise to hydrocarbon migration. These include sediment
fluid is taken as a liquid hydrocarbon immobilized within
an anticlinal structure and the light fluid is taken as a
gas. If the anticlinal structure is filled with the liquid
hydrocarbon and a later generated gas phase enters the
reservoir, the liquid hydrocarbon can be totally displaced
by the gas. A similar type of displacement is shown in
Figure 16.19for salt water and fresh water in a synclinal
structure where the salt water is not in motion and the
displacement is assumed to be immiscible. Normally,
highly saline water would tend to lie at the bottom of
such synclines and less dense moving fluids would tend
to rise and pass over or around the more dense mass,
giving the pattern shown in Figure 16.19.

Figure 1619 Position of salt water in syncline under con. Figure 16.20 Relationship between the tilt of the o i l -
ditions of freshwater movement (from Hubbert, 1953). Re- water interface and the intensity of ground-water tlow (from
printed by permission. Hubbert, 1953). Reprinted by permission.
340 Cbapter 16 / Mass Transport in Ground-Water Fbw: Geologic Systems

as a function of time in a hypothetical basin. They con-

clude that the potential for migration is greatest in reser-
voir rocks during the active sedimentation period. Bon-
ham (1980) considers progressive compaction wherein
during the early years of deposition, water is expelled
upward from the older beds to the younger ones. With
increased time, the older sediments achieve their mini-
mum porosity with no further expulsion of the pore
waters. In the later stages of compaction, no additional
water for migration is provided from the already com-
pacted units. Thus the driving force diminishes. Bethke
(1985) examines these conditions with a numerical
model (see Chapter 9).
The compaction-heat drive model merely recognizes
that the specific volume of water increases with increas-
ing depth (temperature) (Figure 16.21) and the direction
Depth (feet)
of movement of this water coincides with the direction
Ffgure 16.21 Specific volume of water as a function of of movement caused by isothermal compaction.
depth for specified geothermal gradients (from Magara, The compaction-gravity drive system has been dis-
1978). Reprinted from Compaction and Fluid Migration: Prac- cussed by Coustau and others (1975) along with several
tical Petroleum Geology, by K.Magara, 1978, with kind prr- other investigators and is based essentially on the evolu-
mission from Elsevier Science, NL, Sara Burgerhartstmat 25, tion of abnormal pressure basins to mature gravity-flow
1055 KV Amsterdam, Netherlands. systems as demonstrated in Chapter 8. The compaction
system is thought to be the cause of hydrocarbon accu-
compaction and rebound, buoyancy, gravity flow, con- mulation in traps. The gravity flow superimposed on
fined flow, gas expansion, thermal expansion of liquids, this system can flush these hydrocarbons, except in the
molecular diffusion, and osmosis. Most of these driving centrally located parts of the basin. Toth (1988) presents
forces have been previously discussed. Tdth (1988) then an interesting modification of this system based on the
arranges basin migration models into four classes ac- hydrogeologic regimes shown in Figure 8.4. Given a suit-
cording to the dominant force driving the fluids. These able cap rock, no vertical or lateral escape of hydrocar-
include (1) compaction, (2) compaction-heat, (3) com- bons is possible in the vicinity of converging flow fields.
paction-gravitational,and (4) gravitational. This type of flow is exemplified by the convergence of
The compaction models selected for review are those continental gravity flow and marine compaction flow.
of Jacquin and Poulet (1970, 1973), Bonham (1980), and Figure 16.22 demonstrates the various hydraulic, hy-
Bethke (1985). Jacquin and Poulet (1970, 1973) set out drodynamic, and structural traps that can develop within
to determine the flux, the pressure and the flow direction the gravity-driven flow systems. Structural traps occur

Artesian basin

- ""'1
Oil
accumulation
- --+ Migrating hydrocarbons

Water flow line

___-
kI=>r=I Source rock and aquitard
Reservoir rock, carrier bed
andaquifer

Figure 1622 Hydraulic and hydrodynamically aided geologic traps in regionally uncon-
fined fields of gravity flow of formation water (from T&h, 1980). Reprinted by permission.
 16.4 Self-Organixation in Hydrogeologic Systems 341

where faulting result in offset of permeable horizons, dersaturated ground water is flowing. Zones of initially
causing a potential reservoir rock to be bounded by an higher porosity and hydraulic conductivity become pref-
impermeable one. Hydrodynamic traps can occur in anti- erentially more porous and permeable as a result of disso-
clinal structures whereas hydraulic traps occur where lution. In turn, flow is increasingly directed toward those
two flow systems converge. zones, and dissolution can be enhanced there (Figure
16.23). The feedback loop thus established is integctl to
self-organization. Numerical modeling of a generic. sin-
-
164 SelJ-Organization in Hydrogeologic gle-component system by Steefel and Lasaga (1988) indi-
Systems cates that the propagation of preferential channels in
zones of dissolution, or patterning of porosity and hydrau-
As illustrated in the previous sections on mixing zone lic conductivity, is a function of the rate of reaction rela-
phenomena and economic mineralization, the interac- tive to the rate of fluid flow. When the reaction is suffi-
tions between fluid flow and chemical reactions in dis- ciently rapid relative to flow, the length of such channels
equilibrium can result in changes in the types, amounts, is proportional to the ratio of advective to dispersive
and locations of minerals and solutes within a flow sys- transport, which is embodied in the Peclet number. That
tem. These changes Fall into the general category of self- is, when there is no competition between flow and re;ic'-
organization, defined as the patterning of one or more tion, the Damkohler number plays no role in describing
descriptive variables that result from reaction-transport the process. Where the reaction is slow relative to flow,
feedbacks (Ortoleva and others, 1987a,b). Self-organiza- channel length directly depends on the ratio of the reac-
tion in ground-water flow systems may commonly involve tion rate constant to the flow rate, embodied in the Dam-
patterning associated with dissolution, with precipita- kdhler number. Not surprisingly, in both situations, the
tion, and with interactions between dissolution and pre- relative spatial extent of patterning is dependent on the
cipitation and other processes. scale of transport being considered.
An obvious natural analog to the single-component
Patterning Associated with Dissolution system is a carbonate medium. (It should be remembered
that dissolution of the single-phase aragonite creates ;I
Dissolution-induced patterning is the simplest of the self- multicomponent problem involving precipitation; this
organizational mechanisms noted and, as such, has been more coniplicated case will be touched on later.) For a
most widely modeled, both numerically and in the labora- calcitic limestone without a relativel!. inert matrix (its
tory. Consider the most basic case, that of a medium would be provided by quartz grdins), so that dispersion
containing a single reactive component into which un- is minimal, significant channel lengths should be possi-
ble. The formation of karst conduits is a logical conse-
quence. In a related vein, petroleum engineers have con-
ducted experimental studies of flow enhancement by
Region I Region I1
acid injection into carbonate reservoir rocks. Hoefner
and Fogler (1988) introduced dilute hydrochloric acid
Increased into limestone and dolomite cores, then, after evacuating
rate of change rate of change
-c
of hydraulic of hydraulic the cores, injected an alloy into each to make a casting
conductivity conductivity of the channel network resulting from dissolution. E;ich
core was subsequently dissolved in concentmted acid,

1 leaving only the casting. The relatively rapid dissolution

of calcite at moderately acidic pH was reflected in the
limestone castings by the persistence of a single. long
Captured
flow Decreased channel. For dolomite, whose dissolution was sigiiiti-
flowthrough cantly slower, pronounced channelization was evident
only at low flow rates. These results, while forced hy
lower values of pH than those typical of natural systems,
confirm the relations of patterning to the Peclet and

i-
Danikohler numbers obtained by Steefel and Lasaga
(1 988).
Increased net Decreased net
rate of grain rate of grain
dissolution dissolution
Patterning Associated with Precipitation
and Mixed Phenomena
Figure 16.23 Dissoiution-trdnsport feedback loop for a re-
gion (I) of initially higher porosity and hydraulic conductivity It is conceivable that repetitive patterns of precipitates
adjacent to a region (11) (from h e r . J. Sci.. Ortoleva and oth- and, in turn, of reduced porosih and hydraulic conductiv-
ers, 1987a. Reprinted by permission of Amer. J. Sci.). ity could result from ground-water mixing i n it porous
342 Cbapter 16 / Mass Transport in Ground-WaterFlow: Geologic Systems

medium without concurrent dissolution. This situation bers), plugging and diversion of initial channels were min-
might exist, for example, where different flow systems imized.
meet at a common discharge zone. However, given the Beyond the simple cases of reaction-transport feed-
prevalence of multimineralic media, it is possible that backs discussed thus far, self-organization phenomena
precipitation-induced patterning linked to local dissolu- may extend to multiple, interactive precipitate bands
tion is more widespread. The systematic inversion of and dissolution channels in “impure” media. Mechanical
aragonite to calcite is one such water-rock interaction. feedbacks can additionally come into play, as dissolution
The supersaturation - nucleation-depletion cycle (Orto- may induce compaction, thereby potentially reducing
leva and others, 1987a) represents a theoretical mecha- porosity. Other phenomena operative with fluid flow
nism for repetitive banding of precipitates. Consider wa- may include buoyancydriven convective cells resulting
ter containing an aqueous species A as infiltrating a from density differences associated with dissolution and
reactive medium, dissolving a species B from dissemin- precipitation. Dewers and Ortoleva (1988) have consid-
ated grains of a mineral X, so that supersaturation results ered the role of mechanical feedbacks and convective
with respect to a solid AB. Precipitation of AB at a site cells in the generation of relatively impermeable Seals in
within the medium can repress nucleation of the solid sedimentary basins.
at another site down-gradient in favor of continued Self-organization phenomena may also extend to reac-
growth of the solid at the first site. Eventually, though, up- tions that are strongly dependent on temperature. For
gradient diffusion of B will become negligible, permitting heterogeneous chemical processes, it can be argued that
nucleation and growth at a site even farther down-gradi- it is not possible for the overall rate of reaction to increase
ent, and so forth. without limit if heat is added. If surface reactions are
The localization of precipitation by up-gradient diffu- normally small compared to diffusion rates at a given
sion may yet be overwhelmed by considerations of hy- temperature, an increase in temperature can cause an
draulic efficiency. If precipitation of a relatively insoluble exponential rise in the rate of surface reaction, biit will
solid significantly reduces the available pore space, initial have a small effect on the mass transport controlled by
dissolution channels can be abandoned in favor of form- diffusion. Surface kinetics will then speed up and eventu-
ing new, adjacent channels. Rege and Fogler (1989) have ally overhaul and become fast with respect to transport,
demonstrated this process by injecting an acidic solution thereby forcing the reaction process from the kinetic
of ferric chloride into limestone cores, inducing calcite to the diffusion-controlled regime (Frank-Kamenetskii,
dissolution and ferric hydroxide precipitation. Rege and 1969). Homogeneous liquid-liquid reactions, on the
Fogler (1989) observed fluctuations in hydraulic conduc- other hand, may increase without limit in high-tempera-
tivity during runs by monitoring changes in the pressure ture environments. These same ideas apply to exother-
gradient along each core. After each run, following mic reactions taking place in the absence of outside en-
Hoefner and Fogler (19881, a casting of the reaction- ergy. At least in theory, chemically controlled exothermic
induced channel network was created. At low initial flow reactions can pass from the kinetic to the diffusion re-
rates (high Damkohler numbers), precipitation caused gime, whereas endothermic reactions, wliich are self-
eventual refocusing of flow into a few channels in which regulating in a thermal sense, cannot. An overall view of
flow rates increased sufficiently to prevent further plug- self-regulatory or cybernetic chemical systems is given
ging. For high initial flow rates (low Damkohler num- by Frank-Kamenetskii (1969).

ProbCerns
1. Consider the following reactions: closingdip of the anticline is (10ft/mile. Can the structure
kc.’,>
hold oil? Can the structure hold gas?
A (solid) eB (solution) c- D (solid) 3. From the following figure, hl,,may be taken as a point
’1
on a piezometric surface and z as a point on a structure
a. In the absence of the solid D, write an equation for contour map. After multiplying these by appropriate den-
the equilibrium concentration C,, if there is equilib- sity amplification factors, the difference gives the head at
rium between A and B. this point in an oil body (h,,). Explain (hint: see Eq. 16.5).
b. Provide an expression for the concentration when 1-
the net dissolution rate is zero.
2. For oil and gas in an anticline consider the following:

--PI,, - 1 0 -=P U ’
pw - Po -
Pus P ,
The hydraulic gradient is 10 ft/mile and the leeward
 Problems -
343

4. Develop a hydrodynamic approach for petroleum c. If one of the fluids is a liquid hydrocarbon and the
exploration that would take advantage of the typical other is a dense salt water in motion, sin a < 0 when
kinds of reservoir data that become available once a sedi- the hydrocarbon is trapped in an anticlinal structure
mentary basin becomes extensively drilled. and the interface slopes downward in the direction
of flow. The slope will be steeper with increasing q 2 .
5. Assume an interface between two fluids of different
d. If one of the fluids is a gaseous hydrocarbon and the
density in a homogeneous, isotropic medium where
other is a moving salt water, sin (Y < 0 when the
pr > p, . Comment on whether or not the following state-
hydrocarbon is trapped in an anticlinal structure iind
ments are true. Include the reasoning you apply in arriv-
the interface slopes downward in the direction o f
ing at the conclusion.
flow. The slope, however, will be less steep tllan
in (c).
a. A sloping interface always indicates movement of e. A light fluid will always displace a heavier fluid pro-
one or both of the fluids. vided the heavy fluid is not in motion and the dis-
b. A horizontal interface is only possible if q, > q?. placement is done immiscibly.
 ChaDter 17

Introduction to Contaminant Hydrogeology

1 7.1 Sources of Ground-Water Contamination

17.2 Solute Plumes as a Manvestation of Processes
17.3 Design and Quality Assurance Issues in Solute Sampling
17.4 Sampling Methods
1 7.5 Indirect Methods for Detecting Contamination

The next three chapters examine theoretical and practi- The behavior of NAPLs in saturated and unsaturated
cal concepts related to the occurrence of contaminants fluid systems is sufficiently complicated that w e have
in the subsurface. The term contaminant is used gener- reserved Chapter 19 for a more comprehensive discus-
ally to refer to dissolved constituents or nonaqueous sion. Chapter 17, then, focuses nir.inly on plumes of
phase liquids (NAPLs) (gasoline, oil, or industrial sol- contaminants dissolved in water and techniques available
vents) added to the water as a consequence of man’s ac- for assessing the spatial distribution of such contami-
tivities, nants. Chapter 18 presents examples of analytical and
To an important extent, the complexity of a contamina- numerical procedures available to model the spreading
tion problem is determined by whether or not NAPLs of dissolved contaminant plumes.
are present. Figure 1 7 . 1 is
~ an example of the simplest
case of ground-water contamination where there is only
a plume of dissolved contaminants and no NAPLs. Con- 17.1 Sources of Ground-Water
tamination by NAPLs raises the level of complexity be-
Contamination
cause contaminants can migrate as a separate liquid
phase, a dissolved phase, and a vapor phase. Three important attributes distinguish sources of ground-
NAPLs are classified according to whether they are water contamination: (1) their degree of localization, (2)
more or less dense than water. Dense nonaqueous phase their loading history, and (3) the kinds of contaminants
liquids (DNAPLs) have a specilic gravity greater than one emanating from them. Given the large number of ways
and sink through water. Light nonaqueous phase liquids of contaminating ground water, there is a spectrum of
(LNAPLs) have a specific gravity less than one and float on source sizes ranging from an individual well to areas
water. Figure 17.l b illustrates contamination developed of 100s km2 or more. In practice, the termspoint and
due to a DNAPL spill. DNAPL, present as a residual fluid nonpofnt describe the degree of localization of the
within the pores and/or “ponded” on low-permeability source. A point source is characterized by the presence
layers, dissolves into the flowing ground water to create of an identifiable, small-scale source, such as a leaking
a large plume. Within the unsaturated zone, volatilization storage tank, one or more disposal ponds, or a sanitary
promotes the spreading of DNAPL as a vapor phase. landfill. Usually, this source produces a reasonably well-
344
 1 %1 Sources of Ground-Water Contamination 345

Dissolved Contaminant Plume j -1 Vapor

:;.I 1.1 Residual D N A P L

(6)

Figure I % I Conceptualization of contamimtion problems of varying complexity. In Panel

(a),waste dissolved in water leaking from a lagoon creates a large plume of dissolved con-
taminants. In Panel @), the presence of DNAPLs results in a much more complex problem.
where the contamination occurs ;ISa piire organic liquid. a vapor in the soil zone. and a dis-
solved phase in tlie groirnd water.

defined plume. A nonpoiiit problem refers to larger-scale, present at a source can dissolve ;it a slow rate over mimy
relatively diffuse contamination originating from many decades to provide another type of continuous source.
smaller sources, whose locations are often poorly de- Most sources of long-term leakage cannot be described
fined. Examples of nonpoint contaminants could include in terms of a constant loading function. For example.
herbicides or pesticides that are used in farming, nitrates the concentration of chemical wastes added to ;I storage
that originate as effluents from household disposal sys- pond at an industrial site can vary with time due to
tems, salt derived from highways in winter, and acid rain. changes in a manufacturing process. seasonal or cco-
Typically, there are no well-defined plumes in these cases nomic factors, or the addition of other reactive w;istes
but a large enclave of contaniination with extremely vari- (for example, Figure 17.2~).Leaching rates for solid
able concentrations. wastes at a sanitary landfill site could be controlled by
The loading history describes how the concentration seasonal factors related to recharge. or by a decline in
of a contaminant or its rate of production varies as a source strength as components of the waste (for ex;ini-
function of time at the source. A spill is an example of ple, organics) biodegrade. This latter source be1i;ivior
pulse loading, where the source produces contaminants could result in the loading history s1iou.n on Figure 17.2d.
at a fured concentration for a relatively short time (Figure Typically, increasing complexity in the loading function
1 7 . 2 ~ )This
. loading could occur from a one-time release translates directly into increasingly rari;Ible concentra-
of contaniinants from a storage tank or storage pond. tion distributions.
Long-term leakage from a source is termed continriozrs An amazingly large list of potential activities c;in result
source loading. Figure 17.26 is one type of continuous in ground-water contamination (Table 1'. 1). As ;I consc-
loading where the concentration remains constant with quence of these many different industrial. agricultural.
time. This loading might occur, for example, when small and domestic activities, tlie list of potential contaniin;ints
quantities of contaminants are leached from a volumetri- can number in the tliousands or tens of thousands of
cally large source over a long time. LNAPLs or DNAPLs compounds. The question of how to orginize this list
'1
346 Cbapter 17 / Introduction to Contaminant Hjldrogeology

that are included in ( 5 ) naturally subdivide into LNAPLs

and DNAPls.
.- All of these contaminants have the potential to produce
L
E
L * health problems. In moving through the groups, we will
0)
point out the most serious ones. As a broad generaliza-
s
0 c) tion, too much of anything in water can produce health
problems in humans. For some contaminants, particularly
the radionuclides, increasing exposure results in increas-
to *', ing health consequences. Thus, any exposure above
Time Time
(a)Spill or Pulse Loading ( b ) Continuous Source Loading
background levels can be of concern. For other contami-
Constant Concentration nants, such as the major ions Na' or CI-, there is often
a threshold below which no serious health effects will
t occur.

Radioactive Contaminants
The nuclear industry is the main generator of radioactive
contaminants. Potential sources occur throughout the
nuclear fuel cycle, which involves the mining and milling
4
a 4,
of uranium, uranium enrichment and fuel fabrication,
Time Time power plant operation, fuel reprocessing, and waste dis-
(c) Continuous Source Loading ( d ) Continuous Source Loading posal. The kinds of contaminants depend on the type of
Variable Concentration Decaying Concentration reactor and the extent of spent-fuel reprocessing. For
Figure 17.2 Examples of functions used to characterize example, in the United States, Japan, France, and Ger-
contaminant loading from (a) a spill or (b, c, and d ) long many, there are large numbers of light-water reactors
term leakage. (LWRs) that utilize enriched uranium (YJ) as the pre-
dominant fuel source, and possibly 2'9Pu and TJ. In
for study is a difficult one. One approach has been to Canada, heavy-water reactors (HWRs) use natural or en-
concentrate on a subset of this list and to pay special riched uranium as a fuel, and heavy water (DzO) as a
attention to contaminants that commonly occur in efflu- coolant and moderator.
ents and drinking water, that produce adverse health During mining when the raw ore is processed r'wU,
effects, or that persist within the food chain. An example 2.UTh,""Ra, and '"Rn( gas) are potential contaminants
is the U.S. Environmental Protection Agency's list of 129 along with nonradioactive contaminants that include
priority pollutants, containing 114 organic compounds trace constituents, and major ions such as SO:- or CI-.
and 15 inorganic species, mainly trace metals (Table Areas where this contamination occurs are Colorado,
17.2). New Mexico, Texas, Utah, Wyoming, northern Saskatche-
The organic priority pollutants are subdivided into four wan, and Ontario. The enrichment and fuel fabrication
groups: volatiles, base-neutral extractables, acid extract- step treats raw uranium concentrate to increase the con-
ables, and pesticides (Table 17.2). The method of analysis centration of 'j5U in the fuel relative to the more abundant
is different for each group but all compounds are finally ')"U, and actually produces UOz,which is the actual fuel.
examined using combined gas chromatography- mass The most common contaminants are 'W, zjsU, %r,
spectrometry (GC-MS). The inorganic group is analyzed and W s .
with other techniques. One advantage of working with Many radionuclides are generated as fission products
a standard list (for example, Table 17.2) is that highly from the decay of '"U or 'j9Ppu and from neutron activa-
automated procedures screen water samples for subsets tion of stable elements in the coolant or metallic compo-
of priority pollutants at a reasonable cost. Custom analy- nents of the reactor. Fission is the power-generating part
ses of other species are often more costly. of a nuclear reaction whereby the heavy nucleus is split
The way that we organize the contaminants is by into nuclei of lighter elements and neutrons. Neutron
grouping according to reaction type and mode of occur- activation is a process wherein neutrons are added to
rence. Thus, dissolved compounds that are affected simi- the nucleus of a stable isotope to produce a radioactive
larly by chemical, nuclear, or biological processes are one. These processes produce radionuclides like "'Cs,
grouped together. Contaminants that occur as an NAPL '%s, %o, "Cr, 54Mn,55Fe,'H,and 'j'1. Overall in excess
are a separate subgroup. The major groups of contanii- of 75 radionuclides could potentially be produced. Fortu-
nants include (1) radionuclides, (2) trace elements, (3) nately, most of these radioisotopes, except for remain
nutrients, (4) other inorganic species, ( 5 ) organic con- in the spent fuel or the reactor and are not a serious
taminants, and (6) microbial contaminants. The NAPLs source of contamination at this stage in the fuel cycle.
 17.1 Sources of Ground-Water Contaminatfon 347

Table 171 Sources of Ground-Water Contamination

CATEGORY I-SOURCES DESIGNED TO DISCHARGE Open burning and detonation sites

SUBSTANCES Radioactive disposal sites
Subsurface percolation (e.g., septic tanks and cesspools) CATEGORY 111-SOURCES DESIGNED TO RETAIN
Injection wells SUBSTANCES DURING TRANSPORT OR
Hazardous waste TRANSMISSION
Nonhazardous waste (e.g., brine disposal and drainage) Pipelines
Nonwaste (e.g., enhanced recovery, artificial recharge, solu- Hazardous waste
tion mining, and in situ mining) Nonhazardous waste
Land application Nonwaste
Wastewater (e.g., spray irrigation) Materials transport and transfer operations
Wastewater by-products (e.g., sludge) Hazardous waste
Hazardous waste Nonhazardous waste
Nonhazardous waste Nonwaste
CATEGORY 11-SOURCES DESIGNED TO STORE, TREAT, CATEGORY IV-SOURCES DISCHARGING SUBSTANCES
AND/OR DISPOSE OF SUBSTANCES; DISCHARGE AS CONSEQUENCE OF OTHER PLANNED ACTIVITIES
THROUGH UNPLANNED RELEASE Irrigation practices (eg., return flow)
Landfills Pesticide applications
Industrial hazardous waste Fertilizer applications
Industrial nonhazardous waste Animal feeding operations
Municipal sanitary De-icing salts applications
Open dumps, including illegal dumping (waste) Urban runoff
Residential (or local) disposal (waste) Percolation of atmospheric pollutants
Surface impoundments Mining and mine drainage
Hazardous waste Surface mine-related
Nonhazardous waste Underground mine-related
Waste tailings
Waste piles CATEGORY V-SOURCES PROVIDING CONDUIT OR
Hazardous waste INDUCING DISCHARGE THROUGH ALTERED FLOW
Nonhazardous waste PATERNS
Materials stockpiles (nonwaste) Production wells
Graveyards Oil (and gas) wells
Animal burial Geothermal and heat-recovery wells
Above-ground storage tanks Water-supply wells
Hazardous waste Other wells (nonwaste)
Nonhazardous waste Monitoring wells
Nonwaste Exploration wells
Underground storage tanks Construction excavation
HdZdf'dOUS Waste CATEGORY VI-NATURALLY OCCURRING SOURCES
Nonhazardous waste WHOSE DISCHARGE IS CREATED AND/OR
Nonwaste EXACERBATED BY HUMAN ACTIVITY
Containers Ground water-surface water interactions
Hazardous waste Natural leaching
Nonhazardous waste Saltwater intrusion/brackish water upconing (or intrusion o f
Nonwaste other poorquality natural water)
From Oftice of Technology Assessment (1984)

Reprocessing treats spent fuel t o remove '.%J and '"Pu. damage increases as a function of dose without a thresh-
However, the other radionuclides remain to produce the old. Other health problems associated with exposure to
high-level waste problems common to many countries radiation include, for example, cataracts. nonmalignant
of the world. skin damage, depletion of bone marrow, and infertility.
The health hazards associated with ionizing radiation
are well known. An exposed individual can be affected
Trace Metals
by cancer or by genetic defects that could affect his or
her offspring. These risks are difficult to assess at low The next major group of contaminants is the trace metals
levels of exposure. However, most risk assessments con- (Table 17.3). As a group, trace metals contain the largest
sider that the probability of inducing cancer or genetic proportion of elements found on the periodic table. The
348 Cbapter I 7 / Introduction t o Contaminant Hydrogeology

Table 17.2 Environmental Protection Agency List of Priority Pollutants.Organic Compounds Are Subdivided into
Four Categories According to the Method of Analysis
BASE-NEUTRAL EXTRACTABLES VOLATILES
Acenaphthene Diethyl phthalate Acrolein I ,Z-Dichloroethane
Acenaphthylene Dimethyl phthalate Acrylonitrile 1, I -Dichloroethylene
Anthracene 2,4-Dinitrotolueiie Benzene trans- 1,2-Dichloroethylene
Benzidine 2,GDinitrotoluene Bis(chloromethy1) ether 1,2-Dichloropropane
Benzo[a]anthracene Di-n-octyl phthalate Broniodichloromethaiie cis-1,3-Dichloropropene
Benzo[b]fluoranthene 1,2-Diphenylhydrazine Bromoform trans-1,3-Dichloropropene
Benzo[k]fluoranthene Fluoranthene Bromomethane Ethylbenzene
Benzo[ghl]perylene Fluorene Carbon tetrdchloride Methylene chloride
Benzo[a]pyrene Hexachlorobenzene Chlorobenzene 1,1,2,2-Tetrachlormthane
Bis(2-chloroethoxy)methane Hexachlorobutadiene Chloroethane Tet rachf oroethene
Bis(2chloroethyl) ether Hexachlorocyclopentadiene 2-Chloroethyl vinyl ether Toluene
Bis(2chloroisopropyl) ether Hexachloroethane Chloroform 1,1, I-Trichloroethane
Bis(2ethylhexyl) phthalate Indeno[1,2,3-cd] pyrene Chloromethane 1, I ,2-Trichloroethane
4-Bromophenyl phenyl ether Isophorone Dibroniochloromethane Trichloroethylene
Butyl benzyl phthalate Naphthalene Dichlorodifliiorometlinne Trichlorofluoromethane
2-Chloronaphthalene Nitrobenzene 1,I-Dichloroethane Vinyl chloride
4-CNorophenyl phenyl ether N-Nitrosodimethytarnine PESTICIDES
Chrysene N-Nitrosodiphenylamine Aldrin Dieldrin PCElOl6"
Dibenzo[a,h]anthracene N-Nitrosodi-n-propylamine a-BHC a-Endosulfan PCE 122 1"
Di-n-butyl phthalate Phenanthrenc 8-BHC 0-Endosulfan PCR1232"
1,2-Dichlorobenzene Pyrene y-BHC Endosulfan sulfate PCBl242"
1,3-Dichlorobenzene 2,3,7,8Tetrachlorodibenz(~~ioxin GBHC Endrin PCU-I248"
1,4-Dichlorobenzene 1,2,4-Trichlorobenzene Chlordane Endrin aldehyde PCB-1254"
3,3'-Dichlorobenzidine 4,4'-DDD Heptachlor PCB-1260"
ACID EXTRACTABLES 4,4'-DDE Heptachlor epoxide Toxaphene
pChloro-m-cresol 2-Nitrophenol 4,4'-DDT
2-CNorophenol 4-Nitrophenol "notpesticides
2,4-Dichlorophen01 Pentachlorophenol
2,4Dimethylphenol Phenol INORGANICS
Antimony Chromium Nickel
4,GDinitro-ocresol 2,4,GTrichlorophenol
2,4-Dinitrophenol Total phenols Arsenic Copper Selenium
Asbestos Cyanide Silver
Berylliiim Lead Thallium
Cadmium Mercury Zinc

most common sources of contamination include (1) ef-

fluents from mining, (2) industrial waste water, (3) run-
off, solid wastes, or waste water contributed from urban
areas, ( 4 ) agricultural wastes and fertilizers, and ( 5 ) fossil
fuels. Readers can refer to an excellent survey of sources Table 17.3 Examples of Trace Metals
and background concentrations in a book by Forstner Occurring in Ground Water
and Wittman (1981).
Aluminum Gold Silver
Trace metals can be toxic and even lethal to humans Antimony Iron Strontium
even at relatively low concentrations because of their Arsenic Lead Thallium
tendency to accumulate in the body. Some studies have
found positive correlations between the concentration Barium Lithium Tin
of trace metals in water (for example, Be, Cd, Pb, and Beryllium
Ni) and death rates from some cancers. Bioaccumulation Boron Manganese Titanium
of trace metals in the food chain has produced the most Cadmium Mercury Uranium
well-known cases of metal poisoning (for example, in Molybdenuni Vanadium
Minamata, Japan). Organisms higher up the food chain Chromium Nickel Zinc
progressively accumulate metals. Eventually, humans ilt
Cobalt Selenium
the top of the chain can experience severe health Copper
problems.
 17.1 Sources of Ground-Water Contanlittation 343

Nutrients health problems at relatively low Concentrations. With

fluoride, an increase in concentration to as little as 7 o r
This group of potential contaminants includes those ions 8 times the levels for combating tooth decay c;in c;iusc
or organic compounds containing nitrogen or phospho- skeletal fluorosis.
rus. By far, the dominant nitrogen species in ground
water is nitrate (NO;), then to a lesser extent ammonium
ion (NHf ). Agricultural practices including the use of Organic Contaminants
fertilizers containing nitrogen, cattle feeding operations, Contamination of ground water by organic compounds is
and the cultivation of virgin soils (leading to the oxidation a logical consequence of the large quantities of unrefined
of large quantities of nitrogen existing in organic matter petroleum products and man-made organic compounds
in the soil) are important sources of contamination. used today. Of the list of sources we considered pre-
Other sources are sewage that could enter ground viously, almost every one either is known to Contribute
water from septic tank systems, or irrigated waste water. or has the potential to contribute organic contaniin;tnts
According to Bouwer (1985), effluent from sewage treat- to ground water.
ment plants in the United States contains 30 mg/L of Table 17.4 lists the most important families of organic
nitrogen, mainly as NHf , as well as organic nitrogen in contaminants. The following sections examine these con-
the form of nitrobenzenes or nitrotoluenes. stituents in detail.
The main health effects related to contamination by
nitrogen compounds are (1) methemoglobinemia, a type Petroleum Hydrocarbons and Derivatives
of blood disorder in which oxygen transport in young Petroleum hydrocarbons and derivatives are made up of
babies or unborn fetuses is impaired or (2) the possibility carbon and hydrogen that are derived from crude oil.
of forming cancer-causing compounds (for example, ni- natural gas, and coal. The organic compounds in cntcle
trosamines) after drinking contaminated water. Typically, oil can be divided into three main groups (Zemo and
phosphorus contamination is considered together with others, 1993). The alkanes or paraffins are present in
nitrogen. However, it is less important because of the most petroleum products. Common alkanes (n-alkanes)
low solubility of phosphorus compounds in ground wa- have the general formula C,,H2,,-;and include i?-but;ine.
ter, the limited mobility of phosphorus due to its ten- iz-pentane, n-hexane, and so on. The cycloalkanes have
dency to sorb on solids, and the lack of proven health carbon atoms arranged in a circle containing either five
problems. The major sources of phosphorus are again or six carbon atoms. Examples include cyclopentane aiitl
soil-applied fertilizers and waste water. cyclohexane. The second major group, the ;ilkenes o r
olefins, are not constituents of crude oil but formed dur-
ing the refining process. These molecules have the gen-
Otber Inorganic Species eral formula C,,H2,, and include compounds like ethenc
This miscellaneous group includes metals present in non- and propene (see examples in Section 1 1 .?I.
trace quantities such as Ca, Mg, and Na plus nonmetals The third major group of compounds in crude oil is
such as ions containing carbon and sulfur (for example, the aromatic hydrocarbons. These compounds cont;iin
HCO;, HS-, CO:-, SO:-, and H2COJ or other species at least one benzene ring (that is. C,H,). Examples of these
such as CI- and F-.Many of these ions are major contribu- compounds include benzene (C,H,), toluene (C,H,CHO.
tors to the overall salinity of ground water. Extremely ethylbenzene (CHH,,J.and q l e n e (C,,H,(CH,)J. These so-
high concentrations of these species make water unfit called RTEX compounds are both extremely soluble in
for human consumption and for many industrial uses. water and toxic.
The health-related problems are not as serious as those The polynuclear aromatic hydrocarbons (PAHs) ;ire
caused by the other contaminant groups. However, high also components of concern in petroleuni hydrocarbons.
concentrations of even relatively nontoxic salts, particu- As indicated in Section 11.7,these compounds foriii from
larly Na', can disrupt cell or blood chemistry with serious a series of benzene rings. Examples include anthracene
consequences, At lower concentrations, an excessive in- and phenanthrene. The physical and chemic;ll properties
take of Na+ may cause less serious health effects such for the most important organic compounds in fuels are
as hypertension. summarized in Table 17.5.
The potential sources of major ion salinity include (1) The fuels are produced from crude oil through ;I retin-
saline brine that is produced with oil, (2) leachate from ing process. Distillation separates crude oil into different
mine tailings, mine spoil, or sanitary landfills, and (3) fractions according to the temperature of boiling. Zenio
industrial waste water that often has large Concentrations and others (1993) provide the following general tlescrip-
of common ions in addition to heavy metals or organic tiotis of the chemistry of these frxtions:
compounds. Fluoride is probably the best example of a Gasolines (low boiling fractions) C, to C,; a1k;unes. C,
trace nonmetal occurring as a contaminant. Like the met- to C- alkenes, aromatic compounds including the
ats, trace quantities of these contaminants can produce BTEX compounds, C, benzenes. ;tnd C , benzcnes
3 3 Cbapter 17 / Introduction t o Contaminant Hydrogeology

Tab& 17.4 Important F a d e s of Organic Contaminants Found in

Ground Water
Chemical Family Examples of Compounds

Hydrocarbons and derivatives

fuels benzene, toluene, o-xylene, butane, phenol
PAHS anthracene, phenanthrene
alcohols methanol, glycerol
creosote tncresol, ocresol
ketones acetone
Halogenated aliphatics tetrachloroethene, trichloroethene, dichlo-
romethane
Halogenated aromatics chlorobenzene, dichlorobenzene
Polychlorinated biphenyls 2,4’-PCB, 4,4’-PCB

Middle distillates (kerosene, diesel, home heating oil, Halogenated Aromatic Compounds
jet fuel) Cl0to CZ4alkanes, slightly soluble aromatic These compounds are formed from benzene rings with
compounds including C3 to C5 benzenes, C,, to C, substituted halogens. Examples include chlorobenzene
naphthalenes, and C, to C5 anthracenes, and and dichlorobenzene (DCB), which are used in various
industrial and agricultural applications. Again, by virtue
Residual products (diesel Nos. 4 and 6, Bunker C,
of their relatively large specific gravity, these compounds
motor oil), C2”to G8alkanes, nonsoluble aromatics
occur as DNAPLs in ground water.
(predominantly PAHs)
Polychlorinated Biphenyls
Figure 17.3~2shows most of the major components of
Polychlorinated biphenyls were widely used in the 1960s
crude oil as they would appear on a gas chromatograph
and 1970s in transformers and capacitors. Their environ-
(Senn and Johnson, 1985).The bars indicate the fractions
mental persistence and toxicity have made them an im-
represented in the various petroleum products. The ex-
portant contaminant even though their production has
amples of chromatograms in Figures 17.36and c illustrate
been curtailed. Chemically they consist of chlorine-substi-
that products like gasoline or jet fuel contain a very
tuted benzene rings joined together.
different set of compounds. Similarly, the subtle diffrr-
ences among the chromatograms for fuels provide a basis
Health Eflects
for “geochemical fingerprinting’’ for soiirce identifica-
There are important health effects related to drinking
tion (Senn and Johnson, 1985). water contaminated by organic compounds. However,
Overall, the compositional differences among the fuels
as Craun (1984) points out, it is difficult to establish
have important implications for monitoring and cleanup.
which compounds are most toxic because not all have
For example, the low-boiling fractions like gasoline con-
been tested, and health risks are inferred from studies of
tain many highly volatile components. The oils are much
laboratory animals, poisonings or accidental ingestion,
less volatile. The chemistry of fuels is also complicated
and occupational exposures. Furthermore, there is a seri-
by the various additives designed to improve combustion
ous lack of information on the health effects related to
and to remove combustion deposits from engines.
the combined effect of several compounds and on the
epidemiology of populations consuming contaminated
Halogenated Aliphatic Compounds water. Organic contamination may cause cancer in hu-
These compounds are formed from chains of carbon mans and animals, as well as a host of other problems
and hydrogen atoms where certain hydrogens may be including liver damage, impairment of cardiovascular
replaced by chlorine, fluorine, or bromine atoms. Exam- function, depression of the nervous system, brain disor-
ples of these compounds include tetrachloroethene ders, and various kinds of lesions. More detailed informa-
(PCE), trichloroethene (TCE), and carbon tetrdch~oride tion on health effects related to organic materials can be
(CT). Historically, the use of these solvents or their im- obtained from Craun (1984).
proper disposal has given rise to many of the most serious
problems of contamination encountered in hydrogeo-
logic practice. As Table 17.5 indicates, most of these
Biological Contaminants
compounds have specific gravities greater than 1, and The important biological contaminants include patho-
thus can be found as DNAPLs. genic bacteria, viruses, or parasites. It does not take a
 17.I Sources of Ground-Water Contanrinatioti 3 x

Table 17.5 Properties for Selected Organic Compounds

specific SOlubLlity Vapor Pressure
Compound/Fady Formula Gravity <mg/Q K,, (mm Hg)
Fuels and derivatives
Benzene 0.879 1750 130 60
Ethylbenzene 0.867 I52 1400 7
Phenol 1.071 93,000 29 0.2
Toluene 0.866 535 130 22
o-xylene 0.880 175 890 5
PAHS
Acenaphthene I .069 3.42 I0,000 0.01
knzopyrene 1.35 0.001 2 1.15 X 10" -
Henzoperylene - 0.0007 3.24 X lo(' -
Naphthalene 1.145 32 2800 0.23
Methyl naphthalene 1.025 25.4 13.OOO -
Ketones
Acetone CH,COCH, 0.791 inf 0.6 89
Methyl ethyl ketone CH+COCH,CH+ 0.805 2.68 x 105 1 .a 77.5
Halogenated aliphatics
Bn~modichloromethane CHBrC12 2.006 4400 76 50
Hromoform CHBr, 2.903 3010 250 4
Carbon tetrachloride CCI, 1.594 757 440 90
Chloroform CHCI, 1.49 8200 93 160
Chloroethane CH>CH?CI 0.903 5740 35 lo00
I , I-Dichloroethane CZH,CIz I . 176 5500 62 180
I ,ZDichloroethane C,H,CII 1.253 8520 30 61
1, I-Dichloroethene C2HlCll 1.250 2250 69 495
cis- 1,2-Dichloroethene CIH2CIZ 1.27 3500 5 206
truns-1,2,-Dichloroethene CIHICll 1.27 6300 3 265
Hexachloroethane C,Ch, 2.09 50 39,800 0.4
Methylene chloride CHIC12 1.366 20,000 I9 362
1,1,2,2-Tetrachloroethane CHCI,CHCI, 1.600 2900 250 5
Tetrdchloroethene CC12CC12 1.631 150 390 I4
1,1,1-Trichloroethane CC1, CH3 1.346 1500 320 100
I , 1,2-Trichloroethane CH2ClCHCIZ 1.441 4500 290 19
Trichloroethene CZHCIj 1.466 1100 240 60
Vinyl chloride CH,CHCI 0.908 2670 24 266
Halogenated aromatics
Chlorobenzene 1.106 466 690 9
2-Chlorophenol 1.241 29,000 15 I .42
p-Dichlorobenzene (1,4) 1.458 79 3900 0.6
Hexachlorobenzene 2.044 0.006 1.7 X 10' 1 x lo-'
Pentachlorophenol I .978 14 1.0 x 10' 1 x 10.'
1,2,4-TrichIorobenzene 1.446 30 20,000 0.42
2,4,GTrichlorophenol I .490 800 74 0.012
PCBs
Aroclor 1254 1.5 0.012 1.07 X 10" '.7 x 1 0
Other
2,(Fl)initrotoluene 1.283 1320 100 -
1,4Dioxane 1.034 4.31 x lo' 1.02 j t)
Nitrobenzene 1.203 1900 71 0.15
Tetnhydrofuran 0.888 0.3 6.6 131

Specific gravity at various ternpentures; rcfer to Nyer and others (1991) for details.
inf is infinite solubility.
Vapor pressure about 20°C; I atrn = 7<&rnm Hg.
Mcdified from Nyer and others (1~991). Reprinted by permission of Ground Water Monitoring Review. Copyright 0 1991. All rights r e s x w d
352 Chapter 17 / Introduction t o Contaminant Hydrogeology

6- benzene ous agricultural practices such as the improper disposal

T-toluene of wastes from feedlots.
Eb-ethylbenzene Particulate contaniinants are less mobile than those
i Cl I
I
P&MX-p, m-xylene
0X-o-xylene dissolved in water. For this reason, most reported prob-
lems are of a local nature, related for example to (1)
poor well construction, which enables surface runoff or
sewage to enter the well, and (2) sewer line breaks or
septic tank fields located close to a well.

-
17.2 Solute Plumes as a Manifestation
Gasoline Diesel Lubricating oil - of Processes
-Kerosene
Fuel oil The important processes and parameters that influence
(N) Calibration Mixture the transport of mass should be familiar. Now, we will
T
examine how these processes work together to control
the spread of contaminants. The natural starting place is
with the mass transport processes because they deter-
mine the maximum extent of plume spread and the geo-
metric character of the concentration distribution. The
chemical. nuclear, and biological processes mainly atten-
uate the spread of contaminants, reducing the size of the
contaminated region to a fraction of that attributable
to mass transport alone. Advection is by far the most
( b ) Unleaded Regular
dominant mass transport process in shaping the plume.
n
I L12 Hydrodynamic dispersion is usually a second-order pro-
cess, except in some cases involving fractured rocks.
The magnitude and direction of advective transport are
controlled by ( 1 ) the hydraulic conductivity distribution
within the flow field, ( 2 ) the configuration of the water
table or potentiometric surface, (3) the presence of
sources or sinks (for example, wells), and ( 4 ) the shape
(c) Jet Fuel Type A
of the flow domain. All of these parameters are important
Figure 17.3 Chromatograms for various organic com- in controlling the ground-water velocity, which drives
pounds. Panel (a) is a calibration mixture spanning the range advective transport. We can illustrate this concept with
of compounds found in gasoline. The various ranges indicate the simulation results in Figure 17.4. In these examples,
how boiling selects the various compounds in fuels. Panels the water-table configuration and the other boundary
(0) and (c) are examples of chromatograms for unleaded reg- conditions for the steady-state flow system remain fixed.
ular gasoline and jet fuel (modified from Senn and Johnson, Contaminants are added at the same place in the recharge
1985). Reprinted by permission of Conference on Petroleuni
area by fixing concentration along part of the inflow
Hydrocarbons and Organic Chemicals in Ground Water-
boundary.
Prevention, Detection and Restoration. Copyright 0 1985.
All rights reserved. Because there is no dispersion or reactions, the plumes
have a uniform concentration equal to the source concen-
tration. Their shape depends o n the hydraulic conductiv-
degree in medicine to be aware of the serious health ity distribution. A reduction in hydraulic conductivity
problems from typhoid fever, cholera, polio, and hepati- reduces the extent of plume spread by simply lowering
tis. Other less serious abdominal disorders are often too the ground-water velocity (compare Figures 17.4a and
well known by travelers to countries with poor sanitd- 0).Adding two or three layers also influences plume
tion. As a group, these health effects are some of the shape (Figures 17.4c, d, and e ) because in these cases
most significant related to the contamination of ground both the magnitude and direction of ground-water flow
water. change. Again, there is a link between advective transport
The main source of biological contamination is from and the pattern of flow.
human and animal sewage, or waste water. Ground wa- We will not take time to examine the other three
ters become contaminated due to (1) land-disposal of parameters in the list that influence advection. Earlier
sewage from centralized treatment facilities or septic tank parts of the book, hopefully, have developed a sense of
systems, (2) leachates from sanitary landfills, and (3) vari- how patterns of flow respond to water-table configura-
 17.2 Solute Plumes as a Manifestation of Processes 353

Simulated Distribution of Hvdraulic Head

for ((1)

for ( b )

for ((I)
I I 4
I 1

/ I \ \
for ( e ) I I I
L I

0.0
0.02
X

0.05m
O.025x
ox
ox 0 25x O.W% 0.75% 1.ox
Time [in years] (12) x = 884.5 rn
Ffgure 17.4 Changing plume shapes with time ( r ) in years as a function of the different
patterns of layering. In all cases, advection is the only process operating (modified from
Schwartz, 1975). Reprinted from J. Hydro], v. 27, by F W. Schwartz, On radioactive waste
management: An analysis of the parameters controlling subsurface contaminant transfer, p
51 -71, 1975 with kind permission from Elsevier Science, NL, Sara Burgerhartstraat 25. 1055
KV Amsterdam, The Netherlands.

tion, pumping, and region shape. The extension of these transport parameters except the longitudinal ( a l ) and
basic concepts to advective transport simply involves transverse (ar) dispersivities are held constant. As the
visualizing the movement of a contaminant along stream magnitude of the dispersivities (al,c y T ) increases from
tubes, at least for steady-state flow. (0.0, 0.0) m to (0.03, 0.006) m to (0.3, 0.06) m, and
Adding dispersion to advective transport can cause finally (3.0, 0.6) m (Figures 17.5~1,6, c. and d ) , the size
important changes in the shape of a plume. Consider of the plume increases markedly. However. 3s the plume
the nonlayered system shown in Figure 17.5. All of the size increases, the maximum concentration decreases.
354 Cbupter 17 / Introduction t o Contaminant Hydrogeohgy

Simulated Distribution of Hydraulic Head

A
-
Contaminant wurce
4 1.......
\
"_.---I-. ........ - l,u 1 ",= 0.0
(I = 0.0
L......................... f 1

(0)

".I" "L" , I

0 0%
0.075X
ox

0 0%
O.02Sx
n.
"I

ox 0.25~ 0.50~ 0.75~ I .ox

(4

Figure 27.5 Changing plume shapes as a function of the longitudinal and transverse dis-
persivities: (a) aL= 0.0; a7= 0.0, (b) a, = 0.03 m; a7 = 0.0% m, (c) aL= 0.3 m; a7 =
0.06 m, and ( d ) a, = 3.0 m; a7= 0.6 m (modified from Schwartz, 1975). Reprinted from J.
Hydrol., v. 27, by F. W. Schwirtz, On rdtlioactivr waste management: An analysis of the pa-
rameters controlling subsiirface contaminant tnnsfer, p. 51 -71, 1975 with kind permission
from Elsevier Science, NL, Sara Bugerhai'tstraat 25, 1055 KV Amsterdam, The Netherlands.

Increased dispersion mixes the contaminant with an in- removes a contaminant the smaller the plume will be at
creasing proportion of the uncontaminated water. a given time. Thus, the smallest plume is associated with
Because both aLand a,increase proportionately, we the smallest half-life (Figure 17.7~).
cannot separate their overall effects on concentration Ion exchange can also produce the same dramatic
distributions. A series of model trials published by Frirld attenuation in concentration. The changing plume
and Germain (1986) shows how the shape of a continu- shapes in Figure 17.8 are due to a changing selectivity
ous plume changes only as a function of the transverse coefficient (K,). When K, = 0, there is no exchange,
dispersivity. Flow is downward and laterally away from and the contaminant moves due to mass transport alone.
a source at the left end of the system (Figure 17.6). However, as K, increases over four orders of magnitude,
The transport simulations in Figure 17.6 document the the plume becomes smaller as exchange retards spread-
change in plume shape as values of @,are increased from ing (compare Figures 17.8a,b, c, and d ) . Other parame-
0.001 m, to 0.01 m, to 0.1 m. For comparative purposes, ters that control ion exchange reactions (for example,
the position of the advective front is shown by the dashed cation exchange capacity or ion charge) also influence
lines. As transverse dispersion increases, the plume the distribution of Contaminants. Irrespective of the
changes from one that is compact and not much larger model describing sorption, the process is of paramount
than the stream tube initially containing the contaminant importance in controlling contaminant transport.
(Figure 17.h) to one that has spread laterally into adja- We will not present examples of all reactions that
cent stream tubes (Figure 17.6~).Accompanying this might occur in a contaminant plume because the results
transverse spreading is a reduction in the overall length are generally similar. The greater the tendency for any
of the plume, exemplified by the position of the reaction to remove a contaminant from solution, the
0.5 contour line relative to the position of the advective smaller the plume will be relative to its unaltered size.
front. In some instances, the geochemical processes have the
Another important group of processes includes (1) a capability of immobilizing particular contaminants at the
simple, fint-order kinetic reaction that could account for source. In terms of the transport processes as a whole,
radioactive decay, biodegradation, or hydrolysis, and (2) the reactions are as important as advection in finally
an equilibrium, binary exchange reaction. Let us compare determining the disposition of contaminants.
a series of plumes in which the half-life of the kinetic The loading function also influences plume shape. One
reaction decreases from 30 yr, to 3.0 yr, to 0.33 yr (Figures modeler humorously pointed out that given a proper
17.7a, b, and c). In general, the more rapidly a reaction loading function he could generate a plume that "looked
 17.2 Solute Plumes as a ManfSestatbn of Processes 3 3

25

20

15

10

2"J

15

10

5
n
0 40 80 120 160
ttorimnrai Distance Inieters)
(I9

Figure 27.6 Changing plume shapes as a function of the transverse dispersivit)': ( N ) (Yr =
0.001 m; (6) ar = 0.01 m;and (c) aT = 0.1 m. Dashed curves mark the p s i t i o n of the front
due to advection alone (modified from Frind and Gemain, Water Resources Res., v. 12, p.
1857- 1873, 19%. Copyright by American Geophysical IJnion).

Contaminant source
, I SiinuIat;d D i s ; r i b u r Y r a y l i c H y d I ,,
tllL
J
= 30 yr.

0 05x
0 025x
ox
0.80

0.05~

ox I I I
ox 0.25~ 0.50~ 0 75x 1.ox
(c) x = 884.5 rn
Figure 17.7 Changing plume shapes as a function o f the half-life of a kinetic reaction: (a)
f l l L = 30 yr, ( b ) till = 3.0 yr, (c) tl,? = 0.33 yr (modified from Schwartz, 1975). Reprinted

from J. Hydrol., v. 27, by F. W. Schwartz, On radioactive waste management: An analysis of

the parameters controlling subsurface contaminant transfer, p. 51 -71, 1975 with kind per-
mission from Elsevier Science, NL, Sara Burgerhartstraat 25, 1055 KV Amsterdam. The Neth-
erlands.
3 s Cbapter 17 / Introduction to Contamfnant Hjsdrogeology

Simulated Distribution of Hydraulic Head

Contaminant source
1 1 1 1

0.05x
0.025w R , = 16.0
oxox 0.2% 0.75X
1 ..I I. 1 .air
1

(d u = 884.5 ~ l t

Ffgure 27.8 Changing plume shapes as a function of the selectivity coefficient for a bi-
nary exchange where in (a) K, = 0.01, (0)K, = 0.1, (c) K, = 1.0, and ( d ) KI= 10.0 (modi-
fied from Schwartz, 1975). Reprinted from J. Hydrol., v. 27, by F. W. Schwartz, On rddioac-
tive waste management: An analysis of the parameters controlling subsurface contaminant
transfer, p. 51-71, 1975 with kind permission from Elsevier Science, NL, Sara Burgerharts
traat 25, 1055 KV Amsterdam, The Netherlands.

l i e an elephant." Let us look at some plume shapes in trates a case where a contaminant is added for 5.1 years
relation to nonconstant loading functions. Adding the at a relative concentration of 1.0 and after this time at a
same quantity of mass to a flow system over increasingly relative concentration of 0.0. This plume is compact and
longer times or, in effect, reducing the rate of pulse located toward the discharge end of the system. As the
loading changes the center of mass of the plume and the initial period of loading is lengthened while at the same
internal concentration distribution. Figure 1 7 . 9 ~illus- time reducing the concentration (for example, C,, = 0.5

Simulated Distribution of Hydraulic Head

0.05~
0.025~
OX

0 0%
0 025x
ox

0.05
0.025
O
X

Ffgure 17.9 Changing plume shapes as a function of the nonconstant loading rates indi-
cated on each cross section (modified from Schwartz, 1975). Reprinted from J. Hydrol., v.
27, by F. W. Schwartz, On radioactive waste management: An analysis of the parameters con-
trolling subsurface contaminant transfer, p. 51-71, 1975 with kind permission from Elsevier
Science, NL, Sara Burgerhartstraat 25, 1055 KV Amsterdam, The Netherlands.
 17.2 Solute Plumes as a Mantfestation of Processes 3 3

for 10.2 yr in Figure 17.9b;and C,, = 0.25 for 20.4 yr in Site Location Map
Figure 17.9c),the plume is larger in size but has a smaller
maximum internal concentration, and is located closer
to the source.
With mathematical simulations like these, there was
no real problem in controlling the transport with an
appropriate choice of model parameters. In real prob-
lems where the parameters are not known at all, it may
not be easy to interpret contaminant distributions in
terms of individual processes. For example, a broad zone
of dispersion at the front of a plume might be explained
in terms of a large longitudinal dispersivity and a constant
loading function, or of a small longitudinal dispersivity
and a source concentration that increased with time, at
least initially. To explain the geometry of a plume fully
requires a detailed characterization of the various pro-
cesses and the loading function. - 40 - Water-table contour-Showing
altitude of water table, April 1974.
Dashed where approximately located. ,,o
Contour interval 1 foot. Datum is
Practured and Karst Systems mean sea level.
.,,,,,,,
,s 4 I I,. Landfill deposits
Life would be simpler if all plumes were as “ideal” as
those just discussed. Often, plumes are irregularly shaped
with nonsystematic concentration distributions. These
0 0.5 kilometer
plumes form mainly because of extreme variability in the
pattern of flow. Although such variability can occur in
Figure 17.11 Maps showing the location of the Babylon
landfill and the water-table contours in April 1974 (modified
from Kimmel and Braids, 1980).

Landfill -... A
porous media, karst and some types of fractured rocks
provide a setting where changes in hydraulic conductiv-
ity of many orders of magnitude are possible over very
small vertical and horizontal distances.
The complexity of mass spreading in such a system is
exemplified by karst (Quinlan and Ewers. 1985). Ground-
water flow in a mature karst aquifer is typically conver-
gent toward major conduits. Figure 17.10illustrates how
diffuse recharge moving through the unsaturated zone
gradually converges to a major conduit draining most of
the subbasin. Thus unlike the simulated examples. :I
plume will not gradually spread as it moves down the
flow system but converge toward a cave stream and move
rapidly to the spring or springs at the downstream end
of the basin (Quinlan and Ewers, 1985). The geonietn
of the network of small fractures and conduits determines
Tributary conduit the plume geometry. This problem is in many respects
FGw lines (diftuse flow) more analogous to those commonly encountered in
Ftgure 17.10 The complex pattern of flow and contami- surface water rather than ground water. with rapid turbu-
nant migration in a maturely karsted terrane. Flow converges lent flow in a network of channels and tribu-
to a cave stream, which is the ultimate pathway for contami-
taries.
nant migration. Diffuse flow in the ground-water system is
occurring along bedding planes and other joints (from Quin-
Ian and Ewers, 1985). Reprinted by permission of Fifth Na- Babylon, New York, Case Study
tional Symposium and Exposition on Aquifer Restoration and
Ground Water Monitoring. Copyright 0 1985. All rights re- Now let us consider some real examples of how concen-
served. tration distributions depend on processes. At Babylon.
 Next Page

3 3 Cbapter 17 / Introduction to Contaminant Hydrogeology

-50-

>' '
i,,z Landfill deposits
-
Line of equal chloride concentration,
1974-dashed where approximately located.
Interval 50 and 100 milligrams per liter.

0
1000 2000

500 meters

Figure 17.12 Map of the C1- plume in 1974 at depths below the water table ranging he-
tween 9.1 and 12.1 m (modified from Kimmel and Braids, 1980).
3000 feet

New York (see insert Figure 17.1l), CI-, nitrogen com- which is similar to the velocity reported by Kimmel and
pounds, trace metals, and various organic compounds Braids (1980). Values of dispersivity in the longitudinal
originating from a landfill have contaminated shallow and two transverse directions ( y and z) are estimated to
ground water (Kimmel and Braids, 1980). Landfilling at be 18.6 m, 3.1 m, and 0.6 m, respectively. The tendency
the site began in 1947 with urban refuse, incinerated for a,,to be about six times larger than ai.and for azto
garbage, cesspool waste, and industrial refuse. The refuse be much smaller than either of the others is in line with
is 18 to 24 m thick and often placed below the water expectations from carefully run tracer tests. The rela-
table. The cesspool wastes were treated to some extent tively smooth decline in C1- concentration away from
before being discharged into lagoons on the north and the source suggests a relatively constant rate of loading.
west parts of the property. However, there is no way to establish with the given
The surficial sand aquifer is approximately 27.5 m thick information whether the estimate of a,,.contains a compo-
and has a hydraulic conductivity of 1.7 X m/s. Flow nent due to variable source loading that might inflate the
in the aquifer is generally to the south-southeast (Figure dispersivity value.
17.11). Contaminants include major ions Caz+,Mg", Na', The ratio of NO;-N to total N (Figure 17.13) indicates
K', HCO;, SO:-, and C1-; nitrogen species such :is that at the source most of the N is present as NH;, which
NHt and NO;; heavy metals, particularly iron and manga- is indicative of reducing conditions near the landfill. The
nese; and organic compounds. source of NHt is the microbial decomposition of organic
Shown in Figure 17.12 is the C1- plume between 9.1 wastes containing nitrogen. The increase in the propor-
to 12.1 m below the water table. Because CI- does not tion of NO; -N as a function of the travel distance away
react, the plume is a manifestation of only the mass trans- from the source is probably due to the oxidation of
port processes. Of these, the most important is advection. NH; to NO; ;is mixing brings oxygen into the plume.
Longitudinal dispersion is relatively significant but trans- Mapping the distribution of various nitrogen species of-
verse dispersion is negligible. A three-dimensional analy- ten can be helpful in assessing redox conditions.
sis of the main portion of the CI- plume by Kelly (1985) This reduced zone also explains why metals, particu-
determined an advective velocity of 2.0 X cm/s, larly iron and manganese, are so mobile. The EH-pH dia-

-40

,\,,
$,; Landfill deposits
-
- Line of equal percentage, 1973-number

0 0.5

Figure 17.13 Map showing the ratio of NO;-N to total N expressed as a percent-
age. Low values close to the source reflect the presence of NH: -N as the dominant
N species (modified from Kimmel and Ilrdids, 1980).
is
ratio of nitrate to total nitrogen as percentage.
Interval: 1, 9, and 30 percent.

0 112

1 Ktlometer
1 Mile
Previous Page

17.2 Solute Plumes as a Manvestation of Processes 3.3

gram for Fe (Figure 12.11) illustrates that with moderate I Site Location Map I
reducing conditions in a pH range of 6.0 to 6.5 (as ob-
served near the source), Fe’’ is the stable form of Fe.
Similarly, MnZ’ is the stable form of manganese. The
gradual transition to more oxidizing conditions
down the plume creates a situation where solids (for
example, Fe(OH)3 and MnO(OH)2 are the most stable
form. The mobility of the metal ions is dramatically
reduced.

Alkali Lake, Oregon, Case Study

The second case examines contamination of a shallow
ground-water system at Alkali Lake, Oregon (Figure
17.14) by chlorophenolic compounds (Pankow and oth-
ers, 1984; Johnson and others, 1985). It illustrates im- 0 100 meters
portant features of transport involving dissolved organic 26 May U
0 200 400 feet
compounds subject to sorption. In November 1976, u
25,000 barrels of wastes from the manufacture of 2,4-D
Figure 17.14 Maps showing the site location (inset) and
(2,4-dichlorophenoxyaceticacid) and MCPA (4-methyl-
the configuration of the water table in May 1982 (modified
2-chlorophenoxyacetic acid) were crushed and disposed from Pankow and others, 1984). Reprinted by permission of
of in 12 shallow, unlined trenches. The eight different Ground Water. Copyright 0 1984. All rights reserved.
chlorophenolic compounds listed in Table 17.6 were of
particular interest.
Ground water flows westward in fractured, fine- fected by sorption or other retardation processes and
grained, eolian, or lacustrine deposits with a maximum thus is useful for evaluating the mass transport processes.
thickness of approximately 30 m (Figure 17.14). These Qualitatively, the shape of the plume indicates that longi-
deposits contain 2.4 t 0.3%solid organic matter. Hydrau- tudinal dispersion is more significant than transverse dis-
lic conductivities fall in a range from 0.01 to 0.10 cm/s, persion. An unpublished analysis of this plume has deter-
while effective porosities are from 1 to 5%.These porosity mined a mean flow velocity of 1.25 X lo-‘ cm/s and
values are much less than the measured total porosities values of cr,, q ,and rr, of approximately 700 cm, 70 cm,
of 0.6 to 0.7. With the given hydraulic gradient, the mean and 7 cm, respectively.
ground-water velocity ranges from 3.9 to 1040 Of the eight contaminants studied at the site, 2,4-DCP,
cm/day. 2,6DCP, and 2,4,GDCP were shown through batch ex-
All eight contaminants have spread to some extent. periments to have a K,,of zero. There is thus no tendency
The plumes for 2,4-DCP (Figure 17.15a),2,GDCP, and for sorption on solid organic carbon in the aquifer. In
2,4,6-TCP are the largest. Spreading is less with TeCP ground water of pH 10 at this site, these compounds
(Figure 17.156) and PCP. The CPP plumes are the small- ionize to negatively charged species. They are, therefore.
est, especially CL4D2 shown in Figure 17.1%. Contami- much more mobile than their published K,,,,.values would
nants are moving approximately in the direction of the suggest. This example points out the need to understand
hydraulic gradient. However, because of anisotropism, fully the limitations in predicting K,. values from regres-
the direction of flow and the gradient do not exactly sion relationships. The other contaminants TeCP, PCP,
coincide. The plume is elongated in the direction of the and the three CPPs are also largely ionized in this ground
hydraulic gradient, implying that advection is the domi- water. However, these compounds do sorb to some ex-
nant mechanism of spreading. The 2,4-DCPplume is unaf- tent on organic carbon. Even so, the retardation in these

Table 176 List of Contaminants Investigated by Johnson and Others (1985) at

Alkalai Lake, Oregon
~~~~~ ~ ~~ ~~ ~~

2,4-DCP 2,4-Dichlorophenol CPPs including CL2D2, CL3D3, CL4D2

2.6DCP 2,GDichlorophenoI chlorophenoxyphenol dimers with two.
2,4,GTCP 2,4,6Trichlorophenol three, and four chlorines
TeCP 2,3,4,6Tetrachlorophenol
PCP Pentxhlorophenol
Reprinted by permission of Ground Wttrr. Copyright 0 1985. All rights reserved.
360 Cbapter 17 / Introduction to Contaminant Hydrogeology

(a)Distributionof 2,4-DCP fmgnf

TRL

N I

(c) Distribution of CL4D2 (&L)

Rgure 1715 Maps illustrating the pattern of spreading for (a) 2,4DCP, (6) TeCP, and
(c) CL4D2 at Alkali Lake (modified from Johnson and others, 1985). Reprinted by permis-
sion of Ground Water. Copyright 0 1985. All rights reserved.

plumes is much less than predicted from Kdvalues deter- definition is such an integral part of any field study. A
mined from batch experiments. This behavior may be lesson too often learned the “hard way” is that all the
due in part to fractures in sediments at the site. The dat;r basic theory and background knowledge in the world
from the batch experiments do predict correctly the cannot correct problems with poor measurements. This
relative order of retardation for all eight compounds. section deals with network design and issues of quality
This study is instructive because it is one of the first to assurance in sampling water.
document compounddependent retardation for or-
ganic compounds.
Design of Sampling Networks
-
1 Z3 Design and Quality Assurance A major concern in sampling is the design of the network.
Important considerations in design include the need for
Issues in Solute Sampling close interval point sampling and sample locations that
The problem of how to characterize the distribution of take into account the character and complexity of flow.
contaminantsis of great practical concern because plume With piezometers used for measuring water levels, often
 17.3 Design and Quality Assurance Issues in Solute Sampling 3 3

little attention is paid to screen length. Consider the Nitrate

aquifer in Figure 17.16. As far as hydraulic head is con- O r - - - l - l
cerned, there is little difference in whether we measure
Water table
head in a small volume of the aquifer (point sample) or 5- -
in a larger volume (nonpoint sample). The hydraulic head
is about the same in either case (Figure 1 7 . 1 6 ~because
~)
in a permeable unit the vertical gradients in hydraulic -
head are usually small. Using the piezometers for chemi-
cal sampling, however, can produce dramatically differ-
ent results (Figure 17.166). For this example, only the -
point samples provide concentrations suitable for inter-
preting the contaminant distribution. Because of mixing
-
within the sampling device, estimates of the plume geom-
Multilevel sampler
etry based on the nonpoint samples would be in error. Observation well
Thus, all measurements not concerned simply with pres- November I985
-
ence/absence indications should use point sampling.
Because point sampling usually involves removing a
small volume of water, systems should be designed with -
a minimum internal volume. This consideration requires a
0 5 10 15
small sampling chamber or small-diameter casing (tubing) Nitrogen
connecting the intake system to the surface. Keeping the lrnilligrams per liter)
internal volume small avoids the possibility of mixing a
sample with a large preexisting volume of water of differ- Figure 17.1 7 Demonstration of the need for close-intcnal
vertical sampling. Data for nitrate (modified from Smith ind
ent composition. Such mixing in large-diameter monitor-
others, 1987).
ing wells could be a major source of error. Unfortunately,
a system that is ideal for chemical sampling is usually
not very good for measuring hydraulic heads with an
electric tape. interval of about 25 m (Figure 17.17). The loss in resolu-
The location of sampling points should account for tion in defining the plume with a small number of vertical
the character and complexity of flow. In many respects, samples is obvious. With NO;, four piezometers failed
defining the vertical extent of the plume is most difficult even to detect the plume (Figure 17.17).
because often plumes are not very thick. For example, For determining plume boundaries, wells need to be
Smith and others (1987) compare the vertical variation installed within and closely adjacent to the plume. One
in specific conductance and nitrate concentration as de- line of samplers installed along the midline of the plume
duced from four standpipe piezometers with results from will establish its overall length and thickness. Lines of
15 multiport samplers installed at the same site over an samplers installed normal to the midline serve to define
the plume’s width and thickness. This simple design is
not universally applicable. Care must be taken to design
Measured water levels Measured concentration a system that is appropriate for the overall complexity
P P L L of flow conditions. A good illustration comes from Quin-
:
I I
i : ! 14 1 3 ni&L
lan and Ewers (1985), who looked at the problem of
how to define plumes in karst terranes. As we showed
with Figure 17.13, contaminants from a surface source
could migrate through a succession of larger and larger
conduits to one large conduit possibly 3 to 10 ni wide.
With a simple network design such as we just described,
A I there is almost no chance of intersecting this single con-
duit and collecting samples along the critical pathway
(Lo w for contaminant migration. Quinlan and Ewers ( 1985)
proposed a site-specific monitoring scheme that requires
Figure 1716 (0) In many cases there may not be a sub-
stantial difference in hydraulic head values measured in an the definition of major elements of the karst SySteRl.
aquifer using point and nonpoint sampling procedures. (6) Spring mapping and dye tracing are some of the tools
Concentrations of contaminants can change markedly over that can establish connections among the disposal site,
relatively small vertical distances. In this example, nonpoint springs, and the conduit systems. Ultimately, a few criti-
sampling will yield concentrations that are much less than cal springs and wells on the major conduits provide the
the true concentration. proper monitoring sites.
-
362 C b a p t e r I 7 / Introduction t o Contaniftiant Hydrogeology

Assuring the Quality of Cbemical Data others (19811, Scalf and others (19Sl), Claasen (1982),
and Lico and others (1982). Methods for preserving sam-
Another concern in sampling is with the overall quality ples containing organic compounds are described by
of the chemical data. Reliable chemical data are not the Goerlitz and Brown (1972), Wershaw and others (1973),
result of chance but the care and precautions taken in and Minear and Keith (1984).
study programs. There are four potential problems that Problems can also be related to the handling of samples
can have a serious impact on chemical data, including in the field or laboratory analyses. In many studies, these
(1) contamination of samples with fluids that were used problems go unrecognized and have an important impact
to drill the hole, (2) changes in water quality caused by on the quality of the chemical data. Sample contamina-
the presence of the well, (3) sample deterioration, and tion caused by improper bottle washing, filtering, or the
( 4 ) sloppy field and laboratory practices. use of impure preservatives is the main concern in han-
The first problem has to do with the question of exactly dling samples in the field. It is prudent to check for
what fluid is being sampled. Occasionally, samples will this problem on an ongoing basis by ninning blanks of
contain some of the gas, water, or oil in combination ultrapure distilled water through the various field treat-
with foams, emulsions, and muds that were required ments and submitting these samples to a laboratory. In
to drill the hole (Hull and others, 1984). Once these the case of organic compounds, adsorption onto contain-
exogenous fluids are present in the subsurface, it is often ers or filtering devices, and loss of volatiles from the
difficult to remove them completely. One way to avoid sample, can also pose problems (Gibb and Barcelona,
sampling drilling water is to develop the well. A tracer, 1984;Jackson et al., 1985).
added to the drilling fluid, can be monitored as the Likewise, the quality o f the laboratory analyses should
well is developed to check that development is com- be checked on an ongoing basis. Some common a p
plete. proaches are (1) submitting “spiked” samples with
Changes in the chemistry of the ground water can also known concentrations, ( 2 ) submitting duplicate samples
be related to the presence of the well. The most serious to different laboratories, and (3) submitting replicate sam-
potential problem arises from the use of cement (or other ples. The first two of these checks generally assess the
soluble materials) as a seal above the intake of a well or analytical accuracy of the laboratory, while the third tests
piezometer. High sample pHs (for example, above 9) analytical repeatability. Guidance for developing quality
mostly reflect a continuing interaction between the ce- assurance programs is provided by Wilson (1952), the
ment seal and the water being sampled. This problem American Chemical Society (1980), and Friedman and
can persist for a long time and is difficult to fix. Samples Erdman (1982).
can also be affected when water inside or outside of the
well casing dissolves small quantities of the casing or
glues used at joints. A possible solution to these problems
is to remove several standpipe volumes of water before 17.4 Samplgng Metbods
sampling. However, a proper selection of materials for This section examines the various approaches available
the sampling device is the only sure fix for these to hydrogeologists in collecting ground-water samples
problems. for chemical analysis. The range of possible approaches
Another problem is sample deterioration. Once the is surprisingly broad, from simply pumping water from
temperature, pressure, and gas content of a water sample a well to new and technologically innovative approaches.
change following collection, the chemistry can change
a5 well. For example, the addition of oxygen to an anoxic
sample results in a rapid change in Ell that can cause
Conventtonal Wells or Piezometers
dissolved trace metals to precipitate as solids. Loss of
C 0 2 can raise the pH and decrease the HCO; concentra- Commonly, water samples are pumped or bailed from
tion and in some cases cause CaCO, to precipitate. The wells or piezometers. In most site investigations, a cluster
extent of the change in composition depends in many of wells o r piezometers-referred to as a nest-is in-
respects on the chemistry of the particular sample. These stalled so as to obtain vertically distributed water samples
problems have been addressed by specialized sampling (Figure 17.18~). This vertical coverage is critical in defin-
equipment for high-pressure- high-temperature environ- ing contaminant distributions.
ments (Hull et al., 1984), the use of flow cells at the we:II The most important feature with this approach is that
head (Jackson and Inch, 1980), and methods of sample each well or piezometer is completed in its own bore-
treatment to avoid deterioration. In designing a sampling hole. In spite of technologies that facilitate completion
program, it is a good idea to refer to information on of multiple sampling wells in a single piezometer, there
standard practice, for example, Rainwater and Thatcher is a compelling simplicity with a permanent single-well
(1960), Ball and others (1976), Wood (1976), Gihb and completion. Seals are easier to place as compared to
 17.4 Snmplirig Methods 363

multilevel, and water levels can be measured in a straight- contaminating wells, companies are marketing lines of
forward manner. These installations are also extremely disposable bailers. Bailing, however, may pose problems
durable, with little possibility of failure once the stand- in many applications due to loss of gas or volatile org;inic
pipes have been emplaced. The main disadvantage o f compounds. In most cases, water is pumped from wells
piezometers for sampling is the higher cost of drilling using permanently installed gas drive or submersible elec-
additional boreholes versus the installation of many multi- tric pumps. A detailed survey of all of the common s m -
level samplers in a single borehole. piing devices, and extensive discussion of the appropri-
Figure 17.186 illustrates a conventional standpipe pi- ateness in sampling for various types of c1ieniic;tl
ezometer. The design with filter sand placed around the constituents, is provided by U.S. EPA (1093).
intake is most appropriate for shallower depths. When
these samplers are installed in relatively deep borehdles,
Multilevel Samplers
the seal is set above a metal-petal basket (or similar de-
vice) rather than a sand pack. In deep boreholes, placing Multilevel sampling involves placing several samplers ;it
sand around the intake is much more difficult than at various depths in a single borehole (Cherry, 1985). The
shallow depths. Standpipes placed in deposits that cave advantage of this method is tlie relatively large number
are often installed without a filter sand or seal (Figure of discrete sampling points that can be installed at ;I
17.18~).They are usually emplaced using hollow stem relatively low cost. The method is most economical for
augers, which hold the borehole open until the casing near-surface investigations in noncohesive sand o r silts.
is run in to the proper depth. A variation on this design where filter sands and seals are not required. For cohesive
involves placing a bentonite seal above the intake by deposits, it is necessary to place seals between int;tkes.
passing material through the hollow-stem auger. While Whether or not filter sand is required depends on the
a good idea, this installation procedure is tedious and type of seal. Extra difficulties related t o correctly pktcing
may not work. the seals and filter sands can affect tlie cost and re1i;ibility
Water samples can be collected from these wells most of these installations. The small size of casing typicsll!,
simply using bailers. To avoid the possibility of cross- used in these samplers makes it difficult to develop the
screen and sandpack and may promote plugging. Further,
Concrete, the design of an appropriate sounder for measuring w;iter
levels in these samplers can challenge one's ingenuity.
However, the small internal volume o f these devices is
;i significant benefit. Because small quantities of water ;ire
required for flushing the system and the actual s;imple.
sampling will probably not perturb the contaminant
Seal plume.
A variety o f mdtilevel devices (from Cherry, 198.3) is
Screen
shown in Figures 17.19~1, 6 . c.. cl. and e. These c:tn be
Sand pack installed in cohesive or noncoliesive units, but the ex;ini-
pies show either one or the other type. The first system
(N) Nest of Piezorneters ( b ) Conventional Standpipe Piezometer (Figure 17.1910consists of a large number of small-dianie-
ter standpipes placed in a single borehole. I h e main
requirement for installing this system is patience i n c;ire-
fully placing the filter sand and seals.
Horeholes that cave usually require ;I hollow stem
Surface
iiuger to install the multilevel device. To faci1it:ite the
install;ition, the standpipes (sometimes pol!rth!lene tub-
ing) are bundled around a more rigid casing (Figure
17.199). Another variation on rliis design jkices the tub-
ing inside a solid casing to protect the tubing ;tg;iinst
damage (Figure 17.1%). When there is some possibility
of units not caving to prevent vertical conimunic;itio~iin
(c) Piezorneter Installed in Caving Materials
the borehole, seals need to be provided between tlie
Ffgure 17.18 Basic designs for standpipe piezometers. intakes. Cherry (1983) discusses n.a!.s of inst;ilIing
Panel (a)illustrates tlie concept of how a nest of piezome- these seals.
ters provides spatially distributed concentration. Panels (6) Another innovative procedure involves s;inipling
and (c) are examples of typical piezometers for noncaving through specially designed valved couplings (Figure
and caving materials (modified from Cherry, 1983). 17.19d) that join sections of PVC o r stainless steel c;ising.
364 Chapter I7 / Zntroductfonto Contaminant Hydrogeology

- Bentonite im, Cross section of bundle

e C e n t e r W C pipe
1 grout seal

- Sand
(4Standpipe Type
(from Cherry, 1983)

P
I

I
Plastic tubing

rubber
stopper
( 6 ) Bundle Piezometers (from Cherry, 1983)

b -Sampler point

(cl Suction Type (from Cherry, 1983)

Ffgure 27.29 Examples of multilevel sampling devices. Portions of panel (e) are reprinted by
permission of Ground Water Monitoring Review. Copyright 0 1981. All rights reserved.

The sampling device passes through a valve or port to chamber for flushing o r sampling. The figure illustrates
sample fluids outside of the casing. Another tool con- the completion details for these devices.
taining a down-hole transducer measures hydraulic head
through these same ports. Packers prevent communica-
Solid and Fluid Sampling
tion in the borehole between the sampling ports. To
reduce the complexity of the installation, a special tool Sometimes it is worthwhile to collect samples of both
inflates the packers through the casing. A stainless steel the porous medium and the fluids by drilling and coring.
version of this sampling system is particularly well suited Fluids squeezed out of the sample can be analyzed in
for deep boreholes in bedrock. the normal manner. Another approach is to determine
The final type of multilevel device is what Cherry contaminant concentrations in the dried solid and report
(1983) refers to as gas-drive samplers. They consist of a contaminant concentrations as mass of contaminant per
sample chamber with narrowdiameter tubes connecting unit mass of solid. This latter type of analysis is used most
the device to the surface (Figure 17.19e). Compressed commonly in sampling for LNAPLs and DNAPLs.
gas forced down one of the tubes will empty the sample The main advantages of this sampling technique are,
 17.4 Sampling Metbods 365

x coup1ing

-Packer

-Measurement
port coupling
A

LJ
r9

8-
The approach of solid and fluid sampling can eliminate
this possible problem. There are limitations. First, it is
often quite difficult to retain typical aquifer materials
using split-spoon sampling or thin-wall tube samplers. A
variety of designs to overcome this limitation is presented
in U.S. EPA (1993). A second limitation is the cost of
bringing a rig back onto the site if more than one round
of sampling is required.

Cone Penetrometry
Cone penetrometers are used in standard geotechnical
engineering practice. In its most basic form, the cone
penetration test involves pushing a cone-shaped instru-
Probe located at
measurement ment into the soil and measuring the tip resistance to
port coupling penetration and friction (Chiang and others, 1989). Fig-
ure 17.20 illustrates how a hydraulic ram is used to force
the cone penetrometer or other tools into the ground.
Typically, the information is used for assessments of soil
strengths and stratigraphic interpretations of site condi-
tions.
Refinements to this technology have provided the ca-
pability to make water-pressure measurements (yielding
hydraulic heads) as a function of depth, and to estimate
(d) Westbay-Type Samplers (from Cherry, 1983)
hydraulic conductivity from measurements of the rate
Figare 1Z19d of pore pressure dissipation (Chiang and others, 1989).
Other tools provide a capability of collecting water sam-
ples with depth.
first, the spatial control in being able to relate specific Shown in Figures 17.21a, 6 are two samplers that are
concentrations of contaminant to specific lithologies, used to replace the standard cone penetrometer. With
and, second, the ability to minimize contamination due the BAT@system, the sampling probe is pushed to the
to the sampling method. In all of the direct sampling depth of interest. A sliding sleeve protects the filter tip
approaches we discussed earlier, the possibility exists (that is, intake) during penetration. At the desired depth,
for a “memory effect.” Someof the contaminants brought the filter is exposed and the water sample enters the
into the sampling device at an earlier time may remain sampler up to the septum at the top filter. The sample
sorbed on the internal surfaces. These compounds can is collected by lowering a sampling probe consisting
desorb during later sampling and result in errors. of an evacuated sampling tube and the double-ended

Gas
entry
Connection tube Polypropylene Ga
collection tu

olypropylene tubing
Threaded access cap

(from Morrison
and Brewer, 1981)

Inert check
valve arrangement

Slotted 50-pm polyethylene filter

eii screen
(e) Gas Drive Sampler

Figure 17.19e
-
366 Cbapter 17 / Introduction to antaminant Hydrogeology

provides a rdpid and cost-effective approach for collect-

ing ground-water samples. It can be used in its own right

I
I
Hydraulic rants,
'57
HI
for site investigation or in a reconnaissance mode to assist
in siting conventional water-sampling wells. The BAT@
sampler can be installed permanently for future sampling.
Cone penetrometry methods are limited to unlithified
materials and are difficult to use in units containing gravel
(U.S. EPA, 1993).
New sampling approaches based on this technology
are under active development. Probes have been devel-
oped for soil sampling, electrical resistivity and ground-
penetrating radar measurements, and the direct detection

Cone pnetrometer -
v Water table -
w

- Evacuated
sampling
tube
Septum- -

- Cone penetrometer - Double-ended

Electrical sensor
niwsuring side friction - test probe

- Electrical Sensor
Septum- -
hypodermic
needle

measuringpenetration
resistance

Figure 17.20 Schematic of a cone penetrometer rig (from

Smolley and Kappmeyer, 1991). Reprinted by permission of
Ground Water Monitoring Review. Copyright 0 1991. All
rights reserved.

hypodermic needle shown in Figure 17.21~.Once the

sampling probe is connected to the filter tip, the needles
penetrate the septa, and the sample is drawn into the
sample vial. The sampling probe may then be withdrawn.
V
The method appears to minimize the loss of volatiles
(Torstensson and Petsonk, 1988; Mines and others,
1993).
The HydroPunchm is a stainless steel and Teflon@s a w open
pling tool that also can be used with a cone penetrometer pwrtion
rig to collect water from unconsolidated sediments (Edge
Figure 17.21 Schematic of sampling tools used in conjunc-
and Cordry, 1989; Smolley and Kappmeyer, 1991). As tion with a cone penetrometer rig. Panel (a) illustrates the
before, the sampling tool is pushed closed to the desired B A P tool (from Mines and others, 1993). Panel (b) illus-
sampling depth. Retraction of the sample chamber expos- trates the HydmPunch" (from Smolley and Kappmeyer.
es the intake tube and water enters the sample chamber 1991). Reprinted by permission of Ground Water Monitoring
through a check valve (Figure 17.216). Once the sample and Remediation. Copyright 0 1993 and 1991. All rights re-
is collected, the tool is retrieved. This type of sampling st!t-Ved.
 1 Z 5 Indirect Metbods for Detecting Contarnitratioti 367

Other Sampling Methods Access tube <pressurevaiuurn;

-,.- _,-.-

A variation on the cone penetrometer approach involves

the use of a portable hammer to drive a small wellpoint
attached to a metal casing to various depths. Water sam-
ples can be collected from this well at desired depths,
and the installation can be left in place permanently.
An extremely versatile vehicle-mounted sampling sys-
tem based on this sampling method, called the G e e
probe@, was developed in 1988 by Kejr Engineering,
Inc. This system uses a vehicle-mounted, hydraulically
powered percussion probing hammer to advance the
tool string to depths of 30 m or more. Specialty tools
provide a capability for soil sampling and the collection
of water samples through a stainless steel screen. Water
is collected using a bailer or with plastic tubing and a
peristaltic pump.
The screened auger approach involves the use of a
slotted hollow-stem auger to collect samples periodically
while drilling is in progress (Taylor and Serafini, 1988). Figure 17.22 A typical device for collecting soil-w;iter s;iiii-
ples from the unsaturated zone (from Soil-moisture Equip-
Having reached a desired depth, the screened portion of
ment Cop). Reprinted with permission.
the auger is developed by air-lift pumping. The actual
sample is collected by using a submersible pump inside
the hollow-stem augers. Case studies (Taylor and Serafini, pling. These approaches are successful when contamina-
1988) indicate that this sampling approach is useful for tion is manifested in a secondary way, for example, by
reconnaissance-type investigations of ground-water the presence of volatile organic compounds in the soil
problems. gas found above a spill or the change in electrical proper-
ties of a unit caused by the presence of the plume. A
variety of published case histories points to the usefulness
Dissolved Contaminants
of these approaches in site evaluation. However, like
in tbe Unsaturated Zone
many of the indirect approaches, they do not always
Nearly all sampling techniques in the unsaturated zone provide an unambiguous interpretation.
involve pulling the fluid under vacuum through a porous
ceramic cup into a container (Figure 17.22). Once a suf-
Soil-Gas Characterixation
ficient volume of fluid has been collected, the sample is
removed by suction or by gas displacement depending Characterizing the composition of soil gases has emerged
on how deep the sampler is installed. Signor (1985) de- as an industry-standard technique for tracing volatile or-
scribes varieties of these samplers. ganic compounds in ground water. The appro;ich in-
These devices function in both unsaturated and satu- volves defining zones of ground-water contaniin;itioii
rated zones. This characteristic makes them useful for based on the presence of volatile components in the soil
sampling when the state of saturation changes frequently. gas. With time, volatiles at the capillan fringe p i r t i t i o n
An example might be a facility for artificial recharge that into the soil gas and gradually diffuse upward t o the
is operated with only periodic flooding of the infiltration ground surface. The presence of volatiles is est;iblishcrl
basins. The main problem with this method is the possi- commonly by collecting soil gas from some fixed dcpth
bility for the fluid composition to change from reaction and analyzing the sample with a gas c1iromatogr;ipIi.
with the ceramic cup, or from the loss of gases as the However, this approach does not detect all orgtnic con-
sample is collected under vacuum. These effects can be taminants because not all are volatile. In ;idditioii. the
examined in the laboratory and hopefully accounted for organic compound should not be too soluble. Highly
in the interpretation of results. soluble volatiles moving through the iins;itur;ited zone
will dissolve into any water present. Reisiiiger and others
(1987) illustrate the typical components of gasoline t h t
-17.5 Indirect Methods might be successfully detected in a soil-gas sun.c!. (Fig-
ure 17.23).
f o r Detecthg Contamination A metal probe with a perforated tip is driven from 1
The indirect methods for plume delineation can provide to 4 ni into the unsaturated zone, and a smell sample of'
a rapid and inexpensive alternative to conventional sam- soil gas is extracted by pumping. A variety of coniiiic'r-
368 Chapter 17 / Introduction t o Contaminant Hydrogeology

cially available equipment is available to assist with plat- a conventional sampling program. In some studies, the
ing of sampling probes. Gas samples are usually analyzed information provided by a soil-gas survey may be all that
on site with a gas chromatograph (GC). The GC measure- is required to establish whether or not a contamination
ment provides a quantitative estimate of the mass of a problem exists, or to identify a source.
particular volatile compound per volume of soil gas (for There are some circumstances when these approaches
example, pg/L). When plotted on a map, these data can may not work well. A low-permeability layer in the unsat-
be used to infer zones of contamination in the unsatu- urated zone can inhibit the upward d i f h i o n of vapor
rated and saturated zones. By relating contaminant con- and promote extensive horizontal spreading. This situa-
centrations in the soil gas to measured concentrations in tion will produce an estimated distribution of contami-
the ground water, the soil-gas data can be transformed nants that is much larger than the actual one. Other
to provide a quantitative estimate of concentrations in problems relate to how the contaminant occurs. With
ground water. LNAPLs, vapor concentrations in the unsaturated zone
A variation on this approach involves sampling soil will be larger and easier to detect than if the contaminant
gases by using a grid of static collectors (Malley and is dissolved in the ground water (Reisinger et al., 1987).
others, 1985). The sampling device, which contains acti- Furthermore when contaminants are present in localized
vated charcoal, is buried and left for several weeks, giving fracture zones (for example, in sandstone or limestone),
time for the collector to equilibrate with the soil atmo- the rates of diffusion away from the fractures may be
sphere. A grid of samplers provides an estimate of the so slow that vapor-phase transport is limited in extent
spatial variability. Once the collectors are returned to the (Reisinger et a]., 1987). Thus, care must be exercised in
laboratory, the volatiles are desorbed and analyzed using interpreting the results, and where necessary, conclu-
mass spectrometry. This analytical procedure provides sions should be confirmed using an independent ap-
for the detection and identification of compounds up to proach.
mass 240 (Malley and others, 1985). Maps of mass fluxes Whittmann (1985) describes a case study that demon-
(or relative ion counts) provide an estimate of the strates the potential of soil-gas sampling for delineating
plume position. contaminated ground water. The case considered the
Soil-gassampling provides a rapid and economical way origin of contamination in the Verona Well Field at Battle
of surveying large sites for contamination. Thus, it is Creek, Michigan. Conventional ground-water monitoring
attractive for reconnaissance studies aimed at discovering pointed to the presence of an additional but unidentified
what volatile contaminants are present and where they source. Soil-gas surveying was used to explore a large
are located. This information often assists in designing railyard. The sampling approach involved driving a metal
probe approximately I m into the soil and extracting
the soil gas with a hand pump. Sample analyses were
performed on-site using a gas chromatograph. Figure
17.24 shows PCE results for 43 samples. The survey de-
tected three areas of elevated soil gas in the railyard.
The most concentrated of the three was a small solvent-
disposal area that had gone undetected in previous inves-
tigations. Subsequently, two wells confirmed the PCE
source. Overall, the approach worked well, except that
it did not reveal the full extent of the PCE plume. Migrat-
ing away from the source, the plume mr --eddeeper and
lost touch with the water table. Under such conditions,
volatiles were unable to partition into the soil
gas.

Geophysical Metbods
Exploration geophysics is the science (or art) of inferring
the distribution of physical properties beneath the sur-
Figure ZZ23 The unshaded area shows the range in vapor
pressure and aqueous solubility for the common constituents face, given the measurements of associated fields over a
in gasoline for which soil-gas surveying will be successful surface above-or along-a borehole within the me-
(from Reisinger and others, 1987). Reprinted by permission dium. Such methods can ascertain the presence of some
of First Outdoor Action Conference on Aquifer Restoration, well-defined target (Table 17.7). Targets can range from
Ground Water Monitoring and Geophysical Methods. Copy- characteristics of the geologic environment to properties
right @ 1987. A11 rights reserved. of the fluid(s) within pores.
 I Z 5 Indirect Methods for Detecting Contamfnation 363

Electrical Methods
Electrical methods are commonly applied to studies of
ground-water contamination. The various electrical meth-
ods usually measure either conductivity or resistivity (the
reciprocal of conductivity). Traditionally, resistivity is
measured from the surface by adding electrical current
to the ground with a pair of electrodes and measuring the
resulting drop in voltage with a second set of electrodes.
Apparent resistivity values are either read directly from
the instrument or calculated. In lateral profiling, the sur-
vey is run by maintaining a fixed electrode spacing (pro-
ducing a constant depth of current penetration) and mov-
ing from station to station following a grid. Anomalies
on the apparent resistivity map may coincide with con-
taminant distributions.
Early applications (for example, Cartwright and McCo-
mas, 1968) used electrical resistivity surveys to detect
changes caused by the presence of ground-water plumes
iPT7 Monitoring well 0 200 400 with elevated salinity. In recent years, these types of
-N
Scale surveys have fallen out of favor because terrain conductiv-
+v25 Municipal well
ity methods (that is, electromagnetic methods) are much
-10- PCE concentration contour more rapid and offer possibilities for higher resolution.
(Dashed where inferred)
Electromagnetic methods induce a current in the
9 11 Soil gas sample and PCE
-1 concentration (pg/L) ground with an alternating-current transmitting coil. The
magnetic field around the coil induces an electrical field
Ffgure 17.24 Results of a soil gas survey (July 1984) for in the earth to depths that are largely controlled by the
PCE in a railyard in Battle Creek, Michigan (modified from
background properties of the medium, the moisture con-
Whittmann, 1985). Reprinted by permission of Conference
tent, and the relative difference in the conducting proper-
on Petroleum Hydrocarbons and Organic Chemical in
Ground Water-Prevention, Detection, and Restoration. ties of the medium and the target. In the Geonics EM31,
Copyright 6 1985. All rights reserved. the two coils are mounted 3.7 m apart on rigid boom.
The equipment is highly portable, and an operator can
collect a large number of closely spaced electrical con-
Geophysical methods have been used for many years ductivity measurements in a short time. The depth of
as a standard tool in the petroleum and environmental signal penetration is about 0.75 to 1.5 times the coil
industries. These approaches use a variety of instruments separations (Zalasiewicz and others, 1985). Other instni-
to measure changes in the physical and chemical prop- ments provide larger coil separations and greater signal
erty of materials (geologic or otherwise) and pore fluids. Penetration.
In the next few pages, we will survey the geophysical Terrain conductivity methods have been frequently
approaches that are commonly used in contamination applied to subsurface mapping and particularly to rhe
studies, but only mention those that are used for geologic investigation of contaminated sites. Stewart and Gay
investigations. We will leave it to other texts to discuss (1986) have evaluated electromagnetic soundings for
borehole logging techniques. deep detection of conducting fluids, and Greenhouse and
Shine (1983) have applied the technique to mapping
contaminant migration.
Table I % 7 Relationship Between Geophysical The most important application of electromagnetic
Properties and Fields methods is to detect buried objects or other features at
Physical Properties Fields waste-disposal sites. Examples of buried objects are filled
waste trenches or lagoons, buried steel drums, and lost
Density Gravity underground piping or storage tanks. A case study of
Elasticity: Bulk modulus Seismic wave the application of very-high-resolution electromagnetic
Shear moddus surveying is provided by Jordan and others (1991). The
Magnetic permeability Magnetic method was used to examine a vacant lot that was pre-
Electrical conductivity Electromagnetic
viously used for industrial hazardous-waste processing.
(low and high frequency)
Dielectric permittivity During its operation, several unlined lagoons may have
existed. The survey was designed to identih the filled
370 Cbapter 17 / Introduction to Contaminant Hjidrogeology

lagoons and possibly buried metals. Glacial till covered Ground-Penetrating Radar
the site. Ground-penetrating radar (GPR) is a rapidly evolving
The conductivity survey collected 4823 EM data poitits technology for subsurface investigations (Daniels, 1989).
along lines 3.8 m apart in areas suspected of contamina- The method is used to map the distribution of buried
tion. Stations were about 0.6 to 0.9 m apart. The resulting objects (drums, pipelines, etc.), to define the configura-
conductivity map (Figure 17.25) indicates a broad rdnge tion of the water table and stratigraphic boundaries,
of conductivity. Suspected lagoon locations are indicated and to establish the distribution of LNAPLs. Thus, GPR
by conductivity values greater than 50 mS/m (that is, is well suited for surveying abandoned waste-disposal
milliSiemens per meter), while near-zero conductivities sites.
are thought to represent areas of buried metallic debris. A transmitting antenna at the surface radiates short
The locations of the lagoons correspond with locations pulses of radio waves into the ground. Reflected energy
determined from old aerial photographs. from the subsurface is recovered by an antenna that is

Figure IZ25 Results of a very high resolution electromdgnetic survey. Data were collected along lines approximately 3.8 m
apart (from Jordan and others, 1991). Reprinted by permission of Fifth National Outdoor Action Conference on Aquifer Restora-
tion, Ground Water Monitoring and Geophysical Methods. Copyright 0 1991. All rights reserved. Electronic data file kindly pro-
vided by Geosoft Inc., Toronto.
 Problems 371

moved across the surface. In principle, then, GPR works tion. However, magnetometry is widely used to locate
like reflection seismic methods except that electromag- containers and pipelines that may have contained con-
netic reflections, rather than acoustic returns, are re- taminants. The response of the magnetometer is directly
corded at the surface. Radar energy is reflected due to proportional to the mass of the target and inversely pro-
changes in dielectric constants and electrical conductiv- portional to the cube of the distance to the target. FerrcF
ity. Such changes are usually related to variation in prop- magnetic substances can be detected at depths ap-
erties of the minerals present, the degree of saturation, proaching 20 m. Examples in the application o f
and material density. By moving the GPR along at a slow magnetometer surveys at various waste-disposal sites in-
speed, it is possible to obtain an almost continuous profile clude Gilkeson and others (1986) and Cochnn and Dal-
of the subsurface. Under ideal conditions (that is, dry ton (1995).
sandy units or bedrock), GPR provides a highly resolved
picture of subsurface-site conditions. It is less useful in Seismic Metbods
clayey, moist lake clay or glacial till. Seismic techniques use both reflection and refraction t o
The state-of-practice in radar technoiogy is moving to- measure the travel time of seismic waves that are gener-
ward highly resolved three-dimensional studies involving ated from some surface source (sledgehammer, explo-
a large number of parallel survey lines on the ground sives). Reflection seismic approaches tend to be used
surface. With the help of approprhte visualization soft- most frequently in environmental applications. Refine-
ware, one can interpret and display the three-dimensional ments in seismic methods have permitted highly retined
relationships among various radar reflectors on a site investigations at relatively shallow depths (Hunter and
(Thompson and others, 1995). others, 1984, 1988; Kaida and others, 1995). Reflection
seismic methods are often most useful in reconnaissance-
Magnelometry type studies for determining the top of the bedrock sur-
Magnetic surveys measure variaitions in the Earth's total face, the topology of structural features, and the pattern
magnetic field. NAPLs, unlike ferromagnetic materials, o f stratigraphic layering. An example of the application
d o not affect the magnetic properties of the geologic of seismic methods to determine the continuity o f a clay
medium. Consequently, direct magnetometry is of no use layer underlying a proposed hazardous-waste Facility is
in detecting the various types of ground-water contamina- provided by Shine and others (1990).

Problems
1. Shown on the attached figure is a series of plumes
from ;I sanitary landfill. Examine these plumes in detail,
and answer the following questions.
a. Qualitatively evaluate the extent to which advection
and dispersion are important in controlling contami-
nant spread at the site.
b. Given the type of source and the resulting plume
shapes, what can you say about the type of source
loading?
c. Suggest what processes could be operating to cause
the increasing pH away from the source.
d. Metals, for example, Fe, tend t o be relatively abun-
dant in landfill leachates. However, Fez' tends to be
strongly attenuated relative to mobile species like
C1-. Explain why iron behaves in this way.
2. Assume that some very preliminary studies identified
that the contamination problem in question 1 had devel-
oped in a surficial sand aquifer about 15 m thick. Develop
a plan for a detailed site investigation that involves both
direct and indirect methods of detecting contaminition. 2&Q ;a0
The plan should provide specific details about the types Problem 1.
of instninientation/surys required and plans for water-
quality analyses.
 CbaDter 18

Modeling the Transport of

Dissolved Contaminants

18.1 Analytical Approaches

18.2 Programming the Analytical Solutions for Computers
18.3 Numerical Approaches
18.4 Case Study in the Application of a Numerical Model

~ ~ ~ ~ - ~

Chapter 17 explained the nature of contaminant plumes

as a manifestation of physical and chemical processes.
The approach was a conceptual one, simply illustrating
what effect processes could have on contaminant trans-
port. This chapter develops approaches for modeling the
spread of contaminants dissolved in water. The starting where C is the concentration; COis the source concentra-
point in modeling is with the various mass transport tion; a,, cry, and a,are the dispersivities;x, y,zare spatial
equations that were developed in Chapter 14. The end coordinates; and t is time. If the three onedimensional
point will be a set of powerful mathematical tools for solutions can be found, an approximate solution to a
analyzing problems of contaminant transport. three-dimensional problem in some cases will be

18.1 Analytical Approaches

Analytical models have proven to be powerful tools for In words, the three-dimensional solution is the product
studying transport processes. Because analytical solu- of the one-dimensional solutions. Thus, understanding
tions for contaminant transport abound in the literature, the three-dimensional result follows from knowledge
the discussion here will focus on the more practical ones about the onedimensional results. This is exactly the
that can be used in the field. Normally, this would pre- approach we will follow in this section.
clude all one-dimensional solutions. However, in some
cases, the pertinent features of a onedimensional solu- Advection and Longitudinal Dispersion
tion may be embedded in a more complex threedimen-
sional equation. Take for example three onedimensiotial Advective transport is frequently demonstrated with a
solutions of the form plug-flow model that neglects all longitudinal and lateral
mixing. The contaminant moves with the velocity de-
fined by Darcy's equation corrected for flow through the
pores. In Figure 18.la, water containing a tracer with
372
 18.1 AnulytfculApproaches 3 3

concentration C, enters the flow tube and displaces the

original fluid. The advective front is located at the posi-
tion x = vt. The concentrations are at steady state and
everywhere equal to the source concentration C,, . Such -0
Time Distance 1
behavior is unrealistic in the field or the laboratory, but .Y = 1'1
the concept is useful in describing other processes.
Figure 18.16shows advection with longitudinal disper-
sion. In the absence of lateral or transverse dispersion, Figure 28.2 (a)Breakthrough curve at the end of I col-
the displacing fluid mixes with and displaces the original umn describing the variation in relative concentration IS ;I
function of time and (6) the spatial variation in relative con-
fluid in the x direction strictly within the flow tube. The
centration along the column at one time.
concentration at the advective front x = vt is now less
than the original concentration C,,. Longitudinal disper-
which is a solution to the one-dimensional dispersion
sion causes some of the dissolved mass to move ahead
advection equation introduced in Chapter 14
of the advective front at the expense of material behind
the advective front. However, at some point x << zit,
the concentration is steady and is the same 3s the source (18.1)
C,. This steady-state part of the plume possesses all the
properties of the plug-flow model of Figure 18.la. for the conditions
The situation described above can be reproduced in C(0, t) = C, and C(x, 0) = 0
a laboratory column (Figure 10.5). The resulting break-
through curve (Figure 18.2a) can be thought of as a The first boundary condition states that at x = 0, for all
window fixed at some point x along the column. For time t, the concentration is Ct, (that is, a continuous
some time after the experiment is begun, the concentfa- source). The second states that at all points x at time
tion ratio at this point (that is, C/C,,>will be zero, as only t = 0, the concentration is zero. Just at the start of the test.
water initially present in the column passes by. Prior to the concentration of tracer in the water in the column is
the breakthrough of the advective front, the concentra- zero. An exact solution (Eq. 18.1) for these boundiin
tion ratio C'C,, becomes finite and eqiials 0.5 at the instant conditions is
the advective front breaks through. Eventually, with in-
creasing time, C/C,, = 1, indicating that the original fluid
has been totally displaced and only the tracer of concen-
tration C,, is passing through the column.
An alternative way to demonstrate the column experi-
ment is shown in Figure 18.26. This Concentration distri-
bution is the spatial distribution of the ratio C/C, through- where D is the coefficient of hydrodynamic dispersion,
out the column at one point in time. Here, we see that erfc is the complementary error function, and I' is the
C/C,, = 1, 0.5, and 0 at some x << zyt, x = ut, and s >>
linear ground-water velocity, which in this c;ise is itlenti-
ot, respectively. cal to the velocity of the tracer.
The concentration distribution in a column is com-
The second term in Eq. 18.2 is usually small i n many
pletely described by the Ogata-banks equation (1961),
practical situations so it will be ignored in future discus-
sions. Later, we examine conditions when this simplifica-
tion causes errors. Overall, the simplicity of this approxli
versus absolute accuracy is a reasonable trade-off in ni;iny
applications. Second, the quantity (Dt)f'2can be written
as (Ly,t~t)"iprovided we ignore the diffusive contribution
to hydrodynamic dispersion. In this abbreviated form.
the solution becomes

Further discussions will focus only on the abbrevi;ited

form of Eq. 18.2.
The complementary error function occurs frequently
Figure 28.1 Schematic diagram showing the eSSentid f a - in solutions to the advection-dispersion equation. Like
tures of (a) plug flow,and (6) plug flow accompanied by the various well functions in hydraulic testing, it is ii well-
longitudinal dispersion. tabulated function so that graphs or tables can provide
374 Chapter 18 / Modeling tbe Transport of Dissolved Contaminants

numerical values. It is convenient to express this comple- 18.1 and are given graphically in Figure 18.3.Note on
mentary error function as the figure that erf @) ranges from - 1 to + 1 whereas
erfc Cp) ranges from 0 to +2.Further, erfc Cp) takes on
x - vt
erfc Cp) where numbers greater than one only for negative values of the
= 2(axvt)'"
argument. This feature is not evident o n Table 18.1.For
where p is the argument of the complementary error negative values of the argument a,
erfc ( - p ) may be
function. Thus, as with all well-tabulated functions, if we determined as
know 0, we know erfc @); and if we know erfc @), we erfc (-p) = 1 + erf Cp)
likewise know p.
Values for erfc @) and erf Cp) (or error function of 0) Other useful relationships include
for various values of the argument p are shown on Table erf (-p) = -erf (p, and erfc @) = 1 - erf @)
Now let us examine the physical meaning of the argu-
ment of erfc starting first with the numerator x - vt,
Table 18il Values of erf (p) and where both x and vt are lengths. This expression simply
erfc (P) for Positive Values of identifies the point of observation x with respect to the
position of the advective front ut. The point of observd-
tion can be in one of three possible places: at the advec-
0 0 1.o tive front, in front of the advective front, or behind the
0.05 0.056372 0.943628 advective front. At the advective front where x = vt, the
0.1 0.1 12463 0.887537
argument is zero and erfc (0) = 1 so that from Q. 18.3,
0.15 0.167996 0.832004
0.222703 0.777297
C = 0.5C0.This result is observed on the breakthrough
0.2
0.25 0.276326 0.723674 curves of Figure 18.2.Beyond the advective front where
0.3 0.328627 0.671373 x >> zit, the argument is positive and as it approaches
0.35 0.379382 0.6206 18 infinity, erfc (00) = 0 so that from Eq. 18.3,C = 0.In
0.4 0.428392 0.571608 other words, the observation point x is far in front of
0.45 0.47 5482 0.5245 18 the zone of dispersion so that mixing has not yet occurred
0.5 0.520500 0.479500 and we are observing the concentration of the original
0.55 0.563323 0.436677 fluid. However, in a practical sense, a positive argument
0.6 0.603856 0.396144 equal to 2 or 3 suggests that erfc @) is on the order of
0.65 0.642029 0.357971 to loT5(Table 18.1).Hence, for all practical purposes,
0.7 0.677801 0.322199
C = 0 at positive values of /3 greater than 2.
0.75 0.71 1156 0.288844
0.8 0.742 101 0.257899
If the point of observation is behind the advective
0.85 0.770668 0.229332 front, the argument p is negative, and as it approaches
0.9 0.796908 0.203092 negative infinity, erfc (- 00) equals 2 (Figure 18.3).This
0.95 0.82089 1 0.1791W means that C = C,,, Eq. 18.3.However, a negative argu-
1.o 0.842701 0.157299 ment on the order of (-2) suggests that erfc (-2) is
1.1 0.880205 0.1 19795 about 1.999.Thus, for all practical purposes, C = C,,
1.2 0.910314 0.089686 at small negative values of 0. The denominator of the
1.3 0.934008 0.065992 argument has the units of length and may be regarded
1.4 0.952285 0.0477 15 as the longitudinal spreading length or a measure of the
1.5 0.966105 0.033895 spread of the mass around the advective front. The larger
1.6 0.976348 0.023652
1.7 0.983790 0.016210
1.8 0.989091 0.0 10909 2.0
1.9 0.992790 0.007210
1.5
2.0 0.995322 0.004678
4
2.1 0.997021 0.002979 0 1.0
2.2 0.998137 0.001863 5
0.998857 0.001 143 0.5
2.3
2.4 0.9993 1 1 0.000689 fl 0.0
2.5 0.999593 0.000407
2.6 0.999764 0.000236 -0.5
2.7 0.999866 0.000134
-1.0
2.8 0.999925 0.000075 -3 -2 -1 0 1 2 3
2.9 0.999959 0.000041 P
3.0 0.999978 0.m22
Figure 18.3 erf cp> and erfc @> plotted versus @.
 1 8 1 Analytical Approacbes 375

the denominator, the greater the spread about C/C,, = ?'be Retardation Equation
0.5.
Thus, the simple expression Eq. 18.3 tells us something For the case of mass transport accompanied by linear
about longitudinal dispersion. First, from a modeling per- sorption, the governing equation is, as we discussed
spective, longitudinal dispersion merely moves mass earlier,
ahead of the advective front in a given flow tube. It
contributes nothing to the lateral movement of material (18.4)
out of the tube. Second, as the distance of the advective
front from the source increases with increasing time, the where
spread becomes larger around the advective front. Lastly,
at some positionx << ut, the concentration is maintained
at the original source concentration C,, . The Ogata -Banks
R/= 1 + (--)1 - n PSK,, (18.5)

solution is not a field equation but a laboratory one be- The effect of sorption is to decrease the value of the
cause no matter how far the point of observation is from transport parameters D and u. The solution to this eqiia-
a continuous source, the equation predicts that the maxi- tion for the same boundary conditions used with the
mum concentration will eventually be equal to the source Ogata-Banks equation (1961) is
concentration. This condition is unrealistic, but such is
the nature of models with only advection and longitudinal
dispersion as the operative processes.
c
C,,
=(f)erfc [ ( R x - tJa.t)
2(~~,.2~,,tR/)"~
] (18.6)

The example that follows illustrates one of the ways

If Rf = 1, the Ogata-Banks result is recovered ex;ictly.
that analytical solutions can be used in this case to esti- Note that vfcis the velocity of the water, which in this
mate a longitudinal dispersivity from breakthrough data.
case is not the same as the velocity of the species undergo-
ing sorption.
Example 18.1 The relationship between an unretarded species, as
described by the Ogata-Banks equation (1961), and ;I
A nonsorbing species is Sent through a column 30 cm retarded one, as described by Eq. 18.6, is shown in Figure
in length at a velocity of 1 X lo-' cm/s. C/C,, ratios of 18.4. Essentially, the velocity of the Contaminant be-
0.42 and 0.573 are noted at 46.6 and 53.3 min, respec- comes less than the velocity of the ground water. Mathe-
tively, after the test started. What is the longitudinal dis- matically, the following equation relates these two veloc-
persivity? ities:
Using the first breakthrough concentration

Equation 18.7 is known as the retardation equation. It

predicts the position of the front of a plume due to
advective transport with sorption described by a simple
0.84 = erfc @) linear isotherm. The ratio v,,/zJ, describes how many times
faster the ground water(or a nonsorbing tracer) is moving
Now solve for a,, relative to the contaminant being sorbed. Note that when
K,, is zero (no sorption), the velocities are the same. In
0 = 0.14 normal application of Eq. 18.7, z: and if,,can be measured
in any convenient units. However, K,, is normally rc-
ported in m u g and ps in g/cm' so that their product is
m Wcm'. No further conversions are required because of
. 1.8 cm
a.,= 1.o

Unretarded smcies
The second calculation is essentially the same

[30-321
I
I I
0.573 = -erfc c _ _ _

.v2 = I ; ,
2 2(aX32)"' s distance

1.146 = erfc(-P) = 1 + erf@)

0 = 0.13 and a.,= 1.8cni Ffgure 1 8 4 A comparison of concentration distributioiis
for retarded and unretarded tracers.
376 Chapter 18 / Modeling the Transport of Dtssolved Contaminants

the units chosen where there are 1000 cm5 per liter (or The simplicity of this calculation makes it useful for
per lo3 mL). In applications, pr may be taken as 2.65 back-of-the-envelope type analyses of contamination
g/cm3 and (1 - n ) / n is the inverse of the void ratio so that problems. The method obviously breaks down when the
(1 - n)e,/n = pJe, where e is the void ratio introduced in physical setting is complex, or processes other than ad-
Chapter 2. vection or sorption/exchange operate to control contam-
The retardation equation takes on a slightly different inant distributions.
form if instead of defining a Kd value one uses
K,CEC
-- Radioactive Decay, Biodegradation,
- Kd (18.8)
7- and Hydrolysis
where K, is the selectivity coefficient, CEC is the cation Let us now consider mass transport involving a first-
exchange capacity (meq/mass), and r is the total compet- order kinetic reaction. Mathematically, this problem is
ing cation concentration in solution (meq/mass). Thus described by Eq. 14.16. For the same initial and boundary
for a problem involving binary exchange, the retardation conditions of the Ogata-Banks equation (1961), the solu-
equation is tion in one dimension is (Bear, 1979)

The retardation equation has proven useful in evaluat-

C
O =(i)exp{(G)[1-
C
~
(1 +)'/'I} (I8.10)
ing problems of contamination from organic compounds
and to a lesser extent radionuclides and trace metals. edc [
x - vt(1 + ~Ac~JcJ)"'
2(ffXVt)"* I
With rudimentary information about a site, one can begin
to explain differences in the observed distributions of where A is the decay constant, or 0.693 divided by the
various contaminants and predict how contaminants half-life. Although this equation is not sufficiently com-
might spread in the future. To illustrate how the equation plete for describing a field problem, it provides insight
could be applied in practice, consider the following ex- that does apply to field situations. First, if A is zero, Eq.
ample based on an analysis presented by Grisak and Jack- 18.10 reduces to the Ogata-Banks solution (1961). For
son (1978) for the Whiteshell Nuclear Research Establish- this condition, the exponential term in the solution is
ment (WNRE) in Canada. one because e" = 1. Further, for a finite A, the exponent
will be raised to a negative number so that we immedi-
Example 182 ately establish the appropriate limits for the exponen-
tial term
At WNRE, a sequence of glacial drift overlies Precambrian
bedrock. The unit of interest in this problem is the basal e" = 1 e-- = 0
sand, which overlies the bedrock and is confined on top
by clay-loam till. The question to be answered is how The important dimensionless group in the exponential
fast a hypothetical contaminant such as "'Sr might migrate term is 4haJv. As A gets large with respect to the other
in the basdl sand should the primary containment fail. variables in this dimensionless group, the exponential
The seepage velocity through the sand is estimated to term approaches zero, and the concentration approaches
be 1.85 cm/day. Values of other necessary parameters zero. The material is reacting or decaying faster than it
are: p,(l - n ) is 1.9 g/cm', CEC is 1.4 meq/100 g, T is can be transported through the system. As the velocity
4.8 meq/L, n is 0.35, and K, is 1.3. becomes large with respect to the other variables, the
The velocity of "'Sr can be determined by substituting dimensionless group approaches zero, and we approach
the known parameters in consistent units in the retarda- the Ogata-Banks solution (1961). Transporting the mass
tion equation faster than it can degrade or decay makes the decay
ineffective. Competition thus exists between the trans-
port and the kinetics, in particular the decay of a species
and its velocity. As the longitudinal dispersivity axgets
large, the dimensionless group of Eq. 18.10 gets large
1.85 cm/day and decay again dominates. However, the dispersivity
vs,-w = term occurs in two places in the exponential term of Eq.
1+ 1*9g/cm3 0.014meq/g. 1.3 18.10, and these apparently offset each other. In other
0.35 a0.0048 meq/mL
words, any reasonable value of cu, can be used in the
= 0.086 cm/day exponential term and the value of this term will generally
Thus, the velocity of '%r is about 5% of that of the change by less than 1% for a few-orders-in-magnitude
ground water. range in longitudinal dispersivity. For example, for a nu-
clide with a half-life of 1000 yr, a velocity of 4 m/yr, a
 18.1 Analytical Approaches 377

Source retarded velocities favor kinetics over transport. Slow-

moving nuclides or organics have a greater opportunity
to decay or degrade.

Transverse Dtspersion
Figure 18.5 illustrates an advective model with transverse
spreading (lateral dispersion). In the absence of longitudi-
nal dispersion, no mass travels beyond the advective
front. However, the mass is not restricted to a single tlow
Figare la5 Advection accompanied by transverse tube but can spread in a direction transverse to the flow
dispersion.
both laterally and vertically. Concentrations are even-
where less than the source concentration C,,, except at
distance of 5000 m, and cx, = to 30 m, 100 m, and 300
x = 0. Concentrations less than C,, occur because the
m, the exponential term equals 0.5, 0.505, and 0.51, re-
mass is equally distributed throughout the plume, but
spectively.
the volume it occupies increases with increasing distance
Equation 18.10 reflects a steady state when the argu-
from the source.
ment of erfc approaches negative two, that is, the point
In all developments thus far, we specified a source
of observation is behind the modified advective front
concentration but no information concerning the geome-
vt( 1 + 4hcu,/v)i''. This steady-state solution is
try of the source. With transverse dispersion, something
must be stated about the source geometry. Possible geo-
metrical configurations include a point, a vertical line,
or a planar area (Figure 18.6). The equations describing
Although there appears to be a contradiction of terms the parts of a plume controlled exclusively by transverse
when a steady state can be achieved for a decaying spe- spreading for the point or line source are complicated.
cies, the condition can be realized if the source is continu- Thus, the equations that follow are limited to describing
ally renewed. As with the Ogata-Banks equation, the the maximum concentrations along the plane of symme-
steadystate concentration is independent of the longitu- try, that is, along the x axis for 4' and z = 0 or
dinal dispersivity and in this case depends only on advec-
tion and decay. For all practical purposes, Eq. 18.11 de-
scribes the concentration of a species undergoing decay
GQ
cnm = 2v(lra,. (line source) (18.12)
or degradation in a purely advective system. .%-)'I?
There is one more thing to be learned from Eq. 18.11.
The term v is velocity, but what velocity? The answer (point source)
is the velocity of the contarninant. For an unretarded
contaminant, the velocity of the contaminant is simply where Q is a volumetric flow rate with units of L'T-' for
equal to the velocity of the water. For a retarded contami- the line source and L'T-' for the point source so that
nant we can use the retarded velocity directly, or we may CIQ has the units of ML-'T-' and MT-', respectively. The
use v,,./RffromEq. 18.7. Thus, we note the importance of line source result was developed by Wilson and Miller
retardation in problems of decay or degradation in that (1978) and the point source result by Hunt (1978).

Ffgure la6 Geometrical considerations for a contaminant source. Shown in (a)is a vertical
line source, in (6)a point soiirce, and in (c) a finite plane source (from Huyakorn and others.
1987). Reprinted by permission of Ground Water. Copyright 0 1987. All rights reserved.
378 Cbapter 1 8 / Modeling tbe Transport of Dlssolved Contaminants

Equations 18.12 clearly demonstrate the role of the ity, and one kinetic term, as follows:
transverse dispersivity in transport problems. For the line
source, the larger a,,the greater the lateral extent of the
plume and the smaller the concentrations. The point
source result incorporates lateral ayas well as vertical a, where r is defined by some mathematical rate law.
spreading. The source flow rate C,Q reflects the geometry Solutions to Eq. 18.17provide the concentration distri-
of the source, expressed as mass per length per time for bution resulting from continuous or instantaneous
the line source and mass per time for the point source. sources. The following sections will examine these differ-
Further, the concentrations are independent of time, and ent source conditions in detail.
they approach infinity as x approaches zero. Thus, these
equations are useful only for far field conditions where Continuous Sources
x is large. Models that include transverse spreading must incorpo-
The planar source is obviously the more practical one rate information on the source geometry. Wilson and
with lateral spreading in two directions and vertical Miller (1978) provide a complete solution for a nonre-
spreading downward (Figure 18.6). Solutions for this acting species emanating from a line source, and Hunt
source geometry have been developed by Domenico and (1978) presents a threedimensional solution for a nonre-
Palciauskas (1982). For the lateral (y) component of acting species associated with a point source. However,
spreading because the finite planar source is the most realistic ge-
ometry for field problems, it has received the most atten-
tion. Two types of models are available. First there are
finite source models that require numerical integration
where the half source size Y/2 is part of the solution. (for example, Cleary, 1978; Huyakorn and others, 1987).
The z component of spreading is A second type of solution is given in a closed-formformat,
where no numerical integrations are necessary. Two
models are available, one for advection and dispersion
“=(;){erf[-] (z + Z) -erf[-I} (Z - Z) (18.14)
C
O 2(a*x)‘n 2(ff2x)‘R alone (Domenico and Robbins, 1985b), and another for
mass transport together with a first-order reaction (Do-
where the full source size Z becomes part of the solution. menico, 1987).
For spreading directions in both y and z, we adhere to The essence of the Domenico and Robbins model is
the product rule discussed earlier (Domenico and Palci- apparent in Figure 18.7. Because of an assumed one-
auskas, 1982) dimensional velocity, the plume shown in Figure 18.74
can be referred to as a plug-flow model with both longitu-
dinal and transverse spreading (dispersion). The distribu-
tion of concentration (Figure 18.76) has features com-
mon to both the longitudinal (Figure 18.76) and the
transverse spreading models (Figure 18.7~).There is a
frontal zone of dispersion beyond the advective front
For the plane of symmetry (y = z = 0 ) caused exclusively by longitudinal dispersion and a zone
of mass depletion behind the advective front. At some
(18.16) distance behind the advective front, depending largely
on the value of the longitudinal dispersivity, the plume
The source dimensions and transverse dispersivities c o n is at steady state. Because of transverse spreading, the
trol the maximum concentrations that are encountered steady-state concentrations are everywhere less than the
in a steady-state plume. For example, as Y or Z is in- source concentration C,, except at x = 0 . Results of
creased, or a,.and a2decreased, the arguments of the the product rule discussed earlier capture the essential
error function approach +2, and the value of the error features of this plume for the source geometry of Figure
functions approach one; that is, spreading is restricted 18.6~ (Domenico and Robbins, 1985b)
and the concentration is maintained near C,, over small
values of x.

Models for Mu#idimensdonal Transport

Multidimensional transport involves both longitudinal
and transverse dispersion in addition to advection. The
most complex form of the dispersion-advection equa-
tion that is amenable to an analytical solution includes
three dispersive components, a constant advective veloc- (18.18)
 181 Analytical Approacbes 373

(c)

Figure 1 8 8 Idealized contaminant migration geometries

for various transverse spreading directions (from Domenico
and Robbins, 1985b). Reprinted by permission of Ground
Water. Copyright 0 1985. All rights reserved.

z (Figure 18.86), the source dimension Z terms become

Z / 2 . Ifzspreading is eliminated altogether (Figure 18.8c),
Steady state maintained Dispersion zone in
at source concentration C,,. front of v f .
the error functions containing the z terms in Eq. 18.18
Steady state maintained
at concentrations less than
a.
Mass depletion zone
in back of i t
are ignored, and G,/8 becomes C,,/4. If the aquifer thick-
ness is small, the plume may spread vertically and occupy
source concentration C,,. the entire thickness (Figure 18.9). The length x’ over
which this spreading occurs can be approximated as
Figure Z& 7 ldealization of the dispersion process showing
(a)plug flow, ( 0 ) longitudinal dispersion, (c) transverse dis- (18.20)
persion, and ( d ) longitudinal and transverse dispersion (after
Domenico, 1987). Reprinted from J. Hydrol, v. 91, by P. A. where H is the thickness of the aquifer (Domenico and
Domenico, An analytical model for multidimensional trans- Palciauskas, 1982). Hence, Eq. 18.18 is valid for all dis-
port of a decaying contaminant species, p. 49-58, 1987 with tances x equal to or less than (H - ;z>2/cy.-. For distances
kind permission from Elsevier Science, NL, Sara Rugerhart- greater than x‘, the distance x in the denominator of the
straat 25, 1055 KV Amsterdam, The Netherlands. error function of the z term is replaced by x’ (for exam-
ple, Eq. 18.20). Thus, further spreading in x is prohibited
For x << vt, the argument of erfc approaches negative for x > x’. Following is an example that illustrdtes how
two, and we recover the Dornenico-Palciauskas (1982) an analytical solution is used.
steady-state model of Eq. 18.15. For the plane of symme-
tryy = z = 0 Example 1&3
Drums of diethyl ether (de) and carbon tetrachloride (ct)
were buried in a sand aquifer 15 years ago. Calculate the
concentrations of each contaminant along the plane of
{ erf
symmetry of the plume at the point (x = 225 m, J J =
0 m, z = 0 m) at time 15 years (4.73 X 10” s). The
linear velocity of the ground water is 1 X 1 0 - ” m/s. The
This form also shows that as ayand cw, approach zero,
we recover the Ogatd-Banks (1961) Eq. 18.3.
As with all of the transport equations, the velocity ZI
is the contaminant velocity. For a retarded species, the t ZI,
contaminant velocity is used directly or v,,./R,is substi-
tuted directly for u in Eq. 18.19.There is one other feature
about this result that helps to accommodate various field
conditions. As developed, Eq. 18.18 applies to a source
geometry with two transverse and one vertical spreading Ffgure 189 Vertical spreading of a contaminant limited by
direction (Figure 18.8~).For two spreading directions in a low-permeabilityboundary.
-
380 Cbapter 18 / Modeling tbe Transport of Dtssolved Contaminants

retardation factor for de is 1.5 and for ct 27.4.The source which reduces to the Domenico-Robbins (1985b) result
concentration for de is 1 X lo4 p g / L and for ct is 5 X when A = 0. The concentration along the plane of sym-
10' pg/L. The source size in Y is 25 m and in Z is 5 m. metry is
The estimated dispersivities are ax= 1.0 m, ay= 0.1 m,
and a, = 0.01 m. Assume that the plume can spread in
two z directions and two y directions so that

erfc [-
x +
vt(1 4Acu,/~)~/~
2(a.vvt)"'
] (18.22)

For de, and azapproach zero, the result of Bear (1979) is

As a,,
recovered Eq. (18.10). Further, a steady-state form may
be obtained when the argument of the complementary
error function approaches negative two.
Virtually everything that has been stated for the Do-
[ 5 1
erf 4(0.01 X 225)"'
menico-Robbins (1985b) result applies here also. Thus,
Eq. 18.21 can be modified to incorporate different spread-
ing geometries as well as to account for limitations in z
spreading. Further, the velocity v in these equations is
the contaminant velocity so that retardation is already
Because the argument of the complementary error func- incorporated or can be easily accommodated by substitut-
tion has a negative value greater than -2, the plume is ing v = uJR,.
at steady state at this point and the calculated concentra- The models discussed above have been applied to field
tion is a maximum value. problems by Fryar and Domenico (1989) and Ala and
For ct, Domenico (1992). Their main use is in determining the
transport parameters from concentration distributions
observed in the field.

Numedcal Integration of an Analytic Solution

[ 5 1
erf 4(0.01 X 225)"'
Because erfc [(225 - 17.2)/2(1 X 17.2)"'l = erfc
A benefit of the mathematical procedure that we have
just described is the inherent simplicity in the way com-
ponents of the overall solution can be put together and
(25) = 0, it follows that carbon tetrachloride has not yet the ability to manipulate these expressions to form a
reached this point because of greater retardation. The powerful method for parameter estimation (Domenico
maximum concentration that ct will eventually attain is and Robbins, 1985b). Recall that the simplification came
because we worked with a simplified form (Eq. 18.3) of
C,, = (5 X 103(0.93)(0.759) = 3.5 X 10' pg/L the exact solution. This simplification, however, is not
without some cost, which is reflected in some error of
Domenico (1987) uses similar arguments to incorpo- approximation.
rate decay or degradation in the following solution for No rigorous analytical solution to the multidimensional
multidimensional transport: mass transport equation can avoid an extremely compli-
cated numerical integration. Several of these solutions
are available in the literature, including Wexler (1992),
Leij and Dane (1990), and Leij and others (1991). One
of these solutions (Leij and others, 1991) was used by
Leslie Smith to test the accuracy of the approximate
analytical solution developed in this chapter. Figure
18.10 shows that the approximate solution is less accu-
rate close to the origin, and for relatively large values
of dispersivity.

Tbe Instantaneous Point Source Model

The accidental spill, frequently referred to as an instanta-
(18.21) neous or pulse-type problem, represents another poten-
 Next Page

18.1 Analytical Approaches 3-

Direclion of f i w

0I
7'"
I
I
f
I
I
x=o
I
I
I '
I
1
i

.r = t%?,7

Ffgure la11 Plan view of the plume developed from an

instantaneous point source at three different times.

The dimensions of the cloud, assuming it actually started

Distance [rn) as a point, are
Figure la10 Comparison of the Domenico and Robbins qY= (29,t)'/'; u,.= (2D,.t)"'; a, = (2Drt)'"
(1985b) approximate solution (D/R) with an exact solution
(Leij and others, 1991). (Unpublished analysis, kindly pro- where u is the standard deviation so that 3 q , 3a,,and
vided by Leslie Smith.) 30; represent three spreading lengths within which
about 99.7% of the mass is contained.
tial contamination problem. Models for two types are
available: the parallelepiped finite source model of Hunt Example 18.4 (afrer Baetsg, 1969)
(1978) and the point source model of BaetslC (1969).
A leak in a storage tank for radioactive waste results in
The point source of BaetslC (1969) is the more useful one
an accidental release of 1000 ci of 10 years' cooled fission
because the geometry of a spill more closely resembles a
products, along with tritium. The waste contains 400 ci
point, at least when viewed from the far field, than a
parallel piped. Baetsle's model is Cs-137 (tIlz33 yr), 400 ci Sr-90(tlI228 yr), 100 ci Pm-147
(tl122.7 yr), and 100 ci H-3 (tl/?12.26 yr). Assume that n
river exists 100 m from the spill. Determine how much
time is required for each constituent to reach the river
and what the maximum concentrations will be. The lin-
ear velocity of the ground water is 10 cm/day, or 1.16
X cm/s. Other data are given in Table E18.4-1.
(a) Time to reach a point at 100 m
(18.23) From Eq. 18.7 where RJx = v,,t, let s be the distance
traveled by the nuclide. For tritium,
where Cois the original concentration; Vois the original
volume so that the product CoVois the mass involved in lo'
the spill; D,, Dy,D, are the coefficients of hydrodynamic --
t = RJx - (1) * = 0.864 X 10"s = 2.75 yr
v::, 1.16 X lo-'
dispersion; v is the velocity of the contaminant; x,y , and z
are space coordinates; t is time; and A is the disintegration Similar calculations show that Cs-147 will take 112 yrs
constant for a decaying substance. For a nondecaying Sr-90 6.9 yr, and Pm-147 1120 yr.
substance, the term A t is ignored. Again, this solution (b) Peak Concentration at 100 m
can be obtained readily as the product of three one- The half-life for tritium is 12.26 yr so that e-A' = e-"""'
dimensional solutions.
With an idealized threedimensional point source spill,
Table EI&4-I
spreading occurs in the direction of flow, and the peak
or maximum concentration occurs at the center of the Nuclide K,(mL/g) R, D,v cmLIs Dv = D,
"cloud," that is, where y = z = 0 and x = vt (Figure
0-137 10 47.6 lo-' 10
18.11)
Sr-90 0.6 2.885 10.' lo-'
Pm-147 100 476. lo-; lo-;
H-3 0 1. lo-; IOF5
Previous Page

382 Cbapter 18 / Modeling tbe Transport of Dissolved Contaminants

Table 18.2 FORTRAN Code for Evaluating the Analytical Solution for
One-DimensionalTransport with Retardation
CODE FOR E V A L U A T I N G O N E - D I M E N S I O N A L D I S P E R S I O N
WITH RETARDATION

DIMENSION C ( 5 0 )

R E A D N E C E S S A R Y I N P U T DATA

OPEN(5,FILE='IN.DAT')
OPEN(6,FILE='OUT.DAT')
R E A D ( 5 , I O O ) NX
W R I T E ( 6 , l O O ) NX
R E A D ( 5 , I l O ) CO,ALFX,VX,RF,TYR,DELX
W R I T E ( 6 , l I O ) CO,ALFX,VX,RF,TYR,DELX

(complete l i s t i n g p r o v i d e d o n d i s k a s rone.for
inthedirectorytrans).

because A = In 2/half-life and In 2 = 0.693.

1~('2~26x~5xM4m and positions where concentrations are to be calculated.
The value of time for the calculation comes from part (a). Input data are read from the file IN.DAT and output data
are written to the file 0UT.DAT. The analytical solution
is evaluated within a DO loop for various distances from
the source. The function subroutine ERFC evaluates the
error functions. Finally, we write the calculated concen-
trations to the specified output file. The following hypo-
The rest of the results are Cs137 1.08 X lo-'; Sr-900.608 thetical example illustrates how this code is used.
X and Pm-147 negligibly small ci/mL.
Example I b S
An organic contaminant is disposed of continuously in a
1 s Programming the Analytical long narrow trench that fully penetrates a shallow, semi-
infinite aquifer (see Figure 1 3 . 1 ~ ) For
. this aquifer v =
Solutions for Computers 2.31 X m/s, a, = 4.3 m, Rf = 3.0. The contaminant
The closed-form analytical solutions can be applied cli- concentration at the source (CJ remains constant with
rectly to contamination problems. However, an alterna- time at 1.O mg/L. Calculate the contaminant concentra-
tive and more efficient approach is to construct codes tion after 3 years at 10-m intervals from the source.
to perform these calculations. Almost any type of com- This problem is solved with the code listed in Table
puter can be used and the necessary codes can be con- 18.2 and included on the disk as rone.for (see Appendix
structed with little difficulty. B). The data set is shown in Table E18.5-1.
Most codes of this type carry out three operations:
reading values of the transport parameters; solving for
concentration at specified times and locations; and writ-
ing the final results. We can illustrate these steps in a 18
practical way by developing a code for evaluating the 1.0 4.3 .00000231 3. 3. 10.
analytical solution to Eq. 18.1 for onedimensional mass
transport subject to simple sorption. This example is a
demonstration that is not applicable to field problems.
According to the solution Eq. 18.3, suitably modified to For consistency in units, years in the input are converted
account for sorption, three transport parameters ZI, a,,, to seconds in the code. The calculated concentration of
and Rf,and the concentration at the upstream boundary contaminants at the end of 3 years down the system at
(C,) need to be specified. Coding begins (Table 18.2) by 10-m intervals beginning with x = 10 m is shown in
reading and writing these values and defining the time Table E18.5-2.
 Id2 Programming the Analytical Solutions for Computers 383

Table E18.5-2
3. Y E A R S
0.99 0.98 0.96 0.91 0.82 0.70
0.55 0.39 0.25 0.14 0.07 0.03
0.01 0.00 0.00 0.00 0.00 0.00

Step 2
Modeling is one of those activities that must be prac- The analytical solution requires values for ur, a,, a:,
C,,, Z/2,Y/2,and the location of points where we intend
ticed to develop necessary coding skills, the ability to
to calculate concentrations. This code has a similar
formulate a problem in terms of boundary conditions and
scheme for data entry as the previous one except for
data, and the ability to use a computer. Let us consider
requiring additional dispersivities and the dimensions o f
another, more complex problem.
the source. Concentrations are calculated for a grid of
points on a horizontal plane passing through the middfe
Example 18.6 of the source at z = 0.
For the problem in Example 18.3, develop a computer
Step 3
code to predict the concentration of diethyl ether (de) Construct the code (included as rthree.for and rthreexxe
along a horizontal plane through the middle of source on the disk in the directory trans.). A grid provides a
(that is, I = 0 m). convenient way to define the concentration of points o n
There are five steps necessary to solve this problem: the plane through the center of the plume. NX and NY
(1) formulate the problem mathematically and select an are the number of points in the x andj’ directions. DELX
appropriate analytical solution, (2) establish what input and DELY are the spacings between points.
parameters are required, (3) construct the computer
code, ( 4 ) run the code using known or estimated informa- xdir: NX = 10 y-dir: NY = 7
tion about the site, and (5) interpret the results. DELX = 75.0 m DELY = 15.0 ni
Step I
Step 4
The information provided for this problem points to
Run the code using the following set of input data (on
three-dimensional spreading of a plume in a one-dimen-
disk as in 3.dat).
sional velocity field. Diethyl ether with fixed initial con-
The resultingset ofinput data isshown in Table E18.&1.
centration of 10,000pg/L enters the ground water from
All units are consistent in meters and seconds. Because
a source of finite size. This problem Cdn be solved using
R is dimensionless we can use another set of internally
Eq. 18.18. This form of the transport equation accounts
for dispersion in three directions, advection in one direc-
tion, and retardation due to sorption. The downstream Table E I d 6 - I
side of the disposal site acts as the finite source with an
area YZ and specified nonzero concentrations of each 10 7
10000. 1. 0. I .01 1 .OE-(X> 1.5
contaminant. The initial condition is that the concentra-
15. 75. 15. 0.0
tion of either diethyl ether (C&)is zero within the semi-
5. 25.
infinite domain.

Table E 1 8 6 - 2

15. YEARS
0.00 0.00 2486. I 2 9575.76 2486.12 0.00 0.00
0.00 5.95 2757.84 a319.54 2757. a4 5.95 0.00
0.00 34.59 2700.07 71 37.75 2700.07 34.59 0.00
0.07 60.30 iaa6.02 4514.41 iaa6.02 60.30 0.07
0.00 1.21 21 -63 47.73 21 -63 1.21 0.00
0.00 0.00 0.00 0.00 0.00 0.00 0.00
0.00 0.00 0.00 0.00 0.00 0.00 0.00
0.00 0.00 0.00 0.00 0.00 0.00 0.00
0.00 0.00 0.00 0.00 0.00 0.00 0.00
0.00 0.00 0.00 0.00 0.00 0.00 0.00
384 Chapter 18 / Modeling the Transport of Dissolved Contaminants

consistent units. The simulated distribution of de is viding the region also serves to define the nodes at which
shown in Table E18.6-2. concentrations are calculated. The way in which nodes
are defined and how the domain is subdivided (for exam-
ple, squares, rectangles, triangles, stream tube segments,
etc.) depends on the specific numerical technique. The
-
18.3 Numerical Approaches Otis Air Rase case study presented later shows how one
goes about subdividing a region and assigning bound-
The numerical approaches are a family of computer- ary conditions.
based techniques for solving the contaminant transport There is rcaly no difference in the boundary conditions
equations. They approximate forms of the advection- and transport parameters required in the numerical a p
dispersion equation as a system of algebraic equations,
proaches as compared to the analytical approaches. Com-
or alternatively simulating transport through the spreitd plicating the situation with numerical models is the need
of a large number of moving reference particles. What-
to provide boundary conditions at a large number of
ever the procedure used, it invariably has to be coded node points or nodal blocks (areas around the nodes).
for solution on a high-speed computer. For example, values of concentration or loading rates
Numerical approaches deal easily with variability in the
defining various boundary conditions now need to be
flow and transport parameters (for example, hydraulic specified for all nodes located along the boundary of the
conductivity, porosity, dispersivity, cation exchange ca- domain. Initial conditions and transport parameters are
pacity, etc.). This flexibility in representing parameters specified for all nodes except in some cases for nodes
facilitates modeling of layering or other, more complex
with constant concentrations.
geometries in two and three dimensions. Thus, one can
It is no real problem to assign transport parameters to
simulate the complex plume shapes that often develop nodal areas, other than perhaps not knowing what the
in natural systems. The analytical approaches cannot ac- values are. Velocity values are an important exception
count for variability in transport parameters. Thus, nu- because they are more difficult to specify. In most real
merical approaches are readily adapted to site-specific
systems with pumping and injection wells and a variable
problems, which makes them particularly useful in
hydraulic conductivity field, velocities can be extremely
practice. variable. Both the magnitude and the direction of flow
Other important features of the numerical approaches
can change in space and time. In terms of modeling,
are the flexibility in implementing complex boundary
measured or guessed velocity values are not adequate.
conditions and in accounting for a variety of important
Continuity considerations in all numerical solutions re-
mass transport processes. Codes are now available to
quire a smooth and accurate representation of the veloc-
model (1) radionuclide chains in which radioisotopes
ity field.
and their daughters decay, each with a variable half-life,
This kind of field can be obtained only by simulation
(2) equilibrium reactions such as precipitation, dissolu-
with a flow model. Velocity values are calculated by
tion, and complexation, and (3) competitive sorption or applying the Darcy equation with calculated values of
ion exchange. Developing the increased sophistication
hydraulic head and known parameters, or directly calcu-
to deal with reactions is necessary to accurately model
lating velocity values at the nodes. Figure 18.12 illustrates
contaminant groups such as trace metals or radioni-
how flow and mass transport models are used together
clides. to predict contaminant distributions. Input necessary for
the flow model could include the hydraulic conductivity
distribution, the water-table configuration, and other
A General#zedModeling Approach
boundary conditions. The transport model is coupled
The modeling process can be examined independently to the flow model by the velocity terms as shown in
of a particular solution technique. The starting place in Figure 18.12.
nearly all cases is with one of the various forms of the Flow in many caSes is assumed to be independent of
advection -dispersion equation. Because of the power the mass transport. In other words, the concentration of
of the numerical approaches, the transport equation i s contaminants does not influence the flow by changing
commonly formulated in two or three dimensions. Al- the fluid density. In this situation, there is no requirement
though our emphasis here is with mass transport involv- to solve the flow and transport equations simultaneously.
ing a single contaminant, modeling approaches have For steady flow, the ground-water flow equation is solved
evolved to the point where a coupled set of differential once and a single velocity field applies for all time. If
equations can be formulated for problems involving sev- flow is transient, the velocity field must be calculated
eral reacting species. each time that the contaminant concentration is required.
One of the first steps in developing a computer model However, again, the flow and transport equations are
is to subdivide the region in terms of cells or elements. solved separately.
This process makes it possible to account for varying Contaminant concentrations are sometimes so large
parameter values within the domain. Furthermore, subdi- that the presence of the spreading plume causes the flow
 18.3 Numerical Approaches 33

--c
input -+ Flow model I -
-F
4
4
4
-t
4
--c
/
-c

IVelocity

Figure la12 The relationship between ground-waterflow and mass transport models in
simulating the distribution of contaminants (from Schwartz and others, 1985). Reprinted
with permission of Second Canadian/American Conference on Hydrogeology. Copyright 0
1985. All rights reserved.

pattern to change. In these cases, the ground-water flow 1978) and its succeeding versions. This code has become
depends on the mass transport just as the mass transport a standard for use in studying the most complicated mass
depends on the ground-water flow. In Figure 18.12, the transport problems.
arrow joining the flow and transport models now will The essence of the finitedifference method is to re-
point in both directions. For this fully coupled situation, place the governing differential equation by a set of differ-
the flow and transport equations need to be solved simul- ence equations applicable to the system of nodes. The
taneously. difference equations approximate the first- and second-
order derivatives in the transport equation (that is, dc/dx
or d2C/dx2)by concentration differences between node
The Common Solution Techniques
points. When each node in the network is considered.
The techniques for solving the advection-dispersion the result is a system of algebraic equations. This system
equation are extremely varied and would require almost can be solved with matrix algebra. The interested reader
a complete textbook to discuss them all in detail. Here, will readily discover a voluminous literature on the sub-
we present a brief summary of the most common meth- ject. An important feature of the finite-difference method
ods mainly to introduce the pertinent literature and to is the relative ease in formulating the difference equa-
create an awareness of the basic terminology. There are tions. Of all the numerical techniques available to solve
two general approaches available for solving mass trans- differential equations, finite-difference techniques are
port problems. The direct solution techniques are the easiest to understand from a conceptual point of view.
most common and involve a numerical solution of the Finite-element approaches are common in solving mass
advection-dispersion equation. Included in this group transport equations. Although the approach was first de-
are the well-known finitedifference and finite-element veloped in the petroleum industry, much of the subse-
techniques and the method of characteristics. quent refinements and applications were by hydrogeolo-
Research is continuing to develop new and improved gists. The most important early studies included work b!.
approaches for solving mass transport equations. The Pinder (1973) and Duguid and Reeves (1076) on the
continuing effort to develop numerical approaches is modeling of contaminant transport, and by Rubin and
attempting to reduce the computational effort and to James (1973) on transport with multiconiponent ion
overcome numerical problems endemic to the conven- exchange. Various extensions of the technique ex;ini-
tional approaches. Examples of recent modeling ap- ined density-dependent problems (see HqS;tkorn and
proaches or significant refinements of existing methods Pinder, 1983).
include the principal direction method (Frind, 1982), the Unlike the finitedifference methods. which involve
alternating direction Galerkin method (Daus and Frind, solving the mass transport equation directly. the tinite-
1985), and the Laplace transform Galerkin technique element techniques deal with a mathematically equiva-
(Sudicky, 1989). lent, integral form of the mass transport equation. This
The finitedifference methods have traditionally been integral form can be developed in using the method of
applied to solve flow and transport equations. One of the weighted residuals or the variational method. Hityakom
most important implementations of the finitedifference and Pinder (1983) present a detailed discussion o f these
approach is in the powerful code SWIFT (Dillon et al., approaches and in particular the Galerkin procedure
3 3 Cbapter 18 /Modeling tbe Transport of Dissolved Contaminants

(a special case of the method of weighted resid- exceptions, the possible approaches fall into two catego-
uals). ries. With the “one-step’’ approach, a complete mass
The method of characteristics is another useful ap- transport equation including all of the appropriate reac-
proach for solving mass transport equations. It was first tions is written for each solute species. The entire cou-
applied by Garder and others (1‘964). Subsequently, the pled set of algebraic equations is then solved simultane-
method was used in hydrogeological applications to ously using either a finite-element or finite-difference
study saltwater intrusion (Pinder and Cooper, 19701, method. This procedure is numerically complicated but
ground-water contamination (Reddell and Sunada, 1970; is the most rigorous way to handle reactions in a trans-
Pinder, 1973), and the role of various parameters in con- port framework.
trolling the spread of contaminants (Schwartz, 1975). Most of the more general implementations of this
Konikow and Bredehoeft (1978) implemented this ap- scheme have been for one-dimensional transport. Ex-
proach in a versatile two-dimensional computer code, tending this approach to two-dimensional problems is a
which today is one of the most commonly used codes. formidable task. One of the most sophisticated examples
The method takes the advection-dispersion equation of the onestep procedure is a code developed by Miller
and breaks it down into a set of simpler differential equa- and Benson (1983). Their model and the planned exten-
tions. This formulation in effect provides a frame of refer- sions account for one-dimensional advection and disper-
ence that is moving with the linear ground-water velocity. sion along with a host of equilibrium chemical reactions.
Thus, the advection-dispersion equation, which is rela- Included in the suite of reactions are complexation in
tively difficult to solve, becomes a diffusion equation, the aqueous plrase, exchange of both ions and com-
which is simpler to solve. plexes, and mineral precipitation. Other examples of
The particle tracking method is different from the pre- these models include Willis and Rubin (1 987) and Orto-
vious three numerical approaches. It does not solve the leva and others (1987a,b).
mass transport equation directly but simulates the spread The “two-step” procedure separates the physical and
of mass dissolved in water. This approach is formulated chemical processes. A solution to the advection -
as a classical random-walk problem involving the motion dispersion equation provides an initial estimate of con-
of a swarm of reference particles. It has been applied in centration. A second step corrects these concentrations
modeling atomic particle motion in nuclear reactors, the to account for the partitioning of mass due to a suite of
spread of air pollutants, and oil spills in oceans. Ahlstrom chemical reactions. Operationally, the two-step proce-
and others (1977) introduced the method, which subse- dures require iteration between the steps until some
quently was implemented in several computer codes specified convergence criterion is met. Iteration assures
(Schwartz and Crowe, 1980; Prickett and others, 1981). that the calculated concentrations are solutions to both
The transport of contaminants is simulated by adding the advection-dispersion equation and the material bal-
reference particles and moving them in a prescribed man- ance equations describing the chemical system. This
ner. By varying the number of particles added at the method is simpler computationally although less rigorous
source during any one time step, it is possible to simulate than the one-step procedure. For this reason, the a p
complex loading functions. To account for advection, proach is amenable to two- and three-dimensional prob-
each particle moves in the direction of flow a distance lems. It also provides a way of grafting sophisticated
that is determined by the product of the magnitude of geochemical codes like PHREEQE, MINTEQ, or MINEQL
the velocity and the size of the time step. With a small to transport models.
time step, this particle motion traces a pathline through An example in the development of this approach is
the system. Dispersion is accounted for in the particle described by Cederberg (1985). The transport equations
motion by adding a random component to the determinis- are solved in two dimensions using a Galerkin finite-
tic motion, which is a function of the dispersivities. element procedure and the algebraic equations for the
The mean concentration for each grid block is calcu- chemical reactions using a Newton- Raphson method.
lated as the sum of the mass carried by all of the particles Chemical processes include complexation, as well as
located in a given block divided by the total volume of complex sorption, which considered the effects of sur-
water in the block. By appropriate adjustments to the face ionization and complexation at the solid-solution
quantity of mass carried by each particle, first-order ki- interface. Other examples of two-step models come from
netic reactions or simple linear sorption can be simulated. work by Narasimhan and others (1986) and Liu and Nara-
simhan (1989).
Adding Cbemical Reactions
Going beyond problems involving simple reactions like
18.4 Case Study in the Applkation of a
linear sorption or first-order kinetic decay to those with
Numerical Model
several interacting solutes is much more difficult than A model study of contamination at Otis Air Base on Cape
any numerical approach considered so far. With a few Cod, Massachusetts (LeBlanc, 1984b), illustrates some of
 164 Case Study 387

the steps in using a mass transport model. At Otis Air

Base, approximately 1740 m2/day of treated wastewater
has been disposed in the subsurface since 1936. The
disposal unit is an unconfined sand and gravel aquifer
approximately 35 m thick. A zone of contamination has
developed that is approximately 915 m long, 23 m thick,
and 3.35 km long. This site is being studied by the U.S.
Geological Survey as part of the Toxic-Waste Ground-
Water Contamination Program (LeBlanc, 19841; Franks,
1987; Ragone, ed., 1988).
The treated sewage effluent contains above back-
-
- . r
I aeds I

boron plume in 1978-74.

ground concentrations of Na', CI-, ammonium, nitrate,
phosphate, detergents, and several different volatile or-
Q 1500 3ooO Meters
ganic compounds. Boron, one of the trace metals in the t--cf--t-t
effluent, was selected by LeBlanc (1984b) for a detailed 0 5CoO 10,800 Feet
model study because (1) its concentration has remained
relatively constant in the effluent at 500 pg/L, (2) it has
a relatively low background level in the native ground
water (30 pg/L), and (3) it tends not to react chemically
during transport. The objective in modeling was to guide
the collection of data from the site and to test hypotheses
concerning the character of contaminant migration.
The first step in developing the flow and boron trans-
port models was to define the region of interest and to
establish boundary conditions for flow. LeBlanc ( I 984b)
used the following guidelines, which should apply gener-
ally to most studies. The domain had to be large enough along Nmtrcket Sound
to include all of the existing plunie (Figure 18.13) as well
as providing room for future spreading. The side and Figure 18.13 Map of important hydrologic features at the
bottom boundaries down-gradient of the disposal area site in relation t o the boron plume in 1978-79 (modified
are ponds, rivers, and saltwater bays. These features are from LeRlanc, 1984b).
a good choice for boundaries because in the absence of
detailed hydraulic head data, they provide the best places
to estimate boundary conditions. In effect by assuming replicates the region shape and how the boundary condi-
ground-water discharges at these locations, boundaries tions for flow are included.
for ground-water flow can be estimated in terms of con- LeBlanc modeled the pattern o f ground-water flow
stant head nodes or leakage fluxes. North of Coonames- first. Because the density and viscosity of the contami-
sett and Johns Ponds, there are no natural hydrogeologic nated ground waterare nearly the same as the uncontanii-
boundaries. In this area, flow lines formed the side bound- nated water, the distribution of hydraulic head and hence
aries (Figure 18.13). The flow lines are imaginary bound- the velocity field are unaffected by the migration of the
aries, located so as no&to intersect the contaminant plume. The tlow equation therefore can be solved first
plume. The top or northern boundary was arbitrarily independently of the mass transport equation. Observa-
defined by the 604 equipotential line. This boundary tions at the site indicate further that the gradients in
could have been placed anywhere north of the site with hydraulic head do not change significantly with time.
the proviso again that contaminants from the site could Thus ground-water flow is steady, which means that the
not intersect this boundary. ground-water velocities need to be calculated only once.
The Konikow and Bredehoeft (1 978) model requires Figure 18.15 compares the observed configuration of
that the site be subdivided into a regular finite difference the water table with the "best fit" from a series of simula-
grid. In this example, the grid consisted of 40 rows and 36 tions. Such trials are designed to calibrate the model-a
columns. The nodal blocks are square, with dimensions process of selecting the set of parameters that produces
chosen so that there is not an unmanageably large num- the best simulation of a known histon. The calibration
ber of cells (an upper limit might be 50 or so cells in here took initial estimates of model parameters (for exam-
the row/column directions). However, the blocks are ple, hydraulic conductivity, recharge rates) and adjusted
sufficiently small to ensure that the plume is not localized tthem by trial and error until the model successfully repro-
in just a few cells. The plume in this case has a width of duced the observed configuration of the water t;ible.
about seven cells. Figure 18.14 shows how the model Because of the lack of data, no attempt was made to
388 Cbapter 18 / Modeling tbe Transport of Dtssolved Contaminants

3W.W' 3M2' choosing a value is not a problem because analysis shows

that with active recharge, simulated concentrations are
2 not sensitive to the particular choice of dispersivityvalue.
4 A lack of data may not be so easily dealt with in every
6 model study. Contaminant loading was approximated by
8 fixing the boron concentration at 500 pg/L at the four
10 cells representing the infiltration beds.
t2
Tests with the model attempted to reproduce the his-
14
torical spread of contaminants over a 40-year period be-
tween 1940 and 1978-79. Compared in Figures 1 8 . 1 6 ~ ~
41"38' 16
and 6 are the simulated and the observed distributions
18
of boron after 40 years. As LeBlanc (1984b) points out,
20
the simulated path of the plume agrees reasonably well
22 with that observed in 1978-79, particularly at concentra-
24 tions above 50 &L. However, at lower concentrations,
26 the center line of the simulated plume was located east
2s
30
32
41."36'
34 70034' 70"32
I I
36
31) Boundary of
40 active modeled

modeled a r m )

~Spe"ified-inf'\o"
nods [constant rate
a (teakante
Specified he& stode
set equal to
of iflmat infiltration 1.0 per second:
beds and northern tonstant head during
boundary) sirnulationsl

Figure 1814 Grid system used to model both yround-

water flow and mass transport. For ground-water flow model-
ing, the combination of zero transmissivity cells, leakance
nodes, and specified head nodes define the region shape and
bounary conditions (from LeBlanc, 1984b).

reproduce spatial variability in hydraulic conductivity or

recharge rates. Nevertheless, the simulated and observed
results compare well (Figure 18.15). The model as repre- 41"31
sented by the parameters and boundary conditions is not
unique, Another set of different parameters and boundary
conditions could give the same or a better fit. By keeping
the hydraulic conductivity close to the measured values,
the simulated hydraulic heads and resulting velocity field
should be a reasonable representation of the flow system.
The only information required for the transport sirnuhi- Water-level observation well 0 1000 Meters
tion with a conservative species like boron is the Jongjtu- (site where
water ,eve,scomputed
comparedand observed
ca,ibrationl
l-l-+d
0 4000 Feet
dinal and transverse dispersivities, plus the loading func-
tion at the source. In the absence Of actual dispersivity Figure la15 A comparison between the observed (dashed
data, values were selected from the literature for similar lines) and simulated (solid lines) water table elevations for No-
types of geologic materials. Here, this uncertainty in vember 1979 (from LeBlanc, 1984b).
 18.4 Casestudy 383

of the observed center line. The simulated plume is curs. This problem could be related to the complex
also somewhat longer and wider than the observed interaction between ground waters and surface wdters.
plume. Finally, the model could contain undetected numerical
LeBlanc (1 984b) explained the difference between ob- dispersion.
served and simulated plumes as follows. The observed This study went on to evaluate the suitability of man-
plume could in fact be larger than is represented in Figure agement options for the contamination. With reasonable
18.16b because of uncertainty in establishing the edges confidence in the ability of this model to predict boron
of the plume at concentrations below 100 p a . Inaccura- distributions, LeBlanc (1 984b) examined the consr-
cies in the simulated flow field could have existed, pro- quences of operating the site in the future as it is now
ducing somewhat more divergent flow than actually oc- or eliminating it.

70034' 70'32'

0 1800 2 W Meters
m
ff 4 m m Feet
8

_I 1W - Line of equal boron concentration

(concentrations in micrograms per liter.
Interval varies)
---".------
Eastern boundary of plume redefined by
November 1983 test-well drilling
Area of plume delineated in 1978-79
(denoted by detection of detergents, or concentrations
exceeding 20 milligrams per liter chloride or
100 micrograms per liter boron)
Observation well (drilled in November 1983
to delineate plume boundaries)

Figure 28.26 A comparison between ( a ) the simulated distribution of boron in the aqui-
fer with (6) that observed in 1978-79. Also shown on (a) are indications of the plume size
provided by C1- and detergents (from LeBlanc, 1984b).
330 Cbapter 18 / Modeling tbe Transport of Dissolved Contaminants

Problems
1. In a onedimensional column, a retarded species has through the column is about 36 cm/hr. The length of
moved a certain distance when v,J of a nonretarded the flow tube is 30 cm. The C/C,,= 0.25 point is arrived
species is taken as 28 cm and = 2 cm. The retardation at the end of the column in approximately 0.7 hr. At
factor is two. How far behind the advective front of time equal to 0.992 hr, C/C,, = 0.75,and at time 0.833
the nonretarded species is the advective front of the hr., C/C,, = 0.5.
retarded species? a. Estimate the dispersion coefficient in cm2/hr.
2. In a onedimensional column test, C/C,, is noted to b. Estimate the dispersivity in cm.
be on the order of 0.65.This means: c. Can you use the C/C,, = 0.5 at t = 0,833 hr in the
-a. The advective front has already broken through. preceding calculations? Why?
-b. The advective front has not broken through. 6. In a plan view of a contaminant plume, you notice
-c. The argument of the erfc is negative. that chloride has moved approximately 10oO m whereas
cadmium has moved about 100 m. Assuming that both
-d. The argument of the erfc is positive. constituents were released from the source at the same
-e. a and c are correct. time, find the distribution coefficient for cadmium if the
-f. b and c are correct. porosity is 0.2 and the density ps is 2.65 g/cm3.
3. The erfc of negative beta is 1.146.What is the value 7. Answer the following questions with respect to the
of beta? Ogata- Banks equation.
4. Match the breakthrough curves given on the diagram c =1 x - v t =- x - vt
with the proper description given below. - - erfc - 1 erfc
c, 2 2(D$)”Z 2 2(a,vt)”Z

lrI a. The maximum observed concentration will be equal

to ___.
b. This maximum will occur at some point x where x
is __.
C. At x = vt, the concentration C is equal to -.
d. The concentration approaches zero at some point x
where x is -.
Prove or disprove each of the following.
e. The larger a.r,the more extensive the spread ahead
of the advective front.
Number of Pare Volumes
f. The larger ctr, the farther behind the center of mass
h & & m 4. A pore volume in a column is the volume of
the fluid takes on its maximum concentration.
water that will completely fill all of the voids in the column.
g. The larger a,,the longer the time required to reach
steady state at the column exit.
-This curve was obtained for continuous injection h. When determining a, in the laboratory, it is advisable
in a onedimensional laboratory column where the to use the C/Codata at the column exit when x = vt.
constituents in the displacing fluid were being i. The larger a x ,the lower the concentration at some
sorbed by the medium. point just ahead of the advective front.
-This curve was obtained for continuous injection in 8. A spill from a high-level storage tank releases the
a one-dimensional column with such a low velocity radionuclides shown in the table.
that plug flow (no dispersion) was observed. a. What constituent will be first to arrive at the p r o p
-This curve was observed for a pulse (instanta- erty boundary?
neous) type source with both longitudinal and b. What constituent will be last to arrive at the prop-
transverse dispersion. erty boundary?
What constituent will have the smallest maximum
-This curve was obtained for continuous injection C.
concentration at the property boundary?
in a onedimensional (no transverse dispersion) lab-
d. What constituent will have the largest maximum
oratory column.
concentration at the property boundary?
-This curve was obtained for continuous injection e. What constituent will have the smallest spread
with both longitudinal and transverse dispersion. around the center of mass in the direction of flow?
5. An experiment with a step function input of nonreac- f. What constituent will have the largest spread around
tive tracer is conducted. The velocity of the tracer the center of mass in the direction of flow?
 Problems 331

~~~

A 400 33 8 10-1 lo-5 lo-"

n 400 28 52 lo-' 1O - s lo--
c 400 2.7 40 IO-' lo-'
D 400 20 0.1 lo-"

g. What constituent will have the same spread around 11. Following is a site map and hypothetical plume. At
the center of mass in all directions, that is, in the the F-area and H-area seepage basins marked on the tig-
flow direction, transverse to the flow direction, and ure, waste water containing a variety of contaminants
vertically as well? was disposed of over a 13-yearperiod into an unconfined
9. A hazardous waste facility is to be constructed in a aquifer. Your objective is to model the transport of a
given area. The state requires a buffer zone between the contaminant at the F-area that is moderately sorbed and
waste trenches and the property boundary. Over 500 that is lost via a first-order kinetic reaction. Forniulate
different constituents are to be included in the waste. this problem for a numerical solution by (1) defining :I
some with reasonably large distribution coefficients, oth- local region of interest and appropriate boundary condi-
ers with a likely distribution coefficient of zero. A few tions for flow and transport, (2) selecting appropriate
years after construction, a plume is first noted by observa- forms of the flow and transport equations, and (3) indicat-
tion wells at the waste boundary. The maximum concen- ing what parameter values are required for a simulation.
tration in the plume occurs in a zone of about 6 ni thick 12. A plume of bromoform dissolved in grouiid water
and extends the entire length of the waste boundary originates from two storage ponds with a total length of
or about 60 m. Because of low-permeability bounding 150 m. The ponds penetrate the unconfined aquifer to
materials, the plume is capable of spreading only in Y a depth of 5 m. The total saturated thickness of the aquifer
but not in Z. is 15 m. Use the computer code that was developed in
Calculate the maximum concentration that might be Example 18.6 to solve this problem. Note that it will
expected (in terms of C,,, the plume concentration at the be necessary to modify the code to consider only one
waste boundary) once the plume arrives at the buffer direction of z spreading (that is, Ey. 18. IS). Predict the
zone, which is 150 m from the waste boundary. The concentration distribution along the water table after
transverse dispersion coefficient is 1 X lo-' mz/s and the 51 10 days, assuming a linear ground-water velocity of
linear velocity (flow through the pores) is 1 X lo-' m/s. 1.6 X 10-j; dispersivities in x, y , and I of 5.0m, 0.5 ni,
10. Ground water travels with a linear velocity of 15 and 0.1 m, respectively; C,, = 4000 mg/L; Ax = 0.005;
cm/day. What will the transport velocity be for an organic log K,,,,. = 2.38; porosity 0.30;and a solids density of 2.65
contaminant having a K,, of 6.6 m u g in a medium with g/cm'.
a porosity of 0.35and a solids density (p,) of 2.65 g/cm'?
392 Cbapter 18 / Modeling tbe Transport of Dissolved Contaminants

(from Duffield and others, 1987)

(from Duffield and others, 1987)

Problem 11
 CbaDter 19

M u l t i p h a s e Fluid Systems

19.1 Basic Concepts

19.2 LNAPLs and DNAPLs
19.3 Partitioning
19.4 Fate of Organics in the Unsaturated Zone
19.5 Fate of Organics in the Saturated Zone
19.6 Air-Permeability Testing
19.7 Recognizing DNAPL Sites

Some contamination problems involve two or more flu- s, = v,/v,,,,,. (19.1)

ids. Examples include air, water, and organic liquids in
where V , is the volume of the ith fluid. In a multicomptr
tlie unsaturated zone, or organic liquids and water in an
nent system, the sum of all the saturations is equal to one.
aquifer. Organic liquids that have densities greater than
Wettability is the tendency for one fluid to be attracted
water are referred to as DNAPLs (dense nonaqueous
to a surface in preference to another. According to
phase liquids). These compounds provide a significant
Demond and Roberts (1987), tlie only direct measure of
challenge in both investigating sites and cleaning them
wetting is the contact angle. The idea of the cont;ict
up. Nonayueous phase liquids that have densities less
angle can be illustrated by an experiment in which ;I
than water are called LNAPLs (light nonaqueous phase
surface such as a piece of glass is imniersed in ;I reference
liquids). Commonly, contamination by LNAPLs involves
fluid such as water and a drop of the test fluid is placed
spills of fuels like gasoline and jet fuel.
on tlie surface. The contact angle is established b!, n m -
Problems involving NAPLs are more complex than dis-
suring the tangent to the droplet from :i point where
solved plumes. To understand the main factors control-
all three components are in contact (that is. surfiice.
ling the occurrence and flow of organic liquids will re-
reference liquid, and test liquid: Figure 19.1). A contact
quire several new basic concepts. We will take up these
angle <90° (Figure 19.1) implies that the test liquid is
issues in the following section. Having developed these
wetting. A contact angle >90" implies that the reference
basic principles, the remainder of the chapter examines
liquid is wetting and that the test liquid is nonwetting.
concepts and models affecting multiphase flow.
Wettability is unique for given types of solids and tluids.
However, a few generalizations hold:

- Basic Concepts
19.1 Water is always the wetting fluid with respect t o oil
or air on rock-forming minerals (Albertsen and oth-
Saturatiun and Wettability ers, 1986)
Saturation describes the relative abundance of fluid in a Oil is a wetting fluid when combined with air, but ;L
porous medium as the volume of the ith fluid per unit nonwetting fluid when combined with water (Al-
void volume. For a representative elemental volume, bertsen and others, 1986)

333
-
394 Cbapter 19 / MuUtphase Fluid Systems

T
, est liquid capillary pressure, P,., and is given as
-----.
P,. = P,,,. - Pu, (19.3)
where P,,,,,is the pressure of the nonwetting fluid (NAPL),
e > 90"
and Pll.is the pressure of the wetting fluid. For circular
e < 90"
Test liquid Test liquid
openings with a perfectly wetting fluid
is wetting is nonwetting
2a
Figure 19.1 Definition of wetting on the basis of the coil- P,. = - (19.4)
r
tact angle (modified from Demond and Roberts, Water Re-
sources Bulletin, v. 23, p. 617-628, 1987). Reproduced by Thus, capillary pressure is directly proportional to the
permission of Amer. Water Resources Association. surface tension and inversely proportional to the radius
of curvature.
Capillary pressure is a measure of the tendency of a
Oil is the wetting fluid on organic matter (for example, porous medium to imbibe the wetting phase or to repel
peat or humus) in relation to either water or air the nonwetting phase (Bear, 1972). As it is difficult to
(Albertsen and others, 1986) push a nonwetting fluid into a pore filled with wetting
The wetting character of organic contaminants re- fluid, capillary pressure can also be thought of as the
mains uncertain. pressure required to move a particle of nonwetting fluid
into a pore filled with a wetting fluid. Thus, as small pores
provide resistance to entry due to capillarity, nonwetting
Interfacial Tension and Capillary Forces NAPLs will tend to move through the coarser, more per-
When a liquid is in contact with another subst-ance(liquid meable zones of a heterogeneous medium. This behavior
or solid), there is a free interfacial energy between them, traps globules of water. In the unsaturated zone, capillary
due largely to the difference between the inward at- pressure (expressed as a negative pressure head) is re-
traction of that molecule in the interior of each phase. ferred to as suction.
A surface that contains such free energy would like to
contract, so that interfacial tensional forces are estab-
lished. The tensional nature of these forces can be com-
Imbibition and Drainage
pared with those set up in a stretched membrane. Interfa- Imbibition and drainage are the dynamic processes by
cial tension is defined as the work required to separate which fluids displace one another. Imbibition is the dis-
a unit area of one substance from that of another and is placement of the nonwetting fluid hy the wetting fluid,
expressed as a force per unit length. Values range from and drainage is just the opposite (Albertsen and others,
zero for completely miscible liquids to 72 dynes/cm 1986). Water being added to a dry soil or oil being gradu-
(0.072 N/m), which is the value for water at 25°C (Lyman ally displaced by water from a water-wet oil reservoir
and others, 1982). Values for most DNAPLs range he- are examples of imbibition. The entry of a nonwetting
tween 15 and 50 dynes/cm (0.015 to 0.050 N/m) (Cohcn organic liquid into a wat .=wet aquifer is drainage.
and Mercer, 1993). Interfacial tension permits nonwct- Capillary pressure curves are frequently given as a func-
ting DNAPLs to form globules in water-saturated mate- tion of the degree of saturation with respect to the wet-
rials. ting and nonwetting phases. In each case, the phase
As mentioned in Chapter 1, capillarity results from a initially saturating the porous medium is slowly displaced
combination of interfacial tension and the ability of ccr- by the other. As shown in Figure 19.2, the curves are
tain liquids to wet the surfaces with which they come frequently "L" shaped. The drainage or drying curve
into contact. Interfacial tension is frequently determined starts off at a wetting-fluid saturation of 100%but at some
by measuring the height a liquid will rise in a capillary finite value of capillary pressure. For the displacement
tube. When the contact angle between a liquid and the to take place, this value milst be exceeded. This threshold
wall is very small, as it is for water, the relationship value is called the displacement or entry pressure, which
between interfacial tension and capillary rise is given ;IS would be required for a DNAPL to enter a water-wet
aquifer. As noted, capillary pressure becomes more nega-
(19.2) tive as the wetting saturation decreases. With this in-
creased suction, the displacement process eventually
where rn is the interfacial tension, h is the height o f stops. The saturation value at which it stops is known
capillary rise, r is the radius of the tube, p is the liquid as the residual wetting saturation and is designated S,,,, .
density, and g is the acceleration due to gravity. Here, the capillary pressure tends to infinity. Table 19.1
The pressure discontinuity across any curved interface shows some estimates of displacement pressures calcu-
separating two immiscible fluids is referred to as the lated by Hubbert (1953).
 13.1 Basic Cotrcepts 335

Nonwetting saturation 1001 I I I I

Wetting saturation

Figure 19.2 Observations: wetting or imbibition curve. 0 25 50 75 180

Water Saturation
Figure 19.3 Tetrachloroetheiie-~~.ater drainage f,(.S,,)
The imbibition or wetting curve depicts initial satura- curves determined for seven sands o f varying hytlraiilic con-
tion of a nonwetting phase that is slowly displaced by a ductivity (from Kueper and Frind. Water Resources Rex, v.
wetting phax. It does not follow the same pathway as Z7, p. 1040- 1058, 1091. Copyright 0 by American Gcopli)si-
the drainage curve due to hysteresis. In this displacement, cal Union).
the capillary pressure becomes less negative as the wet-
ting saturation increases. At zero capillary pressure, a
point of residual nonwetting saturation is reached, desig- layer and obtain the capillary pressure from the curve.
nated as S,,,,,r on the figure. At this point, there is no This capillary pressure can be compared to a measured
further displacement of the nonwetting phase. or estimated entry pressure for the fine-grained lqw.
Kueper and Frind (1991) show capillary pressure-
water saturation curves for a tetrachloroethylene-water
Relative Permeability
system for sands of differing permeability (Figure 19.3).
The entry pressure decreases with increasing permedbil- The concept of relative permeability accounts for the
ity of the sands. Given such curves at a contaminated tendency for fluids to interfere with one another as t h e y
site, one can assess the potential for DNAPL t o enter flow (Demond and Roberts, 1987). This idea can be ex-
underlying fine-grained material (Kueper and plained in relation to a Darcy's law expression for
McWhoner, 1991). The first step is to determine the multifluid flow. For onedimensional flow in a Iiomoge-
saturation of the soil immediately above the fine-grained neous medium, Darcy's equation written in terms o f pres-

Table 29.1 Threshold Entry (Displacement) Pressures in Sediments with Various Grain Sizes Based on a (tr cos 4)
Value of 25 dynes/cm (0.025 N/m) and Equation 4-5
Threshold Entry Equivalent Equivalent
Mean Grain Size Capillary Capillary Rise of Capillary Rise
Medium Diameter (mm) Pressure (Pa) Water (cm) of Water (ft)
Clay <0.0039 > I OO,OOO > 1000
Silt 0.0039 -0.063 6300- 100,000 65- 1000
Sand 0.063-2.0 200-6300 2.1-65
Coarse sand 2.0-4.0 100 - 200 1 .o- 2. I

From Huhhcri (1053).

336 Chapter 19 / Multiphase Fluid Systems

sure gradients is (Muskdt and Meres, 1936) SW,.

0.0
1.0 I I I I I I
ki I I I
= -- (VP, - PigVh) (19.5) I to = wetting I
PI #I = nonwetting i
I
where qiis the flow of the ith fluid per unit area of the I
I
medium, k, is the effective permeability of the medium I
I
to the ith fluid, p is viscosity, P is pressure, p is density,
g is the gravitational acceleration constant, and h is eleva-
tion. The relative permeability to the fth fluid is
k
krl= -! (19.6)
k
where kl as before is the effective permeability and k is
the intrinsic permeability. Substitution of Eq. 19.6 into
Eq. 19.5 yields the general form of Darcy’s equation for
multifluid flow (Demond and Roberts, 1987)

(19.7)
0.0 0.5 1.o
In multifluid systems,k,.ranges between zero and one. %
The product kk,,represents the reduction in the intrinsic Figure 19.5 Typical relative permeability of curves (from
permeability because two or more liquids are present in Dernond and Roberts, Water Resources Bulletin, v. 23, p.
the system. With a single-fluid system, krl is one arid 617-628, 1987. Reproduced by permission of h e r . Water
Eq. 19.7 reduces to the more familiar form of Darcy’s Resources Association).
equation. Exactly how kHvaries between zero and one
is a complex function of the relative saturation, whether
the fluid is wetting or nonwetting with respect to the Figure 19.5 shows typical curves of relative permedbil-
solids, and whether the system is undergoing imbibition ity as a function of the saturation for wetting and nonwet-
or drainage. The role of relative saturation is shown in ting fluids. Fixing the relative saturation of the wetting
Figure 19.4.Note that at 100%saturation with NAPL, the fluid (for example, S,.= 0.8)implicitly defines the relative
relative permeability is one but decreases to zero as the saturation of the nonwetting fluid (for example, S,, =
water content within the pores increases. Similarly, at 0.2). For these relative saturations, k,,, = 0.47 and k,, =
100% saturation with water, the relative permeability 0.04 (for the imbibition curve, Figure 19.5).
is one. There are several features common to most relative
permeability curves (Demond and Roberts, 1987). First,
when both fluids are present, the relative permeabilities
rarely sum to one. For our example saturations, the sum
is about 0.5. Second, as shown in the figure, the relative
permeabilities of both the wetting and nonwetting fluids
approach zero at finite saturations. In other words, some
quantity of either wetting or nonwetting fluid in the pore
system cannot move below some saturation threshold.
In Figure 19.5, the relative permeability of the wetting
fluid becomes zero at S,,.= 0.2 and krlz= 0 at S,,,,, = 0.1.
These saturations are characteristic parameters known
as residual saturations. The pattern of residual saturation
shown by the example (that is, Sr,,, > S,,,)is observed
frequently. A fluid at residual saturation is not capable
of flow because at the low levels of saturation the fluid
is not connected across the network of pores.
Residual saturation of the wetting fluid is sometimes
called pendular saturation (Albertsen et al., 1986). The
C
Water Content fluid is held by capillary forces in the narrowest parts of
0) 4100% the pore space. Figure 19.6a provides an example of
NAPL Content
lOO%t. 10 pendular saturation with water as the wetting fluid at
Figure 19.4 Water-NAPL relative permeability. residual saturation and air as the nonwetting fluid. Resid-
 19.1 &sic Concepts 397

leum studies authors typically present only imbibition

curves for water displacing oil (Demond and Roberts,
1987). Hysteresis is evident because the nonwetting fluid
occupies a different pore network during imbibition than
during drainage.
The shape of the relative permeability curves depends
on several variables such as (1) intrinsic permeability,(2)
(a)Pendular Saturation ( b ) Insular Saturation pore-size distribution, (3) viscosity ratio, (4) interfacial
tension, and ( 5 ) wettability. Space is not sufficient here
Ffgure 29.6 Examples of different types of fluid saturation.
In (a) the wetting fluid is at residual saturation; in (b) the for us to discuss these factors. Readers can refer to an
nonwetting fluid is at residual saturation. excellent article by Demond and Roberts (1987) and the
text by Dullien (1979).
The volumetric retention capacity R, has been intro-
ual saturation of the nonwetting fluid is sometimes called duced by de Pastrovich and others (1979) as a measure
insular saturation. This saturation is illustrated in Figure of the potential of the unsaturated zone to trap NAPL
19.6&,where a small quantity of organic liquid is at resid-
R, = 1000 S,n (19.8)
ual saturation in a water-wet aquifer. The fluid occurs as
an isolated blob in the center of the pore. where R, is liters of residual NAPL per cubic meter of
A common situation is for water, air, and a contaminant media, n is porosity, and S, is the residual saturation of
such as oil to be found together in the unsaturated zone. the NAPL. Retention capacity values and residual satura-
Water is the wetting fluid, and air is the nonwetting fluid. tions in the unsaturated zone generally increase with
Oil has intermediate wetting properties. It is nonwetting decreasing moisture content, effective porosity, and in-
with respect to water but wetting with respect to air trinsic permeability. Residual saturation values in the un-
(Wilson and Conrad, 1984). The oil at residual saturation saturated zone generally range between 0.1 and 0.2 (Co-
is caught as blobs and pendular rings, trapped between hen and Mercer, 1993). In the saturated zone, where
the water in the small parts of pores (Wilson and Con- NAPL is usually the nonwetting fluid, residual saturation
rad, 1984). values range from about 0.1 to 0.5.
Because a contaminant such as oil behaves differently
in the unsaturated versus the saturated zone, the residual
Solubiltty and Enective Solubility
saturations differ. For uniform glass beads packed in a
column, the residual saturation is two to five times larger Solubility refers to the quantity of mass that will go into
in the saturated zone, depending on bead size (Table solution per unit volume of solution, such as mg/L, and
19.2). is frequently taken as the maximum concentration that
Another feature of relative permeability curves is that can be achieved in a contaminated body. Table 19.3
for comparable saturations k,, > km. For the curves we shows that solubilities can range over several orders of
show, at S, = 0.4 and S, = 0.6, the value of k, is about magnitude. However, concentrations are seldom, if ever,
0.1 (for imbibition) with k,, = 0.04. The reason is that as high as the solubilities would indicate. Several factors
the wetting fluid occupies the smaller pores that contrib- are responsible for this, including heterogeneous flow
ute least to flow. The nonwetting fluid occupies the larger behavior, complex kinetics in the dissolution process,
pores that contribute most to flow. and mixing of waters of different concentrations. An
The last feature of note about the relative permeability effective solubility for a given component can be approxi-
curves is the hysteretic character of the nonwetting fluid. mated by multiplying the mole fraction of the chemical
Many times this behavior is not evident because in petro- by its pure phase aqueous solubility, that is

Table 29.2 Residual Saturation Data for a Hydrocarbon in a Uniform Glass Bead
Column for Three Fluid (Vadose Zone) and Two Fluid (Saturated Zone) Systems

Residual Saturation (010)

Mean Bead k
Diameter (mm) (darcy) Vadose Zone Saturated Zone

0.655 147 3 14
0.327 85 5 14
0.167 22 8 14
~~~~ ~ ~ ~ ~~~ ~~ ~~ ~~ ~~ ~~ ~ ~~ ~ ~

Modihed from Wilson and Conrad (1984). Reprinted by permission of Petroleum Hydrocarbons and Organic
Chemicals in Ground Water-Prevention, Detection, and Restoration. Copyright 0 1984. AU rights reserved.
338 Cbapter 19 / Mukipbase Fluid Systems

Table 19.3 Examples of SolubUty Limits As wiil be demonstrated later in the chapter, Raoult's
law in combination with the ideal gas law will be used
Solubility to calculate the mole fraction.
Constituent Llmit (mgiL)
~~ ~

Benzene 820 - 1800

Bromoform 30 10- 3 190 -
19.2 LNAPLs and DNAPLs
Carbon tetrachloride 785
Chlorobenzene 448-500 Consider a spill of LNAPLs or DNAPLs on the ground
Chlorodibromomethane - surface. With time, free product percolates downward
Chloroethane 5740 through the unsaturated zone toward the water table.
2-Chloroethylvinyl ether 15,000 The most important process influencing downward
Chloroform 8200 movement of the free product is gravity-driven flow. As-
Dichlorobromomethane - sume that in the unsaturated zone, water on the solid
1 , l ,-Dichloroethane 5500 grains of the soil is the wetting phase and that the non-
1,a-Dichloroethane 8690 aqueous phase liquid (NAPL) is the wetting phase with
1,l-Dichloroethylene 400 respect to air on the water film. Because water wets the
1,2-Dichloropropane 2700 solids, the NAPL does not displace the water from the
1 ,3-Dichloropropylene 2700- 2800
surface but moves from pore to pore once saturation
Ethylbenzene 152
exceeds the residual saturation. The NAPL displaces pore
Methyl bromide 900
Methyl chloride 6450-7250 water that is not strongly held and soil gas.
1,1,2,Z-Tetrachloroethane 2900 Several important factors control the flow of NAPLs.
Tetrachloroethylene 150-200 In the case of a noncontinuous source (Figure 19.71,the
Toluene 534 volume of free product gradually decreases because some
1,2-trans-Dichloroethylene boo of the liquid is trapped in each pore at residual saturation.
1 ,1,1-Trichloroethane 480-4400 Thus, if the spill is relatively small, downward percolation
1,1,2-Trichloroethane 4500 in the unsaturated zone will stop once the total spill
Trichloroethylene 1100 volume is at residual saturation. Another pulse of NAPLs
Vinyl chloride 1.1-60
is necessary to move the product downward. The main
Phenol (total)
threat to ground water from such small spills is the oppor-
Cyanide (total) 0.0054
2,4,GTrichlorophenol
tunity for continuing dissolution of the NAPL by infiltra-
Arsenic 2.1E- 06 tion, or vapor phase migration in the vadose zone.
Iron 0.0098 The NAPL in Figure 19.7 also tends to spread horizon-
Copper 0.1054 tally as it moves downward. This spreading is due to
Lead 20.58 capillary forces, which operate together with gravity
Zinc 2.0638-03 forces to control migration. The presence of layers of
Mercury 2.307E- 2 1 varying hydraulic conductivity also influences NAPL
Nickel - spread (Farmer, 1983). Even a relatively thin, low-perme-
Barium 0.0563 ability unit (Figure 19.8) will inhibit downward percola-
Cadmium tion and force the free product to move laterally. If such
Chromium 2.61OE- 04
Manganese N/A
Silver 3.468- 09
Selenium Solublc: NAPL spill at to
r . . . I

'Unstable in water.
Solubility limits for inorganics are calculated using EQ3-Pitzer.

s; = x,s, (19.9)
where S; is the theoretical upper-limit dissolved-phase
concentration of a constituent in equilibrium with
i + Product at residual saturation

ground water, X , is the mole fraction of a component i Saturated zone

in a DNAPL mix, and Si is the pure phase solubility of
component i. Thus, if the mole fraction of 1, l-dichloro- Figure 19.7 The downward percolation of a NAPL in the
ethane is 0.1, the effective solubility is (Table 19.3) unsaturated zone. As the contaminant moves downward, the
quantity of mobile fluid decreases and an increasing quantity
5500 X 0.1 = 550 mg/L is trapped as residual saturation.
 19.2 LffAPLs arid DNAPLs -
399

Source still made to true versus apparent thickness, wc will

I 01 describe this conceptual model briefly.
Assume that LNAPL occurs as a thin zone of s;itimtioii
near the top of the capillary fringe (Figure 19.9). l1ntlt.r
this condition, LNAPL product would flow into the wcll
and fill the casing from the air-NAPL interfnce to the
water table. Thus, the apparent thickness reflects the
thickness of the capillary fringe in addition t o the true
product thickness. If the product is thick. it could also
Figure 19.8 The presence of zones of low hydraulic con- push down the water level in the well. The ratio o f
ductivity within the unsaturated zone can cause NAPLs to apparent to true product thickness is considered to range
mound and spread laterally. In this case, some of the contam- from two to about four, depending on site conditions.
inant is able to move around the ends of discontinuous units More recent studies (Farr and others, 1990; Lenhartl
(modified from Farmer, 1983). and Parker, 1990; Huntley and others, 1992) havc shown
this conceptual model to be erroneous. 1nste;id of oc-
curring as a discrete layer at the top of the capillar),
a layer is continuous, the NAPL will spread only within fringe, NAPL will be distributed throughout and even
the unsaturated zone. If the layer is discontinuous, the above the capillary fringe. Water saturations within the
NAPL will eventually spill over and continue to move capillary fringe usually reniain high, and NAPL satiiratioiih
downward toward the water table. are relatively low. This pattern of saturation nieans that
the actual volume of NAPL in the capillary fringe is r e h
Conceptual Models tively small (Huntley and others, 1992). The ob.wmition
well, however, will still provide the impression of ;I much
for the Occurrence of LNAPLs
greater volume of NAPL present than is likely there.
Chapter 17 described a conceptual model for LNAPL Figure 19.10 from Huntley and others (1 992) illustriites
flow in the subsurface (Figure 17.la). As LNAPL flowed this alternative conceptualization. The LNAPI. distrihu-
downward toward the water table, it would encounter tions actually observed in cores collected from ;I fine-
the capillary fringe. Because of increasing water satura- grained unit are different than expected from the well
tions, relative permeability to the LNAPL declines, and measurement. Furthermore. the NAPL saturations ;it this
buoyancy forces become important. Thus, less dense flu- site are quite low, indicating a lack o f mobility. Thus.
ids (especially gasoline, kerosene, and other fuels) will the more realistic model for NAPL saturation would pro-
float at the water table. vide for much less mobility than is suggested for high
Classical models held that the free product would accu- NAPL saturation within a single layer (Huntley and oth-
mulate along the top of the capillary fringe (Figure 19.9). ers, 1992).
Once some critical thickness was achieved, NAPLs would In recent years, there has been interest in using a1tern;i-
flow down-gradient just above the water table. This con-
ceptual model for the behavior of LNAPLs was useful
because it provided a simple explanation about "true"
versus "apparent" thickness of NAPLs at the capillary
- 200
C
Borehole 11

fringe (Figure 19.9). As we will see, this concept has

outlived its usefulness. However, given that reference is

I I Unsaturated zone

fringe .,"
0.00 0.05 0.10 0.15 0 20 0.25 0.30
table Hydrocarbon Saturation

Saturated zone /Screened sampling well Figure 19.10 Measured hydrocwbon satur;itioii protile
compared to LNAPL distributions obtained froni mi olwwa-
Ffgure 19.9 The concept of true and apparent thickness tion well (From Huntley and others. 1992). Reprinted hy per-
for LNAPLs. Product near the top of the capillary fringe mission of Conference on Petroleum Hydroc;irhoiis and Or-
moves down to the free surface in the well, making the ap- ganic Chemicals in Ground Water-Prevention. Iktcction.
parent thickness much greater than the true thickness. and Restoration. Copyright CG 1992. All rights resrmetl.
4 4 Cbapter 19 / MuUipbase Fluid Systems

tive approaches to obtain information on LNAPL distribu- 70r

tions in the subsurface. The most promising approaches wrs
involve well tests (Gruszczenski, 1987; Hughes and oth-
ers, 1988; and Sullivan and others, 1988) or dielectric
well logging (Keech, 1988).
v 2 5 hours
'\
'\.
/25

.'
0
h'

The flow of NAPLs also depends on the way the spill

occurs (Farmer, 1983).Releasing a relatively large volume
of contaminants over a relatively short time causes rapid
downward and lateral migration (Figure 19.1la). Spread- /17 hours

ing is maximized and a relatively large volume of residual

i
contamination remains in what Farmer (1 983) refers to
as the descent cone (Figure 19.1la). With slow leakage
I
-7
over a long time, the contaminant moves along the most
permeable pathways (Figure 19.116). These pathways
can be a single channel or a more complex array of
smaller channels, arranged in a dendritic pattern. Both
the extent of lateral spreading and the volume of product
V = Water-table elevation
I
0.1
I
0.2
I
0.3
I
0.4
I
.5
Organic Liquid Saturation
held at residual saturation are considerablyless than with Figure 19.12 Results from a laboratory experiment show-
a large volume spill. Overall, more of the mobile liquid ing how NAPL distribution changes in response to water-
will reach the water table from slow leakage. table rise (from Lenhard and others, 1993). Reprinted from
The way LNAPLs interact with the capillary fringe de- J. Contam. Hydrol., v. 12, by R. J. Lenhard, T. G.Johnson,
pends on the rate at which product is supplied (Farmer, and J. C. Parker, Environmental observations of non-aqueous-
1983). A slow rate of supply has little effect on the capil- phase liquid subsurface movement, p. 79- 101, 1993, by kind
lary fringe or the configuration of the water table (Figure permission from Elsevier Science, NL, Sara Burgerhdrtstraat
19.1la). Alternatively, a large volume of fluid, reaching 25, 1055 KV Amsterdam, The Netherlands.
the capillary fringe over a relatively short period of time,
collapses the fringe and depresses the water table (Figure
19.116). The extent of depression depends on the quan-
tity of product and its density. Even with this loading,
the fluid occurs mainly within the capillary fringe
(Figure 19.11). For spills that reach the capillary fringe,
spreading continues until the total spill is at residual satu-
ration.
Another feature of a ground-water system that influ-
ences LNAPL distributions is water-table fluctuations.The
complexities in LNAPL saturation outlined above and the
&&&011I I - lack of mobdity may conspire to trap LNAPLs below a
Water table Capillary fringe ' rapidly rising water table. Lenhard and others (1993)
present experimental results showing the effect of water-
( a )Sudden, Large-Volume Loss
table rise over an LNAPL zone. Figure 19.12 shows how
even in a relatively simple system a rising water table
can trap significant LNAPL. Lenhard and others (1993)
estimate that approximately one-half of the product pres
ent before the water table rose remained trapped. Even
larger volumes may end up trapped if the water table
rises into low-permeabilityunits, Continuing water-table
t fluctuations over a long time will redistribute the free
product across a larger zone.
( 6 ) Slow Leak The possibility for LNAPL to occur below the water
table is illustrated by detailed saturation data developed
Figure 19.11 Pattern of flow is determined by the nature
of the spill. In (a),a sudden, large-volume loss results in max- for a single transect at a crude-oil spill near Bemidji,
imum spreading, large residual saturation, and collapse of the Minnesota (Hess and others, 199213). The distribution of
capillary fringe. In (b) a slow leak causes the product to fol- LNAPL at this site appears to be influenced by sediment
low a set of channels, a large volume of product to reach heterogeneities and water-table fluctuations. In particu-
the water table, and minimal disturbance of the capillary lar, Figure 19.13 illustrates how LNAPL may be trapped
fringe (modified from Farmer, 1983). due to water-table fluctuations. Possibly, high-hydraulic-
 19.2 LNAPLs and DNAPLs 401

981 Observation well and identifier

I

cI
Oil thickness in wells installed in 1983

+ Center of fluid-content sample

Ffgure 29.13 Longitudinal trdnsect through the southern oil pool at the Bemidji site.
Shown on the figure is the location of core samples, the oil thickness in wells, and the oil
content as cni’/cm‘ (from Hess and others, 1992). Reprinted from J. Contam. Hydrol., v.
10, by K. M. Hess, W. N. Herlkelnth, and H. I. Essaid, “Determination of subsurface fluid
contents at a crudeail spill,” p. 75-96, 1992 with kind permission from Elsevier Science,
NI., Sara Rurgerhanstraat 25, 1055 KV Amsterdam. The Netherlands.

conductivity pathways in the vicinity of the water table plexities added by fracturing. Thus, in complex geologic
induced the downward flow of LNAPLs (Hess and 0th- settings, predicting migration pathways for DNAPLs CJn
ers, 1992b). be relatively difficult.

Secondary Contamination Due to NAPLs

Occurrence of DNAPLs in Ground Water
I

NAPLs in ground water can serve as important sources

For a simple geologic setting, I>NAPLshave the potential
of secondary contamination. Problems develop when or-
to move downward to the base of the aquifer. Dow-nward-
ganic contaminants present as the free or residually satu-
moving DNAPLs displace water because they have a spe-
rated products partition into the soil gas through volatil-
cific gravity much greater than that of water. DNAPL
accumulating on low-permeability units will move down-
hill following the topography of the boundary (Figure Source Ground surface
19.14). This flow in many cases will be in a direction
that is different from that of the ground water. Spreading
Ill
continues until the spill is at residual saturation. Within
both the saturated and the unsaturated flow systems,
these zones of residual saturation are a source of dissolved Direction of
contamination as long as DNAPL remains. As was the case ground-water flow Contaminant at
___t
--<esidual saturation
with LNAPLs,this simple conceptual model disguises the in descent zone
complex patterns of saturation in zones of “free
product. ”
Not surprisingly, few sites are as simple as that de-
Low permeability unit
picted in Figure 19.14. Heterogeneities of all kinds may Mobile liauid
occur within the saturated ground-water system to prc- contaminant
moving down dip
duce complex patterns of NAPL distribution. Figure 19.15
illustrates how subtle variations in permeability and the ~i~~~~29.14 pattern of DNAPL flow. In this case, the liq
attitude of heterogeneities can lead to complex distribu- uid moves along the bottom of the aquifer in a direction th:it
tions of DNAPI, “pools”. The figure also illustrates corn- is opposite to that of the ground water.
402 Cbapter 19 / MuUipbase Fluid Systems

Source ,Volatile organic spill

I \

41.. Volatilization

Residual
saturation Ground-water flow

- DNAPL
__t

Ground-water
\ \ I - Figure 19.16 NAPLs in the vadose zone as a source of
ground-water contamination. Volatilization of organic com-
pounds leads to vapor phase migration and dissolution into
the ground water (modified from Mendoza and McAlary,
1990). Reprinted by permission of Ground Water. Copyright
DNAPL flow 0 1990. All rights reserved.
along fractures

Figure 29.15 Complex distribution of DNAPLs in the vicin-

ity of a surface source. Heterogeneities in hydraulic conduc- chlorinated solvent vapors moving downward and away
tivity cause pooling. At depth, DNAPL flows into the bed- from a spill (Mendoza and Frind, 1990a). What is surpris-
rock along fractures. ing about this transport is the speed and extent to which
volatiles are able to spread through the soil gas system.
Simulation results presented by Mendoza and Frind
ization or into the ground water through dissolution. (1990a) showed vapor migration of up to several 10s
Figure 19.16 illustrates how even a small volume spill of meters in a few weeks. This tendency for volatile
of a volatile organic liquid in the unsaturated zone can NAPLs to contaminate soit gas will be enhanced as the
produce a plume of dissolved contaminants more signifi- spill grows in size and spreads along the capillary
cant than the original spill (Mendoza and McAlary, 1990). fringe.
Volatiles spreading laterally and downward away from NAPLs in the subsurface also have the opportunity to
the source can partition into available soil moisture. dissolve into mobile water that moves through the zone
Through continued infiltration, this water ultimdtely con- of contamination. Figure 19.17 illustrates the plume of
taminates the aquifer. dissolved contaminants that could develop in conjunc-
Mendoza and Frind (1990a,b) simulated the vapor tion with a free LNAPL moving along the capillary fringe.
transport of volatile organic solvents in the unsaturated Figure 19.18 shows the plume from DNAPLs at residual
zone. Their model considered (1) transport due to diffu- saturation within the cone of descent and from the free
sion and advection related to density gradients and vapor product. Experimental studies by Anderson and others
mass generation at the source, and (2) attenuation related (1987) and work they review show that water moving
to dissolution into soil moisture. Of particular importance through DNAPLs at residual saturation requires only
is densitydriven advection caused by density gradients 10 cm or so of contact distance before saturation is
that may develop with dense (that is, relative to soil gas) reached in the ground water. Although the solubilities

’- Drssolved Contaminants
Direction of GmnctWatesflow
Ffgure 19.17 LNAPL dissolves into flowing ground water to form a contaminant plume
(from Walther and others, 1986). Reprinted by permission of Conference on Petroleum Hy
drocarbons and Organic Chemicals in Ground Water-Prevention, Detection, and Restora-
tion Copyright 0 19%. All rights reserved.
 19.2 LNAPLs and DNAPLs 403

along fractures
Figure 29.18 DNAPL dissolves into flowing ground water from residually saturated free
product and pools to form a contaminant plume.

of volatile organic compounds (for example, benzene, porosity of 25%, a residual saturation of 0.1, a DNAPL
carbon tetrachloride, trichloroethane) in water can be density of 1500 kg/m', and a gravitational constant of
relatively low, the quantity of dissolved contaminant in 9.8 m/d.
water is often several orders of magnitude greater than Consider a spill volume of approximately 0.3 m'.
the current standards permit (Anderson and others, The volume of aquifer containing residual is
1987). Thus, these secondary sources of contamination
are important in a regulatory sense.

Conceptual Models and Assuming the spill took place over a 2 m X 2m

Quantitative Methods area, the depth containing residual is
Several points have been brought out in the discussions
above, largely with regard to the movement of NAPLs in
the subsurface, heterogeneity, and residual saturations
that hinder this movement, NAPLs as a source of second- If this spill took place over this same area containing
ary contamination due to dissolution and volatilization, a 5OOmicron fracture (5 X lo-' m), the depth of
and entry pressures leading to displacement. These have residual in the fracture is
been discussed largely from a qualitative point of view
because they are complex processes. However, if some L=
v*,a
assumptions can be made, some of these observations
-
S, frac-thick frac-length
may be quantified as long as we keep in mind that a -
- 0.3 m'
complex heterogeneous situation is being replaced by a 0.1 X 5 X 10-.'m X 2 m
geologic medium that is assumed to be the simplest type
conceivable. Researchers at the Waterloo Centre for = 3000m
Groundwater Research have developed a series of illustra- What is the height of a DNAPL pool supported by
tive conceptual models that illustrate these processes a low-permeability layer with an average pore size
in differing geologic environments. Cohen and Mercer of mm m)?
(1993) address these methods within the framework of
conceptual models that lead to a series of questions to 2 u cos 4
HD =
be addressed. We follow their lead, in part, by posing a rhS,g(P,l - PI,.)
given situation and presenting some simple equations 2 X 0.04 N/m X 0.848
that describe some of the more pertinent aspects of the -
-
10-" m X 9.8 m/s2 X (1500 - 1000) kg/m.'
situation. Because of space limitations, these equations
are summarized in Table 19.4. = 13.8 X N s'/kg = m
(Note: A Newton is a force and is equal to
Example 19.1 mass X acceleration)
In all the problems below, assume an interfacial tension A DNAPL pool is 2 m in width, 2 m in height, and
of 0.040 N/m, a wetting contact angle of 32", an effective 10 m long" and has an average
" concentration of 100
 Next Page

404 Chapter 19 / Multiphase Fluid Systems

Tub& 29.4 Quantitative M e t h o d s (In the equations below, 0‘ is the interfacial tension, t$ is the
wetting contact angle, r is a pore radius, p is a density of NAPL n or water w , g is the acceleration
due to gravity.)
1. Volume of aquifer containing residual

2. The depth L of the residual given

above
3. Depth of residual in a fracture

4. Height of a DNAPL pool that can be

supported by low-permeabilitylens
5 . Height of DNAPL pool that can be If residual DNAPL overlies the DNAPL in
supported by a fractured capillary the saturated zone, replace r by the
barrier aperture b in the equation above
6. Thickness of DNAPL on the capillary
fringe to cause DNAPL to enter the
saturated zone
7. Critical DNAPL thickness for entry Replace r by the aperture size b in the
into a fracture at the top of the above equation
saturated zone
8. The time required to dissolve a t=- M
DNAPL pool in the saturated zone, vn,C,,.A
where M is mass, v is the average
ground-water velocity, C,. is the
dissolved DNAPL concentration, and
A is the cross-sectional area of
DNAPL through which ground water
flows

g/m3 (mass = 400 g). Ground water with a DNAPL Gasoline occurs at the water table beneath and east
concentration of 10 g/m3 is moving through this of Station A, but apparently bypasses areas where the till
pool at a velocity of 4 X m/s. The time required is at elevations above the water table (Figure 19.20).
to dissolve this pool is Thus, the spill, while generally moving down the gradient
of the water table, is strongly influenced by the irregular
t=--M - 4000 g
topography of the till surface. The shaded areas (Figure
vn,C,.A 4 X m/s X 0.25 X 10 g/m’ X 4 in2
19.21) illustrate places where the surface of the till is
= 1 X 10ns= 3yr above the water table. The arrows are pathways of migra-
tion where the water table is located in gravel. This
complex pattern of flow explains why gasoline vapors
A Case Study of Gasoline Leakage were detected in only some of the buildings and why it
was difficult to establish whether other stations were
The first case study (from O’Connor and Bouckhout, leaking. It also added to the difficulty in designing the
1983) concerns a spill of gasoline from underground collection system.
storage tanks in an urban area. This case is typical of
the large number of spills involving petroleum products.
More than 75,000 liters of gasoline leaked from buried
underground storage tanks at Station A and possibly oth-
Hyde Park LandlJlU Case Study
ers nearby (Figure 19.19). The spill was first discovered The second case study, taken from Faust (1985) and
when fumes leaked into sanitary sewers and the base- Cohen and others (1987), examines the complex migra-
ments of houses. In the western half of the study area, tion of both DNAPLs and associated aqueous phase liq-
from 4 to 8 m of gravel and sand overlie till (Figure 19.20). uids (APLs) from the Hyde Park Landfill at Niagara Falls,
To the east, the gravel is thinner and capped by a thin New York (Figure 19.22). Disposal of chemical wastes
silt or fine-sand layer. began at the site in 1954 by the then Hooker Chemical
Previous Page

19.3 Partitioning 405

The landfill also contains an estimated 0.6to 1.6tons o f

2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD). which is ;I
potent carcinogen.
Ten meters or less of overburden overlies bedrock.
Overburden consists of silt, clay, silty clay till, and :I zone
of sand. The uppermost bedrock units (Figure 19.22)
include the Lockport Dolomite, Rochester Shde. and Iron-
dequoit and Renales limestones. The Lockport I>olomite
ranges in thickness from approximately 40 ni ininiedi-
ately southeast of the landfill to approximately 20 111 ;it
the Niagara Gorge. Bedrock dips southward.
Ground water flows in a northwesterly direction to-
ward the Niagara River in both overburden and bedrock.
Service stations
The hydraulic head is higher in the overburden than the
Lockport Dolomite, providing a downward conipotient
of flow. The fractured upper 3 to 10 ni of the Lockport
Dolomite provide a highly permeable pathway for
ground-water flow. Deeper parts of the Lockport ;ire
fractured but less permeable. The Rochester Shale is con-
sidered to be a low-permeability barrier to downw;ird
0 Monitoring boreholes 50 0 50
1 - 4 flow. Ground water discharges into the Niag;it;i River
0 Recovery wells m
along the gorge.
Artificial vents (active)
Figures 19.23~ and 0 illustrate the extent of contaniina-
Figure 29.29 Site plan showing the location of buried tion by DNAPLs and dissolved compounds in the overbur-
tanks, basements in which gasoline fumes were detected, den and Lockport Dolomite. By far tlie greatest m i s s
the location of monitoring and recovery wells, and vents of contaminants is represented by the DNAPLs in the
(from O’Connor and Bouckhout, 1983). Lockport Dolomite. Dense liquids from tlie landfill ;ire
localized within the Lockport Dolomite to the top o f
the Rochester Shale. Lateral spreading of both dissolved
and Plastics Corporation. Ultimately, Hyde Park Landfill contaminants and DNAPLs within tlie overburden has
received approximately 80,000 tons of chemical wastes been relatively limited. Much of the contamillation in the
in both solid and liquid forni. The wastes consist generally overburden (Figure 19.23~1) is thought to be related to
of chlorinated benzenes (21%), chlorinated toluenes surface runoff more than lateral subsurface transport.
(16%), chlorofluorotoluenes (1 1%), hexachlorocyclopen- NAPLs in the landfill and those in the Lockport Dolo-
tadiene (C-56)(75%), trichlorophenol (4%),chlorinated mite are the source for dissolved contaminants that have
organic acids, pesticides, metal chlorides, and various been transported more than 750 m in the Lockport Dolo-
other chemicals. Some of the contaminants, particularly mite. Contaminants discharged into the Niagar;i River
chloroform, polychlorinated biphenyls, and mirex, are northwest of the site. However, the DNAPLs in tlie Lock-
of concern because of their toxicity and persistence. port Dolomite are not moving in the same direction ;IS
the ground water and the dissolved contaminants. These
dense liquids are moving southward following the rc-
Southwest
Leakage at Station A
/ Northeast
gional dip of the bedrock. This complex pattern of
spreading helps to explain the differences in the DNAPL
distribution and plumes of dissolved contaminants. Het-
erogeneity due to fracture development within the upper
and lower parts of the Lockport Dolomite also results i n
localized variability in tlie contaminant distributions.

Gaioline
Sewers
-
19.3 Partitioning
We have already learned some things about partitioning
Ffgure 29.20 Cross section in the direction of spreading
showing the general geology of the site and the pattern of in Chapter 12.Included among the partition coefficients
gasoline spreading. The gasoline is moving down-gradient discussed there are Henry’s dimensionless constant, the
along the water table, moving around the till highs (modified octonal-water partition coefficient, the soil-adsorption
from O’Connor and Bouckhout, 1983). coefficient, and the distribution coefficient. These coef-
406 Cbapter 29 / Multfpbase Ruid Systems

t Arrows show till channels

0 Monitoring boreholes
50 0 50
Recovery wells
Artificial vents (active) rn

Figure 29.21 The pattern of gasoline migration in relation to the gradi-

ent on the water table. In general, the gasoline moves through the gravel,
avoiding areas (hatched) where the till is at a higher elevation than the wa-
ter table (from OConnor and Bouckhout, 1983).

ficients tell us something about how organics partition phase, whereas advection of the water phase is enhanced
between phases, that is, solid-liquid or liquid-gas parti- by infiltrating rain water. The gas phase can also be in
tioning. Henry’s dimensionless constant gives an under- transport, driven by diffusion and densitydriven advec-
standing of partitioning between the gas and liquid tion. In this case, there is vaporization at the residual
phases and is useful for problems involving a gas compo. source. As the vapor migrates, it partitions to soil mois-
nent. The distribution coefficient or the soil adsorption ture and is trdnsported both to the ground water and to
coefficient is useful for understanding partitioning be- the atmosphere. As noted, partitioning to air, water, and
tween liquids and solids. soil are dominant processes.
Figure 19.24 sums up the pertinent aspects of parti- Several complicating factors are not shown on this
tioning in the subsurface. Four phases can be present in diagram. First is the fact that we are dealing with a multi-
the unsaturated zone: air, water, NAPL,and solid (soil). component mixture in a transient system that is undergo-
In the saturated zone, the phases include water, NAPL, ing advection as well as diffusive transport. The mass
and solid. As noted, partitioning in all cases is reversible. transfer process may be equilibrium or kinetically con-
Partitioning to air depends on volatility and partitioning trolled in a varying temperature environment. Last, soil
to water is controiled by solubility, whereas partitioning water content may increase with depth or be irregu-
to soil depends on sorptivity. The important role of parti- larly distributed.
tioning will be brought out in the next few sections.

Volatilixation
-
19.4 Fate of Organics
Some of the concepts of significance to volatilization
in the Unsaturated Zone were discussed in Chapter 12. Henry’s constant, in partic-
Figure 19.25 schematically illustrates the active processes ular, is important in this discussion, as is Raoult’s law.
in the unsaturated zone. NAPL can move as a separate The higher Henry’s constant, the more likely the com-
 19.4 Fate of Organics in the Unsaturated Zone 407

Site Location Map

Hyde Park landfill

Figure 29.22 Location and geologic setting o f the Hyde Park landfilJ (modified from
Faust, 1085).

pound will partition to the gas phase. Among the group the soil gas may migrate in response to a diffusive process,
of compounds termed volatile organic compounds (VOC) may be absorbed and therefore retarded by the solid
are the chlorinated solvents and chlorinated aliphatic particles, may be dissolved in ground water, or may es-
and aromatic hydrocarbons. The aromatic hydrocarbons cape to the atmosphere. Densityilriven gas flow m;iy
include the water-soluble components of gasoline, ben- result in contamination of an underlying ground-water
zene, toluene, ethylbenzene, and xylene (BTEX). The body (Falta and others, 1989).The nature of the problem
partitioning of these volatile organic chemicals in the is shown in Figure 19.25.
unsaturated zone depends not only on volatility and As mentioned in Chapter 12, vapor pressure represents
solubility but on soil moisture content, the presence of a compound’s tendency to evaporate and is essentially
organic matter, and temperature. Adsorption of vapor a measure of the solubility of an organic solvent in gas.
increases with decreasing soil moisture content and in- Thus obtaining a measure of a compound’s propensity
creasing organic content. Volatilization increases with to volatilize requires Henry’s law for partitioning of dis-
increasing temperature. solved organic solutions from water to air and Raoult’s
With the movenient of a volatile NAPL through the law for partitioning between the pure solvent (NAPL)
soil zone, some will remain as an immobile residual, and air (see Problem 3, Chapter 12). On the other h;ind.
providing the potential for long-term contamination by transport of the vapor takes place by both an advectivt.
infiltrating ground water. Organic substances released to and diffusive process.
4 2 Chapter 19 / Mukipbase Fluid Systems

Zone Phases Partitioning

Unsaturated Air NAPL air ++

Water NAPL c* water

NAPL Water air

--
++

Soil Water c* soil

NAPL soil

-
Saturated Water NAPL water
NAPL Water ++ soil
soil NAPL soil
Volatility NAPL air ++

Water air ++

Solubility NAPL c* water

Sorptivity NAPL soil ++

Water c* soil
Figure 19.24 Partitioning between phases.

tion statement required here was given already in Chapter

14, modified somewhat for the presence of gas in a par-
tially saturated media
a(C8nJ
-div J,, = - at (19.10)

where J,, is a diffusive flux, C, is the gas concentration,

n8 is the part of the total porosity occupied by the gas
phase, and t is time. This equation states that the mass
outflow rate per unit volume equals the time rate of
change of mass within the unit volume. For a one-dimen-
sional system,

(19.11)

The diffusive fluxJDin one dimension is given by a modi-

fied form of Fick's law

(1 9.12)

(0)

Figwe 19.23 NAPL and dissolved contaminant plumes in Infiltration

(a) the overburden and ( b ) the Lockport Dolomite (modified S l t l l l i
from Faust, 1985).
---

Gas Transport by mflusion

Dtssaluttnn into ',
The complicating factors that arise from gas transport in sat1 nKIursturt?
the unsaturated zone are due to the fact that the system
generally consists of a multicomponent mixture of dif-
fering density in a transient state, subject to temperature Dissolution into c--
Ground Water Ground-Water Ffow
variations and reactions that may be equilibrium or kinet-
ically controlled. Additionally, the transport may take Figure 19.25 Mechanisms of contaminant transport in the
place by both advection and diffusion. In this section, Unsaturated zone (from Mendoza and Frind, Water Resources
we simplify matters by considering diffusive transport of ., 0 bv American Geoohvsi-
Res.. v. 26 no. 3. 1990. CoDvrinht .I 1 .

a single component subject to retardation. The conserva- cal Union).

 19.4 Fate of Organics in the Unsaturated Zone 4 2

where Di is an effective diffusion coefficient. Combining Porosity

Surface d
Eqs. 19.11 and 19.12 and considering the porosity n,c 0
I I I I I I
constant gives I I I t

(19.131

1 1’4
If gas transport by diffusion is subject to retardation, this I
becomes (see developments leading to Eq. 14.27)

(19.14)
Transition
where R,, is a retardation factor. zone
I
Equation 19.13 is frequently used in simulations where
the source is given a concentration of unity and the
relative concentrations are determined spatially for vari-
Capillary
ous times for both covered and uncovered ground. More fringe
realistic simulations have incorporated both advective
and diffusive transport (Mendoza and Frind, 1990a,b), I Satuiated
densitydriven gas movement (Falta and others, 1989), zone
and water-gas partitioning (Sleep and Sykes, 1989; 1
Raehr, 1987). Figure 19.26 Water-air distribution in the iiiisatiir;ited
Transport time to the saturated zone is strongly depen- zone.
dent on water saturation, giving rise to both water trans-
port and air transport regimes. Part of this can be ex-
plained by the manner in which the gas-diffiision
coefficient is treated. As mentioned, the diffusion coeffi-
(19.17)
cient in the unsaturated zone is an effective diffusion
coefficient where the diffusion coefficient in air 4 is
where pI,is the bulk density of the solids, Kt, is the distrihu-
modified to account for the effects of tortuosity
tion coefficient, taken as the product K,Jn (Chapter 121,
Di = rDx (19.15) and H is Henry’s dimensionless constant. Retardation due
to equilibrium partitioning to the aqueous phase is
where T is frequently taken from the empirical develop- given as
ment of Millington and Quirk (1961)
(19.18)
(19.16)
where n,,,is the part of the porosity containing water.
The diffusion coefficient in free air varies for different The combined form represents retardation during diffu-
constituents and can be as high as 1 X lo-‘ cm2/s for sion transport
methylene chloride to 8.11 X lo-?cm’/s for trichloro-
ethene. In most cases it ranges between 7 and 9 X lo-’ (19.19)
cm*/s. Thus, the diffusion coefficient decreases markedly
with increased saturation, that is, small values of n c .This Of the variables in Eq. 19.19, the distribution coefficient
suggests the presence of diffusion “barriers,” such as can vary over a few orders of magnitude. depending on
perched water, low-permeability material, and the capil- J w . Note also that large values of Henq‘s constant, which
lary fringe (Figure 19.26). However, with large satura- reflect a propensity to partition to the gas form. impede
tions, the contaminant can partition to the water, adding retardation of gas diffusion transport.
to the dissolved phase. Additionally, lowering the water
table in a pump-and-treat system will transform a fully
Equilibrium Calculations
saturated region into a partially saturated one, signifi-
of M a s s Distribution
cantly increasing the diffusion coefficient.
The time for diffusive transport to the saturated zone As mentioned in Section 19.3,four pliases can he present
is also sensitive to partitioning of vapor to the solid and in the unsaturated zone: air, water (dissolved), NAPL.
aqueous phases. Retardation due to partitioning to the and solid (soil). Petroleum hydrocarbons. for example.
solid phase can be expressed as can exist in the subsurface as a mobile free product.
-
41 0 Chapter 19 / Mukipbase Fluid Systems

immobile residual, vapor, dissolved in ground water, and The mass in the NAPL phase is given as the product
sorbed onto the solid particles. The distribution of mass of the residual NAPL and the density of the NAPL
among these phases will depend on the three key pro-
cesses described in Figure 19.24, volatility, solubility, and MN= n,, X p, = 0.05 X 1.48 g/cm3 X 1 X lo6cm3/m3
sorptivity, as well as the hydrogeologic conditions at a = 74,000 g/m3 (1 9.26)
particular site. Jury and others (1990) present methods
that allow calculations for the distribution of mass in the The mass in the gas phase is simply the number of
subsurface. Many of the parameters required in these moles N times the molecular weight
calculations can be found in handbooks of physical and
chemical properties of organics (molecular weight, dcn- M, = N X Mu, (19.27)
sity, vapor pressure, solubility, adsorption coefficient), where, from the equation of state for an ideal gas
whereas other parameters require field data (porosity,
residual water content, residual NAPL, mass density of N = -PV
the minerals making up the soil, normally 2.65 g/cm3 for (1 9.28)
RT
most sandy soils, and temperature). For the calculations
below, consider a soil with a porosity n of 30%,a residual where P is the gas pressure, V is the gas volume, R is
water content n,,,, of 0.1, residual NAPL nnrof O.OS, a gas the gas constant, and T is temperature. Units are critical
content of n, of n - (n, + nnr)= 0.15, organic content here. As N is the number of moles and temperature T is
f w of 0.004, mass density pI of 2.65 g/cm3, and a tempera- expressed in degrees Kelvin, the gas constant R can take
ture Tof 293°K. For trichloroethene, the following prop on various units depending on the chosen units for pres-
erties are noted: molecular weight M, of 130 g, density sure and volume
of p,, of 1.48 g/cm3, a partial vapor pressure Pv of 58
R = 0.08315 bar liter mol-' OK-'
mm Hg, solubility of 1 1 0 0 mg/L, and a soil adsorption
coefficient K, of 139 mL/g. = 8.3 15 Pa m3mol-' OK-*
The mass M, contained in the aqueous phase is deter- (19.29)
= 8.3 15 X bar m3mol-' OK-'
mined as the product of residual water saturation and sol-
ubility = 8,315 X lo6mPa m3mol-' OK-'

0.1 x 1100 mg/L x 10' L/m3 = Some common conversions include 1 atmosphere equals
Ma = nwrX S = g/n,3
1 X 103mg/g 760 mm Hg and 1 bar equals 0.987 atmospheres.
If we wish to choose pressure in atmospheres and
(1 9.20) volume in m3, the number of moles per cubic meter is
The solubility is used here whereas for a mixture, a better determined from Eq. 19.28
choice might be the effective solubility (Eq. 19.9). (58/760)atm X 0.15 x 1 m3 x lo3L/m3
The mass contained in the sorbed phase is expressed N=
0.08315 bar L mole-' OK-'X 293°K X 0.987 atm bar-'
MI = pbCs = 1.85 C, (19.21) = 0.4 moles (per m3) (19.30)
where C, is the mass sorbed on the solids (mg/g). It will The mass in the gas phase, from Eq. 19.27, is
be recalled that
MR= N X M , = 0.4 m-' X 130 g = 52 g/m3 (19.31)
Kd = K, fw = CJC, (mL/g) (19.22)
To summarize, then, in grams per cubic meter
where C ,is the concentration in solution (mg/mL) so that
M a = 110
cs = KC€fwCw (19.23)
M, = 1132
If a pure phase is present, C, is replaced by the solubility
S (or by the effective solubility for a mixture). Thus, M, = 74,000
0.004 X 139 mL/g X 110 mg/L Mg = 52
C, = K,fwS =
1 x l @ mL/L x 1 x 103mg/g Two points are worth noting here. The drinking water
= 6.12 x 10-4 (19.24) standard for this constituent is about 5 pg/L (5 X
g/m3). Dividing the mass in grams by 5 X gives the
Equation 19.21 then becomes volume of water in m3 that could be contaminated by
the mass in each phase. The second point is that most
M, = pbCr= 1.85 g/cm3 X 6.12 X X 1 X lo', cm3/m3
of the mass resides in the NAPL residual and, to a lesser
= 1132 g/m5 (19.25) extent, soil phase, two of the more difficult phases to
 19.5 Fate of Organics in the Saturated Zone 411
remediate. Vapor extraction or soil water suction tech- The partial pressure of benzene is determined
niques may be reasonably successful in dealing with the
water and vapor phases, but this hardly seems worth-
while in view of the amount that would remain. [
503 p g L-' X loy6g p g - ' X 0.0.08315 bar
L X 293°K X 0.987 atm bar-' mole-''K-'
78.I 1 g mole-'
1
Example 19.2 Calculations of mass distributiotr which equals 1.5 X atm
in the unsaturated zone The mole fraction of benzene is
This example is a little more elaborate than the preceding 1.5 X lO-'atm
discussion. Consider a total porosity of 0.2, residual NAPL = 1.5 x lo-'
0.1 atm
of 0.01, a residual water content of 0.1, and a gas content
of 0.2 - (0.1 + 0.01) = 0.09. The compound is benzene The mass of benzene in the aqueous phase is
with a molecular weight of 78.11, a density of 0.88 M,,= n,,, X S X X = 0.1 X 1780 mg/L X 1.5
g/cm3, a vapor pressure of 0.1 atm for the pure organic
solvent, and a solubility of 1780 mg/L. In the unsaturated X lo-' X IO'L/m' = 27 mg/m'
zone, the gas concentration and the concentration in the Mass of VOC in Sorbed Phase
sorbed phase vary, but we will take their average values
at 503 pg/L and 100 ,ug/kg, respectively. Mass VOC in sorbed phase = bulk density X concentra-
tion in sorbed phase = p,, X C,, where is the bulk
Mass of VOC in Gas Phase (Benzene) density (1.41 g/cm') and C, is the concentration in the
sorbed phase. The mass of VOC in the sorbed phase is
M, = volumetric gas phase content
M,= 1.41 g/cm4 X IO-'kg/g X lOOpg/kg
X concentration in gas phase
X 10"cm'/mi = 141 mg/m'
= nRX C, = 0.09 X 503 @g/L)
Mass of VOC in the NAPL Phase
X lo3(l/m3) = 45.27 (mg/m')
Mass of VOC in NAPL phase = density of the NAPL
Mass of VOC in Aqueous Phase
X residual NAPL = p,, X n,,,
Mu= residual water saturation X effective solubility
where p,, is the density of the NAPL and n,,,is the residual
= n,, X S; NAPL content. The mass of VOC in the NAPL state is
The effective solubility is determined by multiplying the M , = 0.88 g/cm4 X 0.01 X lo6 cmi/m4 = 8800 g/m'
pure-phase solubility of the organic constituent by the
mole fraction of that component in the mixture. Raoult's
law can be used to calculate the mole fraction
-
19.5 Fate of Organics
in the Saturated Zone
Organic constituents in the saturated zone can occur
where X is the mole fraction of the component of a in the dissolved, NAPL, and solid phases. The dissolved
mixture, P,,.nis the partial vapor pressure of the organic component may be the result of infiltration through a
in the gas phase, and p;lq is the vapor pressure of the residual NAPL in the unsaturated zone, dissolution of
pure organic solvent. The partial vapor pressure of the residual DNAPL in the saturated zone, or direct infiltn-
organic can be obtained from the ideal gas law tion of contaminated recharge. Movement of the dis-
solved contaminant coincides with the direction of
NRT ground-water movement (unless the density contrasts are
Po).#= -
V quite Large). DNAPL movement is far more complex. In
the capillary fringe, DNAPL will tend to spread until the
where N is the number of moles of gas, R is the propor-
gravitational pressure at the base of the DNAPL exceeds
tional gas constant (0.08315 bar X liter mole-' X OK-'),T
the displacement or entry threshold pressure. At this
is the absolute Kelvin temperature, and V is the volume
point the DNAPL will displace the water and continue ;I
of gas. This can be restated as
downward vertical movement until it reaches a capillary
C,RT barrier. The accumulated DNAPL will move in the direc-
Po, = - tion of the slope of the barrier. The height of a DNAPL
m
pool that can be supported by a low-permeability capil-
where m is the molecular weight. lary barrier was given in Table 19.4. If the low-permeabil-
4 2 Chapter 19 / Multtpbase Fluid Systems

ity lens has fractures, DNAPL migration into the fracture mL/g, and p,. = 2.65 g/cm". Equation 19.35 becomes
will commence, the potential for which increases in di-
rect proportion to overlying DNAPL thickness and Totdl mass = Cfl,X 0.3 0.56 x cfl,x 1.85
DNAPL- water density contrasts (Table 19.4). Downwml +
= 0.3 C,lm 1.036 Cll.= 1.336 C,,
migration through a capillary barrier is aided, not only
by fractures with large apertures, but by root holes, strati- Mass in sorbed phase - 1.036 clp
- = 77%
graphic windows, unsealed geotechnical boreholes, Total mass 1.336 Cl,.
poorly constructed monitoring wells, and old uncased
that is, 77%in sorbed phase, 23%in solution.
or abandoned water-supply wells.
As a consequence of the above process, DNAPL in the
saturated zone will often be present in pools supported
by low-permeability barriers, pools in large-aperturefrac-
tures surrounded by capillary barriers, and trapped as
- Air-Permeability Testing
19.6
discontinuous globules and ganglia. If the source is finite A vapor-extraction system typically consists of a blower
and the fractures more or less continuous, residual satura- connected to a series of wells completed in the unsatu-
tion can be achieved, immobilizing the DNAPL. rated zone. The performance of a venting system will
Two- and three-phase flow simulation has long been depend largely on the chemical composition of the con-
the province of petroleum engineers interested in tlie taminant, vapor flow rates through the unsaturated zone,
flow of hydrocarbons to wells in reservoirs. In more and the flow path of carrier vapors relative to the location
recent years, multiphase codes have been developed for of the contaminants (Johnson and others, 1990). Typi-
NAPL migration (see Abriola, 1988, for a review of multi- cally if the permeability to air is less than one darcy
phase codes). It is generally conceded, however, tlie m2), flow rates may be too low to achieve successful
immense data base required for simulation renders the removal in reasonable time frames (Cohen and Mercer,
use of these models impractical if the ultimate goal is 1993). Consequently air-permeabilitytests are frequently
prediction. They have, nonetheless, found some practical required in the predesign studies.
use in process analysis for uncalibrdted situations. Figure 20.26 shows a typical setup for an air-permeabil-
ity test. The objective with the test is to remove air
from the extraction well and make measurements of the
Equilibrium Calculations transient subsurface pressure distribution. Thus, the pro-
of Mass Dtstribution cedures are quite similar to those employed in well hy-
The total mass per unit volume in the saturated zone draulics. It is important here to test the interval that
may be expressed as will be vented during the actual operation. For shallow
settings, pressure in a vapor monitoring probe is de-
Total mass = amount in solution + amount sorbed scribed mathematically (Johnson and others, 1990) as

or
+ amount in NAPL

Total mass = Cl,.(n- nIJ + Cspb+ nrI,.p,,

(19.32)

(19.33)
p' =
4nm(k/p)
[0.5772 - In (s) + In t ] (19.36)

where Q is a constant volumetric gas flow rate, P' is a

where all terms have been previously defined. Recogniz- gauge pressure measured at some distance r and time t,
ing that m is the thickness of the unit, r is the distance from the
CS vapor extraction well, nR is the air-filled porosity, t is
Kc,= - (19.34) time, p is the viscosity of air (1.8 X g/cm-s), P, is
CP
l
the ambient atmospheric pressure (1 atm = 1.013 X 10"
Eq. 19.33 becomes g/cm-s), and k is the permeability to air. Note that the
Total mass = Cl,.(n- nIJ + K,,C,,,p,,+ nn,.p,,
(19.35) term & / p simplifies &pg/p, where pg is assumed equil
to one.
Calculationsare then carried out in much the same man- Johnson and others (1 990) suggest two procedures to
ner as for the unsaturated zone. analyze test results. In all cases, Eq. 19.36 indicates that
a plot of P' versus In t would provide a straight line with
Example 19.3 slope A and y intercept B where
Consider an absence of NAPL in the saturated zone for
trichloroethylene, and calculate the percentage of mass (19.37)
in the sorbed and solution phases. The properties of the
organic and soil mediums are as given for the calculations
in the unsaturated zone; n = 30%,K,, = K,,& = 0.56
B= -0.5772 - In -]):::( (19.38)
 19.7 Recognizing DNAPL Sites 413

If Q and m are known, Eq. 19.37 is used directly to water interface probes, inspection of the pumped
calculate the air permeability. This calculation is illus- fluid or samples collected by transparent bottoni-
trated with Example 20.6. If Q and m are unknown loading bailers, or depthdiscrete samplers.

k= (2)(z + exp 0.5772) ( 19.39)

3. Anomalously high dissolired concentriitions ;ire of-
ten observed at depth in the vicinity of a possible
DNAPL source. A useful yardstick for determining
During the air-permeability tests, at least one pore vol- whether the measured concentration of an organic
ume of vapor should be removed. This assures that all contaminant is anomalous requires an estimate of
the free vapor in the pores will be removed prior to the effective solubility. Recall from Section 19.1 that
setting up the vapor extraction system. Normally, extrac- for a mixture of organic compounds, the effective
tion rates will be initially very high, diminishing consider- solubility is the maximum concentration of ;I particu-
ably as the total withdrawal approaches one pore volume. lar compound that one would find at equilibrium.
These latter lower rates of extraction are somewhat indic- The effective solubility is determined using Eq. 19.9
ative of how the system will actually perform during or S; = X i s i .
the reniediat ion. One may infer the presence of DNAPLs if the dis-
The amount of vapor V, contained in one pore vol- solved phase Concentration is greater than 1% of
ume is S; (U.S. EPA, 1992). Thus, for example, if the S: for
TCE was 110 mg/L, the presence of DNAPLs woiild
V, = n,nR'H ( 19.40)
be suggested by dissolved concentrations gre;iter
where R is the radius of the contaminated zone and H than 1.1 nig/L.
is its thickness. The time to remove one pore volume
4. Soil sampling is another direct approach for NAPL
can be determined by dividing the quantity by the extrac-
identification. Soil samples collected from ;ibove and
tion rate Q.
below the water table can be examined for the prcs-
ence of NAPLs. Cohen and Mercer (1 993) discuss
- Recognizing DNAPL Sites
19.7 several approaches to NAPL detection at a site includ-
ing OVA analysis, visual detection, W fluorescence
Often, success in cleaning up a contaminated site de- analysis, and various shake or dye-shake tests. They
pends on recognizing whether NAPLs are present or not. point to UV fluorescence tests (for fluorescent
This task is not as simple as it might appear, especially NAPLs) and hydrophobic dye methods ;ISbeing most
with DNAPLs. At many DNAPL sites, one also may be effective. The rule of thumb is that the presence o f
limited in the number of wells that can be installed be- NAPLs is indicated when laboraton analysis exceed
cause of the danger of creating permeable vertical path- 10,000 mg/kg ( > I % of soil mass) (Cohen and Mer-
ways that could facilitate further product migration. A cer, 1993).
catalog of direct and indirect approaches for recognizing Another approach involves taking the nie;tsured
potential DNAPL sites has been developed mainly concentration of the organic compound in the satu-
through efforts of researchers at the Waterloo Centre for rated soil (pg/g) and determining w1i;it the tlieoreti-
Groundwater Research. Following below are the most cal concentration would be if the cont;iniin;int were
commonly used criteria to establish whether pure-phase present as a dissolved phase (Feenstra and others,
organic compounds are present. 1991). The calculation, however. needs to ;iccoiint
1. Site use and histoy ,as compiled during a site investi- for the possibility of sorption onto the solid p1i;ise.
gation, provide useful indirect evidence of the possi- The theoretical porephase concentr:ition is &'r1vcn ;is
bility that DNAPLs could be discovered. Clearly, if (Feenstra and others. 1991)
records show that DNAPLs have been used at a site,
or the type of industrial processes indicate an applica- (l9.41)
tion for solvents, then these conipounds could show
up as contaminants. A following section will illustrate where C , is the theoretical conce1itr;ition in the pore
how this approach has been systeniatized for site water, M is the total mass of the contaminant mea-
studies. sured in the soil sample (pg/g d ~\\eight).
, K,, is the
2. Ground-water zone sampling f o s free product in- distribution coefficient (cm'/g). n,, is the w;itcr-filled
volves sampling of monitoring wells to detect free porosity (volume fraction), and p,, is the bulk density
product. DNAPLs, present, say, in pools, can be dis- (g/cni3). One compares the theoreticill concentra-
covered when product is sufficiently mobile to flow tion (C,,) to the effective solubility (S;).If C,, > S: ,
into and to accumulate in monitoring wells. Ac- then DNAPLs are likely present. If C,, < S:, then
cording to U.S. EPA (19921, the presence of DNAPLs DNAPLs are likely not present. Following is an ex;ini-
in monitoring wells can be confirmed by NAPL- pie that illustrates this calculation.
414 Cbapter 19 / Multipbase Fluid Systems

Example 13.4 (1) C, = 3000 and Cu,= 780, which is about the same
as Sy, possible DNAPL site
A solvent spill is comprised of 75% TCE (mole fraction) (2) C, = 300 and C,,, = 78, which is much less than
and 25% PCE. Soil samples from the site contained TCE S;, not DNAPL site
concentrations of 3000,300, and 30 mg/kg. The organic (3) C, = 30 and C,,, = 7.8, which is much less than
carbon content of the soil is thought to be relatively high. S;, not DNAPL site
Using values for the unknowns given below, determine
which of these three sampling sites is likely to have TCE
present as DNAPL.

S; = (0.75)(1100 mg/L) = 825 mg/L Systemic Screening Procedure

= 200 and log K,,,,,
KO,,, = 2.3 The U.S. EPA has developed a systematic approach for
identlfying the possibility of DNAPL contamination at a
log K,,= 2.3 - 0.21 = 2.09 site. It is based on the screening criteria and calculations
K,< = 124 mL/g; plj = 2.0 g/mL outlined above. The approach combines historical infor-
mation about site use (Figure 19.27) with site-specific
J x = 0.03 mg/mg; O,,, = 0.30 information (Figure 19.28) to form a summary evaluation
Kd = 124 * 0.03 = 3.72 mL/g (Figure 19.29). This evaluation provides three categories
including: (I) DNAPLs are confirmed or the potential is
c,* 2.0 high, (11) there is a moderate potential for DNAPLs, or
c, =
[3.72 2.0 + 0.301 = O.26Cl (111) there is a low potential for DNAPLs at the site.

[ Does Historical Site-Use Information Indicate Presence of DNAPL? 3

industry type suggest a high YES

probability of historical I
I NSTR UCTl ON S
DNAPL release?

a
c
._ < > A Doesa

process or waste
practice employed at the site
suggest a high probability of
___(
YES

z historical DNAPL release?

..-
D
(see Guide 21 1
0

E
U
il

DNAPL-related chemicals

(> 10-50 drumdyear)

I NO I GO 10
Figure 19.2X

Figure 19.27 Assessment of the potential for DNAPLs to be present based on information
on how the site has been used (modified from U.S. EPA, 1992).
 17.7 Recognizing DNAPL Sites 4 2

Do Site Characterization Data Indicate Presence of DNAPL?

1

YES
observed in soil cores, or physically INSTRUCTIONS
observed in the aquifer?

cy
+
L
m
S
0
C
.-
c
D
x
I
-1 analyses of ground water YES
n 4
a DNAPL at the site?
z
a
L

8
E

Is it likely that
the existing field program could
miss DNAPL at the site?
(see Guide 6)

Extensive Field Program) I GO f 0

Figure 19.29

Figure 17.28 Assessment o f the potential for DNAPLs to be present based on site chanc-
terization data (modified from U.S. EPA, 1992).

DNAPL. Catqcry
Do Characterization Data Indicate 1
I \ Pre.sense of DNAPL? (Chart 21 I

Figure 29.29 Summary evaluation of the potential for

DNAPL occurrence (from U.S. EPA, 1992).
-
41 6 Chapter 19 / Multfpbase Fluid Systems

Problems
in the aqueous phase. Assume a molecular weight of 100
1. There are 100 g/m' of some organic constituent in g/mole.
an aquifer. The solubility of the constituent is 1000 pg/ 8. Benzene has a Henry's law constant of 5.5 X loT5
L and the drinking water standard is 5 pg/L. Calculate atm-m3/mole.From the data presented in Example 19.2,
the potential volume of water that can be rendered un- calculate the retardation of benzene due to partitioning
drinkable. to the aqueous phase. If the concentration of the water
2. Two immiscible fluids are present in the subsurface. was 200 pg/L and the mass sorbed on the solids was
The saturation of the wetting fluid is 0.8. From Figure 400 pg/kg, calculate the retardation due to partitioning
19.5, for the imbibition curve determine to the solid phase.
a. the relative permeability with respect to the wet- 9. On diagrams (a) and (b) shown in the figure for this
ting fluid problem, illustrate the contaminant distribution for the
b. the relative permeability with respect to the nonwet- organic compounds with the specified chemical proper-
ting fluid ties. Consider all important spreading mechanisms.
c. at what saturation the relative permeability of the
wetting fluid equals zero.
d. at what saturation the relative permeability of the
nonwetting fluid equals zero. , Large-volume source-long time

3. Given a spill of 0.1m.' of product, 35% porosity, arid

a residual NAPL saturation of 5% calculate the volume
of aquifer containing the residual.
4. Assuming a 10 X 10 cm cross-sectional flow path,
calculate the depth of the residual.
5. Consider an aquifer contaminated with an aqueous
phase and a sorbed phase. The organic content J K is
0.0005, the porosity is 35% bulk density is 1.6g/cm5,and
K,K.is 126mL/g. For a unit volume of aquifer, calculate the
percentage of mass in the solution (aqueous) phase and (a)Gasoline Leak over Fissured Clay
Two soluble and volatile components of concern in gasoline
the percentage of mass in the sorbed phase. benzene-log ,K = 2.04: solubility 1780 mg/L
6. Consider an organic compound with the following ethylbenzene-log KO, = 3.15: solubility 140 mg/l

properties: molecular weight I 53 g/mol, density 1.59 g /

cm', partial vapor pressure in the gas phase 90 mm Hg, , Large-volume spill-five years ago

solubility 785 mg/L. Consider the following properties

for a soil contaminated with this contaminant. Porosity
0.3, residual water content 0.1,residual DNAPL 0.05,
gas phase content 0.15, solid-phaseorganic content (J,J
0.001, bulk density 1.5g/cm', K,. 430 mL/g, temperature
20°C (293°K).Using the actual solubility (785 mg/L) in
the calculations, determine the mass in the NAPL phase,
the gas phase, the aqueous phase, and the sorbed phase
for a bulk volume of 1 m'. Note: one atmosphere = 760
mm Hg.
7. An organic solvent has a vapor pressure of 0.1atmcc
spheres. It has a solubility of 1500mg/L. Within a contam- (b)PCB Spill over Sand and Gravel Aquifer
inated soil, a gas concentration of 600 p g / L is noted. PCBs slightly soluble in water; log K, = 4 to 6; not volatile:
specific gravity = 1.5
The soil has a residual water content of 0.12.Calculate
the effective solubility and the mass of this constituent Problem 9
 Chapter 20

Remediation: Overview and

Removal Options

20.1 Containment
20.2 Management Options
20.3 Overview of Methods for Contaminant Removal
20.4 Pump and Treat
20.5 Interceptor Systems for NAPL Recovery
20.6 Soil-Vapor Extraction
20.7 Air Sparging
20.8 Case Studies in Site Remediation

Thus far, we have examined the causes of ground-water Slurry Walls

contamination, how contaminants are transported, and
how to characterize transport in terms of key parameters. Slurry walls are low-permeability barriers emplaced in
What is left for this and the following chapter are ques- the subsurface. They confine contaminants by either sur-
tions of how to correct a contamination problem. There rounding the entire spill (Figure 20.1) or by removing
are generally four main alternatives for dealing with con- the potential for flow through the source. With the trench
tamination problems: (I) containing of the contaminants method, a trench is dug through a bentonite slurry to
the desired depth (Figure 20.2). The wall is solidified
in place, (2) attenuating the possible hazard by institu-
tional controls, (3) removing contaminants from the either by incorporating excavated material with the ben-
ground altogether, and ( 4 ) treating the contaminants in tonite or by adding cement to the bentonite slurry. Walls
situ. This chapter will consider the first three in varying are from about 0.5 to 2 m wide and can be installed t o
detail. In situ methods for contaminant treatment, espe- depths of about 50 m.
cially methods for biorernediation, will be discussed in Walls also can be constructed using various specialized
Chapter 21. Table 20.1 summarizes the remedial tech- equipment. With the vibrating-beam method, 3 steel piate
nologies that we will discuss in this and the following is forced into the ground. As this plate is gradually re-
moved, bentonite is injected through the plate to fill the
chapter.
space that was created. The slurry walls created in this
way are narrower in width than a traditional wall and
-
20.1 Containment emplaced at a shallower depth (Knox and others, 1984).
Because the walls are thin and placed in segments. the
The containment options are control measures that keep barrier may leak.
the Contaminants in the ground but prevent further A California company has developed the S.M.W.’”cut-
spread. They are often standard geotechnical approaches off wall technique, which is an in situ mixing technology
that have been used historically to control ground-water to repair highly permeable zones that sometimes develop
seepage but have been adapted to problems of contanii- in conventionally installed slurry-trench walls. This ap-
nant control. proach uses three hollow-stem augers both to drill and
4 2 Chapter 20 / Remediatiow Overview and Removal Options

Table 20.1 Tabulation of the Remedial Alternatives

Discussed Here and in the Following Chapter
Containment Contaminant Removal
Slurry walls Excawtion and ex situ
Sheet pile walls treatment
Grouting Pump and treat
Surface seals and surface Interceptor syslems for NAPL
drainage recovery
Hydrodynamic controls Soil-vapor extraction
Stabilization and solidification Air sparging
Management Options In Sltu Destruction
(See Table 20.2) Biodegradation
Reactive harriers

to inject slurry. Paddles on the augers mix the bentonite The distance of grout penetration varies from site to
slurry with the native soils. Walls can be installed or site but can be relatively small, requiring closely spaced
repaired to depths of about GO m. injection holes (for example, 1.5 m). In practice, holes
are drilled in two or three staggered rows to ensure a
more or less continuous barrier. Typical grouting fluids
Sheet Pile Walks include cement, bentonite, or specialty fluids like silicate
Sheet pile cutoff walls are constructed by driving inter- or lignochrome grouts (Rogoshewski and others, 1983).
locking piles into the ground to form a continuous barrier Grouting techniques are most useful for sealing frac-
(Rogoshewski and others, 1983). Most commonly, steel tured rocks, where it is not possible to construct a low-
piles are used. The joints between conventional piles permeability barrier in any other way. The high cost and
usually leak until fines migrate and fill the joints between potential contaminant/grout interactions, however, limit
piles. A new technique, however, has been developed the applicability. In addition, most grouts can be injected
at the University of Waterloo for sealing the joints in steel only in materials with sand-sized and larger grain sizes
piles. Field tests at Canadian Forces Base Borden have (Knox and others, 1984).
shown sealable-joint pile walls to be extremely effective
in controlling contaminant migration.
Surface Seals and Surface Drainage
Most waste containment approaches require surface con-
Grouting trols to work effectively. Surface seaIs and drains are
Grout walls can also control the migration of contami- used together to control infiltration moving downward
nants. A grout wall or curtain is constructed by injecting through the contaminant source. The reduction of infil-
special liquids under pressure into the ground. The
grouting material moves away from the zone of injection,
gels or solidifies, and ultimately reduces the permeability
,-Backwile KeysTrench
of the geologic material (Rogoshewski and others, 1983).
Baekiill

Engineer? cover

Figure 20.1 Example of a dissolved contaminant plume

contained by slurry walls (modified from Knox and others, Figure 20.2 Schematic ofthe preparation of a slurry will
1984). (from U.S. EPA, 1985).
 20. I Contaimnent 113

tration passing through contaminated source materials seal is the quality of ongoing management. A seal c;in
decreases the volume of dissolved contaminants that may be damaged by erosion from runoff; cracking due to
be generated. In addition, covering and controlling the desiccation, frost effects, or subsidence; punctures cre-
drainage will reduce the possibility of exposure and ero- ated by burrowing animals or plant roots; and deteriora-
sion of contaminated materials, as well as the rate of tion of synthetic membranes by tearing or exposure to
transfer of volatiles or other hazardous materials (for ex- sunlight. Thus, seals need to he inspected and repaired
ample, asbestos) into the atmosphere. on an ongoing basis.
Surface seals are a common element in many different The surface of seals are graded to provide lateral drain-
types of waste-management problems. Given the relative age to ditches. Water is moved off the surface seal rapidly
ease of constructing a surface seal, a variety of materials to minimize infiltration, while also minimizing erosion.
can be used. Examples of cover materials include com- General specifications for the design of a landscape i n
pacted clay and silty clay; mixtures of natural soils and terms of slope lengths and gradients ;ire provided ly
stabilizers such as cement; bitumen or fly ash; bentonite Rogoshewski and others (1983). They also present guitl-
layered with other natural materials; sprayed bituminous ance in designing, constructing, and maintaining the dif-
membranes; synthetic membranes [e.g., high-density ferent elements of a network including drainage ditches.
polyethylene (HDPE) or polyvinyl chloride (PVC)] again diversions (a combined ditch and dike constructed 'Icross
L

used in combination with other materials; and finally all a slope), grassed waterways. and drainage benches (ter-
sorts of waste materials such as furnace slag, incinerator races built across long slopes to reduce the steepness).
residue, fly ash, and clinkers (Rogoshewski and others,
1983).Figure 20.3 shows examples of these surface liners
used at hazardous waste sites in North America.
Hydrodynamic Controls
The technology for sealing sites is proven. The main Pumping or injection of water can control the spread of
source of uncertainty in the long-term functioning of a contaminants by controlling hydraulic or pressure gradi-
ents. These approaches isolate the contaminants without
removing them. For example, lowering the water table
FMC Vault, Minneapolis, Msnnesola may prevent discharge of contaminants to rivers or lakes
Topsoil and reduce the rate of contaminant generation by desatur-
I ating the waste. Similarly, a plume of dissolved contanii-
-- nants can be confined in a potentiometric low cre;ited
by an appropriate combination of pumping and injection
wells (Rogoshewski and others, 1983).
The design approach is a straightforward extension
i
1
Waste Heavy metals I 121)(1 rnrn
of well-hydraulics concepts developed previously. The
containment strategy will work in niost cases given sufti-
Volatile organis
*
!
Sludge
cornpowids I cient care in designing and monitoring the operation o f
the system. Of the techniques considered so far, this
I Native ciay one requires the greatest ongoing supervision. Wells ;ind
pumps need to be kept properly serviced and pumping
L i f s r i Superfund site, W w s J e w
rates need to be adjusted to respond to changing grouncl-
water conditions. For example, above-average rech;irge
may raise water levels and destroy confinement.
Another example of hydrodynamic control is the use
40 mil of positive or negative differential pressure systems to
HDPE
control vapor migration (O'Connor and others. 19841).
With some hydrocarbon spills, the free product is not as

I much of a problem as the rapid movement o f wpors.

Even when the free product has been removed, the va-
pors may continue to be a problem at a site (O'Connor
and others, 1984a). Vapors migrating into buildings can
Bentonite
Slurry trench pose a health threat when breathed for long periods and
Cutoff wall - if accumulated to sufficient concentrations can explode.
Adding or removing air provides a way to control contmi-
Ffgure 20.3 Examples of engineered covers (from inant spread in the vapor phase. Figure 2 0 . 4 ~shows
1 how
Mclelwain and Reades, 1985). Reprinted by permission of to keep volatiles out of a structure using both positive-
Second Canadian/American Conference on Hydrogeology. and negative-pressure systems (O'Connor ant1 others.
Copyright 0 1985. A11 rights reserved. 1984a).
%4 Cbapler 20 / Remedtation: Overview and Removal Options

lbj

Figure 20.4 Control of subsurface gasoline vapors using (a) pos-

itive and (6)negative differential pressure systems (from O'Con-
nor and others, 1984a). Reprinted by permission of Conference
on Petroleum Hydrocarbons and Organic Chemicals in Ground Wa-
ter-Prevention, Detection and Restoration. Copyright 0 1984. All
rights reserved.

Stabilization and Solidification duced from I .8X 1O-l cm/s to 4.2 x lo-' cm/s following
treatment with a cement-clay combination. The method
Solidification and stabilization techniques involve treat- appears to be most effective in the stabilization of metals
ing contaminated soil (commonly lagoon or pond and organic contaminants like PCBs.
sludges) to improve the physical characteristics of the
material (for example, so that it might be moved) and to
reduce the leachability and mobility of the contaminants.
Stabilization attempts to bind the contaminants to the 20.2 Management Options
solid, effectively reducing their solubilities. By making
a more robust solid out of the contaminated material, The management options (Table 20.2) are institutional
additional benefits in material performance can be actions to avoid health-risks problems o r to shut down
achieved by reducing the permeability. a source of contamination without removing the contami-
Materials are usually processed in place with mixing nation from the ground. These alternatives are straightfor-
augers or digging tools adapted to mix the additives with ward and will not be examined in detail.
the contaminated soils. Common additives include (1)
cement, ( 2 ) lime spiked with fly ash or sodium silicate,
(3) asphalt or bitumen, or ( 4 ) various organic polymers.
In some applications, combinations of additives are used
- Overview of Metbods
20.3
together. By adjusting the quantity of additives, one can
for Contaminant Removal
provide for permanent solidification at the site (mono- Contaminant removal is usually an integral part of most
lithic treatment) or for material appropriate for transport practical strategies for site remediation. Four approaches
after treatment (loose structure treatment). Test results are commonly used including excavation, pumping, in-
indicate that the hydraulic conductivity of soils was re- terceptor systems, and soil venting.
 20.3 Overview of Methodsfor Contaminant Removal 421

Table 20.2 Management Alternatives for Dealing with Ground-Water Contamination

Management options: Management options are usually applied either to prevent further contam-
ination or to protect potential exposure points from contaminated ground water. These methods
thus focus on sources and exposure points rather than on the contaminants per se. The methods
also tend to be institutionally based rather than technology based.
Limit/temzinate aquifer use: Limits access or exposure of receptors to contaminated ground
water.
Develop alternative water supply: Involves the substitution of contaminated ground water
with alternative supplies (for example, surface-waterdiversions and/or storage, desalination,
and new wells).
Purchase alternative water supply: Includes bottled water and water imports.
Source remouul: Involves the physical removal of the source of contamination and includes
measures to eliminate, remove, or otherwise terminate source activities; could also include
modification of a source's features (for example, operations, location. or product) to reduce,
eliminate, or otherwise prevent contamination.
5. Monitoring: Involves an active evaluation program with a "wait and see" orientation.
6. Health advi.wries:Involves the issuance of notifications about ground-watercontamination to
potential receptors.
7. Accept increused risk: Involves the decision to accept increased risk; is usually a "no action"
alternative.
Modified from Office of Technology Assessment ( 1981)

Excavation and E x Situ Treatment Interceptor Systems

Excavation is a straightforward corrective action. In its Interceptor systems involve the use of drains, trenches,
simplest form, it is an exercise in digging and trucking. and lined trenches to collect contaminants close to the
The greatest problems are health risks associated with water table. Knox and others (1984) define a drain as a
potential human exposure to vapors and particles, costs line of buried perforated pipe or drainage tiles. A trench
involved, and finding an appropriate place to dispose of is an excavation that is either open or backfilled with
the contaminated soils. A variation on the method, ex gravel to promote wall stability. Typically, interceptor
situ treatment, involves removing the contaminants from systems are installed close to the ground surfdce and 1
the soils and replacing the treated material at the site. or 2 m below the water table. In the case of dissolved
Examples of these ex situ treatment techniques include contaminants, the drains or trenches operate as an infi-
soil washing, biomounding, low-temperature thermal nite line of wells and are thus efficient at removing shal-
desorption, and high-temperature incineration. Given low contamination. We will examine this approach in
our interest here in the hydrogeologic aspects of site detail in Section 20.5, particularly as it relates to the
remediation, we will not discuss these approaches recovery of LNAPLs.
further.

Soil-Vapor Extraction
Pump and Treat Soil-vaporextraction (SVE) removes VOkdtik organic com-
Pump-and-treat schemes remove NAPLS and aqueous- pounds from the unsaturated zone by passing large vol-
phase contaminants from the ground using wells. This umes of air through the zone of contamination. Air flow
technique has been successful in a variety of applications through the spill is maintained by vacuum pumping one
and site conditions. Because of its importance, we will or more wells (Figure 20.5). Surface treatment is usually
discuss issues of both design and operation of pump-and- required to remove contaminants from the ;iir beforc
treat systems in detail in Section 20.4. release to the atmosphere (Rogoshewski ; i d others.
In developing pumping alternatives, there is always 1983).
the problem of what to d o with the contaminants re- SVE is the leading method for cleaning up fuel spills
moved from the ground. Of necessity, on-site treatment and industrial solvents (that is, BETX compounds, and
is required before injecting the water to the subsurface VOCs) from unsaturated settings. It provides one of the
or releasing it to surface-water bodies. Extensive discus- few alternatives to soil removal in the unsaturated zone
sions of physical and chemical-biological treatment where pure organic liquids often occur at residua 1 ..I
situr;i-
methods are given in Knox and others (1984). tion. The method is able to treat relatively large volumes
-
422 Chapter 20 / Remediatiotc Overview and Removal Options

Vacuum much to the chagrin of those paying the bills and the
regulators.
Figure 20.6 depicts two idealized monitoring records
for a well being used to recover dissolved contaminants.
First, consider the reduction in concentration in an ideal
well that is shown as the “theoretical removal” curve.
The concentration eventually falls to zero once the well
removes a volume of water approximately equivalent to
the volume of contaminated water (Figure 20.6). Mathe-
Unsaturated
matically, the time required for a perfect pump-and-treat
zone system to clean u p a plume is given as,
Figure 20.5 Use of a vacuum system for removing volatile v7
t,.= - (20.1)
organic compounds from the unsaturated zone (from Qf
Schwanz, 1988). Reprinted from J. Hydrol, v. 100, by F. W.
Schwartz, Contaminant hydrogeology-dollars and sense, where V, is the total volume of contaminated water and
p. 453-470,1988 with kind permission from Elsevier Sci- Q, is the pumping rate from each of i recovery wells. The
ence, NL, Sara Burgerhartstraat 25, 1055 KV Amsterdam, The following example illustrates how this equation is used.
Netherlands.
Example 20.1 (adaptedfrom Mercer and others, 1990)
A ground-water plume has an areal extent of approxi-
of material with minimal disruption at the surface. Gas
mately 41,490 m2 and is being cleaned up by a three-
stations or airports with fuel spills can be operated in a
well system, with each well being pumped at 0.13 m’/
normal manner as the cleanup proceeds. SVE will he
min. Assume that the plume is 16.7 m thick in an aquifer
discussed in detail in Section 20.6.
having a porosity of 0.3. Ideally, how long will it take to
clean up this plume?
20.4 Pump and Treat
Pump-and-treat schemes are used to clean up (1) dis-
solved contaminants present as plumes, (2) NAPLs, or (3)
sites with both types of contamination. Systems consist of Pump-and-treat systems often exhibit tailing, exempli-
three major components. Recovery wells remove con- fied by the second curve in Figure 20.6. With tailing,
taminated water or NAPLs from the ground. Treatment the concentration of contaminants declines but never
equipment on the surface separates produced water and reaches zero, even after two or more plume volumes
NAPLs and removes dissolved contaminants that remain are pumped.
in the process water. Monitoring wells establish how There are three main causes of tailing: the presence
well the system is meeting design and cleanup objectives. of NAPLs, the effects of sorption, and heterogeneity in
Pump-and-treat systems are designed to effect either hydraulic conductivity (Keely, 1989). Figure 20.7 illus-
(1) substantially complete removal of contaminants from trates a common situation where a pump-and-treat system
the subsurface or (2) hydraulic control of a plume with-
out significant mass removal. With mass removal in mind,
the system cleans up the contamination within some 1
specified time. With hydraulic control systems, the sys-
tem is run in perpetuity to keep contamination from
spreading. If the recovery wells are ever shut down, the
contamination problem will redevelop because most of
the contaminant mass remains in the ground. i-
.5 \,Removal with tailing I
Tbe Problem of Pump and Treat
When used as intended, pump-and-treat schemes are a n
cost-effective alternative for cleaning up plumes of dis- -0 1 2
solved contaminants. They have received bad press in Plume Volumes
recent years because projects that started out as mass- Figure 20.6 Effects of tailing on number of plume volumes
removal schemes actually functioned as hydraulic control that need to be removed by the pumpand-treat system (mod-
systems. Sites were not being cleaned up as expected, ified fmm Mercer and others, 19900;and Keeley, 1989).
 20.4 Pump and Treat 423

/ Undetected DNAPL The efficiency of pumpand-treat systems will also be

at source
reduced in highly heterogeneous media. Over long times,
0
Dissolved contaminant dissolved contaminants will diffuse from high-permeahil-
ity zones into low-permeability zones (Figure 20.8). Once
0 Recovery well
cleanup begins, wells will rapidly remove dissolved ni;iss
from the zones of higher permeability. However, contam-
Figure 20.7 Conceptualization of a pumpand-treat system inants within low-permeability zones will be transported
for DNAPL recovery being operated without knowledge of back out only by the slow process of diffusion. Water
the existence of residual and pooled DNAPL in the vicinity pumped from the wells will contain low levels of contam-
of the source.
ination for relatively long times.
In summary, pump-and-treat systems for dissolved con-
is being used to clean up a dissolved organic contaminant taminants work best when the rate of cleanup is con-
in a large plume. In the vicinity of the source, the pure trolled by advection. If rate-limiting processes become
DNAPL remains undetected, and the bulk of contaminant controlling, such as DNAPL dissolution rates or diffusion
resides in the DNAPL phase. A relatively small proportion rates, then pump-and-treat systems will not be efficient
of the total contaminant mass is dissolved in the plume. at mass removal.
Thus, the pumpand-treat system would only remove a
small proportion of the total contaminant mass and is only Technical Considerations
effective in controlling further spreading. If the recovery with Injection-Recovery Systems
wells were shut off, the plume would form again.
This hypothetical example illustrates why it is so im- Contaminants dissolved in water are relatively easy to
portant to establish whether NAPLs occur at a site. When deal with in terms of well design. Because water con-
NAPLs are present, the source must be isolated using taining dissolved contaminants is being pumped, water
barriers or wells if a pumpand-treat system is to be effec- well technology is usually appropriate. Thus, ideas pre-
tive. A variety of other source-intensive strategies could sented in Chapter 6 on constructing water wells are di-
then be pursued to remove the NAPLs. rectly transferable. Some care is required in selecting
The possibility for contaminants to be sorbed to the casing materials and pumps that will not fail due to chenii-
porous medium also reduces the efficiency of pump-and- cal interaction with the contaminant.
treat systems. Sorption effectively reduces the velocity The cleanup of NAPLs at the water table requires that
of contaminant transport relative to water. Thus, if a the collection device accomplish two things: (1) create
contaminant is strongly sorbed, many “plume” volumes a cone of depression on the water table causing the
need to be pumped before the sorbed mass is fully re- product to flow toward the well, and (2) remove free
moved. A simple modification of Eq. 20.1 provides an product moving into the well. Wells designed for this
estimate of the cleanup time when the contaminant is purpose are usually one of three types. The single-pump
sorbed, or system with one recovery well (Figure 20.9) is the sim-
plest and least expensive (Blake and Fryherger, 1983).
ti = -
RfVJ The pump is positioned in the well so that as it cycles o n
(20.2)
C Qi

where R, is the retardation factor for the contaminant

(defined as Eq. 18.7). Let us return to Example 20.1 and
consider a case with a sorbing contaminant. W

__t ,Low hydraulic

Example 20.2
Assume that the contaminant in Example 20.1 sorbs with

-
a Kd of 0.2 mug. Assume a solid density of aquifer materi-
als is 2.65 g/cm’. Determine by how much the cleanup
Ground-water
time would be increased. flow
The retardation factor from Eq. 18.7 is given as

= 2.23 Ffgure 20.8 Idealization of difficulties in contmiiniint rc-

Thus, the cleanup would now take 2.23 years instead of covery in a heterogeneous aquifer with advrctivr tlushing
1 year. through the high hydraulic conductivity zones :ind dilfusion
from the low hydraulic conductivity zones.
424 Cbapter 20 / Remediatior Overview and Removal Options

i----s----lMinimal spacing

Bentonite - Seal (as required)

Product detection

pump Product detection

.$ ontinuous slot screen

Continuous slot
i
&’ screen

Figure 20.9 Example of a one-well, one-pump system for

(a)
recovering NAPLs (from Blake and Lewis, 1982). Reprinted
by permission of Second National Symposium on Aquifer Res-
toration and Ground Water Monitoring. Copyright 0 1982.
MI rights reserved.
- hydrocarbons

and off, both water and product are recovered. Pumping

depresses the water table and removes the floating prod-
uct. The biggest problem is that a high-speedpump emul-
sifies the oil in water. This mixture requires treatment,
which at the minimum involves separating the oil and
water. However, the water once separated contains solu-
ble hydrocarbons that could require further treatment
(Blake and Fryberger, 1983). Piston pumps or other
noncentrifugal pumps can reduce the treatment because
less fluid mixing is involved.
Other pumping schemes avoid surface treatment by
preventing the oil and water from mixing during pump-
ing. Figure 20.10 provides two examples of two-well,
two-pump systems in which the deep water well
screened below the spill lowers the water table and the
shallow well collects the free product (Blake and Fryber-
ger, 1983). However, installing two wells and two pumps
pushes up the cost per installation. The one-well, two-
pump design is a compromise design that is less expen- (b)
sive and avoids the large-volume treatment problem of
Figure 20.10 Examples of a two-well, two-pump system
the one-well, one-pump systems. Because two pumps
for the recovery of NAPLs. Panel (a)is from Blake and Lewis
and operational monitoring equipment are required in (1982) and is reprinted by permission of Second National
the same well, casing diameters need to be larger than Symposium on Aquifer Restoration and Ground Water Moni-
either of the other two types of systems. Two variations toring. Copyright 0 1982. Panel ( b ) is modified from O’Con-
on the one-well, two-pump design are shown in Figure nor and others (1984b) and is reprinted by permission of
20.11. Conference on Petroleum Hydrocarbons and Organic Chemi-
The operation of recovery systems must be carefully cals in Ground Water-Prevention, Detection and Restora-
monitored. For example, with the one-well, one-pump tion. Copyright 0 1984. All rights reserved.
 20.4 Pump and Treat 9

Water
Blake and Fryberger (1983) provide additional guidance
age on operational and safety details that need to be consid-
1np ered, especially in selecting the proper equipment and
working with potentiaIly explosive liquids.
I/ - - - II Research on the technical aspects of emplacing wells
for the collection of DNAPLs is extremely limited. Ferty
N a t w backfill and others (1 986) studied how the efficiency of removing
DNAPLs from low-permeability bedrock depended on the
pumping system. One pumping scheme involved simply
removing the accumulated DNAPL at the bottom of the
well with an air-activated purge pump. This approxh of
collecting DNAPL through gravity drainage into the well
was not successful. Once the fluid in the well was re-
moved, no other inflow occurred up to a day aftcnvard.
In their second test, a submersible pump installed higher
in the well was used to produce a significant conc of
depression. Further, they monitored inflow of DNAI’L
into the bottom of the well during pumping. I n one case.
after about 1140 min of pumping and a final dr;iwdown
of about 15 m, the level of DNAPL increased by about 1 ni.
Water table Scavenger eontmi In general, experience suggests that one-pump ancl
denrm‘m and D M ~ Dassembly -
Clean FIN
two-pump systems have a role to play in DNAPL recovery.
However, in some settings, the one-pump system may
not be the best choice or may not even work. At first
glance, the tendency for DNAPL and water to move in
different directions suggests that moving large volumes
of water may not be helpful. However, creating a situ;i-
tion for a “natural water drive” can be helpful in moving
DNAPL into a well.

Methods f o r Designing
Pump-and-Treat Systems
The design of a pumpand-treat system attempts to meet
(b) the desired cleanup objective at some minimal cost. Cost
Figure 20.11 Examples of a one-well, two-pump system is minimized by installing the smallest number of recov-
for the recovery of NAPIS. Panel (a) is from Blake and Lewis ery wells that are pumped at some minimal rate. Thus,
(1982) and is reprinted by permission of Second National
the design strategy is to minimize the overall cost of
Symposium on Aquifer Restoration and Ground Water Moni-
installing and operating the wells and treating the Water
toring. Copyright 0 1982. Panel (b) is modified from Yaniga
and Mulry (1984) and is reprinted by permission of Confer- at the surface. In practice, there are generally four basic
ence on Petroleum Hydrocarbons and Organic Chemicals in design approaches: expanding pilot-scale systems. cap-
Ground Water-Prevention, Detection and Restoration. Copy- ture-zone type curves, trial-and-error simulations using
right0 1984. All rights reserved. models, and simulation-optimization techniques.

Expanding Pilot-Scale Systems

system, the pump must be placed at the proper depth This approach involves starting recovery operations with
and the float switch adjusted so that product is being a pilot system and monitoring how the system performs.
pumped toward the end of the pumping cycle. With the Evaluation of the monitoring data provides a basis for
ma-pump systems, the water well must not be over- expanding the pilot by selecting places where the next
pumped to enable product to be pumped through the well will do the most good. In the case of NAPLs, this
water pump (Blake and Fryberger, 1983). Failure of the well could be placed where (1) thick zones of product
product pump in a one-well, two-pump system could also are not being reduced by existing pumping or (2) the
cause product to accumulate in the well and eventually to product is moving in a direction that cannot be tolemted,
reach the intake of the water pump. Part of the job of for example, toward wells, rivers, or buildings. For dis-
the control system is to shut off the water pump when solved contaminants, the pumping wells could be placed
free product is detected by sensors above the pump. where the plume continues to grow, or does not appear
426 Chapter 20 / Remediatiow Overview and Removal Options

to be shrinking in size. Injection wells are located to

provide barriers to flow or to increase the velocity of
flow to collection wells.
Systems designed in this manner rely on experience
and a cautious, incremental expansion of the network
to ensure that the remedy is effective and economical.
This approach is used commonly with spills involving
NAPLs. Although modeling approaches are available to
optimize the design, a lack of available data usually limits
the applicability of these techniques. The situation with
respect to dissolved contaminants is better with possibili- (6)
ties of using several different analytical and numerical
procedures. Figure 20.13 Illustration of (u) a steady-state and (6)a
time-related capture zone associated with a single recovery
Capture Zones well (from Blandford and Huyakorn, 1991).
Most of the quantitative design approaches use the con-
cept of a capture zone. A capture zone is the area sur-
rounding a well that supplies water to that well. In effect, Analytical Approacbes to Oeflning Capture Zones
it is the zone of contribution to the well. Figure 20.12 Javandel and Tsang (1986) describe an analytical a p
illustrates the idealized shape of the capture zone for a proach to the design of contaminant recovery systems.
well in a homogeneous aquifer. The capture zone, indi- The approach provides a way to minimize the number
cated by the shaded area, extends from the vicinity of of injection or recovery wells, the pumping rates, and the
the well far up-gradient.All contaminants in the capture distance between wells. The optimal scheme provides a
zone should be recovered by the well (Figure 20.12). capture zone that is slightly larger than the plume be-
The shape the capture zone develops results from the ing collected.
combined effects of drawdown caused by the well and We know from well hydraulics theory that a well in
uniform regional flow. In most cases, the full extent of an aquifer can be pumped only at some maximum rate
a capture zone (that is, the steady-state capture zone) is determined by the available drawdown. In terms of con-
rarely shown (Figure 20.13). Most often, a capture zone taminant removal,pumping only one well at its maximum
is described in a time context, such as the one-year cap rate may not provide a capture zone that is large enough.
ture zone or the five-year capture zone (Figure 20.13). The solution is to keep adding wells until the pumping
The water in the volume enclosed by the time-related capacity is sufficient to create a capture zone of the
capture zone will be recovered by the well over that appropriate size. However, once more than one well is
specified time. The width of the capture zone depends added, the spacing must be such to avoid contaminants
on the pumping rate. Pump the well harder and the passing between the two wells. Thus, with more than
capture zone gets wider. one well, pumping rates and optimum spacings between
wells are determined.
Javandel and Tsang (1986) use complex potential the-
ory as the basis for a simple graphical procedure to design
a well system capable of creating a capture zone of the
appropriate size. To develop a design requires the type
curves for one to four wells (Figure 20.14), and values
for two parameters-& the aquifer thickness (assumed
to be constant), and U,the specific discharge or Darcy
velocity (also assumed constant)- for the regional flow
system. The method involves the following five steps
(Javandel and Tsang, 1986).
1. A map of the contaminant plume is constructed at
m the same scale as the type curves. The margin of the
Capture zme plume should be indicated along with the direction
of regional ground-water flow.
Figure 20.12 The effects of pumping superimposed on a
regional flow system creates a capture zone (modified from 2. The type curve for one well is superimposed on the
Gorelick, 1987). Reprinted by permission of Solving Ground plume, keeping the x axis parallel to the direction
Water Problems with Models. Copyright 0 1987. All rights re- of regional ground-water flow and along the midline
served. of the plume. Approximately equal proportions of
 20.4 Pump and Treat 427

One well
inon, 1500 rThree wells I

1000 -
400

500 -
r
3 0-
f
-1000
-500 0
I I

500
I

1000
I
1500
I

2000 2500 -500 -

Meters

-1000 -
(a)

TWOwells -1500 I I I I I I I
-500 0 500 1000 1500 2000 2500 3000
Meters

1 nnn
(c)

Four wells

-500 0 500 1000 1500 2000 2500 3000

Meters
(b)

Figure 20.14 Capture-zone type curves for one, two,

three, and four wells (from Javandel and Tsang, 1986). Re- -500 0 500 1000 1500 2000 2500 3000
printed by permission of Ground Water. Copyright 0 1986. Meters
All rights reserved. (If)

Figure 20.14 Continued

the plume should lie on each side of the x axis. The required production is not feasible due to a lack
well on the type curve will be at the downstream of available drawdown, additional wells need to be
end of the plume. The type curve is adjusted so that added (go to step 5).
the plume is enclosed by a single Q/BU curve. 5. Repeat steps 2, 3, and 4 using the two, three, or four
3. The single well pumping rate @>is calculated using well-type curves in that order, until a pumping rate
the known values of aquifer thickness (B) and the is determined for each well that the aquifer can sup-
Darcy velocity for regional flow (0along with the port. The optimum spacing between wells is calcu-
value of Q/SU indicated on the type curve (TCV) lated using the following formulas:
with the equation Q
two wells:
nBU
Q = B * U * TCV (20.3)
4. Next, check whether the well can support the calcu-
lated pumping rate over the estimated time required
for the cleanup. If the production rate is feasible, 1.2Q
-
four wells:
one well is required with pumping rate Q. If the nBU
4 3 Cbapter 20 / Remediation: Overview and Removal Options

nrle Mll
Be sure to account for the interference among the
pumped wells in the drawdown calculations. The wells
are always located symmetrically around the x axis, as
the type curves show.
Reinjecting the treated water produced by the wells
will accelerate the rate of aquifer cleanup. The procedure
is the same as that just discussed except the type curves
are reversed and the wells are injecting instead of pump
ing. The injection wells should be moved slightly u p
stream of the calculated location to avoid causing parts
of the plume to follow a long flow path.
Javandel and Tsang’s rule of thumb is to place a well
or line of wells half the distance between the theoretical
location and the tail of the plume. The followingexample
taken from Javandel and Tsang (1986) illustrates how the
technique is used.

Example 20.3 PMfo

Shown in Figure 20.15 is a plume of trichloroethylene

500
(TCE) present in a shallow confined aquifer having a
thickness of 10 m, a hydraulic conductivity of m/s, E
an effective porosity of 0.2, and a storativity of 3 X s- 0
The hydraulic gradient for the regional flow system is I
0.002, and the available drawdown for wells in the aquifer -500
is 7 m. Given this information,design an optimum collec-
tion system.
Values of B and U are required for the calculation. B
-lm 1
is given as 10 m, but U needs to be calculated from the
Darcy equation:
U = K grad@) or U= X (0.002)

= 2 x 10-7m/~
Figure 20.26 The TCE plume at the matching position
Superposition of the type curve for one well on the plume with one-well and three-well type curves (modified from
(Figure 20.16~)provides a Q/BU curve of about 2500. Javandel and Tsang, 1986). Reprinted by permission of
Using this number and the values of B and U, the single- Ground Water. Copyright 0 1986. All rights reserved.
well pumping rate is
-
Q = B U TCVa or 10 X (2 X X 2500 Check whether this pumping rate can be supported for
the aquifer using the Cooper-Jacob (1946) equation to
= 5 X 103mJ/s
calculate drawdown at the well. Assume Y = 0.2 m and
the pumping period of one year
N
s=- 2.3Q log L
2 25Tt
t 4aT rzS
(20.4)

where Q = 5 X 10-3m3/s, T = KB = lo-’ mz/s, t = 1

yr or 3.15 X lo7 s, Y = 0.2 m, S = 3 X lo-’.
The pumping period represents some preselected
planning horizon for the cleanup. Substitution of the
known values into Eq. 20.4 gives a drawdown of
9.85 m. Even without accounting for well loss, the calcu-
TCE cmentration lated drawdown exceeds the 7 m available. Thus, a
0 500 1mr
n (in ppb)
w multiwell system is necessary.
Figure 20.15 Hypothetical plume of TCE (from Javandel Superimposing the plume on the two-well type curve
and Tsang, 1986). Reprinted by permission of Ground Water. provides a Q/BV value of 1200, which in turn gives a Q
Copyright 0 1986. All rights reserved. for each of the two wells of 10 X (2 X X 1200 or
 Next Page

20.4 Pumpand Treat 423

2.4 X lo-' m3/s. The optimum distance between wells zone for the well. The following example from Larson
is Q/(rBU) or 2.4 X 10-'/[n X 10 X ( 2 X lo-')] = and others (1987) will illustrate this final step.
382 m. Again, we check the predicted drawdown at each Assume that a partially penetrating well is being
well after one year against the available 7 m. Because of pumped at 21.5 gpm in an aquifer having a K,, = 100 ft/
symmetry, the drawdown in each well is the same. The day, K,, = 1.00 ft/day, and a recharge rate of 4 in./yr. By
total drawdown at one of the wells includes the contribu- examining the pathlines in plan view and cross sections,
tion of that well pumping plus the second one 382 m the capture zone can be readily defined (Figure 20.17).
away or Interestingly, the lateral extent of the capture zone is
larger than predicted by the Javandel and Tsang analysis,
while the vertical size is somewhat less (Figure 20.17).
The partially penetrating well, the recharge, and anisotro-
The calculated drawdown is 6.57 m, which is less than pism are complexities that violate the basic assumptions
the available drawdown. However, well loss should be made in the development of the capture-zone type
considered, which makes the two-well scheme unaccept- curve procedure.
able. Moving to three wells, Q/BU is 800 (Figure 20.166),
which translates into a single pumping rate of 1.6 X Simulation-Optimization Techniques
m3/s for each well. Carrying out the drawdown A promising approach to the design of contaminant re-
calculation for three wells located 1.26Q/(rBU) or covery systems involves the combination of ground-
320 m apart provides an estimate of 5.7 m for the center water simulation with mathematical optimization (Gore-
well, which is comfortably less than the available draw- lick and others, 1984;Atwoodand Gorelick, 1985;Lefkoff
down. Thus, three wells are required 320 m apart. Each and Gorelick, 1986; Gorelick, 1987). This procedure
would be pumped at 1.6 X 10--' m'/s. combines the power of numerical flow modeling to ac-
count for the complexities of the hydrogeologic setting
and the capabilities of formal optimization techniques to
This method is extremely useful, but it cannot be ap- provide the best design subject to imposed constraints.
plied to every design problem encountered in practice. Discussion of this sophisticated modeling approach is
There are assumptions of constant transmissivity, fully beyond the scope of this book. This and similar ap-
penetrating wells, no recharge, and isotropic permeabil-
ity built into calculation. We will explore the implications
of the simplifications in a later section. Plan view
West East

Model-Based Approaches for

tbe Design of Recovery Systems
For most real problems, mathematical models are used
for designing well systems for contaminant recovery. The
simplest models are based on analytical theory similar to
Javandel and Tsang (1986). but use numerical schemes
to define pathlines. An example is the WHPA code distrib-
uted by the U.S. Environmental Protection Agency
(Blandfordand Huyakorn, 1991). Numerical models, such
as MODFLOW and MODPATH. are also used for design.
The mathematical approaches rely on the modeler's ex-
perience and a series of model runs to come up with a
- -
3.0 Water level
Ground-flow direction
reasonable design. These designs are not optimized in
the formal mathematical sense, but the modeling proce-
dures are straightforward.
Larson and others (1 987) apply a threedimensional
ground-water flow model with a second program that
calculates ground-water pathlines. A finite-difference
model is used to solve for the steady-state distribution of
Containment volume
hydraulic head on a three-dimensional nodal network.
These values of hydraulic head are input to the second +Ground-water pathline Vertical exaggeration: 20x

program along with known values of hydraulic conduc- Figure 20.27 Capture zone calculated using a three-dimen-
tivity and porosity. Pathlines are constructed by particle sional ground-watermodel (from Larson and others, 1987).
tracking. Given a well pumping in a three-dimensional Reprinted by permission of Solving Ground Water Problems
system, these pathlines can be used to define the capture with Models. Copyright 0 1987. All rights reserved.
Previous Page

-
430 Chapter 20 / Remediation: Overview and Removal Options

proaches have promise of being able to overcome the an important role in determining where the individual
inherent limitations of the analytical approaches in deal- wells were placed.
ing with complex systems and the non-uniqueness of the This analysis assumed that one knew exactly what the
trial-and-errorprocedures. transmissivity field looked like. When the variability in
the transmissivity field is unknown, which is the usual
Issues in tbe Design of Capture Zones case, even more pumping capacity is required to create
The design of well systems to provide a given capture a robust design for any type of heterogeneity. Thus, igno-
zone is carried out using average values of hydraulic rance about the nature of the heterogeneity translates
conductivity, porosity, and the like. The natural heteroge- into greater pumping rates (Gorelick, 1987).
neity of geologic media is not usually considered because Another factor impacting the shape of the capture zone
at most sites there are insufficient data to define the for a given well Is the extent to which the well penetrates
spatial structure in various hydraulic parameters. How- the aquifer. Most of the simple approaches assume a fully
ever, analyses (Gorelick, 1987) have shown that not con- penetrating well irrespective of the aquifer thickness.
sidering the heterogeneous structure of an aquifer can Wells are rarely designed in this manner, especially for
lead to an underdesigned recovery system. thick aquifers. Bair and Lahm (1996) conducted an analy-
Gorelick (1987) generated a series of heterogeneous sis of the effect of partially penetrating wells on the shape
random transmissivity fields with a mean In T of - 3.22, of capture zones. They showed that neglecting the effects
a standard deviation of 1.0, and a correlation length of of partial penetration can lead to substantial errors in
10 m. An example is shown in Figure 20.18 along with assessing the abilities of an individual well to capture
a series of nodes that represent possible locations for a plume.
recovery wells. Capturing the plume shown in Figure Under a given flow regime, screening only a fraction
20.18 with an optimum system would require three wells of the aquifer can lead to flow under the well. This effect
pumping at a combined rate of 110 Us. This pumping is shown in Figure 20.19. Decreasing the percentage of
rate compares to an ideal combined rate of 96 4 s for a well penetration (80%to 40% to 10%)eventually leads
homogeneous version of the same aquifer (that is, no to the loss of containment of the plume (see Figure
spatial variation in transmissivity). Other examples of the 20.1%). A less significant but still important effect of
heterogeneous aquifer required from 2 to 58% higher partial penetration is to increase the width of the capture
pumping rates as compared to the homogeneous case. zone. The need to consider the effects of partial penetra-
Importantly, the local variability in transmissivity played tion become more important as the hydraulic gradient
increases and the aquifer becomes increasingly aniso-
tropic (that is. K, >> KJ (Bair and Lahm, 1996).

-
20.5 Interceptor Systems
for NAPL Recovery
Interceptor systems are useful in recovering NAPLs at
shallow depths. Figure 20.20a,b illustrates two designs
for open trench systems. In the first, a skimmer pump
selectively removes product entering the trench. Because
little water is removed, a barrier needs to be installed to
prevent the product from reentering the subsurface on
the down-gradient side of the trench (Figure 20.20~).
The second design is similar except that a greater volume
of water is removed to contain the product hydraulically
in the trench (Figure 20.20b). This design avoids the
impermeable barrier and speeds up product recovery
because the flow gradient to the trench is increased
(Blake and Fryberger, 1983). However, a greater volume
0 20m of contaminated water must be dealt with at the surface.
U
An alternative design involvesclosed trenches or drains
Ffgure 20.28 Simulated transmissivity map for capture
zone analysis. The closed circles represent possible locations (Figure 20.21). Product moves through gravel or drain
for one to five recovery wells (modified from Gorelick, pipes in the trench laterally toward collection sumps.
1987). Reprinted with permission of Solving Ground Water Product is removed from the sumps by pumping. Again,
Problems with Models. Copyright 0 1987. All rights re- hydraulic control keeps the product from simply passing
served. through the trench (Figure 20.21c,d). Treatment of water
 20.6 Soil-Vapor Extraction 431

Well Oil-water separator

( pr0ducI):torag.e

Skimmer

Product layer

-
Ground-water flow

(0)
--

___.t

___t
-
25
t ' L
f
1 -
Ground-water flow

(b)
-
Ffgure 20.20 Examples of interceptor trenches used t o
control the spread of NAPLs. In (a)product is recovered u.5-
100 ing a skimmer pump with an impermeable biirrier t o control
product leakage. In (b) product and water are recovered
with a floating sludge pump. Potential product leakage is
75 controlled hydrodynamically (from Blake and Frybeger.
K
I
vt
1983).

a
N

spreading or inflow to surface water bodies. Given the

complexity of multicomponent flow, there are no quanti-
tative approaches available for designing a system. The
300 200 loo 0 100 200 300 main considerations in design are to ensure that the
.r Distance (Rj trench is long enough to avoid the product niipcitinp
(rl Verticat exaggeration 4k around the end and deep enough to provide the necc's-
Ffgure 20.19 Cross sections in the direction of flow that sary hydraulic control and to avoid the water table falling
illustrate how partial penetrations of ((0 SO'!& (6) 40%, and beneath the bottom of the trench during dry periods.
(c) 10% influence the vertical extent of the cipture zone. In
these simulations, K,, = K,, with a regional flow gradient of
0.006 (from Bair and Lahm, 1996). Reprinted by permission
of Ground Water. Copyright 0 1996. All rights reserved.
20.6 Soil-Vapor Extraction
Soil-vapor extraction ( W E ) has emerged ;is the lie!. tech-
nology for cleaning up volatile organic compounds in
from the sumps again would be required. The main at- the unsaturated zone. Most often, W E is used with NhPls.
tractions of interceptor systems are their overall simplic- However, it also works with dissolved contaniinantb oc-
ity, availability of construction materials, and speed of curring in soil water or sorbed to solids. SVE uses wciiiini
installation (Blake and Fryberger, 1983). blowers in combination with soil-gas extraction wells t o
Unlike drainage systems for farmland, interceptor sys- create an air flow through the zone of contamination
tems are not commonly installed as a large areal network (Figure 20.22). The contaminants volatilize into this air
to collect product everywhere. More often, they are in- stream at concentrations dictated by their vapor pressure.
stalled across the nose of a migrating spill or at critical VOCs carried to the well are removed. The air stream is
points along the perimeter of the spill to prevent further cleaned up at the surface by carbon adsorption o r some
132 Cbapter 20 / Remediation. Overview and Reinoval Options

A B

(6)

Figure 20.23 Examples of extraction wells used for SVE

systems. Panel (a) is a single well completion (from U.S.
r EPA, 1991). Panel (6)is a multiwell completion.

in an auger hole (Figure 20.23~).The multilevel design

involves two or more screened sections completed in a
single auger hole (20.23b). The choice of design usually
1 i
depends on the vertical extent of contamination. If the
tb)
spill has a limited vertical extent, then the first design is
more appropriate. If the spill is several tens of feet thick
0 ?
or zoned, then the multilevel approach is preferable.
:;:y,-Backfill
-:- Coarse gravel
Good practice dictates that the well screens be kept
:J less than 2 to 3 m to maintain air flow through the
contaminated zones. When screens are longer than
Water table
Coarse gravel 5 m, airflow may be channeled along a few preferential
I

Pipe pathways and bypass much of the contaminated soil.

(C) (4 Monitoring wells are installed to provide vapor-sam-
piing points away from the extraction wells. Normdliy,
Figure 20.21 Schematic representation of two interceptor vapor-monitoring wells (Figure 20.24) are permanently
systems. Water and contaminants move along the feeder
installed, as opposed to, say, soil-gas probes, which are
trench to a sump (a). A pumping system moves the fluids to
the surface for treatment (b). The feeder trenches can be
inserted and removed from the ground. Specialized
filled with gravel (c) or actual drains (a)from Blake and Fry- equipment such as a Geoprobem sampling rig can install
berger, 1983). a permanent sampling point. More complex multilevel
wells can be installed using an auger rig in cases where
the contamination is thick.
type of thermal treatment (incineration or catalytic oxi- Whether installing vapor extraction or monitoring
dation). wells, care must be taken to seal the annulus above
screened sections with bentonite. Any short circuiting
Components of an SVE System
An SVE system consists of extraction wells connected to
the treatment system and the vacuum blower. Typically,
the extraction wells are one of two types (Figure 20.23).
The simplest design is a single screened well installed

Remove water LContaminant d Blovler

treatment

Unsaturated
zone
Ftgure 20.24 Example of a vadose-zone monitoring well
figure 20.22 Schematic of how an SVE system operates (from U.S. EPA, 1991).
 20.6 Soil-VaporExtraction 433

of airflow along the casing will reduce the airflow through nant and the airflow rate through the unsaturated zone.
the zone of contamination and the overall efficiency of These two factors are reflected in the theoretical contami-
the system. This caution applies especially to the conver- nant removal rate per day at a site (ReJ given as
sion of existing wells. In many projects, one might be
(20.5)
tempted to convert water-table monitoring wells to keep
drilling costs down. Often the seals in these wells may where Co,Ris the concentration of the contaminant in
be inappropriate for vapor extraction wells. the gas phase and Q is the gas flow rate. C,, depends on
Surface treatment of the vapor stream usually begins the vapor pressure for the contaminant, and Q depends
by removal of small solids and liquids. This treatment is mainly on the permeability of the medium.
usually accomplished by a “knock out” drum or vapor- For SVE schemes to work, the organic compound or
liquid separator. The next treatment step deals with the mixture of organic compounds in the subsurface must
contaminated vapor. Vapor combustion units incinerate be somewhat volatile. In other words, the rate of volatil-
the contaminants by burning. These units are most effi- ization into the moving air phase must be sufficiently
cient when vapor concentrations are high-on the order large to reduce the total contaminant mass over a mean-
of a few percent by volume (Johnson and others, 1990). ingful time frame. A rule of thumb for considering SVE
They are less economical at lower concentrations be- for a given contaminant is a vapor pressure greater than
cause a supplemental fuel needs to be added. Another 0.015 atm (14 mm Hg at 20°C). Table 20.3 lists commonly
treatment technique involves passing the vapor over a occurring contaminants that are likely candidates for SVE.
catalytic oxidation bed. These catalytic oxidation units By far, most SVE systems are used for cleaning up fuel
work best with vapor concentrations less than 8000 ppm spills. However, fuels include a spectrum of different
(Johnson and others, 1990). With carbon treatment, va- organic compounds, and each type can respond differ-
por passes through a column of activated charcoal, which ently to SVE systems. Gasoline contains a relatively large
preferentially sorbs organic compounds. This treatment proportion of volatile compounds. Thus, gasoline spills
procedure is most economical when vapor concentra- are most responsive to SVE systems. The middle distil-
tions are very low (Johnson and others, 1990). lates, such as jet fuel and diesel, contain a relatively large
The most economical way of dealing with vapors is to proportion of less volatile compounds. Thus, SVE is slow
simply release them into the air using what is called a to clean up these kinds of spills.
diffuser stack. This approach obviously does not treat When an SVE system is first turned on, it is effective
the air stream and, thus, has health and safety implica- in removing the most volatile components of the given
tions. This approach is not commonly acceptable to regu- fuel. For example, in the case of gasoline, the more vola-
lators. tile compounds, such as butane and benzene, are re-
moved in preference to the heavier and less volatile com-
pounds. As the mass of contaminants in the soil
W e n Can SVE Systems Be Used?
decreases, the remaining compounds are enriched in the
The two main factors determining when SVE systems less volatile compounds (Johnson and others, 1990).
can be used are the chemical character of the contami- Thus, with time the residual product will be harder and

Table 20.3 Llst of Contaminants That Can B e Cleaned Up

Using SVE

Volatiles: methanol
benzene acetone
toluene pyridine
xylenes tetrahydrofuran
ethylbenzene carbon tetrachloride
hexane trichloroethane (TCA)
chloroform Semivolatiles:
methylene chloride chlorobenzene
tetrachloroethylene (PCE) dichlorobenzene (DCB)
trichloroethylene (TCE) trichloropropane
dichloroethylene (DCE)
Hydrocarbons:
ethyl acetate
gasoline
cyclohexane
jet fuel
methyl ethyl ketone (MEK)
diesel
inethyl isobutyl ketone (MIBK)
kerosene
heavy naphthas
From U S . EPA (1 98%).
434 Chapter 20 / Remediation: Overview and Removal Options

harder to remove by SVE. This tendency for residual fuel Example 20.4
to remain limits the practical level of cleanup that can
be achieved at a site by SVE (Johnson and others, 1990). A soil-gas extraction well is completed in a medium sand
having an estimated permeability of 1 X lo-- cm' (that
is, 10 darcy). Assume that the radius of influence of this
Estimating Removal Rates well is 12 m (1200 cm), the radius of the well is 5.1 cm,
Application of Eq. 20.5 first requires an estimate of the the vacuum pressure in the well is 0.90 X lo6g/cm-s2(0.9
concentration of the organic compound in the vapor atm), and other pertinent parameters are given above.
phase or C,,, . This calculation was illustrated with Exam- Calculate Q/H for this well.
ple 12.3. For a simple mixture, the only information that Substitute known values in Eq. 20.6 to give
needs to be provided is P(,rx,the vapor pressure of the
10-7
pure organic solvent, the formula weight of the contami- = 3.14 1.8 0.9
nant, and the mole fraction of the organic solvent. H
The airflow rate cQ> is determined by the permeability (1 - (1.0 X 10'/0.9 X lo")?
of the porous medium and the vapor extraction rate. x 10"
ln(5.1/1200)
Given that constraints exist on the ultimate vacuum that
can be achieved in a well (and hence the extraction rate), (1 - 1.23)
= 1.74 X X 0.9 X 106
permeability is the more important variable. Generally, H -5.46
SVE systems work best in materials, such as sand and
gravel, that are highly permeable. U.S.EPA (1991) pro- = 6.6 X 10' cms/cm-s = 0.4 m'/m-min
H
vides the following guidance in relating the potential
success of an SVE system to soil texture:
To make this calculation easier, Johnson and others
Stop and Evaluate Carefully-clay, silty clay, silty clq (1990) have provided the nomogram shown in Figure
loam 20.25. It provides estimates of Q/H in standard volumet-
Some Difficulty-sandy clay, clay loam ric units p / H , which are equal to Q/H(P,,./Pl,,,,J. The
nomogram assumes R,is= 5.1 cm (2 in.) and R, = 12 m. For
Good-sandy clay loam, silt loam other conditions, the following simple corrections apply
Very Good-sand, loamy sand, sandy loam, loam

Creating appropriate vapor flow rates through rocks or

fine-grained sediments requires a natural fracture net-
work connected to wells by hydraulically induced frac-
tures. Increasing water saturations reduces the perme-
ability of the medium to gas flow because of the effects
of multiphase flow.
Johnson and others (1990) provide a detailed method- -11 -s
ology to examine whether an SVE system is likely to -- E

work. They exploit the analogy between the flow of gas

- 1.1
-P
L

W
and water to develop the gas flow rate per unit thickness f
of well screen &?/If) as -
c

b
a
P,,= 0.80 atm = 6.8 ft HO
, - 0.11 2
P,,= 0.90 atm = 3.4 ft HO
,

= 0.95 atm = 1.7 ft HO

f,,, ,
- 0.011
where H is screen length (cm), k is the soil permeability
to airflow (cm?, p is the viscosity of air (1.8 X lo-' g/
cm-s), P,,, is the absolute pressure at the extraction well
(ucm-s?, P,,,,, is the absolute ambient pressure or 1.01
,0001- sands .1 sands
.o 1
I
1
sands

10
sands

100
0.001 1
1000
X lo6 glcm-s' (1 atm), R,,. is the YddiUS of the vapor Soil permeability (darcy)
extraction well, and R, is the radius of influence of the [ft H,01 denote vacuums expressed as equivalent water column heights

vapor extraction well. If k can be estimated, the only Figure 20.25 Predicted steady-stateflow rates (per unit
unknown in Eq. 20.6 is the radius of influence. Va1ut:s well screen length) for a range of soil permeabilities and ap-
of R, can be estimated from a range of about 9 to 30 ni. plied vacuums (from Johnson and others, 1990). Reprinted
Because of the form of the equation, airflow rates are by permission of Ground Water Monitoring Review. Copy-
not particularly sensitive to R,. right @ 1990. All rights reserved.
 20.6 Soil-VaporExtraction 4 2

Rll,= 5.1 cm (2 in.) R, = 7.6 m (25 ft) Vapor flowmeter Vapor treatment
Pressure unit
multiply g ' / H by 1.09 Vacuum !
Vapor sampling.
Rlt = 5.1 cni (2 in.) R, = 23 m (75 ft)
multiply JZ'/H by 0.90
R,. = 7.6 cm (3 in.) R, = 12 m (40 ft)
Vapor extraction well
multiply Q * / H by 1.08
R,. = 10 cm (4 in.) R, = 1 2 m ( 4 0 f t )
multiply Q'/H by 1.15
R,, = 10 cm (4 in.) R, = 7.6 m (25 ft)
multiply Q*/H by 1.27
Figure 20.26 Schematic o f an air-permeability test run in
Removal Rate Calculations the field (from Johnson and others, 1990). Reprinted by per-
As we saw with Eq. 20.5, the contaminant removal rate mission of Ground Water Monitoring Review. Copyright 0
can be calculated from a knowledge of C,,xand the airflow 1990. AII rights reserved.
rate Q. The main design issue is whether an SVE scheme
can recover the contaminants within a realistic time
frame. We will illustrate this idea with the following cal- of the wells located close to the extraction well (Peargin
culation. and Mohr, 1994).
Section 19.6showed how well hydraulics theory could
be modified to interpret the results of air-permeability
Example 20.5
tests. Here, we illustrate how field data can be interpreted
Within the radius of influence of a single extraction well. using the Cooper-Jacob method. Begin by plotting draw-
assume there is 5000 kg of chlorobenzene. The equilib- down versus the log of time. With a constant rate of
rium concentration of this contaminant in air is 55 mg/ withdrawal in a confined and infinite aquifer, the data
L, and the well is capable of providing an airflow of 280 should plot as a straight line. For the gas-flow problem,
Wmin (10 standard cubic feet per minute). Estimate the the measured pressure decline in the wells should follow
cleanup time using this single well. a straight line.
From Eq. 20.5, the recovery rate is calculated
Rest= 55 X 280 = 1.54 X 10' mg/min or 22 kg/day Example 20.6 (adapted fmm Massmann, 1989)
Given the initial mass of contaminants, it can be deter- An air-permeability test was carried out in January 1986
mined that at the Midway Landfill. Air pressure decline versus time
~ ltotal
t ~ ~ l ~= l l l l , ,mass/recovery rate for one monitoring well (as equivalent centimeters of
water) is shown in Table 20.4. Given that the test zone
l,,l,~ =
. ~5000/22
t c ~ ~= 227 days is 25 m thick, the observation well is 38 m away from
This relatively short cleanup time makes the SVE the extraction well, the air removal rate of 105 L/s. and
scheme attractive. the vapor density of the air is 0.001 g/cm'. calculate the
hydraulic conductivity to gas and the permeability o f
the medium.
Field Estimates of Soil Permeability The first step is to plot the drawdown in centimeters
of water versus the log of time (Figure 20.27). From this
Air-permeability testing in the field is an integral part of graph, the drawdown per log cycle is 8.2 cm water or
any site investigation where SVE is being considered. 8.2/0.001 = 8200 cm ofgas. Rearrange the <:ooper-J;icoh
The most common test involves removing soil gas at a equation and substitute the data. or
constant rate from an extraction well and monitoring
pressure declines in adjacent observation wells (Figure
20.26). Usually, a minimum of three monitoring wells is
used. These wells are located at distances of approxi- Permeability is calculated as
mately one to three times the depth to the water table
away from the extraction well (Peargin and Mohr, 1994). k = -Kp
= .0009 X 1.45 X 10.'
The maximum distance is deterniined by the radius of Pf: ,001 x 980
influence of the well, commonly 9 to 30 m. I n lower = 1.33 x 10--cm2= 13.3darcys
permeability settings, more wells are advisable with some
436 Cbapter 20 / Remediation: Overview and Removal Options

Table 20.4 Drawdowns like a bentonite slurry or bentonite-cement grouts and

Determined from Field Data be longer than 1 m. Commonly, the design of tests often
places monitoring wells inappropriately- at the margin
Time Drawdown of the zone affected by the test. To be most effective,
(mid (cm water)
~ ~~
monitoring wells should be located in the zones with
1 0.0 large expected airflows.
2 0.0
3 0.02
5 0.15 Heterogeneity and the
6 0.25 EHdency of SVE Systems
7 0.43
The efficiency of an SVE system is affected considerably
9 0.74
11 1.04 by heterogeneity both in the distribution of the contami-
14 1.52 nant and the permeability of the medium. All of the
18 2.16 analyses presented so far have assumed implicitly that the
21 2.60 contaminants are distributed homogeneously throughout
26 3.22 the region of active airflow. However, the airflow often
31 3.81 does not effectively contact the contaminants in the sys-
41 4.82 tem. Johnson and others (1990) provide conceptual mod-
46 5.23 els of contaminant distributions, which provide slower-
61 6.35 than-expected contaminant recovery (Figure 20.28).
101 8.00
In the first case, a significant fraction of the airflow is
166 9.14
not passing through the spill because of the well location.
306 10.16
The recovery rate of the contaminant for this case be-
Modified from Massmann (1989).

The design and execution of air-permeability tests is

not without problems. Peargin and Mohr (1994) studied
a data base of 80 SVE pilot tests to identlfy common V a w fl
mechanical or procedural problems. Three problems CIC- Il-t.
.-c
cur commonly: (1) submerged extraction well screens, T

(2) insufficient well seals, and (3) vacuum monitoring Side view Top w w
points either too few or too far away. The screen can be (*r)

submerged when the bottom of the screen is set too

close to the water table. When the extraction well is "
- ........
turned on, the water table will mound at the extraction Vapor con~gntration
well due to the decline in pressure. Problems with the
integrity of the seals commonly lead to short-circuiting
of the airflow. Seals should be constructed from materials

-z
+
1-
2-
3-
1
z 4- 8.2
-5 5-

5
U
6-
7-
$
8-
9- it-)

10 - 0
Figure 20.28 Three examples illustrating situations when
I I I I I I I I I 1 I I I I I I I I I I
the contaminant recovery rate may be less than expected
(from Johnson and others, 1990). Reprinted by permission of
Ground Water Monitoring Review. Copyright 0 1990. All
rights reserved.
 20.7 Air Sparging 437

comes Vapors

ir
Air
(20.7)
I , , + , I I 1

where 4 is the fraction of the total airflow that is moving

through uncontaminated soil. For the case illustrated in
Figure 20.286, this fraction would be about 0.3. Effec-
tively, the recovery rate is 70%of what one might expect.
Figure 20.28b illustrates a case where the NAPL has
pooled and reached relatively high saturations. The air
flows parallel to the surface of the pool but not through
the contaminated soil. A sample calculation provided by
Johnson and others (1990) shows that the removal rate
might be less than 10% of the niaximum rate. Figure
2 0 . 2 8 ~is a case where the airflow is mainly through an Air /
adjacent layer separated from the zone of contamination channels

by a so-called dried zone. Contaminant removal is diffii-

Figure 20.29 Conceptual design of an air sparging system
sion controlled, and thus, inefficient.
(from Newell and others, 1995).
Recovery will also be inefficient when a few highly
permeable units attract most of the airflow through the
system. These units may clean up very quickly but adja-
cent lower permeability zones will not. According to nels as depicted in Figure 20.29. Ahlfeld and others
Peargin and Mohr (1994), airflow in low-permeability (1994) discuss the mechanics of channel formation once
materials is largely controlled by preferential pathways. a sparging well is turned on. In essence, a "pocket" of
Cleanup levels and times are directly related to the den- pressurized air will begin following the path of least
sity and regularity of the spacing of these pathways. resistance, which consists of media having the lowest
air entry pressures or smallest capillary pressure. Thus,
channels form preferentially in media having the largest
-
20.7 A i r Sparging pore size, and by inference, the highest permeability
(Ahlfeld and others, 1994). Airflows will bypass lower
Air sparging can recover volatile organic compounds, permeability units. Once air channels are established,
semivolatiles, chlorinated solvents, and some petroleum they will be maintained as long as the air-entry pressure
compounds (C,-C,,,) from the saturated ground-water is maintained.
system. Air is injected into the zone of contamination. It Contaminants are removed because dissolved VOCs
flows through channels and forms an in situ, air stripping partition into the adjacent air phase. With time, the con-
system, as volatile contaminants partition into the airflow taminant concentrations decline adjacent to the air chm-
(Figure 20.29)).The contaminant-laden air arriving at the nel. The resulting concentration gradient in the aqueous
water table is collected by a conventional SVE system. phase promotes diffusion of contaminants toward the air
Besides stripping volatiles froin the zone of coiitamina- channels. The magnitude of interphase transfer is propor-
tion, the addition of oxygen may promote biodegrada- tional to the area of the interface that exists betweeti the
tion. However, many VOCs are not biodegradable under air and water systems (Ahlfeld and others. 1994). The
oxic conditions. greatest mass transfer would be promoted by m;iny small
Sparging wells are constructed using either PVC or channels, evenly distributed as opposed to a few large
metal casings. The well screen and filter pack must be channels spaced far apart.
carefully sealed with bentonite or neat cement grout to Ahlfeld and others (1994 discuss calculations thvt illus-
avoid air leaks. Individual wells are outfitted ;it the surface trate the importance of maintaining small diffusion path-
with a pressure gage and regulator 10 facilitate the moni- ways. For example, the time to reduce TCE concentr;i-
toring and operation of the well. Check valves are typi- tions from 5000 to 5 pg/L by flowing a column o f air
cally included at the well head to avoid the backflow of with a one-inch radius through soil columns having radii
water when the compressor is shut off. Air is supplied of 20 in., 5 in., and 2 in. was 78 years, 1.- years. and 27
to the wells from an oil-free compressor through metal days, respectively. The smaller the cleanup cdius ;iround
pipes or rubber hoses. These robust materials can accom- a given channel, the more rapidly the cont:imin*lIlt con-
6

modate the heat generated as the air is compressed. centration is reduced.

Air sparging works effectively with airflow rates in ;I
range of 3 to 10 standard cubic feet per minute per sparge
Airfrow and Channeling point. The actual pressure required to produce these
Early work depicted the flow of air through an aquifer rates depends in part on the depth of the well below the
as small bubbles. However, airflow mostly occurs in cham water table or the static head above the sparge point.
-
438 Cbapter 20 / Remediatior Overview and Removal Options

Sparging Vapor
point extraction
Air well
injection
rate (y)

-
Radius of
influence

Figure 20.31 Sizing an air-sparging system to match the ex-

tent of contamination.

Pfgure 20.30 Schematic diagram of an air-sparging and Figure 20.30 illustrates, the radius of influence (R,,,) is
vapor-extraction system showing the radius of influence and defined in relation to the volume of influence, which is
volume of influence (modified from Sellers and others, the zone around the well that experiences the airflow.
1992). Reprinted by permission of Conference on Petroleum Thus, the wells should provide a total area of influence
Hydrocarbons and Organic Chemicals in Ground Water- that corresponds to the area of the contaminated zone
Prevention, Detection and Restoration. Copyright 0 1992. (Figure 20.31).
All rights reserved. A variety of measurements are made in the field to
interpret the radius of influence, including
Deeper wells require higher injection pressures. These changes of pressure in the saturated and unsatu-
airflow rates are promoted by sands and gravels. This rated zones
airflow also must find its way to the unsaturated zone. concentration of VOCs in the unsaturated zone
The presence of low-permeability layers above the well the extent of water-table mounds, and
intake may promote the formation of gas pockets and
the lateral spreading of contaminants. Thus, there are tracer-gas responses.
significant geologic limitations as to where the method Lundegard (1 994) used a highly instrumented test involv-
can be used. ing a single sparge well in a simple geologic setting to
examine these interpretive methods critically. Interest-
Designing Air-Sparging Systems ingly, he found that the common approaches listed above
overestimated the radius of influence by a factor of two
The main variable determining the number of sparging to eight. He suggests that in more complex settings it will
wells is the radius of influence for an individual well. As be even more difficult to use conventional approaches.

in the United States. The Hyde Park site is famous in its

20.8 Case Studies in Site Remediation own right as one of the set of large hazardous-waste
Four case studies have been selected to illustrate the disposal sites located in and around ~i~~~~ ~ ~ lN~~ l ~ ,
typical range of problems encountered in the cleanup of york. The idst case study examines a demonstration of
sites. The first case is typical of literally thousands of SVE tecI1nology for solvents (mainly TCO at an industrial
spills of gasoline and oil. The next two case studies ilre site in ~ ~ ~ ~ ~ ~ h ~
larger and much more serious problems resulting from
the disposal of toxic organic chemicals at uncontrolled
Oil Spit2 Calgary, Alberta
sites. The Gilson Road site in New Hampshire is im-
portant historically because it was the subject of the first The case of a large spill of petroleum products in flood-
cooperative agreement signed under the Supefind law plain sediments is related to an asphalt plant and terminal
 20.8 Case Strdies i n Site Rentediatiott 433

at Calgary, Alberta (O'Connor and others, 1984b). The

spill first came to light when petroleum products dis-
chaged at the surface in the area marked by "Early oil
seeps" (Figure 20.32). After plant personnel constructed
an interceptor ditch along the southern plant boundary
(Figure 20.32), it was thought the problem was solved.
However, at this early stage, the full extent of the problem
was not known. Oil discharged subsequently into lagoons
of the Inglewood Bird Sanctuary.
The need to control seepage in the bird sanctuary in
advance of returning geese required emergency action.
A n interceptor drain was installed across the zone of
seepage (Figure 20.32). A horizontal box culvert and
gravel buried in the trench provided lateral drainage of
the oil to a series of sumps. Migration of oil past the
Figure 20.33 Elevation of the water table (meters) follow-
interceptor system w3s eliminated by adding an imperme-
ing startup of the recovery wells. autiinin lg79 (froni O'Con-
able barrier t o the downstream wall of the culvert. Cutoff nor and others, 1984b). Reprinted by permission of Confcr-
walls added to both ends of the interceptor system de- ence on Petroleum Hydrocarbons and Oaanic Chemicals in
flected oil into the box culvert system. This action elimi- Groiind Water-Prevention, Detection and ~eSt(lrJti(11~.
(:op-
nated oil seepage into the north lagoon. right 0 1984. All rights reserved.
The detailed site investigation revealed that oil had
collected o n the water table over a relatively large area
and was migrating toward the bird sanctuary, following nance contributed to reduced product recovery. Never-
the local gradient of the water table. The contaminants theless, during this period of operation, approximately
were spreading in a sand and gravel unit approximately 1,300,000 L of oil were collected. A significant qu:intity
10 m thick. of product remained in the subsurface, especially under
Because of the size of the spill, the interceptor system the plant site and away from the area of niost intense
was not sufficient to contain and recover all of the oil. remedial activity. The increasing thickness of product ;it
A two-well, two-pump extraction system was installed in the plant site pointed to the continued leakage of oil ;it
the bird sanctuary (refer to Figure 20.106). The water the source. Based on the performance of the system t o
wells produced up to 2.5 X 10-' m-'/s and created the 1983, about six more years would be required to collect
necessary potential low to move the oil into the product- the remaining free product.
collection wells (Figure 20.33). Clean water pumped
from the water wells was discharged into the interceptor
ditches constructed earlier at the site.
Gilson Road Nashua, New Hampshire
This collection system was run for five years and its The Gilson Road Superfund site in New Hampshire is
performance reviewed in 1983. A lack of well mainte- important historically because it was the subject of the
first cooperative agreement signed under the Superfund
law in the United States. Chemical sludge and organic
wastes in drums were placed together with refilse in ;I
sand and gravel pit. In addition, some 4 . ~ ~ 0 0 . 0 0L 0o f
liquid chemical wastes were discharged into the ground
at a nearby trench (Figure 20.34: A y e s and others. 198.3).
The most hazardous contaminants included he;ivy metals
and volatile and extractable organic compounds. Sprcud-
ing of contaminants was facilitated by the presence o f 6
to 35 ni of permeable sand and gravel. which overlie till
Early oil seeps 01 and fractured bedrock. Initial studies in 1980 defined ;I
natural stag&
contaminant plume more than 450 m long and up to 33
m thick (Ayres and others. 1983). By Ileceniber 1980.
some contaminants from the front of the plume had
Figure 20.32 Site of an oil spill at Calgary, Alberta (from begun to discharge into Lyle Reed I3rook. located ;it the
O'Connor and others, 1984b). Reprinted by permission of western edge of the study area. Following completion
Conference on Petroleum Hydrocarbons and Organic Chemi- of the site investigation in 1981, a feasibility study est:th-
cals in Ground Water- Prevention, Detection and Restora- lished a strategy for remedial action. The possible ;iltern;i-
tion. Copyright 0 1984. All rights reserved. tives included combined hydrologic containment and
440 Cbapter 20 / Remediatiom-Overview and Removal Options

the surface, the system was to provide further contain-

Trailer park I
, ment for contaminants. With this design, ground water
only flows into the area surrounded by the grout wall.
This control could be affected by pumping a total of 1.14
m'/min, treating 0.19 m'/min to drinking water standards
for discharge to Lyle Reed Brook, and returning 0.95
m2/min of partially treated water back to the ground
via trenches.
Temporary

H y d e Park Xand$ll= N i a g a r a Falls, New York

. . Coming out of the Settlement Agreement for Hyde Park
0 400'
Landfill in 1982 was a provision to evaluate the available
Figure 20.34 Layout of the Gilson Road hazardous waste- data and to develop a plan for remedial action. The reme-
disposal site at Nashua, New Hampshire, showing the landfill dial measures finally agreed upon were reviewed in detail
and liquid waste-disposal areas as well as facilities for control- by Faust (1 985) and are described here as an illustration
ling contaminant migration (modified from Ayres and others, of the variety of factors that must be considered when
1983). Reprinted by permission of Third National Sympo- dealing with a complex site and a variety of different con-
sium on Aquifer Restoration, and Ground Water Monitoring. taminants.
Copyright 0 1983.
The overall strategy for remediation involved three
parts: (1) source control, (2) control and collection of
ground-water treatment as well as total removal (Ayres dissolved and aqueous phase contaminants from the over-
and others, 1983). The overall strategy that was adopted burden, and (3) control and collection of contaminants
involved isolating the most contaminated p a s of the from the Lockport Dolomite. The complexity of the geo-
plume using a cutoff wall and surface seals, as well as a logic setting, and uncertainty about how well certain
withdrawal-treatment system to remove the contami- remedial strategies might work, necessitated a phased
nants from the ground water. approach that involved using some prototype systems
Before the remedy could be implemented in 1982, along with extensive data collection. Proceeding in this
monitoring showed that a highly Contaminated portion way avoids "the situation where remedial measures are
of the plume was about to discharge into the brook. To installed and later to be found ineffective without any
delay the arrival of this front, an emergency recirculation simple recourse to correct the deficiencies" (Faust,
system was designed to remove contaminated ground 1985).
water from four wells at the nose of the plume (Figure The main reason for source control was to reduce the
20.34) and to return it without treatment using a tempo- quantity of Contaminants moving out of the landfill. A
rary recharge trench. A two-dimensional computer model surface seal will reduce infiltration from precipitation
was developed and calibrated against field data to assist and the flux of water and contaminants. Collection wells
in the design of this system (Ozbiligin and Powers, 198~i). will be used to remove fluids from the actual landfill
Subsequent monitoring indicated that pumping the wells itself. Initial designs are for this to be a prototype system
at a total of 0.25 m'/min was successful in temporarily involving two wells.
retarding the advance of the plume (Ozbiligin and Pow- Remedial action in the overburden will (1) contain the
ers, 1984). lateral spread of contaminants and (2) provide a way
The conventional cutoff wall ranged in thickness from of collecting them. The measure that is proposed is a
0.5 to 1.25 m (Ayres and others, 1983; Schulz and others, subsurface drain around the area where free organic liq-
1984). It surrounded an area of about 80,000 m2to depths uids are known to occur. Contaminants not collected by
ranging from 10 to 33 m. The performance of the wall the drain will move down into bedrock, where other
was evaluated using various mathematical models. The collection systems are in place. These bedrock systems
quantity of contaminated water passing through the site include an injection-withdrawal scheme for DNAPLs
was reduced from 284 m'/day to 114 m'/day during sea- (Figure 20.35) and a withdrawal scheme for collecting
sons with low rainfall (Ozbiligin and Powers, 1984). Con- dissolved contaminants close to the Niagara Gorge. The
tainment was being lost not through the grout wall but system proposed for DNAPLs is analogous to a secondary
through fractures and fracture zones in the bedrock (Oz- recovery scheme for oil (Faust, 1985). The advantages
biligin and Powers, 1984). of the injection-withdrawal scheme as compared to
The literature we examined discussed the design of pumping alone include: (1) it avoids dewatering of the
the system for collecting, treating, and recirculating wa- upper part of the Lockport Dolomite, which is contami-
ter but not its implementation. Computer modeling nated and requires ground water to circulate for cleanup
played an important role in the design of the well system. to occur, and (2) larger volumes of water can be circu-
In addition to simply collecting water for treatment at lated through contaminated parts of the plume. Both two-
 20.8 Case Studies in Site Remediation 111
Before NAPL plume Hyde Park Oil-Gas Oil-Gas
Landfill Abandoned brrte dip
remedial I separator separator leaching field
leaching field
\ manhole \

Figure 20.35 Conceptual diagram of the injection/

withdrawal system for removing DNAPLs from the Lockport
Dolomite at the Hyde Park Landfill (from Faust, 1985).

and three-dimensional flow modeling was conducted to

evaluate the effectiveness of the prototype system, which
involves four collection and two injection wells. Pumping
rates of 20 gallons per minute for each withdrawal well
and 10 gallons per minute for each injection well should
provide a capture zone larger than the DNAPL plume.
After further study, an operational system will be put in
N-
place. The second withdrawal system will collect dis-
solved contaminants in the Lockport Dolomite. These Figure 20.37 Concentration (ppm) of contaminants in the
contaminants are sufficiently far from the site that they soil at the Groveland site (from U.S. EPA, 19892).
will be in general unaffected by the cone of depression
created by the first DNAPL removal wells. Flow, however,
will eventually cause these contaminants to discharge
Groveland We& Site, Massachusetts
along the Niagara Gorge. The plan is to install a line of
interceptor wells across the plume west of the site. The last case study involves Terra Vac, Inc.'s in sitii vac-
In the remedial system at Hyde Park Landfill, monitor- uum extraction process, which was demonstrated as part
ing is an integral part of ongoing operations. Monitoring of the Superfund Innovative Technology Evaluation
will establish whether a particular remedial alternative (SITE) Program ( U S EPA, 1989~).The site selected for
has met specific performance goals, provide criteria for the demonstration was contaminated by a leaking storage
logically expanding the prototype systems, and safeguard tank and improper disposal practices. Approxinutel!.
the public against accidental exposure to the contami- 3,000 to 30,000 Ib of TCE were spilled.
nants. A generalized geologic cross section at the site (Figure

Exlstlng
bui lding

Evaluationsrefer to USGS datum

Figure 20.36 Generalized elst-west cross section through the zone of contamination
(modified from EPA, 1989).
442 Cbapter 20 / Remediatiow Overview and Renroval Options

Table 20.5 Distribution of TCE as a Function of Depth 20.36) shows shallow outwash sand underlain by silty
Before and After Treatment clay and clay till. A second silt and sand unit occurs below
these units. Most of the contamination is in the shallow
TCE Conc. ppm unit where TCE concentrations reach a maximum of ap-
Depth Permeability
(ft) (cm/s) PlT Post proximately 2500 mg/kg. Relatively minor contamination
was detected in the deeper sand unit as well. Figure
0-2 lo-' 2.94 ND 20.37 shows the areal distribution of contaminants and
2-4 lo-' 29.90 N I1 the location of the extraction wells (EW 1 -Ew4) used
4-6 10-5 260.0 39 to demonstrate the SVE system.
6-8 10-5 303.0 9
8- 10 10-1 35 1 .O ND A multiwell design was utilized with two 2-in. wells
10-12 10-4 195.0 ND completed in each borehole in both the sand zones. The
12-14 10-4 3.14 2.3 system was tested for 59 days and removed a total of
14-16 10-" ND ND 590 kg of contaminants. Table 20.5 compares observed
16- 18 ND ND soil concentrations of TCE in the vicinity of J W 4 before
18-20 ND ND and after the test. At other wells, the reduction in TCE
20-22 10-4 ND ND concentrations in the soil was less-ranging from 9 to
22-24 lo+ 6.71 ND 30%. Overall, this test showed the potential of SVE to
M d i e d from U.S. EPA (1989~). clean up VOCs.

Problems
1. The capture-zone type curve for a single well is super- and equilibrium between the phases. Express your
imposed on an outline of a dissolved contaminant plunie answer in both mg/min and kg/day.
as shown in the figure for this problem. c. Compare your answer in (b) with that obtained by
the nomogram.
1OOO r 1
3. In Chapter 19 through discussion, we described the
case of a gasoline spill in gravel along the Trans Canada
Highway. Develop two feasible strategies in qualitative
terms for remediating the site. Be sure to explain the
positive and negative factors that would influence the
choice of a particular strategy.
4. An air-permeability test was conducted using an ex-
traction rate of 75 Ws. The drawdowns shown below
I were measured in an observation well located 5 m away.
Given that the vapor density is 0.001g/cm3 and the zone
I I i I -- I is 2.5 m thick, calculate the hydraulic conductivity to gas.
0 500 1000 1500 2001) 2500
Meters
Problem I Time Drawdown
(min) (cm water)
a. Given that the thickness of the aquifer is 10 m and
the Darcy flow velocity is 1 X lo-' m/s, calculate 10. 69.2
the single-well pumping rate. 100. 108.4
b. For an aquifer transmissivity of 0.005 m2/s, storativity 1000. 147.6
of 1 X and available drawdown of 10 m, de-
termine whether the calculated production rate is 5. An underground storage tank at an industrial facility
feasible. leaked approximately W,OOO L of gasoline over a 5-year
2. A soil-vapor extraction system is proposed to clean period into a shallow sand aquifer. An aggressive program
up pure chlorobenzene spilled at a site. of product recovery collected approximately 35,000 L
a. Knowing that the formula weight of chlorobenzene of free product.
is 113 g/mol and its vapor pressure 0.012 atm at a. Explain what potential problems would occur by
20°C, estimate the vapor-phase concentration (iS simply leaving the residually saturated fraction of the
mg/L) in equilibrium with the pure liquid. spill in the ground.
b. Estimate the rate of removal of chlorobenzene from b. What are the feasible alternatives for removing this
the ground assuming an airflow rate of 140 Wmin volume of the gasoline?
 Cbalbter 21

In Situ Destruction and Risk Assessment

21.1 Intrinsic Bioremediation

21.2 Biouenting and Bioslurping
21.3 Abiotic Chemical Destruction
21.4 Risk Assessment
21.5 Fernald Case Study

~~ ~~

Techniques for degrading, detoxifying, or immobilizing that break down contaminants to less harmful constit-
contaminants in situ rely on biological and chemical pro- uents. To a lesser extent, we will consider techniques
cesses to get the job done. With biodegradation, bacteria that chemically transform contaminants to an immobile
mineralize or transform contaminants using organic con- phase.
taminants as an energy source. Under favorable condi- The chemical schemes are specific to particular types
tions, biodegradation produces harmless compounds like of compounds. For example, destruction of PCE and TCE
water and CO&) from complex organic molecules. One can be achieved through reductive dechlorination, cata-
new alternative, intrinsic bioremediation, evaluates the lyzed by zero-valent iron. Similarly,TCE can be abiotically
natural capabilities of a hydrogeologic system to biode- transformed using oxidizing agents like potassium per-
grade contaminants without active treatment. The chal- manganate or ozone. Overall, the chemical technologies
lenges with this approach are to demonstrate that are less well developed than the biological approaches,
cleanup will be achieved and that the monitoring scheme with demonstration projects just beginning to emerge.
is appropriate.
Other common approaches to biorernediation are
more proactive. They are aimed at enhancing the growth
potential of existing populations of microorganisms
21.1 Intrinsic Bioremediation
7

through the addition of nutrients like nitrogen and phos- Intrinsic bioremediation relies on the natural system to
phorous, as well as electron acceptors like oxygen in the affect the in situ degradation of contaminants without
case of aerobic bacteria. There also has been promising engineering efforts to stimulate the natural populations.
work on introducing specially engineered bacteria to This approach is particularly well suited for sites with ;I
metabolize hard-to-treat contaminants like the chlori- low risk of impacting sensitive off-site receptors. Many
nated solvents. AII of these approaches will be discussed believe that intrinsic bioremediation is a no-action alter-
in the following sections. native. However, pursuing this strategy requires (1) an
Chemical degradation methods treat contaminants in investigative effort to demonstrate that the natural system
situ through the addition of treatment agents. In this is capable of eliminating the specific contaminants, (2)
chapter, we will be concerned especially with techniques a detailed analysis to demonstrate that significant health
413
-
444 Cbapter 21 / In Situ Destruction and Risk Assessment

risks do not exist, or adjacent properties or aquifers state concentration distribution (Buscheck and others,
would not be impacted, and (3) a long-term monitoring 1993).
effort to verify that the expected cleanup is occurring Fitting an exponential model to monitoring well data
OVRC, 1993). should yield an estimate of the decay rate from Eq. 21.1,
Intrinsic bioremediation is mainly used to clean up or k/u from Eq. 21.2. If an estimate of v is available, then
petroleum spills. The tendency for petroleum hydrocar- the latter equation can also provide an estimate of k.
bons to biodegrade is reasonably well understood, and However, evidence of a temporal or spatial reduction in
a variety of site studies have documented the capabilities concentration by itself is not firm evidence that biodegra-
of natural systems to degrade contaminants. Observations dation is operative. Independent confirmation should be
made at the site of a pipeline break at Bemidji, Minnesota, provided by examining various parameters such as dis-
are particularly instructive. Crude oil contaminated a solved oxygen or other electron acceptors, EH measure-
shallow unconfined aquifer. Monitoring of the site ments, or the enumeration of microPsl populations
(Baedecker and others, 1989) showed that dissolved com- attached to the porous medium (Buscheck and others,
pounds had migrated initially some 200 m. However, 1993).
since 1987, the plume has not advanced. Additional evi- This theory was applied to a storage facility at Fairfax,
dence for intrinsic bioremediation comes from a detailed Virginia (Buscheck and others, 1993). Benzene data are
evaluation of chemical components in the plume. plotted in Figure 21.1 for wells located along a p a t h h e
It is difficult to determine that biodegradation is oc- at distances approximately 0, 25, 175, and 505 ft from
curring at a site, and to quantify the rate of degradation the contaminant source. The besi-tit line through the
in fate and transport analyses. Dispersion and sorption data provides a k / v estimate of -0.01 1. With a velocity
lead to concentration reductions, but not to an overall estimate of 0.05 &/day, the decay rate for benzene is
reduction in the contaminant mass. Buscheck and others 5.5 X lo-' day-' or 0.055%day-'. Toluene, ethylbenzene,
(1993) describe a method for using monitored well data and xylene values are 0.045, 0.045, and 0.04% day-',
to quantify patterns of contaminant attenuation as a func- respectively. An analysis of the indicator parameters pro-
tion of either time or distance. The method attempts to vides corroborating evidence of biological activity.
fit simple first-order kinetic models to concentration data Table 21.1 is a summary of data o n a variety of sites
from one or more monitoring sites. studied by Buscheck and others (1993). Overall, most of
The concentration versus time analysis establishes the the decay rates fall in a range of from 0.1 to 0.46%day-'.
rate of concentration reduction over some period of rcc- These rates represent half-lives of approximately 150 to
ord at a single monitoring well. Mathematically, a first- 693 days, which is comparable to half-lives due to biodeg-
order model provides the concentration as a function of radation (Buscheck and others, 1993).
the initial concentration and decay rate as There are data available to indicate that chlorinated
aliphatic hydrocarbons may biodegrade naturally under
(21.1) some conditions (Major and others, 1991; McCarty and
Wilson, 1992). One recent study by Semprini and others
where C(t) is the contaminant concentration at any time (1995) examined solvent contamination in a sand aquifer
> 0, C,,is the initial concentration at t = 0 , and k is the near St. Joseph, Michigan. The study found that TCE,
decay rate constant (t-').Such a decrease in contaminant
concentration could be observed at or near the fringe
of a shrinking plume. In wells closer to the source, an
C, = 2529.45
exponentially decreasing concentration could suggest y 1000 &? = 0.886
significant source decay (Buscheck and others, 1993). 300

d
l.ll
L
The concentration versus distance model uses data 0 MW-4
from several monitoring wells, collected at the same time
along a pathline. By recasting time as distance over veloc- 0 MW-37

ity, Eq. 21.1 can be rewritten as w= 3

d 10 100 200 300 400 500 600

Distance from MW-15 (ft)

Figure 22.2 Change in benzene concentration along a

where C(x) is concentration as a function of distance x, pathline at a site in Fairfax, Virginia (modified from
and u is the linear ground-water velocity (no retardation Buscheck and others, 1993). Reprinted by permission of Con-
assumed). This model conceptualizes the plume being ference on Petroleum Hydrocarbons and Organic Chemicals
at steady state. In other words, dissolved contaminants in Ground Water-Prevention, Detection and Restoration.
being advected and biodegraded will achieve a steddy- Copyright 0 1993. All rights reserved.
 21.2 Bioventfng and Bioslurping 4 3

Table 21.1 Summary of Estimated Intrinsic lation of an existing population of bacteria in aerobic
Biodegradation Rates situations. Other research has focused on methods that
deal with anaerobic or co-metabolic systems, which ;ire
Site Compound k (010 per day)
required for the breakdown of chlorinated solvents like
Campbell, CA Benzene 1.20 PCE and TCE.
Ethylbenzene 0.67 Both bioventing and bioslurping are used primarily to
Xylene 1.12 clean up fiiels from the unsaturated zone and the vicinity
Benzene 0.42 of the water table. They rely on air from the atmosphere
Palo Alto, CA Benzene 0.30 to provide oxygen necessary for biodegradation. I n sonic
Virginia Beach, VA PCE 0.46 respects, bioventing is similar to SVE methods in that
TCE 0.30
a vacuum-extraction well moves air through the spill.
Montrose County, CO Benzene 0.42
However, unlike SVE systems that strive t o maximize
Provo, UT Benzene 0.23
San Jose,CA knzene 0.16 the volatilization of low-molecular-weight conipouncls.
Benzene 0.10 bioventing systems are operated to maximize the in situ
Chemical facility Toluene 0.39 biodegradation of compounds. In other words, SVE SJS-
PCE 0.34 tems involve volatilization with minor biodegr;idation.
TCE 0.26 and bioventing systems involve biodegradation with mi-
nor volatilization.
From Buscheck and others (1993). Reprinted by permission of Confrr-
ence o n Petroleum Hydrocarbons and Organic Clirmicals in Ground A system for in situ bioventing, thus, is the same ;IS
Water-Prevention, Detection and Restoration. Copyright 0 1993. All that in Figure 20.5. The main difference, however. is the
rights reserved. airflow rate. In SVE systems, rates are high to maximize
phase partitioning of the contaminants. In bioventing
systems, rates are much lower because the system needs
which had been present in the aquifer for more than a
to supply oxygen to bacteria only in the zone of contanii-
decade, had degraded to cis-1,2-dichloroethene (c-DCE),
nation. In general, the rates of oxygen utilization by bacte-
vinyl chloride (VC), and ethene. The distribution of these
ria are relatively small so that airflow rates can be kept
compounds was correlated with sulfate and methane con-
small.
centrations. The transformation of TCE to DCE is associ-
Bioventing destroys contaminants in the subsurface.
ated with sulfidogenic conditions, while reductions to
Thus, air removed by the extraction well should contain
VC and ethene require methanogenic conditions (Seni-
relatively low vapor-phase concentrations of contnmi-
prini and others, 1995). Given the relatively slow reaction
nants and may not require treatment nt the surfxc. I n
rates involved and the incomplete nature of the reductive
large projects, reducing the costs for long-term tre;itnient
dechlorination reactions, it is unlikely that a case can be
can dramatically affect the overall economics.
made for applying intrinsic bioremediation schemes to
Recnuse bioventing systems rely on bacterial destruc-
problems with chlorinated solvents. However, these
tion, they can be used to clean up relatively nonvol;itile
studies do provide some hope that systems contaminated
fuels. For example, middle petroleum distillates. like
by solvents that are not being treated could have some
diesel fuel, conrain a smaller proportion of volatile
capacity for intrinsic remediation.
compounds compared to gasoline. Thus. SVE s!atems
are relatively inefficient. These compounds. however.
are commonly biodegradable and will respond to bio-
21.2 Bioventing and Bioslurping venting.
Bioremediation schemes promote the development of Bioslurping is an approach that combines two remedial
natural bacterial populations to transform organic con- technologies: free product recovery and bioventing. Free-
taminants into less hazardous compounds. Unlike other phase LNAPLs are recovered using \-acuuni-enIivicecl
cleanup technologies, biodegradation can reduce con- pumping. The vacuum system also contributes to hie-
tamination to extremely low levels. For example, clean- venting in the unsaturated zone. A bioslurper well con-
ing up a gasoline spill by pumping leaves behind a rela- sists of an exterior casing and well screen that is installecl
tively significant proportion of the spill. This fraction is across the water table (Figure 2 1.2). h one-inch PVC drop
the gasoline held at residual saturation in the pores. If tube enters the well through a vacuum-tight seal ;ind
left in place, this remaining contamination will continue extends to the ground-water-product interface (Kittel
to be a source for dissolved compounds. and others, 1994). When free product is present in the
A variety of schemes have been proposed to utilize well, the vacuum system removes it. However. much o f
natural bacteria to destroy contaminants. In this respect, the time the system pumps air. Equipment on the surfrice
both bioventing and bioslurping have a proven track separates water and free product and treats contaniin;ints
record for cleaning up fuel spills. They facilitate the stimu- in the air stream (Kittel and others. 1994).
446 Cbapter 21 / In Situ Destruction and Risk Assessment

Va$ve The feasibility for degrading halogenated aliphatic

f i T -
TPP

Compression Valve 6 nheader

compounds in aerobic settings improves as the degree
screws, of halogenation decreases (NRC, 1993). Typically, degra-
dation is achieved using a mixed culture of bacteria,
Metal Rubber
plates gasket which grows by oxidizing methane (usually added) and
rl I I h co-metabolizes the halogenated aliphatic compounds
2' tee (Wilson and Wilson, 1985; Fogel and others, 1987).
valve
Another important class of halogenated organic con-
taminants is the halogenated aromatic compounds. These
compounds contain one o r more halogen-bearing ben-
zene rings (NRC, 1993). Examples of these compounds
include chlorinated benzenes, polychlorinated biphenyls
(PCBs), and pentachlorophenol. Highly chlorinated com-
pounds do not appear to degrade aerobically, although
Screen
they are biodegradable to some extent by anaerobes.
Again, as the degree of halogenation is reduced, com-
-Water table pounds are susceptible to aerobic degradation (NRC,
1993).
Because the methods for dealing with compounds de-
Figure 22.2 Schematic diagram of a bioslurper well design graded under anaerobic settings are under development,
(from Kittel and others, 1994). Reprinted by permission of the remainder of the discussion in this section will be
Conference on Petroleum Hydrocarbons and Organic Cherni- concerned mostly with the remediation of petroleum
cals in Ground Water- Prevention, Detection and Restord-
spills.
tion. Copyright 0 1994. All rights reserved.

Requirementsfor Success
wit b Bioventing Systems
Applicability of the Technology
to Contaminant Groups For aerobic bioventing methods to succeed, reasonably
large volumes of air, containing oxygen, must be moved
The essence of a biotransformation reaction is the oxida- into the subsurface. The air permeability of a medium
tion of an organic compound by an electron acceptor depends primarily on the type of geologic material pres-
such as O&), NO,, SO:-, or COz in a reaction whose ent and/or thc extent of fracturing. Soil moisture is an-
rate is controlled by bacteria. To date, the most common other important variable affecting air permeability. High
application of biodegradation schemes is for cleaning up soil-moisture levels can effectively reduce the air perme-
petroleum hydrocarbons. The monoaromatic and ali- ability of the medium due to multiphase effects. In a
phatic constituents in these hydrocarbons are mineral- performance analysis of bioventing at 103 test locations,
ized in aerobic reactions as follows: Miller and others (1993) found that bioventing was infea-
sible at only three sites due to a combination of high
C & , + YO2+ 6C0, + 3H20 (2 1.3) water table, high moisture contents, and fine-grained
soils.
Representative of petroleum hydrocarbons and deriva- One absolute requirement to make a bioventing system
tives are gasoline, fuel oil, polycyclic aromatic hydrocar- work is the presence of a microbial population that is
bons (PAHs), creosote, ethers, alcohols, ketones, and es- capable of degrading the organic contaminant. For many
ters (NRC, 1993). The greatest problem in degrading applications, where the susceptibility of the contaminant
these compounds comes with complex molecules with to degrade is well known (for example, gasoline), this
a strong tendency to sorb onto aquifer solids. requirement is usually met. For other less well known
Another class of potentially treatable organic contami- contaminants, laboratory treatability testing (as outlined
nants is the halogenated aliphatic compounds (Lee and in Chapter 13) and field testing are necessary to demon-
others, 1987). Given their abundance and relative mobil- strate biodegradability. A field procedure for treatability
ity, schemes to treat these compounds are under active testing is outlined in the following section.
research. Certain highly chlorinated contaminants, such Miller and others (1'993) identified soil pH, nutrient
as PCE, are resistant to degradation by aerobic microbes level, and temperature as other factors possibly important
(NRC, 1993). However, they may be degraded by anaero- in affecting biodegradation rates. Of this list, temperature
bic organisms. Thus, many of the conventional ap- appears to be most important. Biodegradation rates at
proaches for treating petroleum spills are ineffective for cold sites are lower than at warmer sites. However, even
these compounds. in cold soils (that is, 3"C), biological activity is measurable
 21.2 Bioventing and Bioslurping 447

(Miller and others, 1993). As a rule of thumb, biological Male

quick
activity appears to double for every IO’C, following the
Sampling pump
van Hoff-Arrhenius equation. Nutrient levels appear to
be somewhat important. However, preliminary experi-
ments in the field find no beneficial effects of adding nu-
trients.
surface
II
In Situ Respiration Testing
h
Kittel and others (1993) describe a field test to determine
the efficacy of bioventing. The in situ respiration test
uses observed oxygen utilization rates to estimate in situ
biodegradation rates. These estimates establish whether
bacteria can biodegrade the contaminant in question.
v Soil probe extensions

The test involves (1) flooding the unsaturated zone with

air, containing oxygen and an inert tracer (helium), (2)
cutting off the source of oxygen, and (3) quantifying how Figrrre 21.3 Equipment used to conduct a soil-gis s;ini-
fast bacteria utilize oxygen. pling survey (from Kittel and others, 1993). Reprinted by per-
Running one of these tests involves the following steps: mission of Conference on Petroleum Hydrocarbons and Or-
ganic Chemicals in Ground Water-Prevention, Detection
site evaluation and test-design with the help of a soil and Restoration. Copyright 0 1993. All rights reserved.
gas survey,
installation of instrumentation,
to determine whether O2concentrations are influenced
test setup and monitoring, and finally,
by other processes like diffusion.
data reduction and rate calculations (Kittel and oth- After the air injection stops, the monitoring wells are
ers, 1993). sampled on a regular basis using a soil-gas monitoring
The first step is to decide on locations for the monitor- pump with appropriate analyzers for 02,C 0 2 , helium,
ing wells. Most wells are installed in zones of contamina- and total hydrocarbons. Gasflow rates are kept small -2
tion. One or more background wells are placed away to 4 ft.’/hr-to avoid drawing in air from the atmosphere.
from sites of contamination. Kittel and others (1993) With time, 0:levels decline and C 0 2 levels increase ;IS
recommend that well sites be selected by a soil-gassurvey oxygen-enhanced biodegradation occurs. The test is
with concentration measurements for C 0 2 , 02,and hy- halted when O2 levels reach about 5%. or after 5 days of
drocarbons. For shallow contamination, samplirrg is con- sampling (Kittel and others, 1993).
ducted with hand-driven monitoring points (Figure 2 1.3).
Soil gas can be sampled at depths greater than about 6
m with hydraulically driven probes (Kittel and others,
1993). Figure 2 1.3 illustrates how the different gas detec-
tors are connected t o the system to make the various
Prre ,&
,
r valvt;rt
Gas sampling
h
Rotometer

Regulator
measurements. Kittel and others (1993) list equipment
manufacturers. YRotometei
Air
Using the soil-gasmeasurements, an obviously contami-
nated site can be selected. At such a site, O2 concentra-
tions would fall in a range of 0 to 2% C 0 2 concentrations
Small-dtarneter
in a range from 5 to 20%,and hydrocarbon concentra- probe
tions > 10,000 mg/L, depending on the contaminant.
The control well(s) would be located at an uncontami-
nated site with OL,about 15 to 20%,C 0 2 about 1 to 5%
and hydrocarbons < 100 mg/L.
The next step is to install the various monitoring wells. Screen

Kittel and others (1993) recommend using permanently

installed gasmonitoring wells. Normally, one would col- Figure 21.4 Equipment used to flood the porotis mediun~
lect data from four or so wells in the zone of contamina- with air and an inert tracer in preparation for an in situ respi-
tion and one or two background wells. The test begins ration test (from Kittel and others, 1993). Reprinted by per-
by injecting air with 1 to 2% helium into the background mission of Conference on Petroleum Hydrocarbons and Or-
and monitoring points (Figure 2 1.4). Gas-flow rates of 60 ganic Chemicals in Ground Water-Prevention. Detection
to 100 ft.l/hr are maintained for 24 hr. The helium is used and Restoration. Copyright 0 1993. All rights resewed.
448 Cbapter 21 / In Situ Destruction and Risk Assessment

The resulting data are interpreted graphically to pro-

vide O2 utilization rates. Kittel and others (1993) prefer -- I
to work with OLrather than C02,because fewer factors
influence O2 concentration. The percent change in O2
with time and an assumed stoichiometric relationship for
the reaction of hydrocarbon and oxygen is used in the
following equation to estimate biodegradation rdtes
II

(2 1.4)

where KB is the biodegradation rate (mg hydrocarbon/

kg soil day), K,, is the measured oxygen utilization rate
I I I I 1
(percent per day), A is the volume of air per m3 of soil 0 10 20 30 40 50 ( 0
(L/kg), Do is the density of oxygen gas (mg/L), and C is Time (hours)
the mass ratio of hydrocarbons to oxygen required lor
Figure 21.5 Results of an in situ respiration test at Fallon
mineralization. This latter parameter comes from assum- NAS (modified from Kittel and others, 1993). Reprinted by
ing that the hydrocarbon reacts like hexane as permission of Conference on Petroleum Hydrocarbons and
C6HId+ 9.502+ 6C02 + 7H20 Organic Chemicals in Ground Water-Prevention, Detection
and Restoration. Copyright 0 1993. All rights reserved.
One mole of hexane is 86 g and 9.5 moles of oxygen is
304 g, which yields a hydrocarbon/oxygen ratio of 1/3.5.
The background measurements provide a basis for cor- perdture, altitude, and atmospheric pressure); a hydrocdr-
recting the observed O2 utilization rates for other pro- bon-to-oxygen ratio for hexane of 1/55; and no other
cesses. For example, O2can be utilized in the oxidation processes *affectinggas concentrations. calculate the bio-
of iron. Similarly, helium behavior in all of the wells is degradation rdte.
used to indicate possible short-circuiting of flow. The A plot of the data (Figure 2 1.5) yields an oxygen utiliza-
following example illustrates the calculation. tion rate of -0.23%/hr. The value of A is determined
from the porosity and soil bulk density as 300/1440 or
Example 21.1 0.21 Wkg.

Kittel and others (1993) report results from an in situ

respiration test at Fallon Naval Air Station (NAS) in Ne-
vada. Table 21.2 lists pretest measurements along with
Expressing the utilization rate for O2 in days gives KO a
11 other measurements of O2 and C 0 2 concentrations.
value of 5.52%/day (-0.23%dhr X 24 hr/day). Thus, the
Assuming a porosity of 0.3; a soil bulk density of 1440
biodegradation rate is
kg/m3;an oxygen density of 1330 mg/L (varies with tern-
(0.8)(-5.52) or -4.4 mg/kg day
Table 21.2 S u m m a r y of Monitoring Data for Oz
and COz at Fallon NAS In situ respiration tests have been conducted at a vari-
ety of Air Force sites. Data summarized by Miller and
Percent
others (1993) show O2 utilization rates commonly in a
Time (hr) 0 2 co2 range from 3 to 12%per day (Figure 21.6). These rates
produce hydrocarbon biodegradation rates of approxi-
-23.5 0.05 20.4
mately 1.5 to 7.5 mg fuel/kilogram of soil each day. With
0 20.9 0.05
20.3 0.08
levels of hydrocarbon contamination (TRPH) in the order
2.5
5.25 19.8 0.10 of 2000 to 3000 mg hydrocarbons/kg soil, it will require
8.75 18.7 0.13 about 1 to 5 years to effect a cleanup.
13.25 18.1 0.16
22.75 15.3 0.14
27.0 15.2 0.22 Progress in Solvent Bioremediation
32.5 13.8 0.14 The bioremedidtion schemes in this chapter have been
37.0 12.9 0.23 targeted almost exclusively to fuel spills. While there is
46.0 11.2 0.22
an obvious need for biodegradation schemes in cleaning
49.5 10.6 0.16
up solvents (that is, halogenated aliphatic compounds),
From Kittel and others (1993). there are important technical limitations. These com-
 21.3 AbfotfcCbemfcalDestructfon 413

10 I extent of about 1200 acres and is approximately 150

I
9 feet thick.
8 The demonstration involves two horizontally drilled
7 wells, installed above and below the water table (Figure
$ 6 21.7). A variety of operational modes were tested in-
2 5 cluding;
?
L 4 SVE in the vadose zone (20 days)
3 air sparging through the deeper well, and SVE in the
2 shallow well (33 days),
1 methane and nutrient addition to the air-spargingwell
<1 1-3 3-5 5-7 7-9 9-11 11-15 15-2020-30 >30 (Figure 2 1.7) as,
Oxygen utrl tzation rate (%O,/day)
1% methane addition (107 days)
Ffgure 21.6 Survey of results from a large number of in 4%methane addition (79 days)
situ respiration tests (from Miller and others, 1993). Re-
printed by permission of Conference on Petroleum Hydrocar- pulsed 4%methane addition (94 days)
bons and Organic Chemicals in Ground Water-Prevention, addition of gaseous nutrients in the form of
Detection and Restoration. Copyright 0 1993. All rights re- nitrous oxide, triethyl phosphate in combina-
served. tion with 4%methane in pulses (94 days).
Over the 384 days of the test, approximately 17,000
Ib of solvents were removed. The SVE system removed
pounds are resistant to biodegradation in aerobic environ- about 12,096 Ib and biodegradation led to the mineralim-
ments because of a lack of appropriate degradative en- tion of 4838 lb. The assessment indicated that before
zymes (Chapelle, 1993). In fact, few if any bacteria can stimulation by methane, the microbial populations were
use halogenated aliphatic compounds as a primary energy low and under nutrient stress. The addition of methane
source (Chapelle, 1993). Thus, solvents in the soil and stimulated the growth of methanotrophs, which were
the ground water are extremely persistent. effective in the biodegradation of TCE. Apparently, con-
Experiments (for example, Wilson and Wilson, 1985) taminants were mineralized without the formation of
have shown that compounds like TCE can be biode- intermediates. PCE degradation was accomplished in an-
graded under aerobic conditions. Essentially, methano- aerobic pockets created within the larger aerobic system.
genic bacteria are capable of co-metabolizing aliphatic The pulsed operation of the system, thus, may have pro-
compounds. These bacteria can be stimulated in aerobic moted both aerobic and anaerobic degradation simultane-
settings through the addition of methane gas. The poten- ously (U.S. DOE, 1995).
tial of this approach was validated in a three-year field
investigation at the Moffett NAS in California (Roberts
and others, 1990; Semprini and others, 1990). A variety -
2 1.3 Abiotic Chemical Destruction
of in situ tests showed the possibilities of degrading sol-
Remedial alternatives are now being studied that involve
vents by stimulating naturally occurring populations of
the in situ abiotic destruction of aqueous contaminants.
methanotrophic bacteria.
A full-scaledemonstration of methane injection to pro-
mote the biodegradation of solvents was undertaken by
the U.S. Department of Energy (DOE). The results of Injection point for Catalystc%Hc'
this study are presented in a large number of DOE/OTD airlrnethane
reports. The site for the demonstration was the M Area
at the Savannah River Site, Aiken, South Carolina. At the
M Area, a leaking process water line resulted in contami- de-hhmhAh
compressor I( 11
nation by TCE, PCE, and TCA. DOE estimates that be-
tween 260,000 and 450,000 Ib of organic contaminants
are present in the ground water, with additional mass
sorbed (U.S. DOE, 1995). Prior to the implementation of 0 0 table
the in situ bioremediation scheme, concentrations of TCE Contaminated zone
and PCE in the water ranged from 10 to 1031 pg/L and
3 to 124 pg/L, respectively. The sorbed concentration
of TCE and PCE on the sediments ranged from 0.67 to Ffgure21.7 Schematic of the in situ bioremediation sys-
6.29 mg/kg and 0.44 to 1.05 mg/kg, respectively (US. tem demonstrated at the Savannah River Site (modified from
DOE, 1995). The zone of contamination has an areal U.S. DOE, 1995).
 Next Page

-
450 Cbapter 21 / In Situ Destruction and Risk Assessment

An exciting new approach involves using reduction reac-

tions to transform certain chlorinated solvents. Kriegmm-
King and Reinhard (1992) showed how carbon tetrachlo-
ride could be made to transform to chloroform and CS2
in the presence of pyrite, sulfide, biotite, or vermiculite.
More recently, Gillham and O’Hannesin (1994) were suc-
cessful in using zero-valent iron (iron filings) as an “en-
hancing agent” to promote the dehalogenation of chiori-
nated methanes, ethanes, and ethenes. They found that Figure 21.8 Schematic representation of a reactive barrier
all of the chlorinated compounds that they tested, with system. Ground water passes through the barrier, while con-
the exception of dichloromethane, were transformed at taminants are removed in situ at the barrier.
relatively rapid rates. Reactions generally followed
pseudo-first-order rate laws with half-lives ranging from
0.25 to 432 hr (Gillham and O’Hannesin, 1994). Overall, barrier. In this application, apatites placed in the wall
their experiments indicated that the presence of iron provide small quantities of phosphate that react with
enhanced transformation rates 3 to 13 orders of mag- dissolved lead to produce lead phosphate minerals. The
nitude. resulting precipitates have an extremely low solubility.
Gillham and O’Hannesin (1994) concluded that the
reaction is electrochemical in nature and involves the Funnel-and-Gate Systems
oxidation of iron and the reductive dechlorination of the
organic compounds. Following is their suggested overall Consideration of reactive barrier technologies often leads
reaction describing this process: to the question of how one would actually install one of
these systems. Simply digging out a piece of aquifer and
+
Feo 3H20 + X-CI 4 2Fez++ 3 0 H - + Hr + X-H + CI replacing it with a reactive medium may be feasible for
Somewhat less advanced is work that examines the shallow small spills. However, this approach would be
potential of oxidizing organic solvents using powerful infeasible for plumes that are wide or located deep in
oxidizing agents like hydrogen peroxide (H20J and po- the ground. The funnel-and-gate system described by
tassium permanganate (KMn04).These reactions are par- Starr and Cherry (1994) overcomes these limitations.
ticularly attractive because they appear capable of miner- Their method uses low-conductivity cutoff walls (the fun-
alizing the compound to C 0 2 and water. nel) to focus flow through a gap (the gate), which con-
tains the chemically active zone.
As Figure 21.9 illustrates, the funnel-and-gate system
Reactive Barrier Systems can be deployed in a variety of ways. The choice of the
The relatively short half-lives for abiotic transforma- design is dictated by the types of contaminants involved
tions involving chlorinated organic contaminants have and the types of contaminants present. If the plume is
prompted the development of creative strategies for engi- wide, several gates could be included in the system. If a
neering the systems. The reactive barrier system is one variety of contaminants are involved, different reactor
such example. In its simplest form, the reactive barrier systems could be stacked together (Figure 21.96). With
is a chemically active zone created in the subsurface to LNAPLs, it is likely that a hanging gate would suffice
facilitate the transformation or mineralization of chlod- (Figure 2 1.9~).
nated solvents. Ground water passes through the barrier, The design of the funnel-andgate system needs to be
leaving the contaminant to be transformed within the closely tied to rate of the reaction acting to destroy the
barrier (Figure 21.8). There is the hope that the barriers contlminant (Starr and Cherry, 1994). A sufficiently long
would operate as a passive remedial system. In other residence time in the reactive portion of the system needs
words, once the barrier is constructed, it would work to be provided to reduce the contaminant concentrations
for years without extensive maintenance. At the time of appropriately. Figure 2 1.10 displays potential concentra-
writing, Gillham and co-workers have several field trials tion reductions that are possible for given residence times
under way to evaluate the performance of zero-valent and reaction half-lives. The residence times are design
iron barrier systems under field conditions. variables determined by the length of the cutoff walls
Other possibilities exist for the treatment of contami- and the extent to which the cutoff walls surround the
nants in reactive barrier systems. For example, the barrier zone of contamination (Starr and Cherry, 1994).
could contain dense populations of microorganisms to
biodegrade organic compounds. Work has been under
way at Lawrence Livermore National Laboratory to de-
-2 Risk Assessment
velop this concept. Xu and Schwartz (1994) describe the One of the realities of the site-remediation business is
results of experiments that show how lead and poten- that most contaminated sites will never be cleaned up.
tially other metals could be immobilized in a reactive Overall, the cost of cleaning u p sites far exceeds the
Previous Page

21.1 Risk Assessment 4 3

dollars available for cleanup, especially for orphan sites.

If cleanups are limited by the availability of funding, then
it makes sense to develop criteria to select those sites
that pose the greatest “risk” to public health. Eventu:tlly.
if a site is to be cleaned up, the remedy should reduce
the health risk to some appropriately small level. It is in
these contexts that risk assessment has emerged as a
significant activity involved with the study of contanii-
nated sites. Risk assessment is broadly defined as the
process of characterizing the adverse health effects of
human exposures to environmental hazards (NAS. 1983).
[U)
Beyond evaluating contaminated sites, risk assessment
provides a means to assess the impact of waste4ispos;il
Multiple-gate system ~ U ~ ~ I ~ i ~ . ~ ~
sy3tcrans schemes, like landfills or nuclear-waste repositories.
Though risk assessments are used most commonly t o
examine risk of human health, the approaches can be
applied more generally to other environmental recepton
(for example, birds and fish). The U.S. Environmental
Protection Agency (EPA) (1989b,c) outlined the basic
framework for evaluating hazardous-waste sites to assess
potential health and environmental impacts. The frame-
work may be summarized in the following steps: (1) data
collection and data evaluation; (2) exposure assessment;
(3) toxicity assessment; (4) public health risk character-
ization; and ( 5 ) environmental risk assessment.

Data Collection and Data Evaluation

The objective of data collection and data evaluation is
to provide basic information for risk assessment. This
process includes reviewing available site information,
identifying potential human exposure, defining further
(c) data-collection needs (background sampling, modeling.
Figure 21.9 Configurations for the funnel-and-gatesystem and other sampling needs) and strategy, and actual data
(from Starr and Cherry, 1994). Reprinted by permission of collection. This process will also combine data available
Ground Water. Copyright 0 1994. All rights reserved. from site investigations to evaluate analytical methods.
chemicals, and extent of contamination, and to develop
a set of chemical data and information for use in the
10 risk assessment.
ts I
I
Exposure Assessment
Throughout this book, w e have touched on the health
implications of contamination, but never really empha-
sized the ways in which contaminants could move from
ground water into humans. In part, our particular focus
on geologic systems is less of an oversight and more a
recognition that the main role of a ground-water hydrolo-
gist is to provide background data for a risk assessment.
However, to do this job properly requires basic knowl-
I I 1 I edge of what this assessment is trying to accomplish.
0 1 2 3 4 5 The concept of an exposure pathway is fundamental
Residence time [days) to a risk assessment. An exposure pathway is the route
Figure 21.10 Effect of residence time and reaction half-life that contaminants follow from the point o f release t o
on concentration reductions achievable in the gate (from some point in the human body where the dose is deliv-
Starr and Cherry, 1994). Reprinted by permission of Ground ered. A reader progressing systematically through the
Water. Copyright 0 1994. All rights reserved. book should have intimate knowledge about how con-
4 s Cbapter 21 / In Situ Destruction and Risk Assessment

Figure 21.11 Conceptual model of a generalized water pathway from a contaminant

source to humans (from FERMCO, 1994).

taminants move from a source along the ground-water and animal food chains. The relationships among these
pathway (that is, geosphere transport). This pathway is pathways are usually captured in a flow chart. Figure
but one of several “water-based’’ pathways available to 21.12 exemplifies the general environmental pathways
contaminants to move away From a source. Further, as for radionuclides.
Figure 21.11 illustrates, the pathway to humans wends A risk assessment requires an estimate of existing con-
its way through the biosphere as parts of complex plant taminant concentrations and sometimes the likely future

Source Environmental Pathway Exposure Pathway

m I I

I
H
I I
Direct exposure
H
1
Ground

I
External
radiation
I

-.-
- On-site
air contamination
DLst
Ratlon.
*
- Dust

Radon t
Inhalation -

I
I
-
Plant Foods I Plants
I

r
I
I
1 . I
I
Meat
Livestock Milk
I,
I

Fish
ingestion

Water
On-site water contamination

-
__c

Soi I
Direct ingestion

Figure 21.12 Schematic representation of pathways by which humans could receive a

dose from radionuclides in soil (from Yu and others, 1993).
 21.4 Risk Assessment 453

concentrations related to each of the exposure pathways Values of Cihave units such as, mg/m’ in air, mg/L in
(for example, dust radon, meat, milk, etc.; Figure 21.12). water, or mg/kg in food. Values of R, have units like W
These concentrations are usually unavailable, even for day in the case of water. In some cases, TI in Eq. 2 1 .S
the analysis of present conditions. Models, then, provide is multiplied by a dimensionless fraction (1 or less) that
predictions of biosphere migration and accumulation. As represents the fraction of the year or the fraction of one’s
an example, we will consider the code RESRAD (Yu and lifetime that a person spends on the site. The total intake
others, 1993) that calculates risk and doses related to (TOin this latter case becomes the average daily exposure
radioactive contaminants. The plant food pathway (Fig- rate expected during an average lifetime. For comparxtive
ure 2 1.12) considers more specific transfer of radionu- purposes, values of TI can be normalized to one kg o f
clides due to (1) root uptake of crops grown in contami- body weight (divide by body weight, BW). Thus,intake or
nated soil, (2) foliar uptake from Contaminated dust exposure levels have units like mg/day or mg/kg-BW day.
deposited on the plant, (3) root uptake of contaminated
irrigation water, and ( 4 ) foliar uptake from contaminated
irrigation water. Example 21.2
These general models are customized to represent de- An individual consumes water, milk, and
tails of the contaminated site (for example, whether sur- beef mg/L, o.oo3 mg/L, and 9.5 10-~
face water is available for imfsation), and the lifestyles mg/kg of the contaminant, respectively. Calculate tile
of individuals living in the vicinity of the site. Thus, assess- exposure level to a receptor who weighs 70 kg and
rnents tend to be quite site specific. To provide some 1.5 wday of water, 0.5 wday of milk, and
basis for comparisons among sites, investigators rely on 0.2 kg/day of beef.
reference scenarios. One of the most well known is the
“born loser” scenario. This scenario presents the case
for a subsistence farmer who spends his or her entire
life on a farm located at the site boundary. This farmer
drinks and bathes in contaminated water, eats contami- (3.0 X 1.5) + (0.003 X (9.5 X lo-’ X 0.2)
0.5) i-
nated food grown on the site, and is exposed to contami- TI =
70.0
nated dust. This scenario represents the case for the
maximally exposed individual, although this individual = 0.064 mg/kg-BW day
would not be found at most sites.
Multimedia models are commonly used to track the The greatest intake comes from drinking contaminated
complexities of contaminant fate and transport within water.
the biosphere. These models represent environmental
media, water, soil, or plants as homogeneous compart-
ments and estimate contaminant fluxes among compart-
Toxicity Assessment
ments or accumulation within compartments (Renner,
1W5). The models are formulated based on conservation The toxicity assessment evaluates toxicological effects
of mass principles accompanied by appropriate source/ and carcinogenic potentials of chemical and radioactive
sink adjustments related to chemical and biological pro- contaminants. The objectives of the toxicological assess-
cesses. Examples of such models include CalTOX, Chem- ment are to assess the toxicity of the potential contami-
CAN, HAZCHEM, and SimpleBOX (Cowan and others, nants for the site considered and to provide the toxicity
1995). values (reference doses for noncarcinogenic effects;
The biosphere analysis provides an estimate of contam- slope factors for carcinogenic effects) for risk assessment.
inant concentrations available in environmental media. Reference Dose (RfD) is an estimate (with an uncertainty
This information establishes intake or exposure levels typically spanning an order of magnitude) of daily expo-
for each relevant environmental medium, plus a total sure (mg/kg day) to the general human population (in-
exposure. The intake equation for air, water, milk, and cluding sensitive subgroups) that is likely to be without
food has the following form an appreciable risk of deleterious effects during a lifetime
of exposure. Slope Factor (SF) (or called risk factor, KF)
(21.5) is the slope of the dose-response curve in the lowdose
region. Source materials for developing toxicological pro-
where TI is the total intake of contaminants with units files include the EPA’s Integrated Risk Inforniatioii System
muday, and El is the estimated daily intake for each of the (IRIS), the U.S. EPA Health Effects Assessment Summary
environmental media or other exposures (for example, Tables (HEAST), the Agency for Toxic Substance Disease
dermal). EI values for eating and drinking are determined Registry Toxicological Profiles (the Department of Health
by multiplying C,, the contaminant concentration in the and Human Services), the Handbook on t1.e To,\.ico/r,~~~
fth environmental medium, by R,, the rate of intake. of Metals, and the peer-reviewed scientific Iite~iture.
-
454 Cbapter 21 / In Sftu Destruction and Rfsk Assessment

Health-Risk Assessment Table 21.3 Conversion Factors for Common Units of

Radionuclide Activity and Dose
The health impact of a contaminant depends not on the
intake to the human body but on the dose that the body Multiply BY To Obtain
~~

receives. For example, some contaminants may pass pCi lo-” Ci

through the body without accumulating in various or- pCi 0.037 Becquerel (Bq)
gans. One of the simplest ways of determining the dose is mrem 0.001 rem
to multiply the intake by a factor, called the bioavailability mrem 10-5 Sieverts (Sv)
factor. This factor represents the fraction of the daily
intake of the contaminant that is retained to produce the
adverse health effect. The total dose of a contaminant DCF, is the dose conversion factor with units of (mrem/
(TD) is obtained as curie) that converts nuclide activity into dose in mrems.
TD = EIi * Bl + El2 * Bz + * * . + EI,l*B,, (21.7) The dose, mrem (milli Roentgen Equivalent Man), ac-
counts for the relative biological damage due to the type
where B, is the bioavailability factor. Values of B, are
of rddiation involved (FERMCO, 1994). Table 21.3 sum-
available from the medical literature. TDvalues have units
marizes the equivalences between commonly used med-
of mass/time and are usually normalized by dividing by
sures of radionuclide activity and dose.
the body weight (for example, mg/kg-BW day).
Standards for radionuclides are often presented in
Before discussing how these dose calculations are
terms of dose. For example, in the United States, DOE
used, let us consider dose thresholds. The existence of
Order 5400.5 (U.S.DOE, 1990) provides guidelines for
a dose threshold implies that if the contaminant intake
radiation exposure to the public based on internationally
is below some threshold, there is no health effect. When
accepted standards. Figure 21.13 shows how the dose
thresholds do not exist, even a very small dose may have
limits are applied to air and ground-water pathways at
health implications. Cancer risks are typically modeled
the DOE site at Femald, Ohio. The 100 mrem/yr dose
in this manner given that even one molecule of a contami-
accounts for all pathways and is the weighted sum of
nant may impact DNA and lead to the growth of a tumor
the dose in various organs.
(Reichard and others, 1990).
Although intake or dose-based standards are intended
One of the most straightforward approaches to screen-
to be protective of public health, they may not capture
ing health risks where thresholds are applicable is to
compare the total daily dose (TD values) to reference
standards. The hazard quotient (HQ) for the human re-
ceptor and the particular contaminant is calculated as

HQ=-
Rfo
TD
(2 1.8)
-ri Regulations tticrt limit swats
pathway dcss w i d e D
reference point for measuring
the Fwnald site complme.
with an HQ > 1 indicating some risk to the receptor. DOE Order 5400.5 cha?ges
This same approach is followed with radionuclides, 1 0
that no tndrvldual in Me genoral
public shall be exposed 10 lm
where the dose is established depending on the type of mrem per year. from cwnhined
radiation and the location of the radiation (that is, inside sources, is a rewlt d Jcte
or outside of the body). Thus, the dose will vary de- Operations durrng any ye%r.
pending on the body organs in which the radionuclides This order further indicates
accumulate. Dose then can be expressed as an “organ” that IM indwiduai in the @nerrl
dose or some “effective” dose that involves a weighted p u m shall recetve 10 mrsm per
average of the doses to various organs. The weights are year from the air pathway
(excludingradon). This standard
proportional to the total risk associated with the irradia- ISadopted from the Natronal
tion of those organs f l u and others, 1993). Elntsston Standards Iw Hazardous
Air Pvllrstants of the Clean Air Act
Dose conversion factors provide a relatively straightfor-
ward way of estimating a dose from some intake quantit). Finally, the wder mandates that
These factors are derived from detailed modeling of the no pmon in the general Public
shall geewe greater than
decay rates, daughter in-growths, and the metabolism of 4 mrcm per year from drinking
radionuclides in the body. The dose to a particular organ wtu. This standard conforms
from a given nuclide would be calculated as to Nattonat Prrmary Drinking
Water Stzvndards of the Safe
TD EIi * DCFI + EIZ * DCFZ + . + EI,,
* * * DCF, Drrnlrmg water Act.
(21.9)
Ffgure 22.13 Example of dose limits applied to Depart-
where TD is dose with units mrem/yr, EI, as before is ment of Energy sites, in particular the site at Fernald, Ohio
C, Ri (now for example, curies kg-’ X kg yr-‘), and (from FERMCO, 1994).
 21.4 Risk Assessment

health risks like cancer where there is no chronic thresh- Categorization Screening
old. A major problem in assembling the dose-response
data from laboratory experiments is that health effects
I
of chemicals in animals are collected when large doses
are applied. These data need to be extrapolated to the t
U.S.’EPA DOE Enviionmental
t
U.S.’EPA
t
AESRAO Pliot
ti
low doses that are more common in contaminant settings HRS Survey (MEPAS) Baseline Version 5.0 Study
Brookhaven
(Reichard and others, 1990).
Let us briefly consider how lifetime cancer risks are Increasingdata needs, site specificity
calculated in the case of exposure to some radionuclide. More realistic estimates replace default assumptions
Commonly risk is presented as individual lifetime risk, Increasingconfidence in estimating uncertainty
*
the estimated increase in probability that an individual lncreasini confidence in risk estimate
will experience a health effect over a lifetime, due to
exposure (Hamilton and others, 1993). As before, the Figure 21.14 Spectrum of approaches to risk assessment
intake (EI) of the nuclide pCi/yr is multiplied by the dose (modified from Hamilton and others, 1993).
conversion factor in mrem/pCi to obtain a dose. These
effective doses are summed over an assumed lifetime of
70 years as E,,L. The ifldividudl lifetime risks for cancer The system was revised in 1992 to consider relative risks
( I l R ) are determined as to human health by scoring exposure along four path-
ways. Sites scoring above 28.5 may be proposed for the
National Priority List.
ILR = E7,L * RF (21.10)
The HRS evaluation is quite detailed, but makes use
of only basic information about the site. The ground-
where E,,)L is the sum of effective doses over a 70-year water migration pathway examines the likelihood for
lifetime (mrem), and RF is the risk factor (deaths/mrem) releases to the aquifer, the characteristics of the waste
(Hamilton and others, 1993). These risk factors are de- (such as toxicity, mobility, volume), and impact (present
rived from internationally accepted standards. and future) on various “targets” (that is, wells, popula-
Hamilton and others (1 993) explain in simple terms tions, resources). The HRS, however, does not provide
how risk data can be interpreted. Assuming a U.S.popula- a quantitative assessment of potential risks associated
tion of 240 million, a (one in a million) excess risk with contamination at a given site (Hamilton and oth-
would mean that 240 people would die prematurely ers, 1993).
sometime within their approximate 70-year life span. The Figure 2 1.14 identifies “screening” approaches, which
resulting excess 3.4 deaths per year would compare to provide increased complexity and accuracy in the risk
2.1 million deaths each year from all causes. The U.S. assessment. Examples of these are the Multimedia Envi-
EPA standard for excess individual cancer risks is 1 X ronmental Exposure System (MEPAS) model (Droppo and
lo+, or one in 10,000. others, 1990) and baseline risk assessments performed
at Superfund sites as part of Remedial Investigation/Feasi-
bility Studies (RIDS)(U.S. EPA, 1989a).
Types of Risk Assessments The MEPAS model is a quantitative risk assessment
model, designed to assist in the investigation of DOE
Risk assessments can vary in complexity, scope, data sites. It consists of pathway models coupled to a set of
needs, and ultimately the confidence in the overall result, exposure models that form the basis for an assessment
as represented in Figure 21.14 (Hamilton and others, of health effects associated with the site and relative
1993). Categorization studies provide a simple way to rankings among environmental problems. Although this
identify from a collection of contaminated sites those model can be used in a detailed and site-specific manner,
that are of greatest potential concern. For example, with the risk assessment is still quite simplified (Hamilton and
the passage of the Comprehensive Environmental Re- others, 1993). Baseline risk assessments are also consid-
sponse, Compensation and Liability Act (CERCLA; that ered “screening-level” assessment because of consewd-
is, the Superfund Law), the U.S. EPA was required to tive assumptions that are embedded in the analyses and
develop priorities for undertaking remedial action and simplified flow and transport models (Hamilton and oth-
possibly removal actions. The Hazard Ranking System ers, 1993).
(HRS) was implemented in 1982 as a scoring system to Moving to the right in Figure 21.14, the analyses be-
determine whether or not a given hazardous site should come more rigorous and incorporate more of the detail
be placed on the National Priority List for further study of the actual site. The RESRAD code was developed at
and remediation. The original HRS evaluated three migra- Argonne National Laboratory to implement DOE guide-
tion pathways (ground water, surface water, and air), as lines for residual radioactive materials in soils. The model
well as threats from direct contact, fire, and explosions. considers the critical pathways for human exposure, such
4 s Cbapter 21 / In Situ Destruction and RkFk Assessment

as direct exposure, internal doses due to inhalation, and of such a modeling effort are tied strongly to an extensive
internal doses due to eating or drinking contaminated data base of information about the site.
materials.
The Brookhaven Pilot Study exemplifiesa detailed and
Environmental R#sk Assessment
quantitatively rigorous analysis. This kind of assessment
might be carried out at a site that poses significant health The objective of environmental assessment is to deter-
concerns. Unlike the screening assessments,which liely mine if the contaminantsdetected at a particular site have
contain conservative assumptions and safety factors, the the potential to adversely affect the existing biological
realistic approach employs reasonable parameter esti- community (plants, wildlife, and livestock) at or sur-
mates and realistic exposure pathways (Hamilton and rounding the site. The U.S. EPA guideline (1989~)may
others, 1993). Flow and transport models that are pm be used qualitatively for this purpose.

-
21.5 Fernald Case Study absent. The so-called South Plume (Figure 2 1.16) contains
uranium concentrationsas high as 27 to 190 pCi/L (Hamil-
The Fernald Case illustrates the application of two differ- ton and others, 1993) with a relatively large zone with
ent health-risk assessments. Fernald is a DOE facility that concentrations above 13.5 pCi/L (20 p@).
produced uranium metal for defense activities. Produc- FERMCO (1994) describes the results of a risk assess-
tion at the site ceased in 1989, and now activities are ment designed to evaluate whether the site met the DOE
targeted toward cleanup and closure (FERMCO, 1994). guidelines in Figure 21.14. The analysis examined both
The site is located along the Great Miami River, approxi- air and water pathways (Table 21.4) using the CAP-88
mately 27 km northwest of Cincinnati, Ohio. A consider- computer code.
able volume of uranium metal waste (low-levelradioactiv- Shown in Table 21.5 is the calculated dose in 1993 to
ity) is contained in six waste pits. The site also has two some maximally exposed individual, determined mainly
silos that contain wastes from refining pitchblende ore. by where this person lives. As indicated, the greatest
These silos leak radon gas to the atmosphere. A cross dose would come from consumption of contaminated
section (Figure 21.15) shows how the site is situated well water. Comparison with applicable guidelines
on a mainly low-permeability, glacialdrift unit. This unit shows that the total dose of 1 mrem is far less than
protected the underlying Great Miami Aquifer from con- the U.S. DOE guideline. Similar dose calculations made
tamination to a significant extent. However, surface-wa- annually since 1989 show that restoration activities have
ter drainage from the site toward the Miami River (for reduced the annual dose to the maximally exposed indi-
example, along Paddys Run; Figure 21.16) enables con- vidual from about 19 mrem in 1989 to about 1 mrem at
taminants to enter the aquifer where the drift unit is present (FERMCO, 1994).

Sir!

-Shale with interbedded limestone

Figure 21.15 Map showing the location of the South Plume. Contaminationoccurred due
to infiltration of water from Paddys Run and the Outfall Ditch, where the overburden is a b
sent (modified from FERMCO,1994).
 21.5 Femald Case Study 457

Figure 21.16 Fast-west geologic cross section at the Fernald site. The regional water table oc-
curs in the Great Miami aquifer. Locally, the overburden is saturated (modified from FERMCO,
1994).

Deta#led Risk Assessment ters. The threshold for kidney damage was determined
to be in a range of 0.1 to 1 pg-U/g-kidney (Hamilton and
Researchers from Brookhaven National Laboratory and others, 1993).
Lawrence Livermore National Laboratory used the Fer- Uranium also may contribute to the development of
nald site to demonstrate an approach to nonconservative human cancers. The main problem in dealing with health
health-based risk assessments (Hamilton and others, risks is that the environmental exposures are relatively
1993).Their approach emphasizes the use of site-specific low. To develop realistic risk numbers, Hamilton and
data, multidimensional transport codes, and the latest others (1993) examined to what extent uranium accumu-
scientific information describing dose -response relation- lated in bones over a lifetime to facilitate estimating dose
ships. Their analysis examined the implications of radon from intake rates. The estimate of the risk factor came
emissions from the K65 silos and uranium contamination from assessments of the results of international studies.
associated with the South Plume. Our discussion here The risk factor from lowdose radiation was estimated to
focuses on the plume. be 0.05 deaths/Sievert.
The main toxic effect for both long- and short-term The analysis was designed to predict the exposure
exposure is related to the liver, where uranium tends to and risks 70 years into the future. Estimates of uranium
accumulate. However, it has been generally difficult to concentrations in the ground water for 70 years were
characterize the extent of kidney damage. In the absence obtained using the flow and transport code SWIFT I11
of detailed epidemiological or other data, conservative with two conditions of source loading. The first assumed
dose-response relationships are often used (Hamilton that the source of uranium to the South Plume was re-
and others, 1993). The study of Hamilton and others moved after 10 years. The second assumed that loading
modeled uranium transfer within the body with a Monte continued to occur at rates comparable to the modeling
Carlo scheme to account for uncertainty of the parame- period. Details on the grid design and parameter selection
4 3 Cbapter 21 / In Situ Destruction and Risk Assessment

Tab& 21.4 Summary of Pathways for Human Table 21.6 (from Hamilton and others, 1993). Note that
Exposure to Radioactive Contamination from the separate equations are written for toxic intakes and can-
Fernald Site cer-causingintakes. The unit for intake in this latter case
~-
is Becquerel (Bq), which is a unit comparable to pCi.
~

Pathway Description
The total uranium dose intake from the sources above
Air are given as
Airborne emissions Dust, water mists, considers re-
lease points, wind speed, direc- (21.1 1)
tion, and population distribution
Foodstuffs Dose from eating foodstuffs TI, (luglday) = E L + EI,, + EI", + EI,, (2 1.12)
grown within about 5 km of the
site-due to uranium deposited The problems at the end of the chapter will explore a few
in the soil of the calculations with these equations in more detail.
Direct radiation Doses from gamma and X-rays The next steps in the toxicity analysis involve con-
due to radioactive waste stored in verting the intakes to concentrations of uranium in the
K 6 5 silos, and thorium elsewhere kidney and comparing these results to the effect thresh-
Water old discussed previously. Hamilton and others (1993)
Well water Dose provided by the consump
propose the following equation to estimate the concen-
tion of water containing uranium
in areas south and east of the
tration of uranium in the kidney from some chronic
plant intake
Fish from Great Various water pathways let ura-
Miami River nium enter the river. Dose is pro-
vided by eating 4.5 kg of fish
each year. where A is the uranium mass in the liver (pg LJ), TI, is
the uranium intake rate (pg/day),f, is the fractional trans-
fer from gut to blood, a! is the fractional transfer of blood
to kidney compartment 1, and T is the biological half-
are provided by Hamilton and others (1993). The model life of uranium in kidney compartment 1 (days). The
was coupled to the risk assessment in the sense that parameters f,, a,and T were assigned mean values of
individual concentration estimates for each node could 1.03, .063, and 1 1.1 days, respectively. When A is divided
be converted to dose and cancer risks. In effect, Hamilton by the total kidney mass, one has a uranium concentration
and others (1993) were able to describe health risks in (pg-U/g-kidney) that can be compared to toxic threshold
detail as a function of location and time. developed previously. Given that model predictions of
The risk assessment considered three exposure path- uranium concentrations in ground water are available for
ways for contaminated water: direct ingestion in water, 70 years, one could carry out such a calculation to assess
irrigation of crops, and watering of dairy cattle. The multi- future risks as well as present risks.
media analysis led to estimates of uranium concentration
in soil and vegetables and in home-produced food and
Example 21.3
milk. For the liver assessment, the uranium intake in food
from other sources was also included. These concentra- A person drinks 1.2 Wday of water containing a uranium
tions were converted to intake using the equations in concentration of 20 pg/L. Assuming a kidney mass of

Tab& 21.5 Estimates of the 1993 Dose to the Maximally Exposed Individual
Dose Attributable
Pathway to the Site Applicable Guideline
Air
Estimated 1993 emissions 0.016 mrem 10 mrem/air
Foodstuffs grown in Femald area 0.01 mrem 100 rnrem/all pathways
Direct radiation 0.0 mrem 100 mrem/all pathways
Liquid
WeU water in the Femald area 0.7 mrem 4 mrem/drinking water
Fish fromGreat Miami River 0.01 mrem 100 mrem/all pathways

Maximally exposed individual -1.0 mmm 100 mrem/all pathways

From FERMCO (1994).

 21.5 Femald Case Study 453

Tab& 21.6 Summary of Equations Used to Calculate Intake Due to Ground-Water

Exposure
Pathway Equation and Parameters
Ingestion in water C,,, @qn> = C,,,(pg/L.) * 0.68 @Ci/pg> . 0.037 (Bq/pCi)
El,,, (Bq/day) = C,,, (Bq/L) * R,. (Wday)
E L (pg/day) = C,,,,( p f l ) . R,. @/day)
C,,, = uranium concentration expressed as activity (c cancer)
C., = uranium concentration in water (t toxic)
Elu, = daily intake for calculating cancer risk
El,,, = daily intake for calculating risk of liver damage
R,. = rate of water intake
Ingestion in home- C,,,c@q/L) = C,,, (mg/L) 0.68<pCi/ps) * 1000 (pg/mg) * 0.037 (Bq/pCi)
produced milk -
El,,,, (Bq/day) = C,,,,(Bq/L) R,,, (Uday)
E L , (pg/day) = C,,,,(mfl) . 1000 (Wmg) * R," (Uday)
C,,,, = uranium concentration in milk expressed as activity
Cw,f= uranium concentration in milk
EI,,,< = daily intake for calculating cancer risk
EI,,,, = daily intake for calculating risk of liver damage
R,,, = rate of milk intake
Ingestion in home- C,, (Bq/kg) = C,, (mgkg) . 0.68 (pCi/pg) . lo00 (pg/mg) . 0.037 (Bq/pCi)
grown fruit and El,, (Bq/day) = FH . C,, (Bq/kg) . R,. (kg/day)
vegetables EI,, (pg/day) = FH * C,,(mg/kg) loo0 (pg/mg) . R,. (kg/day)
a

C,, = uranium concentration in fruit and vegetables as activity

C,, = uranium concentration in fruit and vegetables
EII, = daily intake for calculating cancer risk
El,,f = daily intake for calculating risk of liver damage
R,. = rate of fruit and vegetable intake
FH = fraction of fruit and vegetables that are homegrown
Ingestion of food E L (pg/day) = FN . R, (Wday)
from other = daily intake for calculating risk of liver damage
sources R,, = rate of food intake from other sources
FN = fraction of fruit and vegetables that are not homegrown (1 - FH)

750 g, no other uranium intakes, and the parameters At Fernald, Hamilton and others (1993) predicted no
given for uranium behavior in the kidney (Hamilton toxic effects due to the South Plume. They estimated
et al., 1993), assess whether this intake could be toxic. that uranium concentrations needed to be of the order
of 500 pg/L t o be capable of producing liver concentrd-
tions at the threshold. The main dose came from drinking
contaminated water, rather than from milk or food in-
takes.
The cancer risk requires integrating intakes over a pe-
riod of time. In effect, a dose estimate is obtained for
intakes over each of 70 years. The sum of these effective
doses (E,,,L [Sv]) is multiplied by the risk factor (deaths/
Sv) to estimate the individual lifetime risks for cancer
mortality. The analysis showed that all cancer risks were
small. The future risk for using water from wells south
Expressing concentration per gram of kidney is 24.9/750
of the facility was 2.2 X lo-$.
or 0.033 pg-U/g-kidney. This concentration is less than
Overall, the results of the two risk assessments for
the threshold for toxicity (0.1 to 1 .O pg-U/g-kidney) and
Fernald yielded comparable results. The health risk for
so the intake is not toxic.
individuals in the vicinity of the site is small.
460 Cbapter 21 / In Situ Destruction and Risk Assessment

Problems
1. An in situ respiration test yielded an oxygen utiliza- average lifetime (that is, multiply b y 5 which in-
tion rate of -O.lbb%/hr. Assuming a porosity of 0.30, a cludes this adjustment).
soil bulk density of 1440 kg/m3, an oxygen density of b. Assuming a bioavailability factor of 0.64, estimate
1330 m a , and hexane as the reference hydrocarbon, the daily dose as mg/kg-BW day.
determine the biodegradation rate. c. The chronic oral exposure limit (RfD) for this con-
2. A 70-kg rancher living close to a contaminated site taminant is 1 X mg/kg-BW day. Estimate the
is exposed to a contaminantthrough ( 1 ) the consumption hazard quotient, and determine whether the contam-
of homegrown meat, and (2) the incidental ingestion of inant poses a health risk.
dust. Assuming the following information: 3. A residential well in the vicinity of a hypothetical
For Beef For Dust site was found to contain uranium at a concentration of
100 kg/L. Estimate the risk of this well causing liver
cb = 9.5 x mg/kg cd = 0.024 mg/kg damage. Assume that exposure comes from drinking con-
Rb = 0.18 kg/day Rd = 28 mg/day taminated water and milk and eating contaminated food.
fb = 0.25 fd = 0.17 The following information is provided in the table.
Note f b multiplies Cab, accounting for the fact that the a. Calculate the intake of uranium from each of the
rancher was exposed for only a frdction of his life, and environmental media (that is, EZwt,EZmt, and EZJ.
not all the beef was homegrown. fd is used similarly, b. Calculate the total intake of uranium (TZJ.
reflecting that not all dust is contaminated and the source c. Using the data concerning the behavior of uranium
was present for a fraction of one person’s lifetime. in the body, estimate the concentration in the kidney
a. Use Eq. 21.5 to estimate the total contaminant intake with a mass of 500 g.
(as mg/day and mg/kg-BW day) expressed over an d. Is this dose toxic?

Environmental Intake Fraction Home

Medfum Concentration Rate Produced
Water 100 pg/L 1.2 L/day na
Milk 0.oob mg/L 0.25 L/day 1.
Food 0.30 mg/kg 0.36 kg/day 0.22
 Answers to Problems

Cbapter 1 5. Q = 0.247 ft3/s

1. k = 1.18 x 10-7s-1 6. Inland distance 4125 ft with seaward flow reduced 8250 ft
2. k = 2.22 x 10-7 s-I
Recharge = 1870 X 106 ft3 Cbapter 6
3. Recession length 7.7 months (Answers subject to interpolation of tabulated values.)
4. Change in storage = 6.561 X lo6 m3 1. T = 1850m2/day
2. T = 520 m2/day,S = 1.8 x
Cbapter 3
3. 1 year 49 m, 3 years 51 m
1. q = 1.89 X 10-'m/s
2. Wective porosity = 0.1, u = 1.9 X lo-' m/s 4. Q = 427 m3/day
Effective porosity = 0.0001,u = 1.9 X m/s 5. Q = 2249 m3/day
7. Total head: 300 m,205 m, 110 m, 15 m 6. Zero drawdown at stream, 3.7 m at midpoint
Pressure head: 200 m, 190 rn, 50 m, 8 m 7. Approximately 39%
Elevation head: 100 m, 15 m,60 m, 7 m
11. a. Approximately 6.8 m
8. K = 3.28 X ft/s, K = 2.12 gal/day ft2,
k = 1X cm2,k = 1.04 X lo-' darcys 12. a. 49,543 m
b. 2.28 m
11. K, = 3.7 X m/s, K, = 2.7 X m/s
13. In order of blanks and selections
12. qx = 1 x iO-'crn/s
100, 6, 40, 4, add, 30, 2, add
16. qx = 5 X lo-' cm/s, downward flow
14. 594 m
17. a. Intlow both aquifers, Q = 1 X m3/s
b. Outflow aquifer A, Q = 1.5 X lo-' m3/s 15. Approximately 27 m
Outflow aquifer B, Q = 5 X lo-' m3/s
c. Outflow area A, 1.5 X lo-'
Cbapter 8
Outflow area B , 0.5 X 10-1 1. a. Elastic compression = 0.042 ft
b. Inelastic compression = 2 ft
Cbapter 4 c. Bulk modulus = 3 X lo7 Ib/ft2
2. a. 300m3/m 3. s = 2.2 x 10-4
b. S = 8.4 X 7. 19%at 2000 m, 60%at 10,OOO m
c. 3.6 X lo-'' m2/N
8. a. 150 m/my
4. a. 6250 Ib/ft2, 4368 Ib/ft2, 1882 Ib/ft* b. 300 m/my
b. 4998 Ib/ft2, 4368 Ib/ft2, 630 Ib/ft' d. 15
d 30ft
11. About 40" above ambient
5. 378 m2/day
13. Change in total stress 450 kbars, change in k i d pressure
Cbapter 5 54 kbars, change in effective stress 396 kbars

(Answers subject to interpolation of tabulated values.) Cbapter 9

3. a At the canal 7.77 m,5.25 m, 4.2 m, 17.9 m 1. 1.9 and 0.8 cal m-'s-I O C - ' for the saturated sandstone
At the divide 21 m, 21 m, 20 m, 17.9 m
b. q' = 1379 m2/yr at 10 years, q' = 250 m2/yr at 100 3. 4 = 2.5 lo-'
Years 4. a k = 7 X 10-9cm2
462 Answers to Problems

Cbapter 10 6. 0.85 mL/g

1. Fractured 2.85 yr; unfractured 2850 yr 9. C = 0.91C,,
2. Approximately 130 years 10. 0.45 cm/day
3. 4 x 10-6cmz/s 12. Input data:
5. Advection 0.1 mg/m2s; diffusion 0.4 X mg/m2s 10 7
6. b. 2 X 10-2cm 4000.00 5.00 0.50 0.10 0.00 5.97
14.00 200.00 50.00 0.00
7. 28.2 m, 8.9 m, and 2.8 m in x, y, z, respectively
5.00 150.00
8. Approximately 0.25 m
Output Data:
9. 10 m 14 Years
10. a. 18 mz 0.00 88.02 2195.20 2283.22 2195.20 88.02 0.00
b. 0.21 m 0.15 179.12 1516.28 1695.10 15 16.28 179.12 0.15
1.55 216.39 1191.31 1404.60 1191.31 216.39 1.55
Cbapter.I 1 4.92 231.56 997.65 1219.36 997.65 231.56 4.92
1. a. 2.3 X M, 0.83 X lo-' M ; 5.5 X 10-3 M 9.33 226.28 028.12 1035.77 828.12 226.28 9.33
M; 4.7 meq/L; 0.83 meq/L; 5.47 meq/L 6.75 104.52 339.73 430.82 339.73 104.52 6.75
0.25 X
0.49 5.48 16.27 20.78 16.27 5.48 0.49
0.25 meq/L
0.00 0.01 0.04 0.05 0.04 0.01 0.00
b. 0.012 M 0.00 0.00 0.00 0.00 0.00 0.00 0.00
c. 0.66 and 0.90 0.00 0.00 0.00 0.00 0.00 0.00 0.00
3. Sl = 0.03 slightly oversaturated
4. Cation/anion balance-0.99; good analysis Cbapter 19
1. 2 X lo4 m3
Cbapter 12 2. a. k,, = 0.47
2. a. 1.7 X lO-'atm b. k, = 0.04
b. Strongly oversaturated - IAP/K = 9.2 c. S", = 0.2
3. a. 2 x atm d. S, = 0.1
b. 0.1 atm 3. Volume of aquifer containing residual = 5.71 m'
4. a. n = 0.75; K = 56 4. Depth of residual = 571 m
b. Kd = 0.60 mL/g 5. 22.4% sorbed, 77.6%in solution
5. a. (i) 3480, (ii) 355, (iii) 5.2 X lo5 6. NAPL phase 79,500 g/m3, gas phase 113.2 g/m'
b. (i) 34.8, (ii) 3.6, (iii) 5180
7. Effective solubility = 2.2 X lo-' g/L,
6. 4.16, 3.36, 4.57 Mass in aqueous phase = 0.26 g/m3
7. a. 5 , 1 , 6 , 2 , 3 , 4 8. Retardation due to aqueous phase = 5.83, solid phase =
b. 4, 3, 2, 6, 1, 5 137.2
8. pe = 4.65; En = 0.27 V
9. Fe3+= 9.1 X lo-?; Fez+= 9.1 X Cbapter 20
10. 13.8 pg/L 1. a. 6 X m3/s
12. 8'0 = - IS0/,,,,; SD = - 135°/,mSMOW b. Feasible-drawdown about 2.5 m after 1 year
2. a. 56mg/L
Cbapter 13 b. 10.8 kg/day
1. 13.9 days c. Agrees
2. 0.092 day-', 7.5 day 4. 1.4 x cm/s
3. a. Acetic acid, and 1,l-DCE Cbapter 21
b. Acetic acid
1. -3.1 mg hydrocarbon/kg soil day
Cbapter 15 2. a. EIb = 4.3 X mg/day or 6.I X lo-" mg/kg-BW/day
7 . 2850 years old EId= 1.1 X 10-7mg/dayor 1.6 X 10-9mg/kg-BW/day
TI, = 1.1 X lO-?mg/day or 1.6 X 10-Ymg/kg-BW/day
8. 24,300 years old
b. TD = 1.1 X mg/kg-BW/day
9. About 1950, 1963, 1970, 1976 c. HQ = 1.1 X
Cbapter 18 3. a EIa,, = 120 @day
E L , = 1.5 W d a y
1. 14cm EC,, = 23.7 &day
3. 0.13 b. TI, = 145 &day
5. a. 36cm2/hr c. A = 0.30 F g / g
b. 1.0 m d. Possibly, at threshold
 ADDendix A

Denovationof the Flow Equation

in a Deforming Medium

In the derivation of the flow equation in Chapter 4 it

was stated that Eq. 4.7 was inadequate in that both the
fluids and the solids are in motion in deforming media,
requiring two conservation statements, one for the fluids If the fractional compression is small, the material deriv;i-
and one for the solids. These conservation statements are tives are replaced by partial derivatives
d(p,,.n>
-div( p,,nv,,.) = -
at
(Al.la) - (-)! n ap,,. 1 a12
div(p,.q) = -- + ___ - (Al.5)
pll. dt (1 - n) i/t
a[Ps(l - n)l (Al.lb)
-div[p,(l - n)v,] = Now we may use the relationships discussed in Chap-
at ter 4 to get the correct form of the diffusion equation.
where v,,is the fluid velocity, v, is the solid velocity, and First, it is assumed that the density does not vary spati;illy
ps is the density of the solids. In this form pa,n is the total so that it can be taken out as a constant on the left-livid
mass of fluids per unit volume and p,(l - n) is the total side of Eq. A1.5. Substituting Darcy's law in the first term
mass of solids per unit volume. In this development v,. of Eq. A1.5 gives us, as described previously
and v, are the average velocities of the fluids and the -div q = KV'b (Al.6)
solids with respect to a stationary coordinate system. The
Darcy flux q is described relative to the solid matrix The first term on the right-hand side of Eq. Al.5 becomes.
by way of Eq. 4.20,
q = n(v,. - v,) (A1.2)
Now we must introduce the idea of a material derivative
that follows the motion of the solid phase, that is, Dn/
Dt = an/& + v, . V n , which we notice becomes equal In the last term of Eq. A1.5 we recognize that h / ( I -
to the partial derivative when the last term is small. Intro- n ) is merely the change in pore volume per unit volume
ducing Eq. Al.2 into Eq. A l . l a and assuming further that d y , / V Twhere,
, in accordance with our definition ofcom-
the solid grains are incompressible, p, = constant, Eq. pressibility (Eq. 4.26) and effective stress, d y , / V , =
A1 .1 becomes - &,cG = &&P = p,,.g&ib. Thus, the last tecni of Eq.
A1.5 becomes
D( pll,n>
-div( p,.q) - pll.ndiv(v,) = ~
(Al.3a) 1 an -
Dt (A1 . X I
1-nat
Dn
(1 - n) div(vJ = - (A1.3b) Collecting Eqs. Al.6, Al.7, and Al.8 gives the diffusion
Dt
equation
Eliminating the divergence of the solids between Eqs.
A1.3 gives the final form for the conservation of fluid
and solid mass (Palciauskas and Domenico, 1980)

463
 Appendix B

About the Computer Disk

A computer disk is provided with the book. It contains rone.for -source FORTRAN code for the one-dim-
the well hydraulics package WELLz and a number of en-
other code segments provided with the book. The stan- sional model, Table 18.2
dard high density disk should be readable on most DOS
in.dat -example data set used in Example 18.5
or Windows-based personal computers. To help organize
the information on the disk, the various code segments rone.exe -needs to be created by compiling and
are installed in separate subdirectories. The contents of linking rone.exe
each subdirectory are listed below.
nhree.for -source code for the three-dimensional
code
aAwoost in3.dat -example data set used in Example 18.6
In Chapter 7, we briefly described the application of rthree.exe -executable code
MODFLOW to the Wooster study site described by
Springer and Bair (1992). This subdirectory contains the Copy these files onto your hard drive. Once installed,
MODFLOW data files necessary to run this program in- the code is run as follows
cluding 1. Click on the START button on the Windows 95 task
woo.bas, woo.bcf, woo.ghb, woo.oc, wooxch, woo.riv, bar or choose the File menu on Windows 3 . 1 ~
woo.sip, woo.we1. 2. Choose the RUN command
This data set was tested using MODFLOW/EM (version 3. Type c:\transbthree.exe
9412) from Maximal Engineering Software, Inc. Two 4. Close the C: window that opens
other files are provided in the data set. The first is 5. Read/print the output (out3.dat) using any standard
woo.mfn, which contains the list of MODFLOW codes word processor
used by MODFLOW/EM. The second is woo.nam, which
is an MFI/EM preprocessor name file containing a list of Normally, one can make changes to the input file (in3.dat)
all files used by the MFI Preprocessor and the I-numbers using a word processor. However, be careful to preserve
assigned to each. Copy these files to your hard drive. the input format as shown in the book. Also when you
have made the changes, save the modified file (in3.dat)
in ASCII format. A FORTRAN code will not be able to
=\trans read the standard files created by a word processor.

Chapter 18 contains two simplecodes for simulating one-

and three-dimensional mass transport. The code listings
are supplied along with the sample dataset. In the inter-
a=\wellz
ests of conserving space, only the three-dimensional code This directory contains the Windows version of WELLz.
is provided as an executable (that is, exe file). We have It has been developed and tested with Windows 95. It
used various versions of Lahey compilers to compile runs with Windows 3.1x, although on some older ma-
these codes. chines it can be somewhat unstable.
464
 Appendix B 4 s

The code installs automatically. Choose the RUN com- I. To determine what contour lines are shown:
mand (under the START button on Windows 95, or under -click ‘Options’and select ‘Contour values’
File menu on Windows 3.1~).Type ’a:\setup’or ’b:\setup’ -you can either keep the default set of contours or
as appropriate. The code operates as a normal Windows provide your own set of values by editing the
program. Details on operating the code are provided in ‘Contour values dialog’ box
Chapter 6. Here we present a few additional details. -accept the values by clicking ‘OK’ and the plot
will reflect your choice of contours
2. To label contours or selected points:
-click ‘Options’and select ‘Labeldrawdown-values’
Files on tbe Disk
-click the mouse to add labels at appropriate points
Several data files are provided on the disk. Each time a 3. To get rid of contour labels or other labels:
user selects File-New, the data set “test” is loaded into If you save a data set and later reopen it and edit
the Windows. This file is edited by the user for the prob- parameters, the labels on the contour line will re-
lem at hand. When you select File-Open, note that test main. The old labels can be removed in one of two
is there as “test.wzf ”. There are two other files present ways. With the contour plot in the Window
on the disk “ex6-lO.wzf” and “ex6-1l.wzf” that are
-click ‘File’ and select ‘New’ this returns you to
the data files to run Examples 6.10 and 6.11. As you
the first window, whereupon you run through
create new data sets, these are saved as with the
the set of Windows and obtain a new plot, or
“wzf” extensions.
-click ‘Editing’and proceed to the last Window
4. Plotting precautions:
For the plot to be prepared correctly, the region to
be plotted must have coordinate locations greater
Adding and Labeling Contour Lines
than zero. Stated another way, the coordinate axis
The code lets you plot data points and label the contour (0, 0) should always be at the bottom left of the
lines for drawdown. For example, run “test” by proceed- contour map. It is possible to use negative well coor-
ing through the various data screens in order. The plot dinates, but the contour plot cannot include these
illustrates a pattern of circular drawdown with no labels. wells.
 Appendix C

Table of Atomic Weights

Atomlc Atomic Atomic Atomic

Element Symbol Number Weight Element Symbol Number Weight
Actinium AC 89 Hydrogen H 1 1.01
Aluminum Al 13 26.98 Indium In 49 114.82
Americium Am 95 Iodine I 53 126.90
Antimony Sb 51 121.75 Iridium Ir 77 192.20
Argon Ar 18 39.95 Iron Fe 26 55.85
Arsenic As 33 74.92 Krypton Kr 36 83.80
Astatine At 85 Lanthanum La 57 138.91
Barium Ba 56 137.34 Lead Pb 82 207.19
Berkelium Bk 97 Lithium Li 3 6.94
Beryllium Be 4 9.01 Lutetiurn Lu 71 174.97
Bismuth Bi 83 208.98 Magnesium Mg 12 24.31
Boron B 5 10.81 Manganese Mn 25 54.94
Bromine Br 35 79.91 Mendelevium Md 101
Cadmium Cd 48 112.30 Mercury Hg 80 200.59
Calcium Ca 20 40.08 Molybdenum Mo 42 95.94
Californium Cf 98 Neodymium Nd 60 144.24
Carbon C 6 12.01 Neon Ne 10 20.18
Cerium Ce 58 140.12 Neptunium NP 93
Cesium cs 55 132.90 Nickel Ni 28 58.71
Chlorine Cl 17 35.45 Niobium Nb 41 92.91
Chromium Cr 24 52.00 Nitrogen N 7 14.01
Cobalt co 27 58.93 Nobelium No 102
Copper cu 29 63.54 Osmium 0s 76 190.20
Curium Cm 96 Oxygen 0 8 16.00
Dysprosium DY 66 162.50 Palladium Pd 46 106.40
Einsteinium Es w Phosphorus P 15 30.97
Erbium Er & 167.26 Platinum Pt 78 195.09
Europium Eu 63 151.96 Plutonium Pu 94
Fermium Fm 100 Polonium Po 84
Fluorine F 9 19.00 Potassium K 19 39.10
Francium Fr 87 Praseodymium Pr 59 140.91
Gadolinium Gd 64 157.25 Promethium Pm 61
Gallium Ga 31 69.72 Protactinium Pa 91
Germanium Ge 32 72.59 Radium Ra 88
Gold Au 79 196.97 Radon Rn 86
Hafnium Hf 72 178.49 Rhenium Rt! 75 186.20
Helium He 2 4.00 Rhodium Rh 45 102.90
Holmium Ho 67 164.93 Rubidium Rb 37 85.47
 AppendixC 467

Atomic Atomic Atomic Atomic

Element Symbol Number Weight Element Symbol Number Weight
Ruthenium Ru 44 101.07 Thallium TI 81 204.3'
Samarium Sm 62 150.35 Thorium Th 90 232.04
Scandium sc 21 44.96 Thulium Tm 69 168.93
Selenium Se 34 78.96 Tin Sn 50 I 18.69
Silicon Si 14 28.09 Titanium Ti --
77 P.90
Silver Ag 47 107.87 Tungsten W 74 IH.3.85
Sodium Na 11 22.99 LJranium U 92 2.58.03
Strontium Sr 38 87.62 Vanadium V 23 50.94
Sulfur S 16 32.06 Xenon Xe 54 13 I .30
Tantalum Ta 73 180.95 Ytterbium Yb 70 173.04
Technetium Tc 43 Yttrium Y 39 M.90
Tellurium Te 52 127.60 Zinc Zn 30 65.37
Terbium Th 65 158.92 Zirconium Zr 40 91.22
 References

Abriola, L. M. 1988. Multiphase flow and transport models for Anderson, M. P., and J. A. Munter. 1981. Seasonal reversals of
organic chemicals: A review and assessment. Electric Power groundwater flow around lakes and the relevance to stagna-
Research Inst., Palo Alto, Calif. tion points and lake budgets. Water Resources Res., v. 17,
Adams, E. E., and L. W. Gelhar. 1992. Field study of dispersion in p. 1139-1150.
a heterogeneous aquifer, 2. Spatial moments analysis. Water Anderson, M.P., and W. W. Woessner. 1992. Applied Ground-
Resources Res., v. 28, no. 12, p. 3293-3307. water Modeling. Academic Press, San Diego, Calif., 381 p.
Ahlfeld, D. P., A. Dahmani, and W. Ji. 1994.A conceptual model Anderson, M. R., R. L. Johnson, and J. F. R~IICOW. 1987. The
of field behavior of air sparging, and its implications for dissolution of dense non-aqueous phase liquid (DNAPL) from
application. Ground Water Mon. Rev., Fall, p. 132-139. a saturated porous medium. Proc. of Petroleum Hydrocar-
Ahlstrom, S. W., H. P. Foote, R. C. Arnett, C. R. Cole, and bons and Organic Chemicals in Ground Water: Prevention,
R. J. Seme. 1977. Multicomponent Mass Transport. Model: Detection and Restoration, National Ground Water Assoc.,
Theory and Numerical Implementation (discrete-parcel- Columbus, Ohio, p. 409-428.
random-walk version). Battelle Northwest Laboratories, Anderson, R. E., and R. L. Laney. 19-j. '[he influence of late
BNWL- 2 127, Richland, Wash. Cenozoic stratigraphy on distribution of impoundment re-
Ala, N. K., and P. A. Domenico. 1992. Inverse analytical tech- lated seismicity at Lake Mead, Nevada, Arizona. U.S. Geol.
niques applied to coincident contaminant distributions at SurveyJ. Res., v. 3, p. 337-343.
Otis Air Force Base, Massachusetts. Ground Water, v. 30, Aris, R., and N. R. Amundson. 1973. Mathematical Methods
p. 212-218. in Chemical Engineering, 11. Prentice-Hall, Englewood Cliffs,
Albertsen, M., and an ad hoc Task Force. 1986. Beurteilung
N.J., 369 p.
und Behandlung von MineralUlschadensfallen im Hinblick
Arizona Department of Water Resources. 1994. Riparian Protec-
auf den Grundwasserschutz, Jeil 1, Die wissenschaftlichen
tion Program Legislative Report. Arizona Department of Wa-
Grundlagen zum Versandnis des Verhaltens von Mineral01
ter Resources, Phoenix, Ariz., p. 209-280.
im Untergrund. Federal Office of the Environment, LTw S
Armstrong, J. M., W. Korreck, L. E. Leach, R. M. Powell, S. V.
no. 20, 178 p.
Alexander, M. 1994. Biodegradation and bioremediation: Aca- Vandergrift, and J. T. Wilson. 1988. Bioremediation of a fuel
demic Press, San Diego, Calif., 302p. spill: Evaluation of techniques for preliminary site character-
Alpay, 0. A. 1972. A practical approach to defining reservoir ization. Proc. of Petroleum Hydrocarbons and Organic Chem-
heterogeneity: J. Petrol. Technol., v. 20, no. 7, p. 841-848. icals in Ground Water: Prevention, Detection and Restora-
American Chemical Society. 1980. Guidelines for data acquisi- tion. National Ground Water Assoc., Columbus, Ohio, p.
tion and data quality evaluation in environmental chemistry. 931 -944.
Anal. Chem., v. 52, p. 2242-2249. Athy, L. F. 1930. Density, porosity, and compaction of sedimen-
Amundson, N. R. 1950. Mathematics of adsorption in beds, 11. tary rocks. Bull. Am. Assoc. Petrol. Geol. v. 14, p. 1-24.
J. Phys. Colloid. Chem., v. 54, p. 812-820. Atkinson, T. C. 1977. Diffuse flow and conduit flow in limestone
Anderson, G. M., and R. W. Macqueen. 1982. Ore deposit mod- terrain in the Mendip Hills, Somerset (Great Britain). J. Hy-
els-Mississippi Valley type lead zinc deposits. Geoscience drol., v. 35, p, 93- 110.
Canada, v. 9, p. 108- 117. Atwater, G. I., and E. E. Miller. 1965. The effect of decreases
Anderson, M. P. 1984. Movement of contaminants in ground- in porosity with depth on future development of oil and
water: groundwater transport-advection and dispersion. gas reserves in South Louisiana. Paper presented at AAPG
Groundwater Contamination, M C Studies in Geophysics, meeting, New Orleans.
National Academy Press, Washington, D.C.. p. 37-45. Atwood, D. F., and S . M. Gorelick. 1985. Hydraulic gradient
 References 463
control for groundwater contaminant removal. J. Hydrol., lead-zinc deposits of the Mississippi Valley Region. Geol. Soc.
v. 76, p. 85- 106. Am. Special Paper, v. 24, 16 p.
Ayres, J. E., D. C. Lager, and M. J. Barvenik. 1983. The first EPA Bathhurst, R. G. C. 1971. Carbonate sediments and their diagen-
Superfund cutoff wall: Design and specifications. In D. M. esis. In Developments in Sedimentology, no. 12, Elsevier.
Nielsen (ed.), Proc. of the Third National Symposium on New York.
Aquifer Restoration and Ground Water Monitoring. National Baumann, P. 1963. Theoretical and practical aspects of well
Water Well Assoc., Dublin, Ohio, p. 13-22. recharge. Trans. Am. SOC.Civil Engrs., v. 128. p. 739-764.
Back, W. 1960. Origin of hydrochemical facies of groundwater Baumann, P. 1965. Technical development in groundwater re-
in the Atlantic Coastal Plain. Proc. Intern. Geol. Cong., Copen charge. In V. T. Chow (ed.), Advances in Hydroscience, v.
hagen, p, 87-95. 2, p. 209-279.
Back, W. 1961. Techniques for mapping of hydrocheniical fa- Bear,J., and Y.Bachmat. 1967. Ageneralized theory on hydrody-
cies. U.S. Geol. Surv. Prof. Paper 42&D, p. 380-382. namic dispersion in porous media: Intern. Assoc. Sci. Hydrol.,
Back, W., and R. A. Freeze. 1983. Chemical Hydrogeology. Symp. Haifa, Publ. 72, p. 7-16.
Benchmark papers in Geology. Hutchinson Ross Pub. Co., Bear, J. 1972. Dynamics of Fluids in Porous Media. Elsevier.
Stroudsburg , Pa. New York, 764 p.
Baedecker, M. J.,D. I. Siegal, P. E. Bennett, and I. M. Cozzarelli. Bear, J. 1979. Hydraiilics of Groundwater. McGraw-Hill, New
1989. The fate and effects of crude oil in a shallow aquifer: York.
1. The distribution of chemical species and geochemical fa- Beck, A. E., G. Garven, and L. Stegena. 1989. Hydrogeologic
des. US. Geol. Surv., Water-Resources Investigations Report regimes and their subsurface thermal effects. Geophysical
88-4220. Monogram 47, IUGG, v. 2.
Baehr, A. L. 1987. Selective transport of hydrocarbons in the Behnke J., and W. Bianchi. 1965. Pressure distributions in Iay-
unsaturated zone due to aqueous and vapor phase parti- ered sand columns during transient and steady state flows.
tioning. Water Resources Res., v. 23, p. 1926- 1938. Water Resources Res., v. 1, p. 557-562.
Baetsle, L. H. 1969. Migration of radionuclides in porous media. Bekins, B. A., E. M. Godsy, and D. F. Goerlitz. 1993. Modeling
In A. M. F. Duhamel (ed.), Progress in Nuclear Energy Series steady-state methanogenic degradation of phenols in ground-
XII, Health Physics, Perganion Press, Elmsford, N.Y., p. water. J. Contam. Hydrology. v. 14, p. 279-294.
707-730. Bell, M. L., and A. Nur. 1978. Strength changes due to reservoir
Bahr, J. M., and J. Rubin. 1987. Direct comparison of kinetic and induced pore pressure and stresses and application to h k e
local equilibrium formulations for solute transport affected by Oroville. J. Geophys. Res., v. 83, p. 4469-4483.
surface reactions. Water Resources Res., v. 23, p. 438-452. Bengtsson, G. 1985. Microcosm for ground water research:
Bailey, J. E., and D. F. Ollis. 1987. Biochemical Engineering: Ground Water Quality, C. H. Ward, W. Giger, and P. L.
New York, McGraw-Hill, 753 p. McCarty (eds.), John Wiley & Sons, New York. p. 330-34 1.
Bair, E. S., and T. D. Lahm. 1996. Variations in capturezone Bennett, R. R.,and R. R. Meyer. 1952. Geologyand groundwater
geometry of a partially penetrating pumping well in an uncon- resources of the Baltimore area. Maryland Board Nat. Re-
tined aquifer: Ground Water, v. 34, no. 5, p. 842-852. sources, Dept. Geology, Mines, Water Resources Bull. 4.
Balkwill, D. L., J. K. Fredrickson, and J. M. Tlioni:is. 1989. Bennion, D.W., and J. C. Griffiths. 1966. A stochastic model
Vertical and horizontal variations in the physiological diver- for predicting variations in reservoir rock properties. Trans.
sity of the aerobic chenioheterotrophic bacteria microflora Am. Inst. Mining Met. Engrs., v. 237, no. 2, p. 9- 16.
in deep southeast Coastal Plain subsurface sediments. Appl. Bentley, H. W., F. M. Phillips, S. N. Davis, M. A. Habermehl.
Environ. Microbiol., v. 55, p. 1058- 1065. P. L. Airey, G . E. Calf, D. Elmore, H. E. Grove, and T. Torg-
Ball, J. W., E. A. Jenne, J. M. Uurchard, and A. H. Truesdell. ersen. 1986. Chlorine 36 dating of very old groundwater: 1.
1976. Sampling and preservation techniques for water with The great artesian basin, Australia. Water Resources Res., v.
a section o n gas collection. Proc. First Workshop on Sampling 22, no. 13, p. 1991-2001.
Geothermal Effluents, EPA-600/9-760111, p. 219-234. Berner, R. A. 1974. E. D. Goldberg (ed.), In the Sea, v. 5 . p.
Banks, H. 0. 1953. Utilization of underground storage reser- 427-450. John Wiley & Sons, New York.
voirs. Trans. Am. SOC.Civil Engrs., v. 118, p. 220-234. Berner, R. A. 1980. Early Diagenesis. Princeton University Press,
Banks, H. O., and R. C. Rictor. 1953. Sea water intrusion into Princeton, N.J., 241 p.
groundwater basins bordering the California coast and inland Berner, R. A,, and J. W. Morse. 1974. Dissolution kinetics of
bays. Trans. Am. Geophys. Union, v. 34, p. 575-582. calcium carbonate in sea water: IV.Theon of calcite dissolu-
Barari, A., and L. Hedges. 1985. Movement of water in glacial tion. Am. J. Sci., v. 274, p. 108-134.
till. Memoires, Hydrogeology of Rocks of Low Permeability Betcher, R. N. 1977. Temperature distribution in deep ground-
Inter. Assoc. Sci. Hydrol., Tucson, Ariz., p. 129-134. water flow systems-a finite element model. .Master’s Thcsis.
Barker, J. A., and J. H. Black. 1983. Slug tests in fissured aquifers. Univ. of Waterloo, Ontario, Canada.
Water Resources Res., v. 19, p. 1558-1564. Bethke, C. M. 1985. A numerical model of compaction driven
Barker, J. F., G . C. Patrick, and D. Major. 1987. Natural attenua- groundwater flow and heat transfer and its application t o
tion of aromatic hydrocarbons in a shallow sand aquifer. the paleohydrology of intracratonic sedimentary basins. .1.
Ground Water Mon. Rev., p. 64-71. Geophys. Res., v. 90,p. 6817-6828.
Barker, C. 1972. Aquathermal pressuring-role of temperature Bethke, C. M. 1986. Hydrologic constraints on the genesis of
in development of abnormal pressure zones. Am. Assoc. Pet- the upper Mississippi Valley Mineral District from Illinois
rol. Geol. Bull. 56, p. 2068-2071. Basin Brines. Econ. Geol., v. 81, p. 233-249.
Bastin, E. S. (ed.). 1939. Contributions to the knowledge of the Bethke, C. M. 1988. Origin of long range brine migration across
470 References

the North American craton (abst.). EOS Trans. hi.Geophys. Bosworth, W. 1981. Strain induced partial dissolution of halite.
Union, v. 49, p. 470. Tectonophysics, v. 78, p. 509-525.
Bethke, C., and S. Marshak. 1990. Brine migrations across Nolth Boulton, N. S. 1954. The drawdown of the water table under
America-the plate tectonics of ground water. Ann. Rev. nonsteady conditions near a pumped well in an unconfined
Earth Planet. Sci., v. 18, p. 287-315. formation. Proc. Inst. Civil Engrs., v. 3, p. 564-579.
Bevin, K., and P. Germann. 1982. Macropores and water in Boulton, N. S. 1955. Unsteady radial flow to a pumped well
soils. Water Resources Res.. v. 18, p. 1311 - 1325. allowing for delayed yield from storage. Proc. Gen. Assembly,
Biot, M. A. 1941. General theory of three dimensional consolida- Rome Intern. Assoc. Sci. Hydrol. Publ., v. 37, p. 472-477.
tion. Appl. Physics, v. 12, p. 155-164. Boulton, N. S. 1963. Analysisof data from nonequilibrium p u m p
Biot, M. A., and D. G. Willis. 1957. The elastic coefficients of ing test allowing for delayed yield from storage. Proc. Inst.
the theoryofconsolidation. J. Appl. Mech., v. 24, p. 594-601. Civil Engrs., v. 26, p. 469-482.
Blake S. B., and R. W. Lewis. 1982. Underground oil recovery. Boulton, N. S., and S. V. Streltsova. 1977. Unsteady flow to a
Second National Symposium on Aquifer Restoration arid pumped well in a fissured water bearing formation. J. Hydrol.,
Ground Water Monitoring, Proc., D. M. Nielsen (ed.); Na- V. 35, p. 257-269.

tional Water Well Assoc., Dublin, Ohio, p. 69-75. Bouwer, E. J., and P. H. McCarty. 1984. Modeling of trace
Blake S. B., and J. S. Fryberger. 1983. Containment and recovery organics biotransformation in the subsurface. Ground Water,
of refined hydrocarbons from groundwater. Proc. Seminar v. 22, no. 4, p. 433-440.
on Groundwater and Petroleum Hydrocarbons Protection, Bouwer, H. 1985. Renovating waste water with groundwater
Detection and Restoration. Petroleum Assoc. for Conserva- recharge. In E. T. Smerdon and W. R. Jordan (eds.), Issues
tion of the Canadian Environment, Ottawa, p. 4-1 -4-47. in Groundwater Management, Texas A&M Univ., College Sta-
Blanchard, F. G., and P. Byerly. 1935. A study o f a well gage tion, p. 331-346.
as a seismograph. Bull. Seismol. Soc.Am., v. 25, p. 3 13-32 1. Box, G. E. P., and C.W. Jenkins. 1976. Time Series Analysis:
Blanchard, P. E., and J. Sharp. 1985. Possible free convection Forecasting and Control. Holden-Day lnc., San Francisco,
in thick Gulf Coast sandstone sequences. Southwest Section 575 p.
of the Am. Assoc. Petrol. Geol. Trans., p. 6- 12. Boyd, F. M.,and C. W. Kreitler. 1986. Hydrogeology of a G y p
sum Playa. Northern Salt Basin, Texas. Bureau of Economic
Blandford, T. N., and P. S. Huyakorn. 1991. WHPA Well Head
Geology, The Univ. of Texas at Austin, Report of lnvestiga-
Protection Area Delineation Code. Unpublished report to
tions No. 158, 37 p.
U.S. Environmental Protection Agency, code avdibdbk from
Brace, W. F. 1980. Permeability of crystalline rocks and argella-
Intl. Ground Water Modeling Center, Golden, Colo.
ceous rocks. Intern. J. Rock Mech. Min. Sci. Geomechanics,
Blatt, H. 1979. Diagenetic processes in sandstone. In P. A.
V. 17, p. 241-251.
Scholle and P. R. Schluger (eds.), Aspects of Diagenesis. SOC.
Brace, W. F., B. W.Paulding, and C. Sholz. 1966. Dilatency in
Econ. Paleontol. and Mineral Special Pub. 26, p. 141 - 157.
the fracture of crystalline rocks. J. Geophys. Res., v. 71,
Bodvarsson, G. 1970. Confined fluids as strain meters. J. Getr
p. 3939-3953.
phys. Res., V. 75, p. 271 1-2718.
Brace, W. F., J. B. Walsh, and W. T. Fangos. 1968. Permeability
Boggs, J. M., and others. 1992. Field study of dispersion in a
of granite under high pressure. J. Geophys. Res., v. 63, p.
heterogeneous aquifer. I , Overview and site description: Wa-
2225-2236.
ter Resources Res., v. 28, no. 12, p. 3281-3291.
Bradbury, K. R. 1984. Major ion and isotope geochemistry of
Boggs. S. 1987. Principles of Sedimentology and Stratigraphy. groundwater in clayey till, northwestern Wisconsin, U.S.A. In
Merrill Pub. Co., Columbus, Ohio. B. Hitchon and E. 1. Wallick (eds.), First Canadian/American
Bogomolov, V., and others. 1978. The principles of paleohydro- Conference on Hydrogeology. National Water Well Assoc.,
logical reconsfruction of groundwater formation. Internat. Dublin, Ohio, p. 284-289.
Geol. Congress, 24th Sec 11. p. 205-226. Bradley, J. S. 1975. Abnormal formation pressure. Am. Assoc.
Bolt, G. H., and P. H. Groenvelt. 1969. Coupling phenomena Petrol. Geol. Bull. 59, p. 957-973.
as a possible cause for non-Darcian behavior of water in soil. Brashears, M. L. 1946. Artificial recharge on Long Island, New
Bull. Intern. Assoc. Sci. Hydrol. 14, p. 17-26. York. Econ. Geol., v. 41, p. 503-516.
Bonham, L. C. 1980. Migration of hyrdrocarbons in compacting Bredehoeft, J. D. 1965. The drill stem test: The petroleum indus-
basins. In W. H. Roberts and R. J. Cordell (eds.), Problems try’s deep-well pumping test. Ground Water, v. 3, p. 31-36.
of Petroleum Migration. Am. Assoc. Petrol. Geol. Studies in Bredehoeft, J. D. 1967. Response of well-aquifer systems to
Geology, v. 10, p. 6 9 4 8 , Tulsa, Okla. earth tides. J. Geophys. Res., v. 72, p. 3075-3087.
Borden, R. C., and P. B. Bedient. 1986. Transport of dissolved Bredehoeft J., and B. Hanshaw. 1968. On the maintenance of
hydrocarbons influenced by oxygen-limited biodegradation. anomalous fluid pressures: 1. Thick sedimentary sequences.
1, Theoretical development: Water Resources Res., v. 22, no. Geol. Soc. J. Am. Bull. 79, p. 1097-1106.
13, p. 1973-1982. Bredehoeft, J. D., and L. F. Konikow. 1993. Editorial-Ground-
Born, S. M., S. A. Smith, and D. A. Stephenson. 1974. The water models: validate or invalidate. Ground Water, v. 31,
hydrologic regime of glacial terrain bakes with management no. 2, p. 178-179.
and planning applications. An Inland Lake Renewal and Man- Bredehoeft, J., and 1. S. Papadopulos. 1965. Rates of vertical
agement Demonstration Project Report. Upper Great lakes groundwater movement estimated from the earth’s thermal
Regional Comm. profile. Water Resources Res., v. 1, p. 325-328.
Born, S. M., S. A. Smith, and D. A. Stephenson. 1979. The Bredehoeft, J., and R. Young. 1970. The temporal allocation
hydrologic regime of glacial terrain lakes. J. Hydrol., v. 43, of groundwater-A simulation approach. Water Resources
p. 7-44. R ~ s . ,V. 16, p. 3-21.
 References 471

Breen, K. J., A. L. Kontis, G . L. Rowe, and R. J. Haefner. 1995. Cady, R. C. I94 1 . Effect upon groundwater levels of proposed
Simulated ground-water flow and sources of water in the surface water storage in Flatland Lake. Montana. 17,s. (;eel.
Killbuck Creek Valley near Wooster, Wayne County, Ohio. Survey Water Supply Paper 849 b, p. 51 -81.
U.S. Geological Survey, Water-Res. Investigations Report California I k p t . Water Resources. 1958. Sea water intrusioti in
94-4 131 . California. Calif. Dept. Water Res. Bull. 63, S;icramento.
Brooks, R. H., and A. T. Corey. 1966. Properties of porous California Dept. Water Resources. 1980. Groundwater basins in
media affecting fluid flow. J. Irrigation and Drainage Div., California. Calif. Dept. Water Res. Bull. I 18- 180. Slcmiiento.
Proc. Am. Soc. Civil Eng., v. 92. IR2, p. 61-88. Cant. D. J. 1982. Fluvial facies models and their ;ipplication. I n
Brown, R. F., and T. L. Lambert. 1963. Reconnaissance of ground P. A. Scholle and D. Spearing (eds.). Sandstone I>eposition;tl
water resources in the Mississippi Plateau region, Kentucky. Environments. Am. Assoc. Petrol. Geol. Mem. 3 I . p. I 15- 138.
U S . Geological Survey Water Supply Paper 1603, 58 p. Carder, D. S. 1970. Reservoir loading and local earthquakes. I n
Brown, R. H., and G. G . Parker. 1945. Salt water encroachment W. M. A d a m (ed.), Engineering Seismolog!. (Eng. (ieol. (:;ise
in limestone o f Silver Bluff, Miami, Florida. Econ. Geol., v. Histories #8), Geol. Soc. Am.
40,p. 235-262. Cardwell, W. T., and R. L. Parsons. 1945.Averagc ~'ernie;ihilitic~
Brown, J. A. 1937. Discussion on "Effect of a sea level c;inal of heterogeneous o i l sands. Trans. Am. Inst. Min. 'Met. Engrs.
on the groundwiiter level of Florida." Econ. Geol., v. 32. V. 169, p. 34-42.
p. 589-599. Carlson, W. D. 1983. The polyniorphs o f C X 0 ; ;ind the amgo-
Browne, P. R. L. 1978. Hydrothermal alteration in active geo- nite-calcite transformation. I n R. J. Reeder (4.). <:;trboti;ites:
thermal fields. In F. A. Donath, F. G. Stehli, and G. Wetherill Mineralogy and Chemistry. Reviews in .\lineralogy. v. I 1.
(eds.), Ann. Rev. Earth Planet Sci., v. 6, p. 229-250. Mineral. Soc. Am., p. 191-225.
Bryan, K. 1925. Geology of reservoir and dam sites. 1J.S. Geol. Carlstoti. C. W., L. L. Thatcher. and E. C. Rhodeh;imel. IOOO.
Survey Water Supply Papers, 597A. Tritium as a hydrologic tool. the Wharton Tract Study. Itit.
Buddemeier, R. W., and J. R. Hunt. 1988.Transport of colloidal
Assw. Sci. Hydrol.. Pub. 52. p. 503-512.
contaminants in groundwater: Radionuclide migration at the
Carothers, W. W., and Y.K. Kharaki. 1978.Aliph;itic acid iitiions
Nevada Test Site. Appl. Geochem., v. 3, p. 535-548.
in oil field water-implications for origin o f natural g;is hm.
Buras, N. 1'966. Dynamic programming in water resources de-
Assoc. Petrol. Geol. Bull. 62. p. 2441 -245.3.
velopment. Advances in Hydroscience, V. T. Chow (ed.),
Carr, P. A,, and G. S. van der Kamp. 1969. Drtrrniining aquifer
vol. 3: New York, Academic Press, p. 367-412.
chamcteristics by the tidal method. Water Resources Res..
Bureau of Reclamation. 1960. Design of Small Dams. 1J.S. Gov.
v. 5, p. 1023-1031.
Printing Office, Washington, D.C.
Carrillo, N. 1948. Influence of artisan wells in the sinking of
Bums, W. A. 1969. New single well test fordetermining vertical
Mexico City. Proc. 2nd Intern. Conf. Soil Mcch. ;tiid Found.
permeability. Trans. Am. Inst. Mining Engrs., v. 246, p.
Eng., V. 7, p. 156- 159.
743-752.
Carslaw. H., and J. Jeager. 1959.Conduction o f Heat in Solids.
Burst, J. F. 1'969. Diagenesis of Gulf Coast clayey sediments
and its possible relation to petroleum migration. Am. Assoc. Clarendon Press, Oxford.
Petrol. Geol. Bull. 53, p. 73-93. Cartwright, K. 1970.Groundwater discharge in the Illinois kisiti
Buscheck, T. E., P. E. Kirk, T. O'Reilly, and R. G . Nelson. 1993. as suggested by temperature anomalies. Water Resources
Evaluation of intrinsic bioremediation at field sites: Proc. of Res., v. 6, p. 912-918.
Petroleum Hydrocarbons and Organic Chemicals in Ground Cartwright, K. 1973. The effect of shallow groundwater flow
Water-Prevention, Detection and Restontion. National systems on rock and soil temperatures. 1 ~ 1 1 . 1 Thesis,
~. Geology
Ground Water Assoc., Columbus, Ohio, p. 367-381. Dept. Univ. of Illinois at Urbana.
Busenberg, E., and L. N. Plummer. 1992. Use of chlorofluorc- Cartwright, K.,and M. R. McConias. 1968. Geophysicil sirn.cy.*
methanes (CCI,F and CCIIF,) as hydrologic tracers and age- in the vicinity of sanitary landfills in Northern Illinois. (iround
dating tools: Example, the alluvium and ternace system of Water, v. 16. no. 5, p. 23-50.
Central Oklahoma. Water Resources Res., v. 28, p. 2257- Cathels, L. M. 1977. An analysis of cooling of intrusivrs hy
2283. groundwater convection which includes boiling. Econ.Grol..
Busenberg, E., E. P.Weeks, L. N. Plummer, and R. C. Bartholo- V. 72, p. 804-826.
may. 1993. Age dating ground water by use of chlorofluoro- Cathels, L. M., and A. T. Smith. 1983. Thcrnial constmitits oii
carbons (CCI,F and CCIZF,), and distribution of chlorofluoro- the formation of Mississippi Valley type lead-zinc deposits ;ind
carbons in the unsaturated zone, Snake River Plain Aquifer, their implications for episodic basin dewatering :itid deposit
Idaho National Engineering Laboratory, Idaho: U.S. Geol. genesis. Econ. Geol., v. 78. p. 983- 1002.
Surv. Water Resow. Invest., 93-4054, 47 pp. Cederberg, G. A. 1985. TRANQL: h ground-mter mass-trans-
Butler, R. D. 1984. Hydrogeology of the Dakota aquifer system, port and equilibrium chemistry model for multicomponent
Williston Basin, North Dakota. In D. G. Jorgensen and D. systems. P1i.D. dissertation, Stanford lhiiv.. Stanford. (Atif..
C. Signor (eds.), Proc. of the Geohydrology Dakota Aquifer 117 p.
Symposium, p. 12-23. Water Well JournaI Pub. Co., Wor- Cedergren H. R. 1967. Seepage, Drain:ige. and Flow Nrts. j o h n
thington, Ohio. Wiley & Sons, New York.
Byerlee, J. D. 1967. Frictional characteristics of granite under Chamberlin, T. C. 1885. The requisite and quali&ing conditions
high confining pressure. J. Geophys. Res., v. 72, p. 2629- ofartesian wells. U.S.<;eel. Survey 5th Ann. Rept. p. I3 I - 1'3.
2648. Champ, D. R., J. Gulens, and R. E. Jackson. 19-9. Oxitlation-
-
472 References

reduction sequences in ground-water flow systems. Can. J. Cohen, P., 0. Frank, and B. Foxworthy. 1968. An atlas of Long
Earth Sci., v. 16, no. 1, p. 12-23. Island’s water resources. New York State Water Res. Comm.,
Champ, D. R., J. L. Young, D. E. Robertson, and K. H. Abel. Bull. GW62.
1984. Chemical speciation of long-lived radionuclides in a Cohen, R. M., and J. W. Mercer. 1993. DNAPL Site Evaluation.
shallow groundwater flow system. Water Pollut. Res. J. Can., CRC Press, 2000 Corporate Blvd., Boca Raton, Fla.
V. 19, p. 35-54. Cohen, R. M., R. R. Rambold, C. F. Faust,J. 0.Rumbaugh, 111, and
Chapelle, F. H. 1993. Ground-Water Microbiology and Geo- J. R. Bridge. 1987. Investigation and hydraulic containment of
chemistry. John Wiley & Sons, New York, 424 p. chemical migration: Four landfills in Niagara Falls. Civil Eng.
Chapelle, F. H., J. L. Zetibor, Jr., D. J. Grimes, and L. L. Knobel. Pract., Spring, 1987, p. 33-58.
1987. Bacteria in deep coastal plain sediments of Maryland: Coleman, J. M., and D. B. Prior. 1982. Deltaic environments of
A possible source of CO, to groundwater. Water Resources deposition. In P. A. Scholle and D. Spearing (eds.), Sandstone
Res., v. 23, no. 8, p. 1625-1632. Depositional Environments. Am. Assoc. Petrol. Geol. Mem.
Chapman, R. E. 1980. Mechanical versus thermal causes of 31, p. 139-178.
abnormally high pore pressure in shales. Am. Assoc. Petrol. Collins, R. E. 1961. Flow of Fluids Through Porous Materials.
Geol. Bull. 64,p. 2179-2183. Reinhold, New York.
Chapman, R. E. 1981. Geology and Water. Martinus NijhofW Collins, W. D. 1923. Graphical representation of water analysis.
Dr. W. Junk Pub., Boston. Ind. Eng. Chem., v. 15, p. 394.
Charlax, E., E. Guyon, and N. Rivier. 1984. A criterion for Collins, W. D. 1925. Temperature of water available for individ-
percolation threshold in a random array of plates. Solid State ual use in the United States. U.S. Geol. Surv. Water Supply
Communications, v. 20, p. 999-1002. Papers 520-F, p. 97-104.
Chebotarev, I. I. 1955. Metamorphism of natural water in the Combarnous, M. A., and S. A. Uories. 1975. Hydrothermal con-
crust of weathering. Geochim. Cosmochim. Acta, v. 8, vection in saturated porous media. In V. T. Chow (ed.), Ad-
p. 22-48, 137-170, 198-212. vances in Hydroscience. Academic Press, New York, v. 10,
Cherry,J. A. 1983. Piezometers and other permanently-instakd p. 231 -307.
devices for groundwater quality monitoring. Proc. Seminar Conkling, H. 1946. Utilization of groundwater storage in stream
on Groundwater and Petroleum Hydrocarbons Protection, system development. Trans. Am. Soc. Civil Engrs., v. 1 1 1.
Detection, Restoration. Petroleum Assoc. for Conservation p. 275-305.
of the Canadian Environment, Ottawa, p. IV-1 -IV-39. Cooper, H. H., J. D. Bredehoeft, I. S. Papadopulos. 1967. Re-
Chia, Y. P. 1979. Digital simulation of compaction in sedimen- sponse of a finite diameter well to an instantaneous charge
tary sequences. Ph.D. thesis, Geology Dept., Univ. of llli- of water. Water Resources Res., v. 3, p. 263-269.
nois, Urbana. Cooper, H. H., and C. E. Jacob. 1946. A generalized graphical
Chiang, C. Y., K. R. Loos, R. A. Klopp, and M. C. Beltz. 1989. method for evaluating formation constants and summarizing
A real-time determination of geological/chemical properties well field history. Trans. Am. Geophys. Union, v. 27, p.
of an aquifer by penetration testing. Proc. NWWA/API Con- 526-534.
ference Petroleum Hydrocarbons and Organic Chemicals in Cooper, H. H., and others. 1965. The response of well aquifer
Ground Water-Prevention, Detection and Restoration. Na- systems to seismic waves. J. Geophys. Res., v. 70, p. 3915-
tional Water Well Association, Dublin, Ohio, p. 175-189. 3926.
Chiou, C. T., D. W. Schmedding, and M. Manes. 1982. Pani- Cooper, H. H., and M. I. Rorabaugh. 1963. Groundwater move-
tioning of organic compounds in octanol-water systems. Envi- ments and bank storage due to flood stages in surface streams.
ron. Sci. Technol., v. 16, p. 4-10. US. Geol. Survey Water Supply Paper 153GJ,p. 343-366.
Chun, R. Y. D., L. R. Mitchell, and K. W. Mido. 1964. Groundwa- Corey, J. D., D. R. Nielsen, and J. W. Biggar. 1963. Miscible
ter management for the nation’s future-optimum conjunc- displacement in saturated and unsaturated sandstone. Soil
tive operation of groundwater basins. J. Hydrauls. Div. Am. Sci. SOC. Am. Proc., v. 27, p. 258-262.
SOC.Civil Engrs. HY4, v. 90,p. 79-105. Coustau, H., and others. 1975. Classification hydrodynamique
Claasen, H. C. 1982. Guidelines and Technologies for Obtaining des basins sedimentaires utilisation combinke avec d’autres
Water Samples that Accurately Represent the Water Chemis- methodes pour rationalister I’exploration dans des bassins
try of an Aquifer. U.S. Geol. Surv. Open-File Report 82-1024, non produotifs. Proc. 9th World Petrol. Congr. Tokyo. Sec-
49 p. tion 11, p. 105-1 19.
Cleary, R. W. 1978. Report on 208 Long Island pollution study. Cowan, C. E., and others. 1995. The Multimedia Fate Model:
Water Resources Program, Princeton University. Princeton, A Vital Tool for Predicting the Fate of Chemicals. SETAC
N.J. Press, Pensacola, Fla.
Cochran, G. 1968. Optimization of conjunctive use of ground- Cozzareli, I. M., R. P. Eganhouse, and M. J. Baedecker. 1988.
water and surface water for urban supply. M.S. Thesis in Civil The fate and effects of crude oil in a shallow aquifer: 11.
Eng., Univ. Nevada at Reno. Evidence of anaerobic degradation of monoaromatic hydro-
Cochran, J. R., and K. E. Dalton. 1995. Using highdensity mag carbons. U.S. Geol. Survey Toxic Waste Ground-Water Con-
netic and electromagnetic data for waste site characteriza- tamination Program, U.S. Geol. Surv.Water-Resources Investi-
tion. Proc. of the Symposium on the Application of Gec- gations Rept., 88-4220, p. 21 -33.
physics to Engineering and Environmental Problems, Envi- Craig, H. 1961. Isotopic variations in meteoric water. Science,
ronmental and Engineering Geophysical Society, Englewood V. 133, p. 1702-1703.
Cola., p. 117- 127. Craun. G. F. 1984. Health aspects of groundwater pollution.
 References 473

In G. Bitton and C. P. Gerba (eds.), Groundwater Pollution Fussell. 1979. Protection of ground water forni oil pollution.
Microbiology. John WiIey & Sons, New York, p. 135- 179. CONCAWE, The Hague.
Croff, A. G., T. F. Lomenick, R. S. Lowrie, and S. H. Stow. 1985. dePastrovich, T. L., and others. 1979. Protection ofgroundwater
Evaluation of five sedimentary rocks other than salt for high from oil pollution: CONCAWE, The Hague, 61 p.
level waste repository siting purposes. Oak Ridge Ndt. Lab. DeBoer, R. B. 1977. On the thermodynamics of pressure solu-
ORNL/CF-85/2/V2. Oak Ridge, Tenn. tion-interaction between chemical and mechanical forces.
Cuevas, J. A. 1936. Foundation conditions in Mexico City. Proc. Geochim. Cosmochim. Acta, v. 41, p. 249-256.
Intern. Conf. Soil Mech., v. 3, Cambridge, Mass. Deere, D. U., and F. D. Patton. 1967. Effect of pore pressure
Dagan, G. 1967. A method of determining the permeability and on the stability of slopes. Geol. SOC.Am. Soc. Civil Engrs.
effective porosity of unconfined anisotropic aquifers. Water Symp., New Orleans, La.
Resources Res. v. 3, p. 1059-1071. Deju, R. A. 1973. A worldwide look at the occurrence of high
Dagan, G. 1982. Stochastic modeling of groundwater flow by fluid pressures in petroliferous basins. Special publication of
unconditional and conditional probabilities. 2. The solute Gulf Research and Development Co.
transport. Water Resources Res., v. 18, no. 4, p. 835-848. Delhomme, J. P. 1979. Spatial variability and uncertainty in
Dagan, G. 1984. Solute transport in heterogeneous porous for- groundwater flow parameters: A geostatistical approach. Wa-
mations. J. Fluid Mech., v. 145, p. 15 1 - 177. ter Resources Res., v. 15, no. 2, p. 269-280.
Dagan, G. 1989. Flow and Transport in Porous Formations. Demond, A. H., and P. V. Roberts. 1987. An examination of
Springer-Verlag, New York, 465 p. relative permeability relations for two-phase flow in porous
Dagan, G., and J. Bear. 1968. Solving the problem of local media. Water Resources Bull. 23, no. 4, p. 617-628.
interface upconing in a coastal aquifer by the method of Dept. of Energy. 1980. Statement of position of the United
small perturbations. J. Hydrol. Res., v. 6, p. 15-44. States Department of Energy, proposed rule making on the
Daines, S. R. 1982. Aquathermal pressuring and geopressure storage and disposal of nuclear waste. Washington, D.C.
evaluation. Am. Assoc. Petrol. Geol. Bull. 66, p. 931-939. Dept. of Energy. 1986. Environmental assessment. Reference
Daniels, J. J. 1989. Technical review: ground penetrating radar.
repository location, Hanford Site, Washington. DOE/RW-
In Proc. (2nd) Symposium on the Application to Engineering
0070.
and Environmental Problems, SOC.Engineering and Mineral
Dept. of Energy. 1988. Consultation draft site characterizdtion
Exploration Geophysicists, Golden, Colo., p. 62- 142.
plan, Yucca Mountain site, Nevada. US DOE OCRWM, 1-111.
Darcy, H. P. G. 1856. Les fontaines publiques de la Ville de
Washington, D.C.
Dijon. Victor Dalmont, Paris.
Desaulniers, D. E., J. A. Cherry, and P. Fritz. 1981. Origin.
Darton, N. H. 1909. Geology and underground waters of South
age, and movement of pore water in argillareous quaternary
Dakota. U.S. Geol. Sum. Water Supply Papers 227.
deposits at four sites in southwestern Ontario. J. Hydrol., v.
Daus, A. D., and E. 0. Frind. 1985. An alternating direction
50, p. 231-257.
Galerkin technique for simulation of contaminant transport
DeSitter, L. U. 1956. Structural Geology. McCraw-Hill, New
in complex groundwater systems. Water Resources Rex, v.
21, no. 5, p. 653-664. York.
Davidson, M. R. 1984. A green-ampt model of infiltrdtion in a Dewers, T., and P. Ortoleva. 1988. The role of geochemical self-
cracked soil. Water Resources Res., v. 20, p. 1685-1690. organization in the migration and trapping of hydrocarbons.
Davidson, M. R. 1985. Numerical calculations of saturated-unsat- Appl. Geochem., v. 3, p. 287-316.
urated infiltration in a cracked soil. Water Resources Res., v. Dickinson, G. 1953. Geologic aspects of abnormal reservoir
21, p. 709-714. pressures in Gulf Coast Louisiana. Am. Assoc. Petrol. Geol.
Davis, G. H., and others. 1959. Groundwater conditions and Bull. 37, p. 410-432.
storage capacity in the San Joaquin Valley, California: U.S. Dienes, J. K. 1982. Permeability, percolation. and statistical
Geol. Survey Water Supply Papers 1469. crack mechanisms. In Issues in Rock Mechanics, Proc. 22nd
Davis, G. H., J. 8. Small, andH. B. Counts. 1963. Land subsidence Symp. on Rock Mechanics, Berkeley, Calif.
related to decline of artesian pressure in the Ocala limestone Dillon, R. T., R. B. Lantz, and S. B. Pahwa. 1978. Risk Methodol-
at Savannah, Georgia. Eng. Geol. Case Histories 4, Geol. SOC. ogy for Geologic Disposal of Radioactive Waste. The Sandia
Am., p. 1-8. Waste lsolation and Flow (SWIFT) Model. NUREG/CR-O424,
Davis, J. C. 1986. Statistics and data analysis in geology: John Nuclear Regulatory Comm., Wdshington. D.C.
Wiley & Sons, New York, 646 p. Domenico, P. A. 1972. Concepts and Models in Groundw:iter
Davis, S. N. 1969. In R. J. M. Dewiest (ed.), Porosity and Perme- Hydrology. McGraw-Hill, New York.
ability of Natural Materials in Flow through Porous Materials. Domenico, P. A. 1977. Transport phenomena in chemical rdte
Academic Press, New York, p. 54-89. processes in sediments. In F. A. Donath. F. G. Stehli. and
Davis, S. N., and R. J. M. De Wiest. 1966. Hydrogeology. John G. Wetherill (eds.), Ann Rev. Earth Planet Sci.. v. 5, 1’.
Wiley & Sons, New York, 463 p. 287-3 17.
d e Swaan, A. 0. 1976. Analytical solutions for determining Domenico, P. A. 1983. Determination of bulk rock properties
naturally fractured reservoir properties by well testing. SOC. from groundwater level fluctuations. Bull. Assoc. Eng. (ieol.
Petrol. Eng. J., v. 16, p. 117- 122. 20, p. 283-287.
d e Marsily, G. 1985. Flow and transport in fracture rock. Mem- Domenico, P. A. 1987.An analytical model for multidimensional
oires. Hydrogeology of Rocks of Low Permeability. Intern. transport of a decaying contaminant species. J. Hydrol., v.
Assoc. Hydrogeol., p. 267-277, Tucson, Ariz. 91, p. 49-58.
de Pastrovich, T. L., Y.Baradat, R. Barthel, A. Chiavelli, and D. R. Domenico, P. A., and M. D. Mifflin. 1965. Water from low
 permeability sediments and land subsidence. Water Re- Dunbar, C. O., and K. M. Waage. 1969. Historical Geology. John
sources Res., v. 4, p. 563-576. Wiley Sr Sons, New York.
Domenico, P. A., M. D. Miflin, and A. L. Mindling. 19(6. Geo- Dunham, K. C. 1970. Mineralization by deep formation waters:
logic controls on land subsidence in Las Vegas Valley. Proc. a review. Inst. Min. Metall. Trans., v. 79, p. B127-Bl36.
Ann. Eng. Geol., Soils Eng. Symp., Moscow, Idaho, p. Dunkle S. A., L. N. Plummer, E. Busenberg, P. J. Phillips, J. M.
113- 121. Denver, P. A. Hamilton, R. L. Michel, andT. B. Coplen. 1993.
Domenico, P. A., and V. V. Palciauskas. 1973. Theoretical analy- Chlorofluomhydrocarbons (CC13F and CCI,F,) as dating tools
sis of forced convective heat transfer in regional groundwater and hydrologic tracers in shallow groundwater of the Del-
flow. Geol. SOC.Am. Bull. 84, p. 3803-3814. niarva Peninsula, Atlantic Coastal Plain, United States. Water
Domenico, P. A., and V. V. Palciauskas. 1979. Thermal expan- Resources Res., v. 29, p. 3837-3860.
sion of fluids and fracture initiation in compacting sediments. Durant, N. 11.. L. P. Wilson, and E. J. Bouwer. 1995. Microcosm
Geol. SOC.Am. Bull. Part 11, 90,p. 953-979. studies of subsurface PAHdegrading bacteria from a former
Domenico, P. A., and V. V. Palciauskas. 1982. Alternative bound- manufactured gas plant. J. Contam. Hydrol., v. 17, p.
aries in solid waste management. Ground Water, v. 20, p. 21 3-237.
303- 3 1 1. Eakin, T. A. 1964. Groundwater appraisal of Coyote Spring and
Domenico, P. A., and V. V. Palciauskas. 1988. The generation Kane Spring Valleys and Muddy River Springs area, Lincoln
and dissipation of abnormal fluid pressures in active deposi- and Clark Counties. Nevada. Groundwater Resources Recon-
tion environments. In W. Rack, J. S. Rosenshein, and P. R. n .Ii .w
. ,n c e Series Rept. 25, U.S. Geol. Survey, Carson City, Nev.
Seaber (eds.), The Geology of North America, Hydrogeology, Eakin, T. A. 196% A regional interbasin groundwater system in
v. 0-2, p. 435-445. the White River area, southeastern Nevada. Water Resources
Domenico, P. A., and G. A. Robbins. 1984. A dispersion scale Res., v. 2, no. 2, p. 251-271.
effect in model calibrations and field tracer experiments. J. Eakin, T. A., and others. 1976. Summary appraisals of the Na-
Hydrol., v. 7, p. 121-132. tion's groundwater resources: Great Basin. U.S. Geol. Survey
Domenico, P. A., and G. A. Robbins. 1985a. The displacement Prof. Paper 813-G.
of connate water from aquifers, Geol. Soc. Am. Bull. 96, Earlougher, R. C. 1977. Pressure buildup testing. Advances in
p. 328-335. Well Test Analysis. Monograph 5, SOC. Petrol. Engrs. Am.
Domenico, P. A., and G. A. Robbins. 1985b. A new method of Inst. Mech. Engrs., New York.
contaminant plume analysis. Ground Water, v. 23, p. Eaton, J. P., and K. J. Takasaki. 1959. Seismological interpreta-
476 - 485. tion of earthquake induced water level fluctuations in wells.
Donaldson, I. G. 1962. Temperature gradients in the upper Bull. Seismol. Soc. Am., v. 49, p. 227-245.
layers of the earth's crust due to convective water flows. J. Edge, R. W., and K. Cordry. 1989. The Hydropunch'": An in
Geophys. Res., v. 67, p. 3449-3459. situ sampling too1 for collecting ground water from unconsol-
Downey, J. S. 1984. Geology and hydrology of the Madison idated sediments. Ground Water Mon. Rev., v. 9, no. 3,
limestone and associated rocks in parts of Montana, Nebraska, p. 177-183.
North Dakota, South Dakota, and Wyoming. U.S. Geol. Suw. Eliott, T. 1978. Deltas. In H. G. Reading (ed.), Sedimentary
Prof. Paper 12736, 152 p. Environments and Facies. Blackwell, Oxford, p. 97- 142.
Dozy, J. J. 1970. A geologic model for genesis of lead zinc ores Elkins, J. W., T. M. Thompson, T. H. Swanson, J. H. Butler,
of the Mississippi Valley. U.S.A., Inst. Min. Metal. Trans., v. R. D. Hall, S. 0. Cummings, D. A. Fisher, A. G. Raffo. 1993.
79, p. B163-Bl70. Decrease in growth rates of atmospheric chlorofluorocar-
Drever, J. I. 1982. The Geochemistry of Natural Waters. Pren- bons 1 1 and 12: Nature, 364, 780-783.
tice-Hall, Englewood Cliffs, N.J., 388 p. Elzeftawy, A., R. S. Mansell, and H. M. Selim. 1976. Distribution
Dreybrodt, W. 1990. The role of dissolution kinetics in the of water and herbicide in Lakeland Sand during initial stages
development of karst aquifers in limestone: A model simula- of infiltration. Soil Sci., v. 122, p. 297-307.
tion of karst evolution. J. Cieol., v. 98, p. 639-655. Endoe, H. K. 1984. Mechanical transport in two dimensional
Driscoll, F. G. 1986. Groundwater and Wells. Johnson Division, networks of fractures. Ph.D. Dissertation. Univ. of Califor-
St. Paul, Minn. nia, Berkeley.
Droppo, J. G., Jr.,J. W. Buck, D. L. Strenge, and M. R. Siegcl. Energy Research and Development Administration (ERDA).
1990. Analysis ofHealth Impact Inputs to the U.S. Department 1976. A bibliography of geothermal resources-exploration
of Energy's Risk Information System. Pacific Northwest L a b and exploitation. Washington, D.C.
ratory, PNL-7432, Richland, Wash. Ernest, L. F. 1969. Groundwater flow in the Netherlands delta
Duffield, G. M., D.R. Buss, D. E. Stephenson, and J. W. Mercer. area and its influence on the salt balance of the future Lake
1987. A grid refinement approach to flow and transport mod- Zeeland. J. Hydrol., v. 8, p. 137-172.
cling of a proposed groundwater corrective action at the Evans, D. 1966. The Denver area earthquakes and the Rocky
Savannah River Plant, Aiken, South Carolina. Solving Ground Mountain Arsenal Disposal Well. Mtn. Geol., v. 3, p. 23-36.
Water Problems with Models, Proc.,Vol. 2: National Water Ewers, R. 0.1982. Cavern development in the dimensions of
Well Assoc., Dublin, Ohio, p. 1087- I 120. length and breadth. Unpublished Ph.D. Thesis, McMaster
Duguid, J. O., and M. Reeves. 1976. Material transport through University, Hamilton, Canada, 398 p.
porous media: A finite element Galerkin model. Oak Ridge Fabryka-Martin, J., D. 0. Whittemore, S. N. Davis, P. W. Kubik,
National Laboratory, ORNL-4928, Oak Ridge, Tenn. and P. Sharma. 1991. Geochemistry of halogens in the Milk
Dullien, F. A. L. 1979. Porous Media: Fluid Transport and Pore River aquifer, Alberta, Canada: Applied Geochemistry, v. 6,
Structure. Academic Press, New York, 3% p. pp. 447-464.
 References

Fabryka-Martin, J., S. J. Wightman, W. J. Murphy, M. P. Wick- the existing models and a new approach: Water Resources
ham, M. W. Cafee, G. J. Nimz, J. R. Southon, and P. Sharma. Res., v. 15, no. 2, p. 399-413.
1993. Distribution of chlorine-36 in the unsaturated zone at Fontes, J. C. 1980. Environmental isotopes in groundwater hy
Yucca Mountain: an indicator of fast transpon paths. Proc. drology, Chapter 3. In P. Fritz and J. C. Fontes (eds.), Hand-
Focus ‘93: Site Characterization and Model Validation, Am. book of Environmental Isotope Geochemistn, v. 1 . Elsevier.
Nuclear Soc., La Grange Park, Ill., p. 58-68. Amsterdam, p. 75- 140.
Falta, R. W., I. Javandel, K. Pruess, and P. A. Witherspoon. 1989. Ford, D. C., and P. Williams. 1989. Karst geomorphology and
Density drive flow of gas in the unsaturated zone due to hydrology. Unwin Hyman, London, 601 p.
evaporation of volatile organic chemicals. Water Resources Foster, C., and L. Smith. 1988. Groundwater flow systems i n
Res., v. 25, p. 2159-2169. mountainous terrain: 2. Controlling factors. Water Resources
Farmer, V. E. 1983. Behaviour of petroleum contaminants in Res., v. 24, p. 1011-1023.
an underground environment. Proc. Seminar on Groundwa- Foster, C.. and L. Smith. 1989. The influence of groundwater
ter and Petroleum Hydrocarbons, Protection, Detection, Res- flow on thermal regimes in mountainous terrain. J. Geophys.
toration. Petroleum Assoc. for Conservation of the Canadian Res., v. 94, p, 9439-9451.
Environment, Ottawa, 11-1-11-16. Forstner, U., and T. W. Wittman. 1981. Metal Pollution in the
Farr, A. M., R. J. Houghtalen, and D. B. McWhorter. 1990. Aquatic Environment. Springer-Verlag. Berlin, Heidelberg.
Volume estimation of light nonaqueous phase liquids in po- 486 p.
rous media. Ground Water, v. 28, p. 48-56. Foster, M. D. 1950. The origin of high sodium bicarbonate
Farvolden, R. N. 1964. Geologic controls on groundwater stor- waters in the Atlantic and Gulf Coastal Plains. Geochini. Cos-
age and base flow. J. Hydrol., v. 1, p. 219-250. mochim. Acta, v. 1, p. 33-48.
Faust, C. R. 1985. Affidavits in Civil Action No. 79989. United Frank-Kamenetskii, P. A. 1955. Diffusion and Heat Transfer in
States District Court for the Western District of New York, Chemical Kinetics, trans. by J. P. Appleton (1969): New York.
92 p. Plenum Press.
Feenstra, S., D. M. Mackay, and J. A. Cherry. 1991. Presence of Franks, B. J. (ed.). 1987. U.S. Geological Survey Prognm on
residual NAPL based on organic chemical concentrations in Toxic Waste-Ground-Water Contamination: Proc. of the
soil samples. Ground Water Mon. Rev., v. 1 1 , p. 128- 136. Third Technical Meeting, Pensacola, Florida, March 23-27,
1987. U.S. Geol. Surv. Open File Report 87-109.
FERMCO, Femald Environmental Restoration Management Cor-
Fredrickson, J. K., F. J. Brockman, R. J. Hicks, and R. A. Denovan.
poration. 1994. Femald 1993 Site Environmental Report.
1990. Biodegradation of nitrogencontaining aromatic com-
FEMP-2342, 137 p.
pounds in deep subsurface sediments. In C. B. Fliermans and
Ferris, J. G. 195 1. Cyclic fluctuations of water levels as a basis for
T. C. Hazen (eds.), Proc. of the First International Symposium
determining aquifer transmissibility. Int. Assoc. Sci. Hydrol.
on Microbiology of the Deep Subsurface. Westinghoiise Sa-
Pub., V. 33, p. 148- 155.
vannah River Company, p. 627 to 6-44.
Ferris, J. G. 1959. Groundwater. In C. 0. W i s h and E. F. Brater
Freeze, R. A. 1969. The mechanism of natural groundw:iter
(eds.), Hydrology. John Wiley & Sons, New York.
recharge and discharge; 1: one dimensional vertical, unsteady
Ferris, J. G., and others. 1962. Theory of aquifer tests. U.S. Geol.
unsaturated flow above a recharging or discharging ground-
Surv. Water Supply Papers 153GE., p. 69-174. water flow system. Water Resources Res., v. 5, p. 153- 171.
Ferry, J. P., P. J, Dougherty, J. 8. Moser, and R. M. Schuller. Freeze, R. A. 1971. Three dimensional transient saturated-unsat-
1986. Occurrence and recovery of a DNAPL in a low-yielding urated flow in a groundwater basin. Water Resources Res..
bedrock aquifer. Proc. of the NWWNAPl Conference on v. 7, p. 347-366.
Petroleum Hydrocarbons and Organic Chemicals in Ground Freeze, R. A., and J. A. C h e m . 1979. Groundwater. Prenticc-
Water-Prevention, Detection and Restoration. National Wa- Hall, Inc. Englewood Cliffs, N.J.,604 p.
ter Well Assoc., Dublin, Ohio, p. 722-735. Freeze, R. A., J. Masmann, L. Smith, T. Sperling. and H.James.
Fletcher, J. B., and L. R. Sykes. 1977. Earthquakes related to 1990. Hydrogeologic decision analysis: 1. A framework.
hydraulic mining and natural seismic activity in western New Ground Water, v. 28, no. 5. p. 738-7Ch.
York State. J. Geophys. Res., v. 82, p. 3767-3780. Freeze, R. A., and P. A,, Witherspoon. 1%‘. Theoretical analysis
Fogel, S., M. Findlay, A. Moore, and M. Leahy. 1987. Biodegrada- of regional groundwater flow. I: Analytical and numeric;il
tion of chlorinated chemicals in groundwater by methane solutions to the mathematical model. Water Resources Res..
oxidizing bacteria. Proc. of the NWWNAPI Conference o n V. 2, p. 641-656.
Petroleum Hydrocarbons and Organic Chemicals in Ground Freeze, R. A., and P. A. Witherspoon. 1967. Theoreticdl analysis
Water-Prevention, Detection and Restoration. National Wa- of regional groundwater flow. 11: Effect of water table config
ter Well Assoc., Dublin, Ohio, p. 167- 185. uration and subsurface permeability variations. Wxer Rr-
Fogg, G. E., and C. W. Kreitler. 1982. Ground-water Hydraulics sources Res.. v. 3, p. 623-634.
and Hydrochemical Facies in Eocene Aquifers of the East Freyberg, I). L. 1986. A natural gradient experiment on solute
Texas Basin. Bureau of Economic Geology, The Univ. ofTexas transport in a sand aquifer. 2. Spatial moments and the :itlvec-
at Austin, Report of Investigations No. 127, 75 p. tion and dispersion of nonreactive tracers. Water Resourcch
Folk, R. L., and L. S. Land. 1975. Mg/Ca ratio and salinity: two Res., v. 22, no. 13, p. 2031 -2046.
controls over crystallization of dolomite. Am. Assoc. Petrol. Freyberg, D. L. 1988.An exercise in ground-water model c;ilibr;i-
GeOl. Bull. 59, p. 60-68. tion and prediction: Ground Water 26(3). pp. 350- 360.
Fontes, J. C., and J. M. Gamier. 1979. Determination of the Fried, J. J. 1975. Groundwater Pollution. Elsevier. Anistc.rcl:ini.
initial “C activity of the total dissolved carbon: A review of 330 p.
-
476 References

Fried, J. J., and M. A. Combarnous. 1971. Dispersion in porous Survey Toxic Waste Ground-water Contdmination Program,
media. Adv. Hydrosci., v. 7, p. 169-282. U.S. Geol. Surv., Water-Resoures Invest. Report, 91-4034,
Friedman, L. C., and D. E. Erdman. 1982. Quality assurance p. 1-9.
practices for the chemical and biological analyses of water Garder, A. O., A. L. Peaceman, and R. Pozzi. 1964. Numerical
and fluvial sediments. U.S. Geol. Surv. TWRI,Book 5, Chapter calculation of multidimensional displacement by the method
A6, 181 p. of characteristics. SOC.Petrol. Eng. J., v. 4, no. 1, p. 26-38.
Frind, E. 0. 1982. The principal direction technique a new Garrels, R. M. 1960. Mineral Equilibrium at Low Temperature
approach to ground-water contaminant transport modeling. and Pressure. Harper, New York.
Fourth International Conference on Finite Elements in Water Garrels, R. M., and F. T. MacKenzie. 1967. Origin of the chemical
-
Resources, Proc.: New York, Springer-Verlag,p. 13-25 13-42. compositions of some springs and lakes. In R. F. Gould (ed.),
Frind, E. O., and D. Germain. 1986. Simulation of contaminant Equilibrium Concepts in Natural Water Systems. Am. Chem.
plumes with large dispersive contrast: evaluation of alternat- SOC.,Adv. Chem. Ser. 67, Washington, D.C., p. 222-242.
ing direction Galerkin models. Water Resources Res., v. 22, Garrison, A. W., L. H. Keith, and W. H. Shackelford. 1977.
no. 13, p. 1857-1873. Occurrence, registry, and classification of organic pollutants
Fritz, P., and J. C. Fontes. 1980. Introduction. In P. Fritz and in water, with development of a master scheme for their
J. C. Fontes (eds.), Handbook of Environmental Isotope Geo- analysis. In 0. Hutzinger, I. H. Van Lelyveld, and B. C. J.
chemistry, v. 1. Elsevier, Amsterdam, p. 1-19. Zoeteman (eds.), Aquatic Pollutants: Transformation and Bio-
Frind, E. 0. 1982. The principal direction technique: a new logical Effects. Pergamon Press, Oxford, p. 39-68.
approach to groundwater contaminant transport modeling. Garven, G. 1988. Paleohydrology of oil migration in the Alberta
In Proceedings, Fourth International Conference on Finite Basin (abst). EOS, Trans. Am. Geophys. Union, v. 69, p. 360.
Elements in Water Resources. Springer-Verlag, New York, p. Garven, G. 1995. Continental scale ground water flow and
13-25- 13-42. geologic processes. Ann. Rev. Earth Planet. Sci., v. 23, p.
Fryar, A. E.,and P. A. Domenico. 1989. Analytical inverse model- 89- 117.
ing of regional scale tritium waste migration. J. Contam. Hy- Garven, G., and R. A. Freeze. 1984a. Theoretical analysis of the
drol., V. 4, p. 113-125. role of groundwater flow in the genesis of stratabound ore
Gabrysch, R. K., and C. W. Bonnet. 1975. Land surface subsi- deposits. I. Mathematical and numerical model. Am. J. Sci.,
dence in the Houston-Galveston region, Texas. Texas Water V. 284, p. 1085-1124.
Dev. Board Rept. 188. Garven, G., and R. A. Freeze. 1984b. Theoretical analysis of the
Gale,J. E. 1975. A numerical field and laboratory study of flow role of groundwater flow in the genesis of stratabound ore
in rocks with deformable fractures. Ph.D. dissertation, Univ. deposits. 11. Quantitative results. Am. J. Sci., v. 284, p. 1125-
of California, Berkeley. 1174.
Gale, J. E., A. Rouleau, and L. C. Atkinson. 1985. Hydraulic Gary, S. K., and D. R. Kassoy. 1981. Convective heat and mass
properties of fractures. Memoires. Hydrogeology of Rocks of transfer in hydrothermal systems. In L. Rybach and L. J. P.
Low Permeability. Inter. Assoc. Sci. Hydrol., Tucson, A r k , Muffler (eds.), Geothermal Systems, Principles and Case His-
p. 1 - 1 1 . tories. John Wiley & Sons, New York, p. 37-76.
Galloway, W. E. 1978. Uranium mineralization in a coastal-plain Gat,J. R. 1980. The isotopes of hydrogen and oxygen in precipi-
fluvial aquifer system; Catahoula Formation, Texas. Econ. tation. Chapter 1. In P. Fritz and J. C.Fontes (eds.), Handbook
Geol., v. 73, p. 1656-1676. of Environmental Isotope Geochemistry, v. 1. Elsevier, Am-
Galloway, W. E., and D. K. Hobday. 1983. Terrigenous Clastic sterdam, p. 21-47.
Depositional Systems: Application to Petroleum, Coal and Gelhar, L. W., A. L. Gutjahr, and R. ..I Naff. 1979. Stochastic
Uranium Exploration. Springer-Verlag, New York. analysis of macrodispersion in a stratified aquifer: Water Re-
Galloway, W. E., and W. R. Kaiser. 1980. Catahoula Formation sources Res., v. 15, no. 6, p. 1387- 1397.
of the Texas coastal plain: origin, geochemical evolution, Gelhar, L. W., and C. L. Axness. 1983. Three-dimensional sto-
and characteristics of uranium deposits. Bureau of Economic chastic analysis of macrodispersion in aquifers. Water Re-
Geology, The IJniv. Texas, Austin, Report Invest., no. 100. sources Res., v. 19, no. 1, p. 161 - 180.
Gambolati, G., and R. A. Freeze. 1973. Mathematical simulation Gelhar, L. W., A. Mantoglou, C. Welty, and K. R. Rehfeldt. 1985.
of the subsidence ofVenice. I. Theory. Water Resources Res., A review of field-scale physical solute transport processes
V. 9, p. 721-733. in saturated and unsaturated porous media. Electric Power
Gambolati, G., P. Gatto, and R. A. Freeze. 1974a. Mathematical Research Institute EPRI EA-4190 Project 2485-5, 116 p.
simulation of the subsidence of Venice. 2. Results. Water Gelhar, L. W., C. Welty, and K. R. Rehfeldt. 1992. A critical
Resources Res., v. 10, p. 563-577. review of data on field-scale dispersion in aquifers. Water
Gambolati, G., P. Gatto, and R. A. Freeze. 1974b. Predictive Resources Res., v. 28, no. 7, p. 1955- 1974.
simulation of the subsidence of Venice. Science, v. 183, p. Ghiorse, W. C., and J. T. Wilson. 1988. Microbial ecology of
849-85 1. the terrestrial subsurface. Adv. Appl. Microbiol., v. 33, p.
Gangi, A. F. 1978. Variation of the whole and fractional porous 107-172.
rock permeability with confining pressure. Int. J. Rock Mech., Ghyben, W. B. 1899. Notes in verband met Voorgenomen Put
V. 15, p. 249-257. boring Nabji Amsterdam, Tijdschr. Koninhitk, Inst. Ingrs.,
Garabedian, S. P., and D. R. LeBlanc. 1991.Overview of research The Hague.
at the Cape Cod Site: Field and laboratory studies of hydro- Gibson, R. E. 1958. The progress of consolidation in a clay
logic, chemical, and microbial processes affecting transport layer increasing in thickness with time. Geotechnique, v. 8,
in a sewage-contaminated sand and gravel aquifer. U.S. Geol. p. 171-182.
 Referetrees 477

Gibb, J. P., and M.J. Barcelona. 1984. Sampling for organic sance electrical methods to map contaminant migration.
contaminants in ground water. Ground Water. v. 76, no. 5, Ground Water Mon. Rev., v. 3, p. 47-49.
p. 48-51. Greenkorn, R. A. 1983. Flow Phenomena in Porous Media:
Gibb, J. P., R. M. Schuller, and R. A. Griffin. 1981. Procedures Fundamentals and Applications in Petroleum, Water and
for the Collection of Representative Water Quality Data from Food Production. Marcel Dekker, New York, 550 p.
Monitoring Wells. Cooperative Ground Water Report. Illinois Greenkorn, R. A., and D.P. Kessler. 1972. Transfer Operations.
State Water Siirvey and U.S. Environ. Protection Agency, 60 p. McGraw-Hill, New York, 548 p.
Gibson, R. E. 1958. The progress of consolidation in a clay Gretener, P. E. 1972. Thoughts on overthrust faulting in a Ly-
layer increasing in thickness with time: Geotechnique, v. 8, ered sequence. Bull. Canadian Petrol. Geol. v. 20, p. 583-607.
p. 171-182. Gretener, P. E. 1981. Pore pressure: fundamentals, general rami-
Gilkeson. R. H., P. C. Heigold, and D. E. Laynion. 1986. Practical fications, and implications for structural geology (revised).
application of theoretical models to inagnetometer surveys Education Course Note Series 4, Am. Assoc. Petrol. Geol.
of hazardous waste disposal sites-A case study. Ground Tulsa, OMa.
Water Mon. Rev., v. 6, 54-61. Griffin, R. A., and W. R. Roy. 1985. Interaction of Organic
Gillham, R. W., and S. F. O’Hannesin. 1994. Enhanced degrada- Solvents with Saturated Soil-water Systems. Environmental
tion of halogenated aliphatics by zero-valent iron. Ground Institute for Waste Management Studies, Univ. of Alabama,
Water, v. 32, no. 6, p. 958-967. Open File Report No. 3, 86 p.
Gilluly, J., and U. S. Grant. 1949. Subsidence in Long Beach Grisak, G. E., and R. E. Jackson. 1978. An appraisal o f the
Harbor Area, California. Bull. Geol. SOC. Am. v. 60, p. Hydrogeological Processes Involved in Shallow Subsiirfacr
461 -521. Radioactive Waste Management in Canadian Terrain. Scien-
Glasstone, S., and D.Lewis. 1960. Elements of Physical Cheniis- tific Series No. 84, Inland Waters Directorate, Environment
try. D. Van Nostrand Company, Princeton, NJ., 758 p. Canada, Ottawa, 194 p.
Glover, R. E. 1964. The pattern of freshwater flowing in a Grisak, G. E., and J. F. Pickens. 1980. Solute transport through
coastal aquifer. In Sea Water in Coastal Aquifers. U.S. Geol. fractured media, 1. The effect of matrix diffusion. Water
Survey Water Supply Paper 1613-C, p. 32-35. Resources Res., v. 16, no. 4, p. 719-730.
Goerlitz, D. F., and E. Brown. 1972. Methods for analysis of Grove, D. B. 1971. U.S. Geological Survey tracer study, Armar-
organic substances in water. U.S. Geol. Surv. TWRI,Book 5, gosa Desert, Nye County, Nevada. 11. An analysis of the flow
Chapter A3, 4 0 p. field ofa discharging-recharging pair of wells. U.S.Geol. Surv.
Goguel, J. 1976. Geothermics (trans. by A. Rite). McGraw-Hill, Rep. USGS-474-99, 56 p.
New York. Grove, D.B., and W. A. Beetem. 1971. Porosifl and dispersion
Goode, D. J., and C. A. Appel. 1992. Finite-difference interblock constant calculations for a fractured carbonate aquifer using
tnnsmissivity for iinconfined aquifers and for aquifers having the two well tracer method. Water Resources Res.. v. 7.
sn~ootlilyvarying transmissivity. U.S. Geol. Surv., Water- p. 128-134.
Resources Investigations Rept.. 92-4 124, 79 p. Gruszczenski, T. S. 1987. Determination of a realistic estimate o f
Gordan, F. R. 1970. Water level changes preceding the Necker- the actual formation product thickness using monitor wells: a
ing Western Australia earthquake of October 14, 1968. Bull. field bailout test. Proc. of the NWWA/API Conference o n
Seismol. SOC.Am. v. 60,p. 1739-1740. Petroleum Hydrocarbons and Organic Chemicals in Ground
Gorelick, S. M. 1987. Sensitivity analysis of optimal ground Water-Prevention, Detection and Restoration. National Wa-
wwer contaminant capture curves. Solving Ground Water ter Well Assoc. Dublin, Ohio, p. 235-253.
Problems with Models. National Water Well Assoc., Dublin, Gschwend, P. M., and M. D. Reynolds. 1987. Monodispersr
Ohio, p. 133- 146. ferrous phosphate colloids in an anoxic groundwater plume.
Gorelick, S. M., C. I. Voss, P. Gill, W. Murray, M. Saunders, and J. Contam. Hydrol., v. 1, p. 309-327.
M. Wright. 1984. Aquifer reclamation design: The use of Gupta, M. L., and 8. K. Rastogi. 1970. Dinis and earthquakes.
contaminant transport simulation combined with nonlinear Dev. Engr. Geol., v. 11.
programming. WaterResources Res., v. 20, no. 4, p. 415-427. Gupta, M. L., 8. K. Rastogi, and H. Narain. 1972. Common
Goiinaris, V., P. R. Anderson, and T. M. Holsen. 1993. Character- features of the reservoir associated seismic activities. Bull.
istics and environmental significance of colloids in a landfill Seismol. Soc. Am. v. 6, p. 481-492.
leachnte. Environ. Sci. Techno]., v. 27, p. 1381- 1387. Gupta, M. L., and B. S. Suknija. 19-4. Preliminary studies o f some
Graton, L. C., and H. J. Fraser. 1935. Systematic packing of geothermal areas in India: Geotherniics. v. 3. p. 105- 1 12.
spheres with particular relation to porosity and permeability Guth, P. L., K. V. Hodges, andJ. H. Willemin. 1982. Limitations
and experimental study of the porosity and permeability of on the role of pore pressure in graviv sliding. Hull. Geol.
clastic sediments. Geology. v. 43, p. 785-909. Soc. Am. v. 93, p. 606-612.
Green, H. W. 1984. “Pressure solution” creep: some causes Gutjahr, A. L., L. W. Gelhar, A. A. Bakr. and J. R. McMil1;uii.
and mechanisms. J. Geophys. Res., v. 89, p. 4313-4318. 1978. Stochastic analysis of spatial variability in sl1bsurf;lcr
Green, D. H., and H. F. Wang. 1986. Fluid pressure response flow: 2. Evaluation and application. Water Resources Rcs..
to untlrained compression in saturated sedimentary rock. V. 14, p. 953-959.
Geophysics, v. 51, p. 948-956. Giiven, O., R. W. Falta, F. J. Molz, and J. G. Melville. 1986. A
Green, D. W.. and R. H. Perry. 1961. Heat transfer with a fluid simplified analysis of two-well tracer tests in stratified aqui-
flowing through porous media. Chem. Eng. Prog. Synip. Ser. fers. Ground Water, v. 24, no. 1, p. 63-71.
57, l>. 61 -68. Hamilton, L. D., A. F. Meinhold, S. L. Baxter, S. Holtznian. S. (1.
Greenhouse, J. P., and D. D. Shine. 1983. The use of reconais- Morris, R. Pardi, M. D. Rowe, and C. Sun. 1993. Pilot Study
478 References

Risk Assessment at the Femald Environmental Management Harvey, R. W., R. L. Smith, and L. George. 1984. Microbial
Project (FEMP). Brookhaven National Laboratory, BNL-48777, distribution and heterotropic uptake in a sewage plume.
168 p. Movement and Fate of Solutes in a Plume of Sewage-Contami-
Handin, J . , and others. 1963. Experimental deformation of sudi- nated Ground Water, Cape Cod, Massachusetts: U.S. Geol.
rnentary rocks under confining pressure: pore pressure tests. Surv. Toxic Waste Ground-Water Contamination Program,
Am. Assoc. Petrol. Geol. Bull. 47, p. 717-755. U.S. Geol. Surv. Open-File Rept. 84-475, p. 139- 152.
Hansch, C., and A. Leo. 1979. Substituent constants for correla- Hashin, Z., and S. Shtrikrnan. 1962. A variation approach to
tion analysis in chemistry and biology. John Wiley & Sons, the theory of effective magnetic permeability of multiphase
New York, 339 p. materials. J. Appl. Phys., v. 33, p. 3125-3131.
Hanshaw, B., W. Back, and R. Deike. 1971. A geochemical Hassett, J. J., W. L. Banwart, and R. A. Griffin. 1983. Correlation
hypothesis for dolomitization by groundwater. Econ. Geol., of compound properties with sorption characteristics of non-
V. 66, p. 710-724. polar compounds by soils and sediments: Concepts and Iimi-
Hanshaw, B., and E. Zen. 1965. Osmotic equilibrium and tations. In C. W. Francis and S. I. Auerbach (eds.), Envimn-
overthrust faulting. Geol. SOC. Am. Bull. 76, p. 1379-1386. ment and Solid Wastes: Characterization, Treatment and
Hantush, M. S. 1956. Analysis of data from pumping tests in Disposal. Butterworth Publishers, Chap. 15, p. 161- 178.
leaky aquifers. Trans. Am. Geophys. Union, v. 37, p. 702-7 14. Hastings, T. 1986. A theoretical approach for assessing the role
Hantush, M. S. 1957. Nonsteady radial flow to a well partially of rock and fluid properties in the development of abnormal
penetrating an infinite leaky aquifer. Proc. Iraqi Sci. Soc., v. pressure. M.S. thesis, Texas A&M Univ.
1, p. 10-19. Hayes, J. B. 1977. Sandstone diagenesis, the hole truth. In P. A.
Hantush, M. S. 1960. Modification of the theory of leaky aqui- Scholle and R. R. Schluger (eds.), Aspects of Diagenesis. Soc.
fers. J. Geophys. Res., v. 65, p. 3713-3725. Econ. Paleontol. and Mineral Spec. Pub. 26, p. 127-139.
Hantush, M. S. 1961. Aquifer tests on partially penetrating wells. Healy, J. W., and others. 1968. The Denver earthquakes. Sci-
Proc. Am. Soc. Civil Engrs., v. 87, p. 171 - 195. ence, v. 161, p. 1301-1310.
Hantush, M. S. 1964. Hydraulics of wells. In V. T. Chow (ed.), Heard, H. C., and W. W. Rubey. 1966. Tectonic implications
Advances in Hydroscience. Academic Press, New York, p. of gypsum dehydrdtion. Geol. Soc.Am. Bull. 77, p. 741 -760.
281 -432. Heath, R. C. 1982. Classification of groundwater systems of the
Hantush, M. S. 1966a. Wells in homogeneous anisotropic aqui- United States. Ground Water, v. 20, p. 393-401.
fers. Water Resources Res., v. 2, p. 273-279. Heath, R. C. 1984. Groundwater regions of the IJnited States.
Hantush, M. S. 1966b. Analysis of data from pumping test in U.S. Geol. Survey Water Supply Paper 2242.
anisotropic aquifers. J. Geophys. Res., v. 71, p. 421 -426. Helfferich, F. 1966. In J. A. Marinsky (ed.), Ion Exchange, A
Hantush, M. S., and C. E. Jacob. 1955. Nonsteady radial flow Series of Advances. Marcel Dekker, New York p. 65- 100.
in an m i t e leaky aquifer. Trans. Am. Geophys. Union, v. Helm, D. C. 1975. One dimensional simulation of aquifer com-
36, p. 95-100. paction system near Pixley, California. Water Resources Res.,
Hantush, M. S., and R. G. Thomas. 1966.A method for analyzing V. 1 1 , p. 465-478.
a drawdown test in anisotropic aquifers. Water Resources Hem, J. D. 1959. Study and interpretation of the chemical char-
Res., v. 2, p. 281-285. acteristics of natural water. U.S. Geol. Survey Water Supply
Hardie, L. A. 1991. On the significance of evaporites. Ann. Rev. Papers, 1473.
Earth Planet. Sci., v. 19, p. 131-168. Hendry, M. J., J. A. Cherry, and E. I. Wallick. 1986. Origin and
Hardie, L. A., and H. P. Eugster. 1970. The evolution of closed- distribution of sulfate in a fractured till in southern Alberta,
basin brines. Mineral. SOC.Am. Spec. Publ., v. 3. p. 273-290. Canada. Water Resources Res., v. 22, no. 1, p. 45-61.
Harris, J. C. 1982. Rate of hydrolysis. In W. J. Lyman, W. F. Hendry, M. J., and F. W. Schwartz. 1988. An alternative view
ReeN, and D. H. Rosenblatt (eds.), Handbook of Chemical on the origin of chemical and isotopic patterns in groundwa-
Property Estimation Methods. McGraw-Hill,New York. 977 p. ter from the Milk River aquifer. Water Resources Res., v. 24,
Harris, J. F., G. L. Taylor, and J. L. Walper. 1960. Relation of no. 10, p. 1747-1764.
deformational fracture in sedimentary rocks to regional and Hendry, M. J., and F. W. Schwartz. 1990.The chemical evolution
local structure. Am. Assoc. Petrol. Geol. Bull. 44, p. 1853- of ground water in the Milk River aquifer, Canada. Ground
1873. Water, v. 28, no. 2, p. 253-261.
Harvey, R. W., and S . P. Garabedian. 1991. Use of colloid filtra- Herrin, H., and T. Goforth. 1975. Environmental problems asso-
tion theory in modeling movement of bacteria through a ciated with power production from geopressured reservoirs.
contamlnated sandy aquifer. Environ. Sci. Technol., v. 25, Proc. 1st Geopress. Geotherm Conf., p. 311-320.
no. 1, p. 178-185. Henberg B. 1901. Die Wasserversovgung einiger Nordseebaser.
Harvey, R. W., and L. George. 1986. Bacterial distribution and J. Gasbeleucht and wasserversov, v. 44, p. 815-819.
transport in a plume of sewagecontaminated ground water. Hess, K. M., W. N. Herkelrath, and H. I. Fssaid. 1992.Determina-
U S . Geol. Survey Toxic Waste Ground-water Contamination tion of subsurface fluid contents at a crude-oil spill site. J.
Program, U.S. Geol. Surv. Open Fie Rept., 86-481. 1’. Contam. Hydrol., v. 10, p. 75-96.
B25- 826. Hess, K. M., S. H. Wolf, and M. A. Celia. 1992. Large-scale
Harvey, R. W., L. H. George, R. L. Smith, and D. R. LeBlanc. natural gradient tracer test in sand and gravel, Cape Cod,
1989. Transport of microspheres and indigenous bacteria Massachusetts, 3. Hydraulic conductivity variability and calcu-
through a sandy aquifer: Results of natural- and forced-gradi- lated macrodispersivities. Water Resources Res., v. 28, no.
ent tracer experiments. Environ. Sci. Technol., v. 23, no. I , 8, p. 201 1-2027.
p. 51-56. Hewitt, T. A. 1986. Fractal distributions of reservoir heterogene-
 References 473

ity and their influence on fluid transport. Proc. 6lst Annual compaction, and permeability of sands. Am. Assoc. Petrol.
Tech. Conf. Soc. Petrol. Ehg., New Orleans, PaperSPE-15386. Geol. Bull. 61, p. 169-191.
Heyl, A. V. and others. 1970. Guidebook to the Upper Missis Hubben, M. K. 1940. The theory of groundwater motion. J.
sippi Valley basic-metal district. Univ. of Wisconsin Geol. and Geol., v. 48, p. 785-944.
Nat. History Survey, Inf. Circ. No. 16, 49 p.. Madison. Hubbert, M. K. 1953. Entrapment of petroleum under hydrody-
Hickey, J. J. 1984. Field testing of the hypothesis of Darcian namic conditions. Bull. Am. Assoc. Petrol. Geol. 37, p. 1944-
flow through a carbonate aquifer. Ground Water, v. 22, p. 2026.
544- 547. Hubbert, M. K. 1956. Darcy’s law and the field equations of
Hill, M. C. 1990. Preconditioned Conjugate-Gradient 2 (PCGZ), the flow of underground fluids. Trans. Am. Inst. Min. Met.
A computer program for solving ground-water flow equa- Engrs., V. 207, p. 222-239.
tions. U.S. Geol. Surv., Water-Resources Investigations Re- Hubbert, M. K., and W. W. Rubey. 1959. Role of fluid pressure
port, 90-4048, 43 p. in mechanics of overthrust faulting. Part I: Mechanics of fluid
Hill, M. C. 1992. A computer program (MODFLOWP) for esti- filled porous solids and its application to overthrust faulting.
mating parameters of a transient, three-dimensional ground- Bull. Geol. SOC.Am. 70, p. 1 15- 166.
water flow model using nonlinear regression. US. Geol. Surv., Hubbert, M. K.,and D. G. Willis. 1957. Mechanics of hydraulic
Open-File Report 91-484. fracture. Am. Inst. Min. Engr. Trans., v. 210, p. 153-168.
Hillel, D. 1971. Soil and Water: Physical Principles and Pro- Hughes, J. P., C. R. Sullivan, and R. E. Zinner. 1988. Two tech-
cesses. Academic Press, New York. niques for determining the true hydrocarbon thickness in an
Hirasaka, G. J. 1974. Pulse tests and other early transient p r e s unconfined sandy aquifer. Proc. of the NWWA/API Confer-
sure analysis for in-situ estimation of vertical permeability. ence on Petroleum Hydrocarbons and Organic Chemicals
Trans. Am. Inst. Mining Engr., v. 257, p. 75-90. in Ground Water-Prevention, Detection and Restoration.
Hoefner, M. L., and H. S. Fogler. 1988. Pore evolution and National Water Well Assoc., Dublin, Ohio, p. 291-314.
channel formation during flow and reaction in porous media. Huitt, J. L. 1956. Fluid flow in simulated fractures. Am. Inst.
AlChE J., V. 34, p. 45-54. Chem. Eng., v. 2, p. 259.
Hoek, E., and J. W. Bray. 1901. Rock Slope Engineering. Inst. Hull, R. W., Y. K. Kharaka, A. S. Maest, and T. L. Fries. 1984.
of Min. and Met., London. Sampling and analysis of subsurface water. In B. Hitchon and
Homberger, G. M., J. Ebert, and I. Remson. 1970. Numerical E. I. Wallick (eds.), First Canadian/American Conference on
solution of the Boussinesq equation for aquifer-stream inter- Hydrogeology. National Water Well Assoc.. Dublin, Ohio.
action. Water Resources Res., v. 6, p. 6 0 1 -608. p. 117- 126.
Hornberger, G. M., I. Remson, and A. A. Fungaroli. 1969. Nu- Hunt, B. 1978. Dispersive source in uniform groundwater flow.
meric studies of a composite soil moisture groundwater sys- J. Hydrol. Div., Proc. Amer. SOC.Civil Eng.. v. 104, p. 75-85.
tem. Water Resources Res., v. 5, p. 797-802. Hunt, B. W. 1978. Dispersive sources in uniform groundwater
Horner, D. R. 1951. Pressure buildup in wells. Proc.3rd World flow. J. Hydraulics. Div., ASCE, v. 99, no. HYi. p. 13-21.
Petrol. Congress. Sect. 11. E. J. Brill, Leiden, Holland. p. Hunter, J. A,, S. E. Pullan, R. A. Burns, R. M. Gagne, and R. L.
503-52 1. Good. 1984. Shallow seismic reflection mapping of the over-
Horton, C. W.,and F. T. Rogers. 1945. Convection currents in burden-bedrock interface with the engineering seismc+
a porous medium. J. Appl. Phys., v. 16, p. 367-370. graph-somesimple techniques. Geophysics. v. 49, p. 1381 -
Horton, R. E. 1933. The role of infiltration in the hydrologic 1385.
cycle. Trans. Am. Geophys. Union, v. 14, p. 446-460. Hunter, J. A., S. E. Pullan, R. A. Burns, R. M. Gagne. and R. L.
Horton, R. E. 1940. Approach toward a physical interpretation Good. 1988. Applications of a shallow seismic reflection
of infiltration capacity. Soil Sci. SOC. Am. Proc., v. 5, p. method to groundwater and engineering studies. In Proc. of
339-417. Exploration ‘87: Third Decennial International Conference
Howard, A. D., and C. G. Groves. 1995. Early development of on Geophysical and Geochemical Exploration for Minerals
karst systems: 2. Turbulent flow. Water Resources Res., v. and Groundwater, Ontario Geological Survey. Special Vol-
31, p. 19-26. ume, p. 704-715.
Hsieh, P. A,, and J. D. Bredehoeft. 1981. A reservoir analysis of Huntley, D.. R. N. Hawk, and H. P. Corley. 1992. Non-aqueous
the Denver earthquakes. J. Geophys. Res.. v. 86, p. 903-920. phase hydrocarbon saturations and mobility in a finegrained,
Hsieh, P. A,, and J. R. Freckleton. 1993. Documentation of a poorly consolidated sandstone. Proc. of the 1992 Petroleum
computer program to simulate horizontal-flow barriers using Hydrocarbons and Organic Chemicals in Ground Water: Pre-
the U.S. Geological Survey’s modular three-dimensional finite- vention, Detection, and Restoration. National Ground Water
difference ground-water flow model. U.S. Geol. Sum., Open- Assoc., Columbus, Ohio, p. 223-237.
File Report 92-477, 32 p. Huyakorn, P. S., P. F. Andersen, 0.Giiven, and F. J. Molz. 1986.
Hsieh, P. A,, and S. P. Neuman. 1985. Field determination of A curvilinear finite element model for simulating two-well
the three dimensional hydraulic conductivity tensor of anim- tracer tests and transport in stratified aquifers. Water Re-
tropic media. I: Theory. Water Resources Res., v. 21, p. 1655- sources Res., v. 22, no. 5, p. 663-678.
1665. Huyakorn, P., and others. 1987. A three dimensional analytical
Hsieh, P. A,, and others. 1985. Field determination of the three method for predicting leachate migration. Ground Water.
dimensional hydraulic conductivity tensor of anisotropic me- V. 25, p. 588-598.
dia: 11. Methodology and application to fractured rocks. Water Huyakorn, P. S., andG. F. Pinder. 1983. Computational Methods
Resources Res., v. 21, p. 1667-1676. in Subsurface Flow. Academic Press, New York, 473 p.
Hsu, K. J. 1977. Studies ofVentura field, California: 11. I.ithology, Hvorslev, M. J. 195 1. Time lag and soil permeability in ground-
 Next Page

-
480 References

water observations. U.S. Army Corps of Engr. Waterway E:xp. Johnson, A. I., and D. A. Morris. 1962. Physical and hydrologic
Stat. Bull. 36, Vicksburg, Miss. properties of water bearing deposits from core holes in the
Ibrahm, H. A., and W. Brutsaert. 1965. Inflow hydrographs Las Banos-Kettleman City area, California. U S . Geol. Survey
from large unconfined aquifers. J. Irr. Drain. Div., Proc. Am. Open. File Rept. Denver, Colo.
SOC.Civil Engrs., v. 91, p. 21-38. Johnson, K. S. 1981. Dissolution of salt on the east flank of the
Ingerson, E., and J. F. Pearson, Jr. 1964. Estimation of age and Permian Basin in the southwestern U.S.A. In W. Back and R.
rate of motion of ground-water by the “C method. Recent Lc%lle (eds.), J. Hydrol. Special Issue, Symposium on <;eo
Research in the Fields of Hydrosphere, Atmosphere and Nu- chemistry of Groundwater, v. 54, p. 75-94.
clear Chemistry. Maruzen, Tokyo, p. 263-283. Johnson, P. C., C. C. Stanley, M. W. Kemblowski, D. L. Byers,
Ivanovich, M., K. Frohlich, and M. J. Hendry. 1991. Dating very and J. B. Colthart. 1990. A practical approach to the design,
old groundwater, Milk River Aquifer, Alberta, Canada. Appl. operation, and monitoring of in situ soil-venting systems.
Geochem., v. 6, no. 4, p. 367-472. Ground Water Mon. Rev., Spring, p. 159-178.
Jackson, R. E., and J. Inch. 1980. Hydrogeochemical processes Johnson, R. L., S. M. Brillante, L. M. Isabelle, J. E. Houck, and
affecting the migration of radionuclides in a fluvial sand aqui- J. F. Pankow. 1985. Migration of chlorophenolic compounds
fer at the Chalk River Nuclear Laboratories. Environment at the chemical waste disposal site at Alkali Lake, Oregon. 2.
Canada, National Hydrol. Research Instit., Paper No. 7, Sci- Contaminant distributions, transport and retardation. Ground
ence Series, Ottawa, 58 p. Water, v. 23, no. 5, p. 652-665.
Jackson, R. E., and R. J. Patterson. 1982. Interpretation of pH Jones, J. W., and others. 1985. Field and theoretical investigation
and En trends in a fluvial-sand aquifer system. Water Resources of fractured crystalline rock near Oracle, Southern Arizona.
Res., v. 18, no. 4, p. 1255-1268. NUREG/CR-3736. Nuclear Reg. Comm.
Jackson, R. E., R. J. Patterson, B. W. Graham,J. Bahr, D. Belanger, Jordan T. E., D. G. Leask, D. Slain, 1. MacLeod, and T. M. Dobush.
J. Lockwood, and M. Priddle. 1985. Contaminant Hydrogeol- 1991. The use of high resolution electromagnetic methods
ogy ofToxic Organic Chemicalsat a Disposal Site, Gloucester, for reconaissance mapping of buried wastes. Proc. of the Fifth
Ontario: 1. Chemical Concepts and Site Assessment. Environ- National Outdoor Action Conference on Aquifer Restoration:
ment Canada, National Hydrol. Research Instit., Paper No. Ground Water Monitoring and Geophysical Methods, p.
23, Ottawa, 114 p. 849-862.
Jackson, S. A, and Beales, F. W. 1967. An aspect of sedimentary Jouanna, P. 1972. In situ permeability tests under applied
basin evolution: The concentration of Mississippi Valley type stresses. Proc. Symp. Percolation through Fissured Rock.
ores during late stages of diagenesis. Canadian Petrol. Gcol. Jury, W. A., D. Russo, G. Streile, and H. El Abd. 1990. Evaluation
Bull. 15, p. 383-437. of volatilization by organic chemicals residing below the soil
Jacob, C. E. 1939. Fluctuations in artesian pressure produced surface. Water Resources Res., v. 26, p. 13-20.
by passing railroad trains as shown in a well on Long Island, Kaida, Y., M. Matsubara, R. Ghose, and T. Kanemori. 1995. Very
New York. Trans. Am. Geophys. Union, v. 20, p. 666-674. shallow seismic reflection profiling using portable vibrator.
Jacob, C. E. 1940. On the flow of water in an elastic artesian Proc. of the Symposium on the Application of Geophysics
aquifer. Trans. Am. Geophys. Union, v. 22, p. 574-586. to Engineering and Environmental Problems, Environ. and
Jacob, C. E. 1941. Notes on the elasticity of the Lloyd Sand on Engr. Geophys. SOC.,Englewood, Colo., p. 601-607.
Long Island, New York. Trans. Am. Geophys. Union, v. 22, Karickhoff, S. W. 1981. Semi-empirical estimation of sorption
p. 783-787. of hydrophobic pollutants on natural sediments and soils:
Jacob, C. E. 1950. Flow of groundwater. In H. Rouse (etl.), Chemosphere, v. 10, no. 8, p. 833-846.
Engineering Hydraulics. John Wiley & Sons, New York, p. Karickhoff, S. W. 1984. Organic pollutant sorption in aquatic
32 1 - 386. systems. J. Hydraul. Engr. (ASCE),v. 110, no. 6, p. 707-735.
Jacquin, C., and M. Poulet. 1970. Study of hydrodynamic pattern Karickhoff, S. W. 1985. Pollutant sorption in environmental
in a sedimentary basin subject to subsidence. Soc. Petrol. systems. In W.B. Neely and G. E. Blau (eds.), Environmental
Engrs. Am. Inst. Mining, Metal. Petrol. Engs. Paper no. Exposure form Chemicals, Volume I. CRC Press Inc., Boca
SPE2988, 10 p. Raton, Fla., p. 49-62.
Jacquin, C., and M. Poulet. 1973. Essas de Restitution des condi- Karickhoff, S. W., D. S. Brown, andT. A. Scott. 1979. Sorption
tions hydrodynamiques regnant dans un bassin sedimentarire of hydrophobic pollutants on natural sediments. Water Res.,
au coors de son evolution. Revue de L‘Institut Francais du V. 13, p. 241-248.
Petrole, v. 28, p. 269-297. Kasten, P. R., L. Lapidus, and N. R. Amundson. 1952. Mathemat-
Jaeger, J. C., and N.G. W. Cook. 1969. Fundamentals of Rock ics of adsorption in beds, V. Effect of intraparticle diffusion in
Mechanics. Methuen Press, London. flow systems in fixed beds. J. Phys. Chem., v. 56, p. 683-(58.
Jamieson, G. R., and R. A. Freeze. 1983. Determining hydraulic Kazmann, R. G. 1956. Safe yield in groundwater development,
conductivity distribution in a mountainous area using mathe- reality or illusion. Proc. Am. Soc. Civil Engrs., v. 82.
matical modeling. Ground Water, v. 21, p. 168- 177. Keech, D. A. 1988. Hydrocarbon thickness on groundwater by
Javandel, I., and C. Tsang. 1986.Capture-zone type curves: a tool dielectric well logging. Proc. of the NWWA/API Conference
for aquifer cleanup. Ground Water, v. 24, no. 5, p. 616-625. on Petroleum Hydrocarbons and Organic Chemicals in
Johnson, A. I. 1948. Groundwater recharge on Long Island. J. Ground Water-Prevention, Detection and Restoration. Na-
Am. Water Works Assoc., v. 49, p. 1159-1166. tional Water Well Assoc., Dublin, Ohio, p. 275-289.
Johnson, A. I. 1967. Specific yield-compilation of specific yields Keeley, J. F. 1989. Performance Evaluations of Pump-and-Treat
for various materials. U.S. Geol. Survey Water Supply Paper Remediations. U.S. Environmental Protection Agency, EPA/
1662-D, 74 p. 540/4-89/005, 19 p.
Previous Page

References 481

Kelly, F. X.,K.J. Dapsis, and D. A. Lauffenburger. 1988. Effect of homa, Oklahoma State University and Rice University. N o r -
bacterial chemotaxis on dynamics of microbial competition. man, Okla., National Technical Information Service. .+-1 p.
Microb. Ecol., v. 16, p. 115-131. Kohout, F. A. 1960. Flow pattern of fresh water and salt w;iter
Kelly, V. 1985. Field Determination of Dispersivity of Coming in the Biscayne aquifer of the Miami area. Florida: Intern.
ling Plumes: Unpublished M.S. Theses, Texas ASrM Univer- Assoc. Sci. Hydrol., Comm. Subter. Waters. puhl. 52.
sity, College Station. p. 440-448.
Keswick, B. H., and C. P. Gerba. 1980. Viruses in groundwater. .
Kohout, F. A. 1961. Case history of salt water encro‘cc-I inieiit
Environ. Sci. Technol., v. 14, p. 1290-1297. caused by a storm sewer in Miami.J. Am. Water Works Assoc.,
Kharaka, Y. K., and 1. Barnes. 1973.SOLMNEQ: Solution-Mineral V. 53, p. 1406-1416.
Equilibrium Computations. National Technol. Info. Sew. Kohout, F. A., and H. Klein. 1967. Effect of pulse rechrge on
Technol. Rept. PB214-899, 82 p. the zone of diffusion in the Biscayne aquifer. Intern. Assoc.
Kiersch, G. 1973. The Vaiont Reservoir Disaster. In R. Tank Sci. Hydrol. Symp. Haifa, Israel, pub. 72, p. 252-270.
(ed.), Focus on Environmental Geology. Oxford Univ. Press, Kolbel-Boelke,J., E. M. Anderson, and A. Nehrkorn. 1988. Micrtr
New York. bial communities in saturated groundwater environment, 11.
Killey, R. W. D., J. 0. McHugh, D. R. Champ, E. L. Cooper, and Diversity of bacterial communities in a Pleistocene sand aqui-
J. L. Young. 1984. Subsurface cobalt& migration from a fer and their in vitro activities. Microb. Ecol.. v. 16, p. 3 1-48,
low-level waste disposal site. Environ. Sci. Technol., v. 18, Koltermann, C. E., and S. M. Gorelick. 1992. Paleocliniate signa-
p. 148-157. ture in terrestrial flood deposits. Science. v. 256. p. 1775-
Kilty, K., and D. S. Chapman. 1980. Convective heat transfer 1782.
in selected geological situations. Ground Water, v. 18, p. Konikow, L. F., and J. D. Bredehoeft. 1978. Computer modrl
386-394. of two-dimensional solute transport and dispersion in ground
Kimball, B. A. 1984. Ground water age determinations, Piceance water. U.S. Geol. Surv. TWRI, Book 7, I Chap. C2, Reston.
Creek Basin, Colorado. In B. Hitchon and E. I. Wallick (eds.), Va., 40 p.
First CanadiadAmerican Conference o n Hydrogeology. Na- Koons, P. 0. 1988. The evolution of the fluid lhw regime
tional Water Well Assoc., Dublin, Ohio, p. 267-283. during continental collison (abst). EOS. Trans. Am. Geophys.
Union, v. 69, p. 361.
Kimmei, G. E., and 0. C. Braids. 1980. Leachate plumes in
Kreitler, C. W. 1977. Fault control of subsidence, Houston.
ground water from Babylon and Islip landfills, Long Island,
Texas. Ground Water, v. 3, p. 203-214.
New York. U.S. Geol. Surv., Prof. Paper 1085, 38 p.
Kriegman-King, M. R., and M. Reinhard. 1992. Transformation
King, F. H. 1899. Principles and conditions of the movements
of carbon tetrachloride in the presence of sulfide. hiotite,
of groundwater. U.S. Geol. Survey 19th Ann. Rept., pt. 2,
and vermiculite. Environ. Sci. Technol.. v. 26. p. 2198-2206.
p. 59-294.
Krupp, H. K., S. W. Biggar, and D. R. Nielsen. 1972. Relative
Kirkham, D., and W. L. Powers. 1972. Advanced Soil Physics.
flow rates of salt in water in soil. Soil Sci. Soc. Am. I’roc..
Wiley-Interscience, New York.
V. 36, p. 412-417.
Kissen, 1. G. 1978. The principal distinctive features of the Kueper, B. H., and E. 0. Ftind. 1991. Two phase flow in hetertr
hydrodynamic regime of intensive earth crust downwarping geneous porous media, 2. Model development. Water Re-
areas. In Hydrogeology of Great Sedimentary Basins: Pub. of sources Res., v. 27, p. 1049-1058.
the Intern. Assoc. Sci. Hydrol., no. 120, p. 178-185. Kueper, B. H., and D. B. McWorter. 1991. The behavior of
Kittel, J. A., R. E. Hinchee, R. Hoeppel, and R. Miller. 1994. dense, nonaqueous phase liquids in fractured clay and rock.
Bioslurping-Vacuum-enhanced free-product recovery cou- Ground Water, v. 29, p. 716-728.
pled with bioventing: A case study. Proc. of Petroleum Kunii, D., and J. M. Smith. 1961. Heat transfer characteristics
Hydrocarbons and Organic Chemicals in Ground Water- of porous rocks, 11. Thermal conductivities of unconso1id;iterl
Prevention, Detection and Restoration. National Ground Wa- particles with flowing fluids. J. Am. Inst. Cheni. Eng.. v. -.
ter Assoc., Columbus, Ohio, p. 255-270. p. 29-31.
Kittel, J. A., R. E. Hinchee, R. D. Miller, C. Vogel, and R. Hoeppel. Lachenbruch, A. H., and J. H. Sass. 1977. Heat flow in the
1993. hi situ respiration testing: A field treatability test for United States. In J. Heacock (ed.), The Earth‘s Crust. Monogr.
bioventing. Proc. of Petroleum Hydrocarbons and Organic Ser. Am. Geophys. Union, Washington. D.C.. p. 626-6’5.
Chemicals in Ground Water-Prevention, Detection and Res- h i , S. H., and J. J. Jurinak. 1971. Numerical approxim;rtion
toration. National Ground Water Assoc., Columbus, Ohio, of cation exchange in miscible displacement through soil
p. 351-366. columns. Soil Sci. SOC.Am. Proc., v. 35, p. 894-899.
Klotz, 1. M. 1946. The adsorption wave. Chem. Rev., v. 39, Land, L. S. 1973. Holocene meteoric doloniitization of Pleisto-
p. 241 -268. cene limestone in North Jamaica. Setlinientology. v. 2 0 . p.
Klotz, I. M. 1950. Chemical Thermodynamics. Prentice-Hall, 41 1-422.
Englewood Cliffs, N.J., 369 p. Land, L. S., F. T. MacKenzie, and S. J. Could. 196-’.Pleistocene
Knapp, R. B., and J. E. Knight. 1977. Differential thermal expan- history of Bermuda. Geol. SOC.Am. Bull. ‘8. p. 993- I O o 6 .
sion of pore fluids: fracture propagation and microearthquake Land, L. S., and D.Prezbindowski. 1981. The origin ;iiid evolu-
production in hot pliiton environments. J. Geophys. Res., tion of saline formation waters; lower Cretaceous carlmi:ites.
V. 82, p. 2515-2522. South-Central Texas, U.S.A. J. Hydrol.. v. 54, p. 5 I --4
Knox, R. D., L. W. Canter, D. F. Kincannon. E. L. Stover, and Landes, K. K., and others. 1960. Petroleum resources in lme-
C. H. Ward. 1984. State-of-the-art of aquifer restoration. Na- ment rock. Am. Assoc. Petrol. Geol. Bull. 44.p. 1682- IWI.
tional Center for Groundwater Research, University of Okla- Langmuir, D. 1971. The geochemistn of wiic c;trbon;itc
-4 References

ground waters in central Pennsylvania. Geochim. Cos- Proc. of the NWWA/API Conference on Petroleum Hydmcar-
mochim. Acta, v. 35, p. 1023-1045. bons and Organic Chemicals in Ground Water-Prevention,
Langmuir, D. 1978. Uranium solution-mineral equilibria at low Detection and Restoration. National Water Well Assoc., Dub-
temperatures with applications to sedimentary ore deposits. lin, Ohio, p. 167-185.
Geochim. Cosmochim. Acta, v. 42, p. 547-569. Letkoff, L. J., and S. M. Gorelick. 1986. Design and cost analysis
Langmuir, D., and J. Mahoney. 1984. Chemical equilibrium and of rapid aquifer restoration systems using flow simulation
kinetics of geochemical processes in ground water studies. and quadratic programming. Ground Water, v. 24, no. 6,
In B. Hitchon and E. 1. Wallick (eds.),First Canadian/American p. 777-790.
Conference on Hydrogeology. National Water Well Assoc., LeGrand, H. E. 1954.Geology and groundwater in the Statesville
Dublin, Ohio, p. 69-95. area, North Carolina. North Carolina Dept. Conser. Dev., Div.
Lapidus. L., and N. R. Amundson. 1952. Mathematics of adsorp- Mineral Resources, Bull. 68.
tion in beds. VI. the effect of longitudinal diffusion in ion Leij. F. J.. and J. H. Dane. 1990. Analytical solutions of the one-
exchange and chromatographic columns. J. Phys. Chem., v. dimensional advection equation and two or threedimen-
56, p. 984-988. sional dispersion equation. Water Resources Res., v. 26, no.
Lapwood, E. R. 1948. Convection of a fluid in a porous medium. 7, p. 1475-1482.
Cambridge Phil. SOC.Proc., v. 44, p. 508-521. Leij, F. J., T. H. Skaggs, and M. T. van Genuchten. 1991. Analyti-
Larson, S. P., C. B. Andrews, M. D. Howland, and D. T. Feinstcin. cal solutions for solute transport in three-dimensional semi-
1987. A threedimensional modeling analysis of ground water infinite porous media. Water Resources Res., v. 27, no. 10,
pumping schemes for containment of shallow ground water p. 2719-2733.
contamination. Solving Ground Water Problems with Models. Lenhard, R. J., T.G. Johnson, and J. C. Parker. 1993. Experimen-
National Water Well Assoc., Dublin, Ohio, p. 517-531. tal observations of non-aqueous-phase liquid subsurface
Larson, S. P., M.S. McBride, and R. J. Wolf. 1976. Digital models movement. J. Contam. Hydrol., v. 12, p. 79-101.
of a glacial outwasb aquifer in the Pearl-Sallie Lakes area, Lenhard, R. J., and J. C. Parker. 1990. Estimation of free hydro-
West-Central Minnesota. Water Res. Invest., U.S. Geol. Sur- carbon volume from fluid levels in monitoring wells. Ground
vey, p. 40-75. Water, v. 28, p. 57-67.
Lattman, L. A., and R. R. Parizek. 1964. Relationship between Leonards, G. A. 1962. Engineering properties of soils. In G. A.
fracture traces and the occurrence of groundwater in carbon- Leonards (ed.), Foundation Engineering. McGraw-Hill, New
ate rocks. J. Hydrol., v. 2, p. 73-91. York, p. 66-240.
Law,J. 1944.A statistical approach to the interstitial heterogene- Levine, S. N., and W. C. Ghiorse. 1990. Analysis of environmen-
ity of sand reservoirs. Trans. Am. Inst. Min. Met. Engr., v. tal factors affecting abundance and distribution of bacteria,
155, p. 202-222. fungi and protozoa in subsurface sediments of the Upper
Lawrence Berkeley Laboratory, 1984. Panel report on coupled Atlantic Coastal Plain, USA. In C. 8. Fliermans and T. C.
thermomechanical-hydro chemical process associated with Hazen (eds.), Proc. of the First International Symposium on
a nuclear waste repository. Prepared for office of Nuclear Microbiology of the Deep Subsurface. Westinghouse Savan-
Regulatory Commission, Rept. LBL-18250. nah River Company, p. 5-31-5-45.
Leake, S. A., and D. E. Prudic. 1991. Documentation of a Com- Li, Y.-H., and S. Gregory. 1974. D m s i o n of ions in sea water
puter Program to SimulateAquifer-System Compaction Using and in deep-sea sediments: Geochim. Cosmochim. Acta., v.
the Modular Finite-Difference Ground-Water Flow Model: 38, p. 703-714.
Techniques of Water-Resources Investigations, U.S. Geologi- Lico, M. S., Y. K. Kharaka, W. W. Carothers, and V. A. Wright.
cal Survey, 68 p. 1982. Methods for Collection and Analysis of Geopressured,
LeBlanc, D. R. (ed.) 1984a. Movement and Fate of Solutes in a Geothermal and Oil Field Water. U S . Geol. Surv., Water Sup
Plume of Sewage-Contaminated Ground Water, Cape Cod, ply Paper 2194, 21 p.
Massachusetts. U S . Geological Survey Toxic Waste Ground- Lindberg, R. D., and D. D. Runnells. 1984. Ground water redox
Water Contamination Program. US. Geol. Sum. Open-File reactions: An analysis of equilibrium state applied to EH mea-
Rept., 84475, 175 p. surements and geochemical modeling. Science, v. 225, p.
LeBlanc, D. R. 1984b. Digital modeling of solute transport in a 925 -927.
plume of sewage-contaminated ground water. Movement and Linsley, R. K.,M.A. Kohler, and J. L. H. Paulhus. 1958. Hydrol-
Fate of Solutes in a Plume of Sewage-Contaminated Ground ogy for Engineers. McGraw-Hill, New York.
Water, Cape Cod, Massachusetts: U.S. Geological Survey Liu, C. W., andT. N. Narasimhan. 1989. Redoxcontrolled multi-
Toxic Waste Ground-Water Contamination Program. U.S. ple-species reactive chemical transport, 1. Model develop
Geol. Surv. Open-File Rept., p. 1 1 -45. ment. Water Resources Res., v. 25, no. 5, p. 869-882.
LeBlanc, D. R., S. P. Garabedian, K. M. Hess, L. W. Gelhar, R. 0. Lohman, S. W. 1%1. Compression of elastic aquifers. U.S. Geol.
Quadri, K. G. Stollenwerk, and W. W. Wood. 1991. Large- Survey Prof. Paper 424-8, p. 47-48.
scale natural gradient tracer test in sand and gravel, Cape Long, J. C. S., and others. 1982. Porous media equivalents for
Cod, Massachusetts, 1, Experimental design and observed networks of discontinuous fractures. Water Resources Res.,
tracer movement: Water Resour. Res., 27(5), p. 895-910. V. 18, p. 645-658.
Lee, C. H. 1915. The determination of safe yield of underground Longman, M. W. 1982. Carbonate diagenesis as a control on
reservoirs of the closed basin type. Trans. Am. SOC.Civil stratigraphic traps. Am. Assoc. Petrol. Geol. Eduction Course
Engrs., v. 78, p. 148-151. Notes 21.
Lee, M. D., V.W. Jamison, and R. L. Raymond. 1987.Applicabil- Loo, W. W., K.Frantz, and G. R. Holzhausen. 1984. The applica-
ity of in-situ bioreclamation as a remedial action alternative. tion of telemetry to large scale horizontal anisotropic perme-
 ReJerences 483

ability determination by surface tiltmeter survey. Ground Wa- Las Vegas groundwater basin, Nevada, U.S. Geol. Sur. Open
ter Mon. Rev, v. 4,p. 124-130. File Rept.
Lovering, T. S., and H. D. Goode. 1963.Measuring geothermal Manheim, F. T., and J. F. Bischoff. 1969.Geochemistry of pore
gradients in drill holes less than 60 feet deep, East Tintic waters from Shell Oil Company drill holes on the continental
District, Utah. U.S. Geol. Survey Bull. 1172. slope of the northern Gulf of Mexico. In E. E. Angino and
Lundegard, P. D. 1994. Actual versus apparent radius of influ- G. K. Billings (eds.), Chem. Geol. Special Issue, Geochemistry
ence-an air sparging pilot test in a sandy aquifer. In Proc. of Subsurface Brines, v. 4, p. 63-82.
Petroleum Hydrocarbons and Organic Chemicals in Ground Manheim, F. T.,and C. K. Paul]. 1981.Patterns of groundwater
Water- Prevention, Detection and Restoration. National salinity changes in a deep continental-oceanic tranxct off
Ground Water Assoc., Columbus, Ohio, p. 191 -206. the southeastern Atlantic coast of the U.S.A. In W. Back
Lusczynski, N. J.. and W. V. Swarzenski. 1966.Salt water en- and R. Letolle (eds.), J. Hydro]. Special Issue, Symposium on
croachment in Southern Nassau and Southeastern Queens Geochemistry of Groundwater, v. 54, p. 95-106.
Counties, Long Island, New York. U.S. Geol. Survey Water Manov, G. C., R. C. Bates, W. J. Hamer, and S. F. Acree. 1943.
Paper 1613-F. Values of the constants in the Debye-Huckel equation for
Lyman, W. J., W. F. Reehl, and D. D. Rosenblatt. 1982.Hand- activity coefficients: Jour. Amer. Chem. Soc., v. 65,p. 1765-
book of chemical property estimation methods. Environmen- 1767.
tal Behavior of Organic Compounds. McGraw-Hill, New York. Mantoglou, A,, and J. L. Wilson. 1982. The turning bands
Lyngkilde, J., and T. H. Christensen. 1992.Fate of organic con- method for simulation of random fields using line generation
taminants in the redox zones of a landfill leachate pollution by a spectral method: Water Resour. Res., l8(5), 1379-1394.
plume (vejen, Denmark). J. Contam. Hydro]., v. 10, p. Massniann, J. W. 1989.Applying groundwater flow models in
291 -307. vapor extraction system design. J. Environ. Eng., v. 115,no.
Mackay, D. M., D. L. Freyberg, and P. V. Roberts. 19%. A natural 1. p. 129-149.
Matheron, G . 1967. Elements pour une theorie des milieux
gradient experiment on solute transport in a sand aquifer. 1.
poreux. Masson et Cie, Pans.
Approach and overview of plume movement. Water Re-
Matthess, G. 1982.The Properties of Groundwater: New York,
sources Res., v. 22,no. 13,p. 2017-2029.
John Wiley & Sons, 406 p.
Mackay, D. M., and P. J. Leinonen. 1975.Rate of evaporation
-, and A. Pekdeger. 1985.Survival and transport of patho-
of low-solubility contaminants from water bodies to atmtr
genic bacteria and viruses in ground water. Ground Water
sphere. Environ. Sci. Technol., v. 9,no. 8,p. 1178-1180.
Quality, eds. C. H. Ward, W. Giger, and P. L. McCarty: New
Mackay, D. M., P. V. Roberts, and j. A. Cherry. 1985.Tra~sport
York, John Wiley & Sons, p. 472-482.
of organic contaminants in groundwater. Environ. Sci. Tech-
Matthews, C. S.,and D. G. Russel. 1967.Pressure buildup analy-
nol., v. 19,no. 5,p. 384-392. sis. In Pressure Buildup and Flow Test in Wells. Monograph
Mackay, D. M., and T. M. Vogel. 1985.Ground watercontamina- 1. Soc. Petrol. Engrs., Am. Inst. Mech. Engrs., New York.
tion by organic chemicals: Uncertainties in assessing impact. Maxwell, J. C. 1964.Influence of depth. temperature and geo-
Second Canadian/Americdn Conference on Hydrogeology. logic age on porosity of quartzose sandstone. Bull. Am. Assoc.
€3. Hitchon and M. Tnidell (eds.): Dublin, Ohio, National Petrol. Geol. 48,p. 697-709.
Water Well Assoc., p. 50-59. McRride, M. S., and H. 0. Pfannkuch. 1975.The distribution
Macumber, P. G . 1984.Hydrochemical processes in the regional of seepage within lake beds. J. Res., U.S. Geol. Surv., v. 3,
ground water discharge zones of the Murray Basin, Southeast- p. 505-512.
ern Australia. First Canadian/American Conference on Hydro- McCanhy, J. F., and J. M. Zachara. 1989.Subsurface tnnsport
geology. B. Hitchon and E. 1. Wallick (eds.): Dublin, Ohio, of contaminants. Environ. Sci. Technol., v. 23. no. 5. p.
National Water Well Assoc., p. 47-63. 496- 502.
Magara, K. 1975.Importance of aquathermal pressuring effect McCarty, P. I.., M. Reinhard, and 8. E. Rittmann. 1981. Trace
in GulfCoast. Am. Assoc. Petrol. Geol. Bull. 57,p. 2037-2045. organics in groundwater. Environ. Sci. Technol.. v. I S . no.
Magara, K. 1978. Compaction and Fluid Migration: Practical 1, p. 40-51.
Petroleum Geology. Elsevier Science, Amsterdam. McCarty, P. L.. B. E. Rittman, and E. J. Bouwer. 1984.Microbio-
Major, D. W., E. W. Hodgins, and 8. J. Butler. 1991.Field and logical processes affecting chemical transformations in
laboratoryevidence ofin situ biotransforniation of tetrachloro- groundwater. In G. Bitton and C. P. Gerba (cds.). Groundwa-
ethene to ethene and ethane at a chemical transfer facility ter Pollution Microbiology, john Wiley & Sons. New York.
in North Toronto. In R. E. Hinchee and R. G . Olfenbuttel p. 89-115.
(eds.), On-Site Bioreclamation Processes for Xenobiotic and McCarty, P. L., and J. T. Wilson. 1992.Natural anaerobic trrat-
Hydrocarbon Treatment. Butterworth, Stoneham, Mass., p. nieiit of a TCE plume, St. Joseph Michigan. NPL site. In
113-133. Bioremediation of Hazardous Wastes. U.S. Environmental Pnb
Malley, M. J., W. W. Eath, and L. H. Bongers. 1785.A case tection Agency, EPA/6OOR-92/16, p. &-'2.
history: Surface static collection and analysis of chlorinated McDonald, M. G., and A. W. Harbaugh. 1988.A modular thrce-
hydrocarbons from contaminated ground water. Proc. of the dimensional finite-difference ground-water flow model. I1.S.
NWWA/API Conference on Petroleum Hydrocarbons and Geol. Surv., Techniques of Water-Resources Investigations
Organic Chemicals in Ground Water- Prevention, Detection OGAI, 576 p.
and Restoration. National Water Well Assoc., Dublin, Ohio, McDonald, M. G., A. W. Harbaugh, 8.R. Om, and D. J. Ackerman.
p. 276-290. 1971.A method of converting no-flow cells to variable-hrad
Malmberg, G . T. 1960. An analysis of the hydrology of the cells for the t1.S. Geological Survey Modular Finite-Difference
-4 References

Ground-Water Flow Model. U.S. Geological Survey, Open- water contamination caused by organic solvent vapors.
File Report, 91-536, 99 p. Ground Water, v. 28, no. 2, p. 199-206.
McDowell-Boyer, L. M., J. R. Hunt, and N. Sitar. 1986. Particle Mercer,J. W., G. P. Pinder, and I. G. Donaldson. 1975.A Galerkin
transport through porous media. Water Resources Res.. v. finite element analysis of the hydrothermal system at Wai-
22, no. 13, p. 1901-1921. rakei, New Zealand. J. Geophys. Res., v. 80, p. 2608-2621.
McGarr A., and N. C. Gay. 1978. State of stress in the earth’s Mercer, J. W., D. C. Skipp, and D. Griffin. 1990. Basics of p u m p
crust. Ann. Rev. of Earth Planet. Sci., v. 6, p. 405-436. and-treat remediation technology. U S . Environ. Protection
McGowen, J. H., and C. G. Groat. 1971. Van Horn Sandstone, Agency, EPA-600/8-90/003, 31 p.
West Texas: An alluvial fan model for mineral exploration. Meyboom, P. 1960. Geology and groundwater resources of the
Texas Bur. Econ. Geol. Rept. Inv., v. 72, Austin. Milk River Sandstone in southern Alberta. Research Council
Mclelwain, T. A., and Reades, D. W., 1985, Status Report: The of Albert Memoir 2, Edmonton.
use of engineered covers at waste disposal sites. Second Meyboom, P. 1961. Estimating groundwater recharge from
Annual Canadian/American Conference on Hydrology, Proc. : stream hydrographs. J. Geophys. Res., v. 66, p. 1203- 1214.
Dublin, Ohio, National Water Well Assoc., p. 176- 182. Meyboom, P. 1962. Patterns of groundwater flow in the Prairie
McNabb,J. F., and G. Mallard. 1984. Microbial sampling in the environment. Can. Hydrol. Symp., 3rd, p. 5-33.
assessment of groundwater pollution. In G. Bitton and C.
Meyboom, P. 1966a. Unsteady groundwater flow near a willow
P. Gerba (eds.), Groundwater Pollution Microbiology. John
ring in hummocky moraine. J. Hydrol., v. 4, p. 38-62.
Wiley & Sons, New York, p. 235-260.
Meyboom, P. 1966b. Groundwater studies in the Assiniboine
Mead, D. W. 1919. Hydrology. McGraw-Hill, New York.
River drainage basin, pt. 1. The evaluation of a flow system in
Meinzer, 0. E. 1917. Geology and water resources of Big
Smokey, Clayton, and Alkali Springs Valleys, Nevada. U . S . southcentral Saskatchewan. Can. Dept. Mines Tech. Survey.
Geol. Survey Water Supply Papers 423. Geol. Surv. Can. Bull. 139.
Meinzer, 0. E. 1922. Map of Pleistocene lakes of the Basin and Meyboom, P. 1967. Mass transfer studies to determine the
Range Province and its significance. Bull. Geol. SOC.Am. 33, groundwater regime of permanent lakes in hummocky mo-
p. 541-552. raine of western Canada. J. Hydrol., v. 5, p. 117- 142.
Meinzer, 0. E. 1923. The occurrence of groundwater in the Mifflin, M. D. 1968. Delineation of groundwater flow systems
United States with a discussion of prlnciples. U S . Geol. Sum. in Nevada. Desert Research Inst. Tech. Rept. Ser. H-W, no.
Water Supply Papers, v. 489. 4, Reno.
Meinzer, 0. E. 1923. Outline of groundwater in hydrology with Mikels, F. C. 1952. Report on hydrogeological survey for city
definitions. U.S. Geol. Surv. Water Supply Papers, v. 494 of Zion, Illinois. Ranney Method Water Supplies, Inc., Colum-
Meinzer, 0. E. 1927. Plants as indicators of groundwater. L1.S. bus, Ohio.
Geol. Survey Water Supply Papers, v. 577. Miller, C. W., and L. V. Benson. 1983. Simulation of solute
Meinzer, 0. E. 1928. Compressibility and elasticity of artesian transport in a chemically reactive heterogeneous system:
aquifers. Econ. Geol., v. 23, p. 263-291. model development and application. Water Resources Res.,
Meinzer, 0. E. 1932. Outline of methods for estimating ground- v. 19, no. 2, p. 381-391.
water supplies. U.S. Geol. Survey Water Supply Papers Miller, R. N., D. C. Downey, V. A. Carmen, R. E. Hinchee, and
63&C, p. 94- 144. A. Leeson. 1993. A summary of bioventing performance at
Meinzer, 0. E. 1939. Groundwater in the United States. L.S. multiple Air Force sites. Proc. of Petroleum Hydrocarbons
Geol. Survey Water Supply Papers S36-D, p. 157-232. and Organic Chemicals in Ground Water-Prevention, De-
Meinzer, 0. E. 1942 (ed.). Hydrology. McGraw-Hill, New York. tection and Restoration. National Ground Water Assoc., Co-
Meinzer, 0. E., and H. H. Hard. 1925. The artesian water supply lumbus, Ohio, p. 397-411.
of the Dakota Sandstone in North Dakota, with special refer- Millington, R. J. and J. P. Quirk. 1961. Permeability of porous
ence to the Edgeley Quadrangle. U.S. Geol. Survey Water solids. Trans., Faraday SOC.,v. 57, p. 1200- 1207.
Supply Papers 52@-E, p. 73-95. Minear, R. A., and L. H. Keith (eds.). 1984. Water Analysis, v.
Meinzer, 0. E., and N. D. Steams. 1928. A study of groundwater Ill, Organic Species. Academic Press, Orlando, Fla., 456 p.
in the Pomperaug Basin, Conn., with special reference to
Mines, B. S., J. L. Davidson, D. Bloomquist, and T. B. Stauffer.
intake and discharge. US. Geol. Survey Water Supply Papers
1993. Sampling of VOCs with the BAT Ground Water Sam-
V. 597, p. 73-146.
pling System. Ground Water Mon. Remediation, v. 13,
Melnyk, T. W., F. 8. Walton, and H. L. Johnson. 1983. High
level waste glass field burial tests at CRNL: The effect of
no. 1, p. 115-120.
geochemical kinetics on the release and migration of fission Mitsdarffer, A. R. 1985. Hydrodynamics of the Mission Canyon
products in a sandy aquifer. Rep. AECLd836, Atom. Energy, formation in the BillingsNose area, North Dakota. M.S . thesis,
Can. Ltd. Chalk River, Ontario. Geology Dept., Texas A M University, College Station.
Mendoza, C. A., and E. 0. Frind. 1990a. Advective-dispersive Moltz, F. J., A. D. Parr, and P. F. Anderson. 1981. Thermal
transport of dense organic vapors in the unsaturated zone. energy storage in a confined aquifer, second cycle. Water
1. Model development. Water Resources Res., v. 26, no. 3, Resources Res., v. 17, p. 641 -645.
p. 379-387. Moltt, F. J., M. A. Widdowson, and L. D. Benefield. 1986.Simula-
Mendoza, C. A., and E. 0. Frind. 1990b. Advective-dispersive tion of microbial growth dynamics coupled to nutrient and
transport of dense organic vapors in the unsaturated zone. oxygen transport in porous media: Water Resources Res., v.
2. Sensitivity analysis. Water Resources Res., v. 26, no. 3, 22, no. 8, p. 1207-1216.
p. 388-398. Moody, J. B. 1982. Radionuclide Migration/Retardation: Re-
Mendoza, C. A., and T. A. McAlary. 1990. Modeling of ground- search and Development Technology Status Report. Office
 References 485

of Nuclear Waste Isolation, Battelle Memorial Ins., O M - Neuman, S. P. 1975. Analysis of pumping test data from anistr
321, 61 p. tropic unconfined aquifers considering delayed gravity re-
Mook, W. G. 1980. Carbon-14 in hydrogeological studies. Chap- ponse. Water Resources Res., v. 1 1 , p. 329-342.
ter 2. In P. Fritz and J. C. Fontes (eds.), Handbook of Environ- Neuman, S. P., and P. A. Witherspoon. 1969a. Theory o f flow
mental Isotope Geochemistry, v. 1. Elsevier, Amsterdam, in confined two aquifer system. Water Resources Res.. v. 5 ,
p. 49-74. p. 803-816.
Moran, S. R., G. H. Groenwold, and J. A. Cherry. 1978a. Geo- Neuman, S. P., and P. A. Witherspoon. 1969b. Applicability o f
logic, Hydrologic, and Geochemical. Concepts and Tech- current theories of flow in leaky aquifers. Water Resources
niques in Overburden Characterization for Mined-land Res., v. 5, p. 817-829.
Reclamation. North Dakota Geological Survey, Report of In- Neuman, S. P., and P. A. Witherspoon. 1972. Field determiria-
vestigation no. 63, 152 p. tion of hydraulic properties of leaky multiple aquifer systems.
Moran, S. R., J. A. Cherry, P. Fritz, W. M. Peterson, M. H. Water Resources Res., v. 8, p. 1284- 1298.
Somerville, S. A. Stancel, and J. H. Ulmer, 1978b, Geology. Neuzil, C. E. 1986. Groundwater in low permeability environ-
Groundwater Hydrology, and Hydrochemistry of a Proposed ments. Water Resources Res., v. 22, p. 1 163- 1195.
Surface Mine and Lignite Gasification Plant Site Near Dunn Neuzil, C. E., and D. W. Pollock. 1983. Erosional unloading iind
Center, North Dakota: North Dakota Geological Survey, Re- fluid pressures in hydraulically tight rocks. J. Geol.. v. 9,
port of Investigation 61, 263 p. p. 179-193.
Morel, F. M. M. 1983. Principles of Aquatic Chemistry. John Newell, C. J., S. D. Acree, R. D. Ross, and S. G. Hiding, 1995.
Wiley flr Sons, New York, 446 p. Light nonaqueous phase liquids: U.S. Environmental Protec-
Morel, F. M. M., and J. G. Hering. 1993. Principles and Applica- tion Agency, EPA/540/S95/500, 28 p.
tions of Aquatic Chemistry. John WiIey & Sons, New York, Nield, D. A. 1968. Onset of thermohaline convection in a porous
588 p. medium. Water Resources Res., v. 4 , p. 1553-1560.
Morgan, P. V., and others. 1981. A groundwater convective Nielsen, P. H., H. Bjarnadottir, P. L. Winter, and T. H. Chris-
model for Rio Grande Rift geothermal systems. Geotherni.
tensen. 1995. In situ and laboratory studies on the fate of
Resour. Council Trans., v. 5, p. 193- 196.
specific organic compounds in an anaerobic landfill Ieachate
Morganstern, N. R. 1970. The influence of groundwater on
plume. 2. Fate of aromatic and chlorinated aliphatic coni-
stability. Proc. 1st Intern. Conf. on Stability in Open Pit Min-
pounds. J. Contam. Hydrol., v. 20. p. 5 1-66.
ing. Vancouver, British Columbia.
Nielsen, P. H., and T. H. Christensen. 1994. Variability of biologi-
Morrison, R. D., and P. E. Brewer, 1981. Airlift samples for
cal degradation of aromatic hydrocarbons in an aerobic aqui-
zone-of-saturation monitoring: Ground Water Monitoring Re-
fer determined by laboratory batch experiments. J. Contam.
view, v. 1, no. 1 , p. 52-55.
Hydrol., v. 15, p. 305-320.
Muskat, M. 1937. The Flow of Homogeneous Fluids Through
Nielsen, P. H., andT. H. Christensen, 1994. Variability ofbiologi-
Porous Media. McGraw-Hill, New York.
cal degradation of phenolic hydrocarbons in an aerobic aqui-
Muskat, M., and M. W. Meres. 1936. The flow of heterogeiieous
fluids through porous media. Physics, v. 7, p. 346-363.
fer determined by laboratory batch experiments: J. Contim.
Narasimhan, T. N., A. F. White, and T. Tokunaga. 1986. Ground- Hydrol., v. 17, p. 55-67.
water contamination from an inactive uranium mill tailings Nobel, E. A. 1963. Formation of ore deposits by water of com-
pile. 2. Application of a dynamic mixing model. Water Re- paction. Econ. Geol., v. 58, p. 1145 - 1 156.
sources Res., v. 22, no. 13, p. 1820-1834. Nordstrom, D. K., and J. L. Munoz. 1986. Geochemical Themio-
Nathenson, M., and L. J, P. Muffler. 1975. Geothermal resources dynamics. Blackwell Scientific Publications. Palo Alto, Calif..
in hydrothermal convection systems and conduction domi- 477 p.
nated areas. In Assessment of Geothermal Resources of the Norton, D. 1978. Sourcelines, source regions, and pathlinrs in
United States- 1975, U.S. Geol. SUN. Circ. v. 726, p. 104- hydrothermal systems related to cooling plutons. Econ.Grol..
121. V. 73, p. 21-28.

National Academy of Science. 1983. Risk Assessment in the Norton, D., and R. Knapp. 1977. Transport phenomena in hy-
Federal Government: Managing the Process. U.S. Govern- drothermal systems: Nature of porosity. Am. J. Sci.. v. 27.
ment Printing Office, Washington, D.C. p. 913-936.
Nativ, R., and others. 1995. Water recharge and solute transport NRC, National Research Council. 1993. In Situ Bioremediation:
through the vadose zone of fractured chalk under desert When Does It Work. National Academy Press. Washington.
conditions. Water Resources Res., v. 31, p. 253-261. D.C., 207 p.
Neely, W. B. 1985. Hydrolysis. In W. B. Neely and G. E. Blau Nur, A. 1972. Dilatancy, pore fluids and premonitory wri;itions
(eds.), Environmental Exposure from Chemicals, v. I. CRC of ts/tp travel times. Bull. Seismol. Soc. Am. v. 62, p. 121'-
Press lnc., Boca Raton, Fla., p. 157- 173. 1222.
Neretnieks, 1. 1985. Transport in fractured rocks. Proc. of Hy- Nyer, E., G. Boettcher, and B. Morello. 1991. LJsingthe proper-
drogeology of Rocks of Low Permeability. Tucson Arizona ties of organic compounds to help design a treatment system.
International Assoc. of Hydrogeologists, p. 306. Ground Water Mon. Rev., v. 1 1 , no. 4, p. 1 15- 120.
Neuman, S. P. 1972. Theory of flow in unconfined aquifers O'Connor, M. J., J. G. Agar, and R. D. King. 19H-ia. Practical
considering delayed response of the water table. Witer Re- experience in the management of hydrocarbon vapors in the
sources Res., v. 8, p. 1031-1045. subsurface. Proc. of the NWWA/API Conference on Prtro-
Neuman, S. P. 1974. Effect of partial penetration on flow in leum Hydrocarbons and Organic Chemicals in Ground W'I.
unconfined aquifers considering delayed response of the wa- ter-Prevention, Detection and Restoration. National Water
ter table. Water Resoiirces Res., v. 9, p. 1102- 1103. Well Assoc., Dublin, Ohio, p. 519-533.
4 3 References

O’Connor, M. J., and L. W. Bouckhout. 1983. Gasoline spills Palmer, C. D., and J. A. Cherry. 1984. Geochemical evolution
in urban areas: a comparison of two case histories. Proc. of groundwater in sequences of sedimentary rocks. J. Hydrol.,
Seminar on Groundwater and Petroleum Hydrocarbons- V. 75, p. 27-65.
Protection, Detection, Restoration. Petroleum Assoc. for Con- Pankow, J. F., R. L. Johnson, J. E. Houck. S. M. Brillante, and
servation of the Canadian Environment, Ottawa, p. VIII-1- W. J. Bryan. 1984. Migration of chlorophenolic compounds
VIII-34. at the chemical waste disposal site at Alkali Lake, Oregon.
O’Connor, M. J., A. M. Wofford, and S. K. Ray. 1984b. Recovery 1 . Site description and ground-water flow. Ground Water, v.
of subsurface hydrocarbons at an asphalt plant: Results of 22, no. 5, p. 593-601.
a five year monitoring program. Proc. of the NWWA/API Papadopulos, 1. S. 1965. Nonsteddy flow of a well in an infinite
Conference on Petroleum Hydrocarbons and Organic Chemi- anisotropic aquifer. Proc. Dubrovnik Symp. on Hydrol. of
cals in Ground Water-Prevention, Detection and Restora- Fractured Rock, Inter. Assoc. Sci. Hydrol., p. 21-31.
tion National Water Well Assoc., Dublin, Ohio, p. 359-376. Papadopulos, I. S., J. D. Bredehoeft, and H. H. Cooper. 1973.
Office of Technology Assessment. 1982. Summary: Managing O n the analysis of slug test data. Water Resources Res., v. 9,
commercial high level radioactive work. U.S. Congress, Wash- p. 1087- 1089.
ington, D.C. Papadopulos, I. S., and H. H. Cooper. 1967. Drawdown in a well
Office of Technology Assessment, 1984. Protecting the nation’s of large diameter. Water Resources Res., v. 3, p. 241-244.
groundwater from contamination. Office of Technology As- Papadopulos, S. S., and others. 1975. Assessment of onshore
sessment, 0TA-O-233:Washington, D.C., OTA, 244 pp. geopressured-geothermal sources of the United States. As-
Ogata, A., and R. B. Banks. 1961. A solution of the differential sessment of Geothermal Resources in the United States-
equation of longitudinal dispersion in porous media. 1J.S. 1975. U.S. Geol. Surv. Circ. 726, p. 125-142.
Geol. Surv. Prof. Paper 41 1-A. Parker, G. G., and V. T. Springfield. 1950. Effects of earthquakes,
ONe, E. L. 1959. Some consideration in determining the origin rains, tides, winds, and atmospheric pressure changes on the
of ore deposits of the Mississippi Valley Type. Part I, Econ. water in geologic formations of southern Florida. &on. Geol.,
Geol., v. 54, p. 769-789. V. 45, p. 441-460.
Oliver,J. 1986. Fluids expelled tectonically from orogenic belts: Parkhurst, D. L., D. C. Thorstenson, and L. N. Plummer. 1980.
Their role in hydrocarbon migration and other geologic phe- PHREEQE-A Computer Program for Geochemical Calcula-
nomena. Geology, v. 14, p. 99-102. tion. U.S. Geol. Surv. Water Resources Invest. Report. 80-96,
Oreskes, N., K. Shrader-Frechette, and K. Belitz. 1994. Verifica- 210 p.
tion, validation, and confirmation of numerical models in Parsons, M. L. 1970. Groundwater thermal regime in a glacial
earth sciences. Science, v. 263., p. 641-646. -
complex. Water Resources Res., v. 6, p. 1701 1720.
Ortoleva, P.,J. Chadam, E. Merino,and A. Sen. 1987b. Geochem- Patton, F. D., and A. J. Hendron, Jr. 1974. General report on
ical self-organization. 11: The reactive-infiltration instability. mass movements. Proc. 2nd Intern. Congress, Intern. Assoc.
Am. J. Sci., v. 287, p. 1008-1040. Eng. Geol., San Pdulo, Brazil, v. 2, p. V-GR.1 -V-GR.57.
Ortoleva, P., E. Merino, C. Moore, and J. Chadam. 1987a. G e e Peargin, T. R., and D. H. Mohr. 1994. Field criteria for SVE
chemical self-organization. I: Reaction-transport feedbacks pilot tests to evaluate data quality and estimate remediation
and modeling approach. Am. J. Sci., v. 287, p. 979-1007. feasibility. Proc. of Petroleum Hydrocarbons and Otganic
Ozbdigin, M. M., and M. A. Powers. 1984. Hydrodynamic isola- Chemicals in Ground Water- Prevention, Detection and Res-
tion in hazardous waste containment. Proc. of the Fourth toration. National Ground Water Assoc., Columbus, Ohio,
National Symposium and Exposition on Aquifer Restoration & p. 337-350.
Groundwater Monitoring. National Water Well Assoc., Dub- Pearson, F. J., Jr., and B. B. Hanshaw. 1970. Sources of dissolved
lin, Ohio, p. 44-49. carbonate species in groundwater and their effectson carbon-
Paige, S. 1936. Effect of a sea-level canal on the groundwater 14 dating. Isotope Hydrology. International Atomic Energy
level of Florida. Econ. Geol., v. 31, p. 537-570. Agency, Vienna, p. 271 -286.
Paige, S. 1938. Effect of a sea-level canal on the groundwater Perkins, T. K., and 0. C. Johnston. 1963. A review of diffusion
level of Florida-A reply. Econ. Geol., v. 33, p. 647-665. and dispersion in porous media. J. SOC.Petrol. Eng., v. 3,
Palciauskas, V. V., and P. A. Domenico. 1976. Solution cheniis- p. 70-83.
try, mass transport, and the approach to chemical equilibrium Peter, K. L. 1984. Hydrochemistry of lower Cretaceous sand-
in porous carbonate rocks and sediments. Geol. Soc. Am. stone aquifers, Northern Great Plains. In D. G. Jorgenson and
Bull. 87, p. 207-214. D. C. Signor (eds.), Proc. of the Geohydrology Dakota Aquifer
Palciauskas, V. V., and P. A. Domenico. 1980. Microfracture Symposium, Water Well Journal Publ., Worthington, Ohio,
development in compacting sediments: Relation to hydrocar- p. 163-174.
bon maturation. Am. Assoc. Petrol. Geol. Bull. 64, p. 927- Peters, R. R., and E. A. Klavetter. 1988. A continuum model
937. for water movement in an unsaturated fractured rock mass.
Palciauskas, V. V.. and P. A. Domenico. 1982. Characterization Water Resources Res., v. 24, p. 416-430.
of drained and undrained response of thermally loaded reposi- Pfannkuch, H. 0. 1962.Contribution a I’Ctude des deplacement
tory rocks. Water Resources Res., v. 18, p. 281-290. de fluides miscible dans un milieu poreux. Rev. Inst. Fr.
Palciauskas, V. V., and P. A. Domenico. 1989. Fluid pressure Petrol., v. 18, no. 2, p. 215-270.
in deforming porous rocks. Water Resources Res., v. 25. Phillips, F. M., H. W. Bentley, S. N. Davis, D. Elmore, and G.
p. 203-213. Swanick. 1986. Chlorine 36 dating of very old groundwater.
Palmer, A. N. 1991. Origin and morphology of limestone caves. 2. Milk River Aquifer, Alberta. Water Resources Res., v. 22,
Geol. Soc. Am. Bull. 103, p. 1 -21. no. 13, p. 2003-2016.
 References 487

Phillips, F. M., J. L. Wilson, and J. M. Davis. 1989. Statistical Price, M. 1994. A method for assessing the extent o f fissuring
analysis of hydraulic conductivity distributions: A quantitative in double porosity aquifers, using data from packer tests.
geological approach. In F. J. Molz, J. C. Melville, and 0.G w e n IAHS Publ. NO. 222, p. 271-278.
(eds.), Proc. Conf. on New Field Techniques for Quantifying Prichett, W.C. 1980. Physical properties of shales and possible
the Physical and Chemical Properties of Heterogeneous Aqui- origin of high pressures. Soc. Petrol. Eng. J.. v. 20. p. 3-41-348.
fers. National Ground Water Assoc.. Dublin, Ohio, p. 19-31. Prickett, T. A. 1965. Type curve solution to aquifer tests under
Pickens, J. F., R. E. Jackson, K. J. Inch, and W. F. Merritt. water table conditions. Ground Water, v. 3, p. 5- 11.
1981. Measurement of distribution coefficients using a radial Prickett, T. A., T. G . Naymik, and C. G. Lonquist. 1981. A
injection dual-tracer test. Water Resources Res., v. 17, no. 3, Random-Walk Solute Transport Model for Selected Ground-
p. 529-544. water Quality Evaluations. Illinois State Water Surv. Bull. 65,
Pinder, G. F. 1973. A Galerkin finite-elenient simulation of Ch;impaign, 111.. 103 p.
groundwater contamination on Long Island, New York. Wa- Prudic, D. E. 1989. Documentation of a computer program
ter Resources Res., v. 9, no. 6, p. 1657- lM4. t o simulate stream-aquifer relations using a modular, finite-
Pinder, G. 1979. State of the art review of geothermal reservoir difference, ground-water flow model. U.S. Geol. SUN.. Open-
engineering. Rept. LBL-9093, Lawrence Berkeley Lab., Berke- File Rept. 88-729, 1 I3 p.
ley, Calif. Puls, K. W., and R. LM. Powell. 1092. Transport o f inorganic
Pinder, G. F., and H. H. Cooper, Jr. 1970. A numerical technique colloids through natural aquifer material: Implications for
for calculating the transient position of the saltwater front. contaminant transport. Environ. Sci. Technol.. Y. 26. no. 5.
Water Resources Res., v. 6, p. 875-882. p. 614-621.
Piper, A. M., 1944, A graphic procedure in the geochemical Pytkowicz, R. P. 1983. Equilibria, Nonequilibria SL Natiir~lW:i-
interpretation of water analysis: Trans. Amer. Geophys. t e n , Vol. 11. John Wiley SL Sons,New York. 353 p.
Union, v. 25, p. 914-923. Quinlan, J. F. 1990. Special problenis of ground-water monitor-
Pitzer, K. S., and J. J. Kim. 1974. Thermodynamics of electro- ing in karst terranes. In D. .M. Nielson and A. 1. Johnson
lytes: 4. Activity and osmotic coefficients for mixed electro- (eds.), Ground Water and Vadose Zone Monitorings. Am.
lytes. J. Am. Chem. Soc., v. 96, p. 5701 -5707. Soc. Testing and Materials, Philadelphia, p. 275-304.
Plummer, L. N. 1984. Geochemical modeling: A comparison o f Quinlan, J. F., and R. 0. Ewers. 1985. Ground w;iter flow in
forward and inverse methods. In B. Hitchon and E. I. Wallick limestone terranes: Strategy, rationale and procedure for reli-
(eds.), First Canadkdn/Ameflcan Conference on Hydrogeol- able, efficient monitoring of ground water quality in Karst
ogy. National Water Well Assoc., Dublin, Ohio, p. 149- 177. areas. Pmc. Fifth National Symposium and Exposition on
Plummer, L. N., U. F. Jones, and A. H. Truesdell. 1976. Aquifer Restoration and Ground Water Monitoring. National
WATEQF-A Fortran IV version o f WATEQ, A Computer Water Well Assoc.., Dublin. Ohio, p. 107-234.
Program for Calculating Chemical Equilibrium of Natural Wa- Quinlan, J. F., and J. A. Ray. 1981. Groundwater basins in the
ters. U.S. Geol. Surv. Water Resources Invest. 7613, 61 p. Mammoth Cave Region, Kentucky. Occ. Publ. No. I . Friends
Plummer, 1,. N., and T. M. L. Wigley. 1976. The dissolution of of the Karst, Mammoth Cave.
calcite in C02-saturated solutions at 25°C and 1 atmosphere Quinlan, J. F., and D. R. Rowe. 1977. Hydrology and water
total pressure. Geochim. Cosmochim. Acta, v. 40, p. 191- quality in the central Kentucky karst: Phase I. Univ. Kentucky
202. Water Resources Res. Institute, Research Report No. 101.
Poland, J. F. 1961. The coefficient of storage in a region of 93 p.
major subsidence caused by compaction of an aquifer. U.S. Ragone, S. E. (ed.). 1988. U.S. Geological Survey Program on
Geol. Prof. Paper 424-8, p. 52-54. Toxic Waste-Ground Water Contamination. Proc. of the
Poland, J. F.. and G. H. Davis. 1956. Subsidence of the land Second Technical Meeting, Cape Cod, Massachusetts. Octtr
surface in Tulare- Wasco (Delano) and Los Ranos- Kettleman ber 21-25, 1985. U.S. Geol. Surv. Open-File R e p . M~481.
City area, SanJoaquin Valley, California. Trans. Am. Geophys. Rainwater, F. H., and L. L. Thatcher. 1'960. Methods for Collec-
Union, v. 37, p. 287-296. tion and Analysis of Water Samples. US. Geol. SUN. Water
Poland, J. F., and others. 1975. Land subsidence in the San Supply Paper 1454, 301 p.
Joaquin Valley, California as of 1972. U.S. ( k 0 l . .Survey Prof. Raleigh, C. H. 1971. Earthquakes and fluid injection. Am. Ahsoc.
Paper 437-H. Petrol. Geol. Memoirs, no. 18, p. 273-279.
Pollock, D. W. 1989. Documentation of computer programs to Raleigh, C. B., J. H. Healy, and J. D. Rredehoeft. 1972. Faulting
conipirte and display pathlines using results from the U.S. and crustal stress at Rangely, Colorado. Geopliys. Union
Geological Survey modular three-dimensional finite-differ- Monog. Series 16, p. 275-284.
ence ground-water flow model. US. Geol. Surv., Open-File Raleigh, C. B., J. H. Healy, and J. I). Bredehoeft. 1076. A n
Report 89-381, 1 8 8 13. experiment in earthquake control at Rangeley. Colorado. Sci-
Powers, M. C. 1967. Fluid-release mechanisms in compacting ence, v. 191, p. 1230- 1236.
marine miidrocks and their importance in oil exploration. Randall, J. H., and T. R. Schultz. 1976. C1ilorotluoroc;irhons :IS
Am. Assoc. Petrol. Geol. Bull. 51, p. 1240- 1254. hydrologic tmcers: A new technology. Hydro]. W;itrr Rcs.
Pratts, M. 1966. The effect of horizontal flow on thermally Ariz. Southwest, v. 6, p. 189-195.
induced convection currents in porous mediums. J. Geophys. Ranganathan, V., and J. S. Hanor. 1987. A numerical model fir
R ~ s . ,V. 71, p. 4835-4838. the formation of saline waters due to diffusion o f dissolved
Pratts, M. 1970. A method for determining the net vertical NaCl in subsiding sedimentary basins with euporites. J. Hy-
permeability near ;I well from in-situ measurements. Trans. drol.. v. 92, p. 97- 120.
Am. Inst. Mining Engrs., v. 249, p. 637-643. Rasmussen, T. C., J. H. Blanford, and P. J. Sheets. 1989. C1i;ir;ic-
4 3 References

terization of unsaturated fractured tuff at the Apache Leap 1990. A field evaluation of in-situ biodegradation of chlori-
Site: comparison of laboratory/field, matrix/fracture and nated ethenes: Part 1. Methodology and field site characteriza-
water/air data. Paper presented at Nuclear Waste Isolation tion. Ground Water, v. 28, no. 4, p. 591-604.
in the Unsaturated Zone, joint sponsorship by the American Roberts, P. V., L. Semprini, G. D. Hopkins, D. Grbic-Galic,
Nuclear SOC. and the Geological SOC.of America, Las Ve- P. L. McCarty, and M. Reinhard. 1989. In-situ Restoration of
gas, Nev. Chlorinated Aliphatics by Methanogenic Bacteria. U.S. Envi-
Raven, K., and K. S. Novakowski. 1984. Field investigation of ron. Protection Agency, EPA/600/2-89/033, 214 p.
the solute-transport properties of fractures In monzonitic Robin, M. J. L., A. L. Gutjahr, E. A. Sudicky, and J. L. Wilson.
gneiss. International Symposium on Groundwater Resources 1993. Crosscorrelated random field generation with direct
Utilization and Contaminant Hydrogeology, Vol. 11. Atomic Fourier transform method. Water Resources Res., v. 29, no.
Energy of Canada Ltd., Pinawa, Manitoba, p. 507-516. 7, p. 2385-2397.
Reardon, E. J.$and P. Fritz, 1978,Computer modelling of ground Robinson, G. D. 1985. Structure of pre-Cenozoic rocks in the
water 13C and “C isotope compositions: J. Hydrol., v. 36, vicinity of Yucca Mountain, Nye County, Nevada-A potential
p. 201-224. nuclear waste disposal site: U.S. Geol. Survey Bull., p. 1647.
Reddell, D. L., and D. K. Sunada. 1970. Numerical Simulation Robinson, P. C. 1982. Connectivity of fracture systems: A perco-
of Dispersion in Groundwater Aquifers. Colorado State Univ., lation theory approach. Theor. Phys. Div., AERE Harwrll,
Hydrol. Paper 4 1, 79 p. HL82/960.
Regan, L. J., and A. W. Hughes. 1949. Fractured reservoirs of Robinson, T. W. 1939. Earth tides shown by fluctuations of
Santa Maria District, California. Bull. Am. Assoc. Petrol. Geol. water levels in wells in New Mexico and Iowa. Trans. Am.
33, p. 32-51. Geophys. Union, v. 20, p. 656-666.
Rege, S. D., and H. S. Folger. 1989. Competition among flow, Rogoshewski, P., H. Bryson, and K. Wagner. 1983. Remedial
dissolution and precipitation in porous media. AIChE J., v. Action Technology for Waste Disposal Sites. Pollution Tech-
35, p. 1177- 1185. nology Review No. 101. Noyes Data Corp., Park Ridge, N.J.,
Rehfeldt, K. R., and others. 1992. Field study of dispersion in a 498 p.
heterogeneous aquifer. 1, Geostatistical analysis of hydraulic Rojstaczer, S., and J. D. Hredehoeft. 1988. Groundwater and
conductivity: Water Resources Res., vol. 28, no. 12, p. 3309- fault strength. In J. J. Back, J. Rosenshein, and P. R. Seaber
3324. (eds.), Hydrogeology, The Geology of North America. Geol.
Reichard, E., C. Cranor, R. Raucher, and G. Zapponi. 1990. SOC.Am. Pub., p. 447-460.
Groundwater contamination risk assessment: A guide to un- Romm, E. S. 1966. Flow characteristics of fractured rocks (in
derstanding and managing uncertainties. Intl. Assoc. Hydrol. Russian). Nedra, Moscow.
Sci., IAHS Press, Publication 196, 204 p. Roth, J. P. 1969. Earthquakes and reservoir loading. Proc. 4th
Reinson, C. E. 1984. Barrier Island and associated strand plain World Conf. on Earthquake Engineering, Santiago, Chile.
systems. In R. G. Walker (ed.), Facies Models. GeoSciences Royster, D. L. 1979. Landslide remedial measures. Bull. Assoc.
Canada Reprint Series 1, p. 119-140. Eng. Geol. v. 16, p. 301-352.
Reisinger, H. J., D. R. Burris, L. R. Cessar, and G. D. McCleary. Rubin,J. 1968. Theoretical analysisof two dimensional transient
1987. Factors affecting the utility of soil vapor assessment flow of water in unsaturated and partly saturated soils. Soil
data. Proc. First Outdoor Action Conference on Aquifer Resto- Sci. Am. Proc. v. 32, p. 607-615.
ration, Ground Water Monitoring and Geophysical Methods. Rubin,1..and R.V.James. 1973. Dispersion affected transport of
National Water Well Assoc., Dublin, Ohio, p. 425-435. reacting solutes in saturated porous media: Galerkin method
Relyea,J. F. 1982. Theoretical and experimental considerations applied to equilibrium controlled exchange in unidirectional
for the use of the column method for determining retardation steady water flow. Water Resources Res.,v. 9, no.5, p. 1332-
factors. Radioactiv. Manage. Nucl. Fuel Cycle, v. 3, p. 151- 1356.
166. Russell, W. L. 1972. Pressure-depth relations in Appalachian
Renner, J. L., D. E. White, and D. L. Williams. 1975. Hydrother- region. Am. Assoc. Petrol. Geol. Bull. 56, p. 528-536.
mal convection systems. In Assessment of Geothermal Re- Rust, B. R., and E. H. Koster. 1984. Coarse alluvial deposits. In
sources of the United States-1975. U.S. Geol. Surv. Circ. R. G. Walker (ed.), Facies Models. GeoSciences Canada Re-
726, p. 47-56. print Series 1.
Renner, R. 1995. Predicting chemical risks with multimedia Rutter, E. H. 1983. Pressure solution in nature, theory and
fate models. Environ. Sci. Techno]., v. 29, no. 12, p. 556-559. experiment. J. Geol. SOC.,v. 140, p. 725-740.
Richards, L. A. 1931. Capillary conduction of liquids through Sadovsky, M. A., and others, 1969, The process preceding strong
porous mediums. Physics, v. 1, p. 318-333. earthquakes in some regions of Middle Asia: Tectonophysics,
Rittmann, B. E., and P. L. McCarty. 1980. Model of steady-state V. 23, p. 247-255.
bioNm kinetics. Biotech. Bioeng., v. 22, p. 2343-2357. Sass, J. W., and H. Lachenbrach. 1982. Preliminary interpreta-
Rittmann, B. E., and P. L. McCarty. 1980. Evaluation of steady- tion of thermal data from the Nevada Test Site. U.S. Geol.
state biofilm kinetics: Biotech. Bioeng., v. 22, p. 2359-2373. Surv. Open File Rept. 82-973, Denver.
Rittmann, B. E. 1993. The significance of biofilms in porous Sass, J. W., and others. 1976. A new heat flow contour map of
media. Water Resources Res., v. 29, no. 7, p. 2195-2202. the conterminous United States. U.S. Geol. Sum. Open File
Robbins, G. A. 1983. Determining Dispersion Parameters to Rept. 76756, Denver, Colo.
Predict Groundwater Contamination. Ph.D. dissertation, Sauter, M. 1992.Assessment of hydraulic conductivity in a karst
Texas A&M Univ., College Station, 226 p. aquifer at local and regional scale. In J. F. Quinlan and A.
Roberts, P. V., G. D. Hopkins, D. M. MacKay, and L. Semprini. Stanley (eds.), Hydrology, Ecology, Monitoring, and Manage-
 References 183

ment of Karst Terranes Conference. Proc. National Ground ence on Hydrogeology, Proc.: National Water Well Assoc.,
Water Assoc., Dublin, Ohio, p. 39-57. Dublin, Ohio, p. 2-12.
Sauty, J. P., and others. 1982a. Sensible energy storage in aqui- Schwartz, F. W., and K. Muehlenbachs. 1979. Isotope and ion
fers. 1. Theoretical study. Water Resources Res., v. 18, p. geochemistry of groundwaters in the Milk River aquifer. Al-
245-252. berta. Water Resources Res., v. 15, no. 2, p. 259-268.
Sauty, J. P., and others. 1982b. Sensible energy storage in aqui- Schwarzenbach, R. P., and W. Giger, 1985. Behavior and kite
fers. 11. Field experiments and comparison with theoretical of halogenated hydrocarbons in ground water. Ground Water
results. Water Resources Res., v. 18, p. 253-265. Quality, C.H. Ward, W. Giger, and P. L. McCarty (eds.): New
Savin, S. M. 1980. Oxygen and hydrogen isotope effects in low- York, Wiley-Interscience, p. 446-471.
temperature mineral-water interactions. Chapter 8. In P. Fritz Schwarzenbach, R. P., P. M. Gschwend, and D. M. Iniboden.
and J. C. Fontes (eds.), Handbook of Environmental Isotope 1993. Environmental Organic Chemistry. John Wiley & Sons.
Geochemistry, v. 1. Elsevier, New York, p. 283-327. New York, 681 p.
Scalf, M. R., J. F. McNabb, W. J. Dunlap, R. L. Cosby, and Schwarzenbach, R. P., and J. Westall. 1981. Transport ofnonptr
J. Fryberger. 1981. Manual of Ground Water Sampling Proce- lar organic compounds from surface water to groiindwater.
dures. US. Environ. Protection Agency, Reports EPA 660/ Laboratory studies. Environ. Sci. Technol., v. 15. p. 1300-
2-81-160, 93 p. 1367.
Schiff, L. 1955. The status of water spreading for groundwater Schwille, F. W. 1985. Migration of organic fluids imniiscible
replenishment. Trans. Am. Geophys. Union, v. 36, p. 1009- with water in the unsaturated and saturated zones. Second
1020. Canadian/American Conference on Hydrogeology. B. Hit-
Schleicher, D. 1975. A model for earthquakes near Palisades chon and M.Trudell (eds.): National Water Well Assoc., Dub-
Reservoir, Southeast Idaho. U.S. Geol. Survey J. Res., v. 3, lin, Ohio, p. 31-35.
Schwille, F., 1988, Dense chlorinated solvents in porous and
p. 393-400.
Schmidt, V., and D. A. McDonald. 1977. The role of secondary fractural media: Lewis Publishers, Chelsea, Mirh., 146 p.
Scoffin, T. 1987. An Introduction to Carbonate Sediments and
porosity in the course of sandstone diagenesis. In P. A. Scholle
Rocks. Chapman and Hall, New York.
and P. R. Schluger (ed.), Aspects of Diagenesis. SOC. Econ.
Scott, R. F. 1963. Principles of Soil Mechanics. Addison-Wesley,
Paleontol. and Mineral. Spec. Pub. 26, p. 175-225.
Reading, Mass.
Schmorak, S. 1967. Saltwater encroachment in the Coastal Plain
Seaburn, G. E. 1970. Preliminary analysis of rate of movement of
of Israel. Inter. Assoc. Sci. Hydrol. Symp. Hai6. Pub. 72,
storm runoff through the zone of aeration beneath a recharge
p. 305-318.
basin on Long Island. New York. U.S. Geol. S u n . Prof. Paper
Schmorak, S., and A. Mercado. 1969. Upconing of freshwater-
70043, p. 196- 198.
seawater interface below pumping wells. Water Resources
Segol, G., and G. F. Pinder. 1976. Transient simulation of saltwit-
Res., v. 5 , p. 1290-1311. ter intrusion in southeastern Florida. Water Resources Res..
Schneider, R. 1'964. Relation of temperature distribution to V. 12, p. 65-70.
groundwater movement in carbonate rocks of Central Israel. Sellers, K. L., and R. P. Schreiber. 1992. Air sparging model for
Geol. SOC.Am. Bull. 75, p. 209-216. predicting groundwater cleanup rate: in Proceedings Petro-
Scholz, C. H., L. R. Sykes, and Y. P. Aggerwal. 1973. Earthquake leum Hydrocarbons and Organic Chemicals in Ground Water:
prediction. A physical basis. Science, v. 181, p. 803-809. Prevention, Detection and Restoration, National Ground W I -
Schuler, R. W., V. P. Stallings, J. M. Smith. 1952. Heat and mass ter Association, Columbus, Ohio, p. 365-376.
transfer in fixed bed reactors. Chem. Eng. Prog. Symp. Ser. Semprini, L., P. K. Kitanidis, D. H. Kanipbell. and J. T. Wilson.
48, p. 19-30. 1995. Anaerobic transformation of chlorinated aliphatic hy-
Schulz, D.. M. Barvenik, and J. Ayres. 1984. Design of soil- drocarbons in a sand aquifer based on spatial chemical distri-
bentonite backfill mix for the first Environmental Protection butions. Water Resources Res., v. 31. no. 4. p . 1051 - 1062.
Agency Superfund Cutoff wall. Proc. of the Fourth Natioiial Semprini, L., P. V. Roberts, G. D. Hopkins. and P. L. ,Mc<:arty.
Symposium and exposition on Aquifer Restoration Sr Ground- 1990. A field evaluation of in-situ biodegradation o f chlori-
water Monitoring. National Water Well Assoc., Dublin, Ohio, nated ethenes: Part 2, Results of biostimulation and biotfiins-
p. 8- 17. formation experiments: Ground Water. v. 28. no. 5 . p.
Schwam, F. W. 1975. On radioactive waste management: An 715-727.
analysis of the parameters controlling subsurface contami- Senger, R. K., and G. E. Fogg. 1987. Regional underpressuring
nant transfer. J. Hydrol., v. 27, p. 51 -71. in deep brine aquifers, Palo Duro Basin. Texas: 1 . Effects of
Schwartz, F. W. 1988. Contaminant hydrigeology-dollars and hydrostratigraphy and topography. Water Resources Res.. v.
sense: J. Hydrol., v. 100, p. 453-470. 23, p. 1481- 1493.
Schwartz, F. W.. and A. S. Crowe. 1980. A Deterministic- Senger, R. K.,G. E. Fog, and C. Kreitler. I%-. Effect of hydro-
Probabilistic Model for Contaminant Transport. NUREWCR- stratigraphy and basin development on hydrodynamics of
1609, Nuclear Regulatory Comm., Washington, D.C., 158 p. the Palo Duro Basin, Texas. Rept. of Invest. 165. Bureau
Schwartz, F. W., and D. Gallup. 1978. Some factors controlling Econ. Geol., Austin, Tex.
the major ion chemistry of small lakes: Examples from the Senger, R. K., C. Kreitler, and G. Fog. 198'. Regional under-
Prairie Parkland of Canada. Hydrobiologia, v. 58, no. 1, p. pressuring in deep brine aquifers, Palo Duro Basin. Texas:
65-81. 2. The effect of Cenozoic development. Water Resources
Schwartz, F. W., G. L. McClymont, and L. Smith, 1985. On the Res., v. 23, p. 1494- 1504.
role or mass transport modeling. Canadian/American Confer- Senn, R. B., and M. S. Johnson. 1985. Interpretation of &IS
-
490 References

chromatography data as a tool in subsurface hydrocarbon Smith, R. L., R. W. Harvey, J. H. Duff, and D. R. LeBlanc. 1987.
investigations. In Petroleum Hydrocarbons and Organic Importance of close interval vertical sampling in delineating
Chemicals in Ground Water-Prevention, Detection and R e s chemical and microbiological gradients in ground-water stud-
toration. National Ground Water Assoc., Columbus, Ohio, ies. U.S. Geol. Surv. Open-File Rept. 87-109, p. B33-B35.
p. 331-357. Smolenski, W. J., and J. M. Sulflita. 1987. Biodegradation of
Serafin, J. L., and A. del Campo. 1965. Interstitial pressure on cresol isomers in anoxic aquifers. Appl. Environ. Microbiol.,
rock foundations of dams. J. Soil Mech. Found. Div., Am. V. 53, p. 710-716.
Assoc. Civil Engrs., v. 91, p. 65-85. Smolley, M., and J. C. Kappmeyer. 1991. Cone penetrometer
Shante, V. K. S., and S. Kirkpatrick. 1971. An introduction to tests and Hydropunch@sampling: A screening technique for
percolation theory. Adv. Phys., v. 20, p. 352-357. plume definition. Ground Water Mon. Rev., v. 11, no. 2,
Sharp, J. M. 1978. Energy and momentum transport model of p. 101-106.
the Ouachita Basin and its possible impact on formation of Smoluchowski, M. S. 1909. Some remarks on the mechanics of
economic mineraldeposits. Ron. Geol., v. 73, p. 1057- 1068. overthrusting. Geol. Magazine, v. 6, p. 1133- 1138.
Sharp, J. M. 1983. Permeability control on aquathermal pres Smyth, J. R. 1982. Zeolite stability constraints on radioactive
sure. Am. Assoc. Petrol. Geol. Bull. 67, p. 2057-2061. waste isolation in zeolite bearing rocks. J. Geol., v. 90,p.
Sharp, J. M., and P. A. Domenico. 1976. Energy transport in 195- 20 1 .
thick sequences of compacting sediment. Geol. SOC. Am. Snow, D. T. 1968. Rock fracture spacings, openings, and porosi-
Bull. 87, p. 390-400. ties. J. Soil Mech., Found. Div., Proc. Am. SOC. Civil Engrs.,
Sharp, J. M., and J. R. Kyle. 1988. The role of groundwater V. 94, p. 73-91.
processes in the formation of ore deposits. In W. Back, J. Snow, D. T. 1972. Geodynamics of seismic reservoirs. Proc.
Rosenshein, and P. Seaber (eds.), The Geology of North Symp. on Percolation through Fissured Rock, Stuttgart, T2-J.
America, Hydrogeology. Geol. SOC.Am., v. 0-2, p. 461 -483. Solomon, D. K., R. J. Poreda, P. G. Cook, and A. Hunt. 1995.
Sharp, J. C., and Y. N. T. Maini. 1972. In W. Wittke (cd.), Site characterization using 'H/'He ground-water ages, Cape
Fundamental considerations on the hydraulic characteristics Cod, Mass. Ground Water, v. 33, no. 6, p. 988-996.
of joints in rock in percolation through fractured rock. Intern. Sorey, M. L. 1971. Measurement of vertical groundwater veloc-
Soc. Rock Mech., Stuttgart. ity from temperature profiles in wells. Water Resources Res.,
Shi, Y.,and C. Y. Wang. 19%. Pore pressure generation in V. 7, p. 963-970.
sedimentary basins: Overloading versus aquathermal. J. Geo- Sorey, M. L. 1978. Numerical modeling of liquid geothermal
phys. Res.,v. 91, p. 2153-2162. systems. U.S. Geol. Sum. Prof. Papers 1044D.
Signor, D. C. 1985. Groundwater sampling during artificial re- Sposito, G . 1984. The Surface Chemistry of Soils. Oxford Univer-
charge: equipment, techniques, and data analyses. InT. Asano
sity Press, New York, 234 p.
(ed.), Artificial Recharge of Groundwater. Butterworth,
Springer, A. E. 1990. An evaluation of wellfield-protection area
Stoneham, Mass., p. 151-202.
delineation methods as applied to municipal wells in the
Signor, D. C., and others. 1970. Annotated bibliography on
stratifieddrift aquifer at Wooster, Ohio. M.S. thesis, The Ohio
artificial recharge of groundwater, 1955- 1967. US. Geol.
State University, 167 p.
Survey Water Supply Paper 1 9 9 0 .
Springer, A. E., and E. S. Bair. 1992. Comparison of methods
Simon, R. B. 1969. Seismicity of Colorado. Science, v. 165,
used to delineate capture zones of wells: 2. Stratifieddrift
p. 897-899.
Singer, E., and R. H. Wdhelm, 1950. Heat transfer in packed buried-valley aquifer. Ground Water, v. 30, no. 6, p. 908-917.
beds: Analytical solution and design methods: Chem. Eng. Stallman, R. W. 1952. Nonequilibrium type curves for two well
Prog., v. 46, p. 343-357. systems. U.S. Geol. Sum. Groundwater Notes, Open File
Slaine, D. D., P. E. Pehme, J. A. Hunter, S. E. Pullan, and J. P. Rept. 3.
Greenhouse. 1990. Mapping overburden stratigraphy at a Starr, R. C., and J. A. Cherry. 1994. In situ remediation of
proposed hazardous waste facility using shallow seismic re- contaminated ground water: The funnel-and-gate system.
flection methods. In S. H. Ward (ed.), Special Volume on Ground Water, v. 32, no. 3, p. 465-476.
Environmental Physics, SOC.Explor. Geophysics, Tulsa, Okla. Steams, D. W. 1964.Macrofracture patterns on Tetron anticline,
Slattery, J. C. 1972. Momentum, Energy, and Mass Transfer in Northwest Montana (abst). Am. Geophys. Union Trans. v.
Continua. McGraw-Hill, New York. 45, p. 107-108.
Sleep, B. E., and J. F. Sykes. 1989. Modeling the transport of Steams, D. W. 1967. Certain aspects of fracture in naturally
volatile organics in variably saturated media. Water Resources deformed rock. In R. E. Riecker (ed.), National Science Foun-
Res., v. 25, p. 81-92. dation Advanced Science Seminar in Rock Mechanics. Air
Slichter, C. S. 1899. Theoretical investigation of the motion of Force Cambridge Research Lab Spec. Rept., Bedford, Mass.,
groundwater. U.S. Geol. Survey Annual Rept. 19 (part 2), p. 97- 118.
p. 295-384. Steams, D. W., and M. Friedman. 1972. Reservoirs in fractured
Smith, L., and D. S. Chapman. 1983. On the thermal effects of rock. In Fracture Controlled Production. Am. Assoc. Petrol.
groundwater flow: I. Regional scale systems.J. Geophys. Rcs., Geol Reprint Ser. 21, p. 174-198.
v. 88, p. 593-608. Steefel, C. I., and A. C. Lasaga. 1988. The space-time evolution
Smith, L., and F. W. Schwartz. 1981. Mass transport. 2. Two- of dissolution patterns: Permeability change due to coupled
dimensional simulations. Water Resources Res., v. 17, no. 2, flow and reaction. Submitted to Am. Chem. SOC.Symp. Set.
p. 351 -369. Stein, R., and F. W. Schwartz. 1990. On the origin of saline
Smith, R. E., and D. A. Woolhiser. 1971. Overland flow on an soils at Blackspring Ridge, Alberta, Canada: J. Hydrol., 117,
infiltrating surface. Water Resources Res., v. 7, p. 899-913. p. 99-131.
 References

Stewart, M. T., and M. C. Gay. 1986. Evaluation of transient transport in fractured porous media: Analytical solution for
electromagnetic soundings for deep detection of conductive a single fracture. Water Resources Res., v. 17, no. 3, p.
fluids. Ground Water, v. 24, p. 351-356. 555-564.
Stiff, H. A., Jr. 195 1. The interpretation of chemical water analy- Taylor, T. W., and M. C. Serafini. 1988. Screened auger sampling:
sis by means of patterns. J. Petrol. Technol., v. 3, no. 10, The technique and two case studies. Ground Water Mon.
p. 15-17. Rev., v. 8, no. 3, p. 145-152.
Stoiber, R. 1941. Movement of mineralizing solutions in the Tejesen, S. G., 0. Erga, G. Thorsen, and A. Ve. 1961. Phase
Picher Field, Oklahoma-Kansas. Econ. Geol., v. 46, p. boundary processes as rate determining steps in reactions
800-812. between solids and liquids. The inhibitory action of metal
Stollenwerk, K. G., 1991. Simulation of molybdate sorption ions on the formation of calcium bicarbonate by the reaction
with diffuse layer surfacecomplexation model. U.S. Geol. of calcite with aqueous carbon dioxide. Chem. Eng. Sci., v.
Surv. Water Resources Investigations Report, 914034, p. 14, p. 277-289.
47-52. Terzaghi, K. 1950. Mechanism of landslides. In Berkeley, Vol.,
Stoodley. P., D. d e Beer, and Z. Lewandowski. 1994. Liquid Application of geology to engineering practice. Geol. Soc.
flow in biofilm systems. Appl. Environ. Microbiol.. Aug., Am. P ~ b l . p.
, 83-123.
p. 2711-2716. Terzaghi, K.,and R. B. Peck. 1948. Soil Mechanics in Engi-
Streltsova, T. D. 1972. Unsteady radial flow in an unconfined neering Practice. John Wiley & Sons, New York.
aquifer. Water Resources Res., v. 8, p. 1059- 1066. Theis, C. V. 1935. The relation between the lowering of the
Streltsova, T. D., and K. Rushton. 1973. Water table drawdown piezometric surface and rate and duration of discharge of a
due to a pumped well. Water Resources Res., v. 9, p. well using groundwater storage. Trans. Am. Geophys. Union,
236- 242. V. 2, p. 519-524.
Stringheld, V. T., and H. E. LeGrande. 1966. Hydrology of lime- Theis, C. V. 1940. The source of water derived from wells-
stone terranes. Geol. SOC.Am. Spec. Paper 3. essential factors controlling the response of an aquifer t o
Strom, E. W. 1993. A model of the large hydraulic gradient development. Civil Eng., Am. SOC.Civil Eng.. p. 277-280.
at Yucca Mountain, Nevada Test Site, based o n hydraulic
Theis, C. V., R. H. Brown, and R. R. Meyers. 1963. Estimating
conductivity contrasts between Cenozoic and Paleozoic
the transmissibility of aquifers from the specific capacity of
rocks. M.S. thesis, Texas A&M Univ., Dept. Geol. and Geop.,
wells. In Methods of Determining Permeability. Tnnsmissibil-
College Station, Tex.
ity, and Drawdown. U.S. Geol. Survey Water Supply Papers
Stumm, W., and J. J. Morgan. 1981. Aquatic Chemistry, 2nd ed.
15361.
John Wiky & Sons, New York, 780 p.
Thiem, G. 1906. Hydrologische Methode. Gebhardt. Leipzig.
Sudicky, E. A. 1986. A natural gradient experiment on solute
Thomas, H. C., 1944. Heterogeneous ion exchange in a tlowing
transport in a sand aquifer: Spatial variability of hydraulic
system: J. Amer. Chem. SOC., v. 66, p. 1664-1666.
conductivity and its role in the dispersion process. Water
Thomas, H. E. 1951. The Conservation of <;roundwater.
Resources Res., v. 22, no. 13, p. 2069-2082.
McGraw-Hill, New York.
Sudicky, E. A. 1989. The laplace transform Galerkin technique:
A timecontinuous finite element theory and application to
Thompson, C., G. McMechan, R. Szerbiak, and N. Gaynor. 1‘995.
mass transport in groundwater. Water Resources Res., v. 25, 3-D GPR imaging of complex stratigraphy within the Ferron
no. 8, p. 1833-1846. Sandstone, Castle Valley, Utah. Proc. of the Symposium on the
Sullivan, C. R., R. E. Zinner, and J. P. Hughes. 1988. The occur- Application of Geophysics to Engineering and Environmental
rence of hydrocarbon on an unconfined aquifer and implica- Problems, Environmental and Engineering. Geophys. Soc.,
tions for the liquid recovery. Proc. of the MyIwA/API Confer- Englewood, Colo., p. 435-443.
ence Petroleum Hydrocarbons and Organic Chemicals in Thompson, D. G., 0. E. Meinzer, and V. T. Stringtield. 1038.
Ground Water-Prevention, Detection and Restoration. Na- Discussion on “Effect of a sea level canal on the groundwater
tional Water Well Assoc., Dublin, Ohio, p. 135-156. level of Florida.” Econ. Geol., v. 33, p. 87- 107.
Surdam, R. C., S. W. Boese, and L. J. Crossey. 1984. The chemis- Thompson, G. M. 1976. Trichloromethane: A new hydrologic
try of secondary porosity. In D. A. McDonald and R. C. Surdam tool for tracing and dating groundwater. Ph.D. dissenation,
(eds.), Clastic Diagenesis, AAPG Memoir 37, p. 127- 149. Dept. of Geology, Indiana University, Bloomington, 93 p.
Swenson, F. A. 1968. New theory of recharge t o the artesian Thompson, G. M., and J. M. Hayes. 1979. Trichlorofliiorometh--
basin of the Dakotas. Geol. Soc. Am. Bull. 79, p. 163-1832, ane in groundwater: A possible tracer and indicator of ground-
Szymanski, J. 1989. Conceptual considerations of the Yucca water age. Water Resources Res., v. 15, p. 546-554.
Mountain groundwater system with special emphasis on the Thornthwaite, C. W. 1948. A n approach toward a rational classi-
adequacy of this system to accommodate a high level nuclear fication of climate. Geograph. Rev., v. 38, p. 55-94.
waste repository. Dept. of Energy Rept. (unpublished). Ne- Thurman, E. M. 1985. Organic Geochemistry of Natural Waters.
vada operations office, Las Vegas. Martinus Nijhoff/Dr. W. Junk Publishers. Dordrecht. The
Tamers, M. A. 1967. Surface-water infiltration and groundwater Netherlands, 497 p.
movement in arid zones of Venezuela. Isotopes in Hydrology. Todd, D. K. 1959a. Annotated bibliography on artilicial recharge
Vienna, Intl. Atomic Energy Agency, p. 339-351. of groundwater through 1954: U.S. Geol. Sunr. Water Supply
-, 1975, Variability of radiocarbon dates on groundwater. Papers, 1477.
Geophys. Sum., V. 2, p. 217-239. Todd, D. K. 1959b. Groundwater Hydrology. 1st ed. John Wi-
Tang, D. H., E. 0. Frind, and E. A. Sudicky. 1981. Contaminant ley & Sons, New York.
-
492 References

Todd, D. K. 1980. Groundwater Hydrology. 2nd ed. John Wi- U.S. EPA. 1992. Estimating potential for occurrence of DNAPL
ley SCSons, New York, 535 p. at superfund sites. Publ. 9355.4-07FS, 9 p.
Tokyo Inst. Civ. Eng. 1975. Subsidence as of 1974. Tokyo Met- U.S. EPA. 1993. Subsurface characterization and monitoring
rop. Govt. Annual Rept. (in Japanese). techniques: A desk reference guide. EPA/625/RR-93/003a,
Tortensson, B. A., and A. M. Petsonk. 1988. A hermetically Vol. I.
isolated sampling method for ground-water investigations. In U.S. Department of Energy. 1990. Radiation Protection for the
A. G. Collins and A. I. Johnson (eds.), Ground-Water Contm+ Public and the Environment. U.S. Department of Energy Or-
nation: Field Methods. ASTM STP 963, Am. SOC.for Testing der 5400, v. 5, February 8.
and Materials, Philadelphia, Pa., p. 274-289. U.S. Department of Energy. 1995. In Situ Bioremediation Using
Toth, J. 1962.A theory of groundwater motion in small drainage Horizontal Wells. Innovative Technology Summary Report,
basins in Central Alberta. J. Geophys. Res., v. 67, p. 4375- U.S. DOE, DOE/EM-0270.
4387. United States Dept. of the Navy, Bureau of Yards and Docks.
Toth, J. 1963. A theoretical analysis of groundwater flow in 1961. Design manual, Soil mechanics, Foundations, and Earth
small drainage basins. J. Geophys. Res., v. 68,p. 4795-4812. Structures. DM-7 Chap. 4.
Toth, J. 1966. Mapping and interpretation of field phenomena United States Geological Survey Annual Report 19. 1899. U S .
for groundwater reconnaissance in a prairie environment, Geol. Surv. Ann. Rept.
Alberta, Canada. Int. Assoc. Sci. Hydrol., 1 1 Annee, no. 2. United States Geological Survey. 1970. The national atlas of the
p. 1-49. United States of America.
Tbth, J. 1980. Cross formational gravity flow of groundwater-a Vaccaro, J. 1986. Columbia plateau basalt regional aquifer sys-
mechanism of the transport and accumulation of petroleum. tem study. From Ren Jen Sun (ed.), Regional aquifer system
In W. H. Roberts and R. J. Cordell (eds.), Problems of Petro- analysis program of the US. Geological Survey summary of
leum Migration. Am. Assoc. Petrol. Geol. Studies in Geol., projects, 1978-84, U.S. Geol. Sur. Circ. 1002.
no. 10, Tulsa, Okla., p. 121-167. Valocchi, A. J. 1985. Validity of the local equilibrium assumption
Toth, J. 1988. Ground water and hydrocarbon migration. In W. for modeling sorbing solute transport through homogeneous
Back, J. S. Rosenshein, and P. R. Seaber (eds.), The Geology soils. Water Resources Res., v. 21, p. 808-820.
of North America, v. 0-2. Geol. SOC. Am., Boulder, Colo., Van Golf-Racht, T. D. 1982. Fundamentals of Fractured Reser-
p. 485-501. voir Engineering. Elsevier, New York.
Toth J., and T. Corbett. 1986. Post-Paleocene evolution of re- Van Genuchten, M. Th. 1981. Non-equilibrium transport param-
gional groundwater flow systems and their relation to petro- eters from miscible displacement experiments. US. Dept.
leum accumulations, Taber area, southern Alberta. Bull. Cana- Agric. U.S. Salinity Lab., Res. Rept. No. 119, 88 p.
dian Petrol. Geol. v. 34, no. 3, p. 339-363. Van der Kamp, G.1972. Tidal fluctwations in a confined aquifer
Truesdell, A. H., and J. R. Hulston. 1980. Isotopic evidence on extending under the sea. Environ. Canada reprint 242, p.
environments of geothermal systems. Chapter 5. In P. Fritz 101-106, Ottawa.
and J. C. Fontes (eds.), Handbook of Environmental Isotope Van der Kamp, G., and S. Bachu, 1989. Use of dimensional
Geochemistry, v. 1. Elsevier, Amsterdam, p. 179-226. analysis in the study of thermal effects of various hydrogeolog-
Tsang, C. F., T. Buscheck, and C. Doughty. 1981. Aquifer ther- ical regimes, in Hydrogeologic regimes and their subsurface
mal storage. A numerical simulation of Auburn University thermal effects, Beck and others (cds.): Geophysical Mono-
Field experiments. Water Resources Res., v. 17, p. 647-6’58. graph 47, IUGG, v. 2.
ULARA Watermaster. 1994. Watermaster service in the upper Van der Kamp, G., and J. E. Gale. 1983. Theory of earth tide and
Los Angeles River Area, Los Angeles County. ULARA Wa- barometric effects in porous formations with compressible
termaster Report, May 1994. grains. Water Resources Res., v. 19, p. 538-544.
United Nations. 1976. Proc. 2nd U.N. Symposium on Develop- Van Everdingen, R. 0. 1972. Observed chanp. in groundwater
ment and Use of Geothermal Resources. 3 Vols. San Fran- regime caused by the creation of Lake Dietenbaker, Saskatch-
cisco: United Nations. ewan. Can. Dept. of the Environ., Inland Waters Branch Tech.
U.S. Environmental Protection Agency (EPA). 1985. Remedial Bull. 59.
Action at Waste Disposal Sites. EPA/625/@5/006 (NTIS Van Orstrand, C. E. 1934. Temperature gradients. In Problems
PB87-201034). of Petrol. Geol. Am. Assoc. Petrol. Geol., Tulsa, Okla., p. 989-
U.S. EPA. 1989a. Risk Assessment Guidance for Superfund. Vol. 1021.
I. Human Health Evaluation Manual (Part A), Interim Final. Verma, R. D., and W. Brutsaert. 1970. Ur.confined aquifer s e e p
U.S. Environmental Protection Agency, Office of Emergency age by capillaryflow theory. J. Hydraul. Div., Proc. Am. Assoc.
and Remedial Response, EPA/540/189/002, Washington, Civil Engrs., v. 96, p. 1331-1344.
D.C. Verma, R. D.,and W. Brutsaert. 1971. Unsteady free surface
U.S. EPA. 1989b. Terra Vac In Situ System, Applications Analysis groundwater seepage. J. Hydraul. Div., Proc. Am. Assoc. Civil
Report. Environmental Protection Agency, EPA/540/A5 -89/ Engrs., v. 97, p. 1213- 1229.
003, 52 p. Vernon, R. 0. 1969. ?’he geology and hydrology associated with
U.S. EPA. 1 9 8 9 ~Technology
. Evaluation Report: Site Program a zone of high permeability (Boulder Zone) in Florida. SOC.
Demonstration Test Terra Vac In Situ Vacuum Extraction Min. Engr. Preprint 69-AG12, 24 p.
System Groveland, Mass. Environmental Protection Agency, Verschueren, K. 1983. Handbook of Environmental Data on
EPA/540/5-89/003a, 97 p. Organic Chemicals, 2nd ed. Van Nostrand Reinhold, New
U.S. EPA. 1991. Site characterization for subsurface remediation York, 1310 p.
seminar publication, EPA/625/491/026, 259 p. Vogel,J. C. 1967. Investigation of groundwater flow with radio-
 References 433

carbon. Isotopes in Hydrology. Intl. Atomic Energy Agency, Werner, P. W., and D. Noren. 1951. Progressive WJVCS iri i i o i i -
Vienna, Austria, p. 355-368. artesian aquifers. Trans. Am. Geophys. Union. v. 32. p.
Vogel, J. C. 1970. Carbon-I4 dating of groundwater. Isotope 238- 244.
Hydrology. Intl. Atomic Energy Agency, Vienna, Austria. Wershaw, R. L., M. J. Fishman. R. R. Cabbe. and L. E. Lowe.
p. 235-237. 1983. Methods for the determination of organic subst:inces
Vogel, T. M., and Y.L. McCarty. 1987. Abiotic and biotic transfor- in water and fluvial sediments. U.S. Geol. Sun. TWRI. H o o k
mations of 1,1, I-trichloroethane tinder methanogenic condi- 5, Chapter A3, 173 p.
tions, Environ. Sci. Technol., v. 21, p. 1208-1213. Wesson, R. L. 1981. Interpretation of changes in wirer le~rl
Vorhis, R. R. 1955. Interpretation of hydrologic data resulting accompanying fault creep and implications for earthqu;ike
from earthquakes. Geologischer Rundschar, v. 43, p. 47-52. prediction. J. Geophys. Res., v. 86, p. 9259-9267.
Walder, J., and A. Nur. 1984. Porosity reduction and crustal Wexler, E. J. 1992. Analytical solutions for one-. two-. and three-
pore pressure development. J. Geophys. Res., v. 89, p. dimensional solute transport in ground-water systems with
1 1539- 11548. uniform flow. Techniques of Water-Resources Investig;itioiis
Walker, R. G., and D. J. Cant. 1984. Sandy fluvial systems. In of the U.S. Geol. Surv., Book 3, Chapter B?, 190 p.
R. Walker (ed.), Facies Models. GeoSciences Canada Reprint Weyl, P. K. 1959. Pressure solution and the force o f crytaliz;~-
Series I, p. 71-89. tion. A phenomenological theory. j. Geophys. Res.. v. 64.
Wallace, R. E. 1974. Goals, strategy, and tasks of the earthquake p. 2001-2025.
hazard reduction program. U.S. Geol. Survey Circular 701. Wheatcraft, S. W., and S. W. Tyler. 1988. An explanation o f
Wallick, E. I., H. It. Krouse, and A. Shakur. 1984. Environmental scale-dependent dispersivity in heterogeneous aquifers using
isotopes: Principles and applications in ground water geo- concepts of fractal geometry. Water Resources Res.. v. l i .
chemical studies in Alberta, Canada. In B. Hitchon arid E. J. p. 566-578.
Wallick (eds.), First Canadian/American Conference on Hy- Whitaker, S. 1967. Diffusion aiid dispersion in porous medi;i.
drogeology. National Water Well Assoc., Dublin, Ohio, p. Am. Inst. ChenI. Eng. J., V. 13, p. 420-427.
249-266. White, D. E., and D. L. Williams. 1975. Summary and Conclti-
Wallis, P. M., H. B. N. Hynes, and S. A. Telang. 1981. The sion. In Assessment of Geothermal Resources of the United
importance of groundwater in the transportation of allo- States-1975. U.S. Geol. Sun. Circ. 726, p. 147-155.
chthonous dissolved organic matter to the streams draining White, D. E. 1968. Environments of generation of sonic hasic-
a small mountain basin. Hydrobiohgia, v. 79, p. 77-90. metal ore deposits. Econ. Geol., v. 63, p. 301-355.
Walsh, J. B. 1978. The effect of fractures on compressibility, White, W. B.. and E. L. White. 1987. Ordered :incl stc~hiistic
resistivity, and permeability. EOS, no. 59, p. 1193. arrangements within regional sinkhole populations. In H. F.
Walsh, J. R. 1981. Effect of the pore pressure and confining Beck and W. L. Wilson (eds.). Karst Hydrogeology: Engi-
pressure on fracture permeability. Int. J . Rock Mech. Min. neering and Environmental Applications. A.A. R:tlkeni;i Pub-
Soc. and Geomech., v. 18, p. 429-435. lishers, Accord, Mass., p. 85-90.
Walther, E. G., A. M. Pitchford. and G. R. Olhoeft. 1986. A Whittinan, S. G. 1985. Use of soil gas sampling techniclues
strategy for detecting SUbSiirfaceorganic contaminants. Proc. for assessment of ground water contamination. Proc. of the
of the Petroleum Hydrocarbons aiid Organic Chemicals in NWWA/API Conference on Petroleum Hydroc;irhons ;ind
Ground Water: Prevention, Detection and Restoration: Dub- Organic Chemicals in Grorind Water-Prevention, Detection
lin, Ohio, National Water Well Assoc., p. 357-381. and Restoration. National Water Well Assoc.. Dublin. Ohio.
Walton, W. C. 1962. Selected analytical methods for well and p. 291-309.
aquifer evaluation. Illinois State Water Survey, Bull. 49, 81 p. Wierenga, P. J., L. W. Gelhar. C. S. Simmons. G . W. Gee, imrl
Walton, W. C. 1970. Groundwater Resource Evaluation. T. J. Nicholson. 1986. Validation of Stochastic Flow atid Tmis-
McGraw-Hill, New York. port Models for Unsaturated Soil: A Coniprehensi\re Field
Wang, J. S. Y., and T. N. Narasimhan. 1985. Hydrologic mecha- Study. U.S. Nuclear Regulatory Commission. NI JRE(i/(X-
nisms governing fluid flow in a partially saturated fractured 4622.
porous medium. Water Resources Res.. v. 21, p. 1861- 1874. Wigley, T. M. L., L. N. Plummet-. and F. J. Pe;trson. .lr. 19'8.
Wang, J. S. Y., and others. 1978. Transient flow in tight fractures. Mass transfer and carbon isotope evolution in n;itucil wiiter
Proc. Invitational Well Testing Symp., Univ. of California, systems. Geochim. Cosmochim. Acta. v. 12. p. 1 1 17- 1 1 59.
Lawrence Berkeley Lab., Berkeley, Calif., p. 103- 116. Willis, C., and J. Rubin. 1987. Transport o f reacting solutes
Warren, J. E., and H. S. Price. 1961. Flow in heterogeneous subject t o a moving dissolution boundav: Numerical mctli-
porous media. Soc. Petrol. Eng. J., v. 1, p. 153- 169. ods and solutions. Water Resources Res.. v. 2.5. no. 8. 1'.
Watts, E. V. 1948. Some aspects of high pressure in the D7 I561 - 1574.
zone of the Ventura Avenue Field. Am. Inst. Mining Engrs. Wilson, E. B.. Jr. 1952. An Introduction to Scieiltilic Rest.:irch.
Petrol. Div., v. 174, p. 191-200. McGmw-Hill. New York. 373 p.
Way, S. C., and C. R. McKee. 1982. In-situ determination o f Wilson, G., and H. Gnce. 1942. The settlement o f Lontlon
three dimensional aquifer permeabilities. Ground Water, due to underdrainage of the London clay. J . Inst. Civil Engr.
v. 20, p. 594-603. (Londoli), V. 19. p. 100- 127.
Weeks, E. P. 1969. Determining the ratio of horizontal to vertical Wilson, J., and M.J. Noonan. 1984. Microbi;il activity i n iiiotlel
permeability by aquifer test analysis. Water Resources Res., aquifer systems. In G. Bitton and C. P. Gerha (eds.). (;round-
V. 5, 11. 196-214. water Pollution Microbiology. John Wile! 9; Zoiis. New York.
Wenzel, L. K. 1942. Methods for determining permeability of p. 117-133.
water bearing niaterials with special reference to discharging Wilson, J. L.. and S. H. Conrad. 1984. Is ph!-sic;il displ;iceni~-iit
well methods. U S . Geol. Surv. Water Supply P;ipers 887. of residual hydrocarbons ;I realistic possibility i n aquifcr rcsto-
-
494 References

ration? Proc. of the NWWA/API Conference on Petroltxim Woodward and Clyde, Consultants. 1979. Study o f reservoir
Hydrocarbons and Organic Chemicals in Ground Water- induced seismicity, Reston, Virginia. U.S. Geol. Surv. techni-
Prevention, Detection and Restoration. National Water Well cal report contract 14-08-OOOI-16809.
Assoc., Dublin, Ohio, p. 274-298. Worthington, S., and I). Ford. 1995. Borehole tests for mega-
Wilson, J. L., and C. Jordan. 1983. Middle shelf. In P. A. Scholle, scale channeling in carbonate aquifers. Proc. XXVI Congress
D. G. Bebout, and C. H. Moore (eds.), Carbonate Depositional of the International Assoc. of Hydrogeologists, Edmonton, Al-
Environments. Am. Assoc. Petrol. Geol. Mem. 33, p. 297-344. berta.
Wilson, J. L., and P. J. Miller. 1978. Two dimensional plume in Xu, Y., and F. W. Schwartz. 1994. Lead immobilization b y hy-
uniform groundwater flow.J. Hydraul. Div., Proc. Am. Assoc. droxyapatite in aqueous solutions. J. Contam. Hydrol., v. 15,
Civil Eng., v. 104, p. 503-514. no. 3, p. 187-206.
Wilson, J. T., and B. H. Wilson. 1985. Biotransformation of Yagi, S., and D. Kunii. 1957. Studies o n effective thermal con-
trichloroethylene in soil. Appl. Environ. Microbiol., v. 49, ductivities in packed beds. Am. Inst. Chem. Eng. J., v. 3,
p. 242-243. p. 373-381.
Wilson, J. T., M. J. Noonan, and J. F. McNabb. 1985. Riodegrrida- Yaniga, P. M., and J. Mulry, 1984. Accelerated aquifer restora-
tion of contaminants In the subsurface: Ground Water Qual- tion: In-situ applied techniques for enhanced free product
ity, C. H. Ward, W. Giger, and P. L. McCarty (eds.), John recovery/absorbed hydrocarbon reduction via bioreclama-
Wiley & Sons, New York, p. 483-492. tion. NWWA/API Conference o n Petroleum Hydrocarbons
Winograd, I. J. 1962. lnterbasin movement of groundwater at and Organic Chemicals in Ground Water-Prevention, De-
the Nevada Test Site, Nevada. U.S. Geol. Surv. Prof. Papers tection and Restoration, Proc.: Dublin, Ohio, National Water
4 5 K , p. 108- 1 1 1. Well Assoc., p. 421 -440.
Winograd, I. J., and W. Thordarson. 1975. Hydrogeologic and Yoder, H. S. 1955. Role of water in metamorphism. Geol. Soc.
hydrochemical framework, South Central Great Basin, mith Am. Spec. Paper 62, p. 505-524.
special reference t o the Nevada Test Site. U.S. Geol. Surv. Yong, R. N. 1985. Interaction of clay and industrial waste: A
Prof. Papers 712-C. summary review, In 8. Hitchon and M. Tnidell (eds.), Second
Winslow, A. G., and L. A. Wood. 1959. Relation of land subsi- Canadian/American Conference on Hydrogeology. National
dence to groundwater withdrawals in the upper Gulf Coastal Water Well Assoc., Dublin, Ohio, p. 13-25.
region of Texas. Am. Inst. Min. Engrs. Min. Div., Min. Eng., Young, R., and J. Bredehoeft. 1972. Digital computer simulation
p. 1030- 1034. for solving mamgement problems of conjuctive groundwater
Winter, T. C. 1976. Numerical simulation analysis of the intenc- and surface water systems. Water Resources Res., v. 8, p.
tion of lakes and groundwaters. U.S. Geol. Surv. Prof. Pa- 533-556.
pers, 1 0 0 1 . Yu, C., A. J. Zielen, 1.11.Cheng, Y. C. Yuan, L. G. Jones, D. J.
Winter, T. C. 1978. Numerical simulation of steady state three kPoire, Y. Y. Wang, C. 0. Loureiro, E. Gnanapragasam, E.
dimensional groundwater flow near lakes. Water Resources Faillance, A. Wallo 111, W. A. Williams, and H. Peterson. 1993.
Res., v. 14, p. 245-254. Manual for Implementing Residual Radioactive Material
Witherspoon, P. A., and J. E. Gale. 1977. Mechanical and hydrau- Guidelines Using RESRAD, Version 5.0. Argonne National
lic properties of rocks related to induced seismicity. Eng. Laboratory, ANL/EAD/LD-2, Argonne, Ill.
Geol., v. 11, p. 23-55. Zachara, J., 1990, Hydrogeology in relation to microorganisms:
Witherspoon, P. A., and others. 1980. Validity of cubic law lor Proceedings of the First International Symposium o n Micro-
fluid flow in a deforniable rock fracture. Water Resources biology of the deep subsurface, <:. B. Fliermans and
Res., v. 16, p. 1016-1024. T. C. Hazen (eds.), Westinghouse Savannah River Company,
Wolery, T.J. 1979. Calculation of chemical equilibrium between p. 5-3 t o 5-13.
aqueous solutions and minerals: The EQ3/EQ6 software pack- Zalasiewicz, J. A., S. J. Mathers, and J. I). Cornwell. 1985. The
age. UCRL-52658, Univ. of California, Lawrence Livermore application of ground conductivity measurements to geologi-
Laboratory, Livermore, Calif., 41 p. cal mapping. Q. J. Eng. Geol. London, v. 18, p. 139-148.
Wow, R. G. 1970. Field and laboratory determination of the Zaporozec, A. 1972. Graphical interpretation of water quality
hydraulic dmsivity of a confining bed. Water Resources Res., data: Ground Water, v. 10, p. 32-43.
V. 6, p. 194-203. Zemo, D. A., T. E. Graf, and J. E. Bniya. 1993. The importance
Wolff, R. G., and S. S. Papadopulos. 1972. Determination of the and benefit of fingerprint characterization in site investigation
hydraulic diffusivity of a heterogeneous confining bed. Water and remediation focusing o n petroleum hydrocarbons. Proc.
Resources Res., v. 8, p. 1051 - 1058. of the Petroleum Hydrocarbons and Organic Chemicals in
Wood, J. R., and T. A. Hewett. 1982. Fluid convection and mass Ground Water-Prevention, Detection and Restoration. Na-
transfer in porous sandstone-a theoretical model. Geochirn. tional Ground Water Assoc., Columbus, Ohio, p. 39-54.
Cosmochim. Acta, v. 46, p. 1707-1713. Zenger. D. H. 1072. Dolomitization and uniformitarianism. J.
Wood, W. W. 1976. Guidelines for the collection and field Geol. E ~ u c . ,V. 20, p. 107-124.
analysis of ground water samples for selected unstable con- Zheng, C., and Bennett, G. D. 1995. Applied contaminant t r a n s
stituents. U.S. Geol. Sum. TWFU, Rook 1 , Chapter D2, 24 1,. port modeling. Intern. Thomson Pub. Co., New York. 440 p.
Woodbury, A. D., and L. Smith. 1985. On the thermal effects Zimmemlan, R. W., W. H. Somerton, and M. S. King. 1986.
of three dimensional groundwater flow. J. Geophy. Res., v. Compressibility of porous rocks. J. Geophys. Res., v. 91, p.
90.p. 759-767. 12,765- 12,777.
Woodbury, A. D., and E. A. Sudicky. 1991. The geostatistical Zoback, M. I)., and J. D.Byerlee. 1976. A note on the deforina-
characteristics of the Borden aquifer. Water Resources Res., tion behavior and permeability of crushed granite. Inter. J.
v. 27, no. 4, p. 533-546. Rock Mech., Min. Sci., and Geomech., v. 13, p. 291-294.
INDEX

Index Terms Links

Abiotic chemical destruction, in situ 449

Abnormal fluid pressures 172
See also Deformable porous media;
Tectonic processes
irreversible processes 185
isothermal basin loading and tectonic
strain 175
horizontal extension and
compression 178
one-dimensional loading 176
tectonic strain 179
vertical compression 178
mathematical formulation 174
origin and distribution 172
phase transformations 183
rock fractures 182
subnormal pressure 184
thermal expansion of fluids 179
Acid-base reactions 255
alkalinity 257
CO2-water systems 256
natural weak systems 256
weak acid-strong base reactions,
natural ground-water systems
transport
saturated zone chemical reactions 310
unsaturated zone chemical reactions 307
Active depositional environments:
abnormal fluid pressures 172
See also Abnormal fluid pressures
heat transport, forced convection 203
Activity models, aqueous geochemistry 242

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Advection:
dissolved contaminant transport
modeling, analytical approaches 372
solute transport 215
See also Solute transport
Advection-diffusion equation, mass
transport equations 297
Advection dispersion equation, mass
transport equations 297
Age dating, of ground water, natural
ground-water systems transport 319
Air-permeability testing, multiphase
fluid systems 412
Air sparging, contaminant removal 437
Alcohols, aqueous geochemistry 248
Aldehyes, aqueous geochemistry 248
Alkali lake, Oregon, solute plumes,
case studies 360
Alkalinity, acid-base reactions 257
Alkane, alene, and alkyne
hydrocarbons, aqueous
geochemistry 248
Alluvial basins, porosity and
permeability 19
Alluvial fans, porosity and permeability 18
Alluvial valleys, porosity and
permeability 18
Amines, aqueous geochemistry 247
Amino acids, aqueous geochemistry 247
Ancestral seas, porosity and
permeability 22
Anisotropicity:
Darcy's law, ground-water
movement 43
heterogeneity and, ground-water
movement 39

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Aqueous geochemistry 238

See also Chemical reactions
chemical data 250
concentration scales 239
equilibrium models 241
deviations from 243
equilibrium versus kinetic
descriptions 240
gas and solid phases 240
ground-water composition 248 250
kinetic reactions 244
organic compounds 245
overview 238
water structure and mass in water 240
Aquifers:
bounded, hydraulic testing 120
classification of ground-water
movement, hydraulic conductivity 41
confined, ground-water flow
equations 63 65
defined 16
energy resources, heat transport 209
leaky, Hantush-Jacob method,
hydraulic testing 110
mapping of 15
response to pumping. ground water
resource development 136
water table, hydraulic testing,
conventional 112
Aquitatds, defined 16
Aromatic hydrocarbons, aqueous
geochemistry 248
Artificial recharge, ground water
resource development 139
Atomic weights, table of 466

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Babylon, New York, solute plumes,

case studies 357
Bacteria, microorganisms 285
See also Microorganisms
Barrier island environment. porosity
and permeability 20
Base flow, hydrologic cycle 8
Bases. See Acid-base reactions
Basin hydrologic cycle 75
See also Hydrologic cycle
engineering and geologic
implications 97
dams and reservoirs 97
excavations 98
surface features 91
ground-water–lake interactions 93
ground-water–surface-water interactions 95
recharge-discharge relations 91
topographic driving forces 75
carbonate terrain 87
coastal regions 88
early studies 75
models of 76
mountainous terrain 83
Basin migration models, hydrocarbons,
geologic work of ground water 339
Beach environment, porosity and permeability 20
Bernoulli equation, ground-water movement 34
Biochemical techniques, microorganisms 288
Biodegradation:
dissolved contaminant transport
modeling, analytical approaches 376
microbial processes 292
Biofilm kinetics, microbial processes 292
Biofilms, microorganisms 287
Biological contaminants, contaminant
hydrogeology 350 352
This page has been reformatted by Knovel to provide easier navigation.
Index Terms Links

Bioremediation, intrinsic, in situ

destruction 443
Biotransformation 293
complex transformation pathways 295
halogenated aliphatic compounds 294
halogenated aromatic compounds 294
hydrocarbons 293
of organic compounds, oxidation-
reduction reactions 276
polychlorinated biphenyls (PCBs) 295
Bioventing and bioslurping 445
applicability 446
progress in 448
requirements 446
testing 447
Borden tracer experiment, solute transport 234
Borehole tests. See Multiple borehole
tests; Single-borehole tests
Boundary conditions:
equations 68
mass transport equations 300
MODFLOW codes 153
Boundary environments, continental-
marine, porosity and permeability 20
Bounded aquifers, hydraulic testing 120
computer applications 131

Calgary, Alberta, Canada, oil spill,

contaminant removal case study 438
Canal, sea-level, basin hydrologic cycle 98
Cape Cod, colloid transport 285
Capillarity:
ground-water movement, unsaturated zone 53
hydrologic cycle 6
multiphase fluid systems 394
Capture zones, pump-and-treat systems 426 430
Carbon-14, age dating of ground water 319 320

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Carbonate rocks:
geologic work of ground water 326
See also Geologic work of ground water
secondary porosity, uplift, diagenesis,
and erosion 29
Carbonate terrain, basin hydrologic
cycle, topographic driving forces 87
Carbon dioxide-water systems, acid-
base reactions 256
Carboxylic acids, aqueous
geochemistry 247
Cation exchange:
saturated zone chemical reactions 315
unsaturated zone chemical reactions 309
Cenozoic rock group, porosity and
permeability 24
Chemical change agent, mixing as,
natural ground-water systems
transport 303
See also Natural ground-water systems
transport
Chemical data, aqueous geochemistry 250
See also Aqueous geochemistry
Chemical hydrogeology, historical
perspective 3
See also Hydrogeology
Chemical reactions 255
See also Aqueous geochemistry
acid-base reactions 255
alkalinity 257
CO2,-water systems 256
natural weak systems 256
complexation reactions 263
ion complexation and equilibrium
calculations 264
metals, enhancement of mobility 264
organic complexation 265
stability and speciation modeling 263

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Chemical reactions (Cont.)

dissolution and precipitation of
solids 262
dissolved contaminant transport
modeling 386
hydrolysis 277
isotropic processes 277
deuterium and oxygen-18 279
isotropic reactions 278
radioactive decay 277
oxidation-reduction reactions 272
kinetics and dominant couples 275
mobility of metals control 276
organic compounds, biotransformation of 276
oxidation numbers, haif-reactions,
electron activity, and redox
potential 272
solid solubility 262
solution and exsolution 257
surface reactions 266
hydrophobic sorption of organic
compounds 267
multiparametcr equilibrium
models 269
sorption isotherms 266
sorption of metals, Kd-based
approaches to modeling 269
volatilization 259
Chemical rock-water interactions,
diagenesis, porosity and
permeability 26
Chemical wastes, contaminant removal
case study 439
Chemical weathering, porosity and
permeability 16
Chlorine, age dating of ground
water, natural ground-water
systems transport 322

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Chlorofluorocarbons, age dating of

ground water, natural ground-
water systems transport 322
Clays, diagenesis, porosity and
permeability 27
Clean Air Act 4
Clean Drinking Water Act 4
Clean Water Act 4
Coastal regions, basin hydrologic cycle,
topographic driving forces 88
Colloids 282
See also Microorganisms
conceptual model 282
defined 282
ground water transport 284
Compaction:
diagenesis, porosity and permeability 24
lead-zinc deposits, geologic work of
ground water 332
Complexation reactions 263
ion complexation and equilibrium
calculations 264
metals, enhancement of mobility 264
stability and speciation modeling 263
Complex transformation pathways,
biotransformation 295
Comprehensive Environmental
Response, Compensation, and
Liability Act (CERCLA) 4
Compressibility:
of rock matrix, ground-water flow
equations 64
of water, storage properties of
porous media, ground-water flow
equations 63
Computer calculations, hydraulic
testing 128

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Computer programming, dissolved

contaminant transport modeling,
analytical approaches 382
Concentration scales, aqueous
geochemistry 239
Conceptual models, basin hydrologic
cycle, topographic driving forces 76
Conduction, heat transport 191
See also Heat transport
Conductive-convective equation 195
Cone penetrometry, contaminant
hydrogeology, sampling methods 365
Conjunctive use, ground water
resource development 141
Connectivity, tectonism and fracture
formation, porosity and
permeability 31
Conservation of fluid masses, ground-
water flow equations 59
Containment 417
grouting 418
hydrodynamic controls 419
sheet pile walls 418
slurry walls 417
stabilization and solidification 420
surface seals and surface drainage 418
Contaminant hydrogeology 344
dissolved contaminant transport
modeling 372
See also Dissolved contaminant transport
modeling
indirect detection methods 367
geophysical methods 368
soil-gas characterization 367
multiphase fluid systems 393
See also Multiphase fluid systems

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Contaminant hydrogeology (Cont.)

sampling 360
network designs 360
quality assurance 362
sampling methods 362
cone penetrometry 365
conventional wells or piezometers 362
Geoprobe 367
multilevel samplers 363
solid and fluid sampling 364
unsaturated zone 367
solute plumes 352
case studies 357
described 352
fractured and karst systems 357
sources of 344
attributes 344
biological 350 352
inorganic species 349
nutrients 348
organic 349 351
radioactive 346
summary table 347
trace metals 347
Contaminant plumes, tracers and tracer
tests, solute transport 233
Contaminant removal 420
air sparging 437
case studies 438
interceptor systems 430
overview 420
pump-and-treat systems 422
See also Pump-and-treat systems
design methods 425
problem of 422
technical considerations 423

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Contaminant removal (Cont.)

soil-vapor extraction 431
heterogeneity and efficiency 436
indications for 433
soil permeability estimates 435
system components 432
Continental collision, lead-zinc
deposits, geologic work of ground
water 335
See also Tectonic processes
Continental environments. See also
Porosity and permeability
marine boundary 20
porosity and permeability 16
Continuum approach, ground-water
movement, fractured rock 48
Convection, heat transport 191
See also Heat transport
Conventional hydraulic testing 105
See also Hydraulic testing
Crystalline rock, hydraulic conductivity 39
Cubic law, ground-water movement,
fractured rock 50
Curve-matching procedure, Theis
nonequilibrium pumping method,
hydraulic testing 107

Dakota aquifer:
porosity and permeability, marine
environments 22
uplift, diagenesis, and erosion 27
Dakota sandstone, potentiometric
surface of 47
Dams, basin hydrologic cycle 97

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Darcy's law:
anisotropic material, ground-water
movement 43
deformable porous media 159 160
ground-water movement 33
fractured rock 49
unsaturated zone 51 54
heat transport 193 196
Data collection, risk assessment 451
Deep sedimentary environments,
diffusion in, natural ground-water
systems transport 305
Deformable porous media 159
See also Abnormal fluid pressures;
Tectonic processes
land subsidence as one-dimensional
drained response 163
stimulation of 169
time rate of 167
vertical compression 165
one-dimensional consolidation 159
drained response to natural loading
events 163
Row equation development 159
undrained response to natural
loading events 160
three-dimensional consolidation 169
elastic properties 169
flow equations 171
δ O and δD age dating of ground
18

water, natural ground-water

systems transport 322
Dense nonaqueous phase liquids
(DNAPLs) 398
See also Contaminant removal: Multiphase
fluid systems
Deposition, porosity and permeability 17
Deuterium, isotropic processes 279

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Dewatering problem, hydraulic testing 125

Diagenesis:
porosity and permeability, marine environments 24
uplii and erosion, porosity and
permeability 27
Diffusion:
in deep sedimentary environments,
natural ground-water systems transport 305
gas transport by, multiphase fluid
systems, unsaturated zone 408
solute transport, dispersion concepts 218
Diffusion equation, mass transport
equations 296
Dimensional analysis, equations 72
Dispersion:
geostatistical model of, solute
transport 228
See also Solute transport
longitudinal, dissolved contaminant
transport modeling, analytical
approaches 372
mechanical, solute transport,
dispersion concepts 219
solute transport 216
See also Solute transport
stochastic model of, Borden tracer
experiment 235
transverse, dissolved contaminant
transport modeling, analytical
approaches 377
Dispersion coefficient, solute transport 220
See also Solute transport
Dissolution:
hydrogeologic systems, self-
organization, geologic work of
ground water 341
of solids, chemical reactions 262

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Dissolved contaminant transport

modeling 372
analytical approaches 372
advection and longitudinal
dispersion 372
computer programming for 382
multidimensional transport models 378
radioactive decay, biodegradation,
and hydrolysis 376
retardation equation 375
transverse dispersion 377
numerical approaches 384
case study 386
chemical reactions 386
common solution techniques 385
generalized modeling 384
Distance-drandown method,
nonequilibrium equation, hydraulic
testing 109
DNAPLs. See Dense nonaqueous phase
liquids
Dolomitizatioa, carbonate rocks,
geologic work of ground water 330
Dominant couples, oxidation-reduction
reactions 275
Drainage:
multiphase fluid systems 394
surface, containment 418
Drained response, to natural loading
events, one-dimensional
consolidation 163
Drawdowns, MODFLOW codes,
operational issues 152
Drill stem test, single-borehole tests 114

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Earthquake, tectonic processes, pore

fluids, phreatic seismograph 188
See also Tectonic processes
Effective solubility, multiphase fluid
systems, concepts 397
Elastic behavior of porous rock,
equation for 3
Elastic properties, three-dimensional
consolidation, deformable porous
media 169
Electrical methods, contaminant
hydrogeology, indirect detection
methods 369
Electrolytes, aqueous geochemistry 238
See also Aqueous geochemistry
Electron activity, oxidation-reduction
reactions 272
Energy resources, heat transport 209
Environmental law, hydrogeology 4
Environmental Protection Agency
(EPA), contaminant sources 348
Environmental risk assessment 456
Environmental tracers, tracers and
tracer tests: solute transport 233
Eolian deposits, porosity and
permeability 20
Equations 58
boundary conditions and flow nets 68
deformable porous media, flow
equation development 159
dimensional analysis 72
finite difference equation, ground
water resources 145
finitedifference equations,
MODFLOW codes 148
flow equations, in deforming
medium, derivation of 463
This page has been reformatted by Knovel to provide easier navigation.
Index Terms Links

Equations (Cont.)
ground-water flow 58
conservation of fluid masses 59
organization of 58
storage properties of porous
media 62
heat transport 194
of mass transport 296
See also Mass transport equations
retardation equation, dissolved
contaminant transport modeling,
analytical approaches 375
Equilibrium approach:
aqueous geochemistry 241
deviations from 243
carbonate rocks, geologic work of
ground water 327
chemical hydrogeology 3
kinetic descriptions versus, aqueous
geochemistry 240
Equilibrium calculations:
ion complexation and, complexation
reactions 264
mass distribution, multiphase fluid
systems
saturated zone 412
unsaturated zone 409
Equilibrium sorption reactions, mass
transport equations, with reaction 299
Erosion, porosity and permeability 17 27
Esters, aqueous geochemistry 248
Estuarine environment, porosity and
permeability 21
Ethers, aqueous geochemistry 248
Evaporation, volatilization, chemical
reactions 259
Evapotranspiration, hydrologic cycle 7

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Excavation:
basin hydrologic cycle 98
contaminant removal 421
Exposure assessment 451
Ex situ treatment, contaminant
removal 421
Exsolution, chemical reactions 257

Facies, defined 21
Fernald Case, risk assessment case
Study 456
Fickian model, of dispersion, solute
transport 223
Filtration, colloids 284
Finite difference equation:
ground water resources 145
MODFLOW codes 148
First-order kinetic reactions mass
transport equations, with reaction 298
Flow equations:
deformable porous media
one-dimensional consolidation 159
three-dimensional consolidation 171
in deforming medium, derivation of 463
Flow nets, equations 68
Flow stimulation, ground water
resources 142
Fluid injection, seismicity-induced,
tectonic processes, pore fluids 186
Fluid pressure:
porosity and, porosity and
permeability 31
thrust faulting and, tectonic
processes 185
Fluids, thermal expansion of, abnormal
fluid pressures 179

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Fluid sampling, contaminant

hydrogeology, sampling methods 364
Fluvial deposits, porosity and
permeability 17
Forced convection, heat transport 197
See also Heat transport
Fourier’s law, heat transport 192
Fractured rock:
abnormal fluid pressures 182
contaminant hydrogeology, solute
plumes 357
ground-water movement 48
hydraulic testing 122
porosity and permeability 29
solute transport 226
unsaturated zone, ground-water
movement 54
Free convection, heat transport 207
See also Heat transport
Freshwater-saltwater interface, coastal
regions, basin hydrologic cycle 89
Fungi, microorganisms 285
See also Microorganisms

Gas dissolution and redistribution,

natural ground-water systems
transport, unsaturated zone
chemical reactions 306
Gasoline leakage, multiphase fluid
systems, case studies 404
Gas phase, aqueous geochemistry 240
Gas solution, chemical reactions 257
Gas transport, by diffusion, multiphase
fluid systems, unsaturated zone 408
Geochemistry. See Aqueous
geochemistry
Geohydrology 2

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Geologic age, of rock groups 23

Geologic models, prototype, hydraulic
testing 103
Geologic work of ground water 326
carbonate rocks 326
chemical equilibrium 327
dolomitization 330
undersaturation problem 329
hydrocarbon migration and
entrapment 337
hydrogeologic systems, self-
organization 341
mineralization 330
lead-zinc deposits 332
noncommercial 337
uranium deposits, roll-front 331
Geophysical methods, contaminant
hydrogeology, indirect detection
methods 368
Geoprobe, contaminant hydrogeology,
sampling methods 367
Geostatistical concepts, anisotropicity,
ground-water movement 40
Geostatistical model, of dispersion,
solute transport 228
See also Solute transport
Geothermal energy, heat transport 209
Ghyben-Herzberg relation, freshwater-
saltwater interface, basin
hydrologic cycle 89
Gilson Road, Nashua, New Hampshire,
contaminant removal case study 439
Glacial deposits, porosity and
permeability 20
Gradient, Darcy’s law, ground-water
movement 36
Graphical models, basin hydrologic
cycle, topographic driving forces 76

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Gravity flow, lead-zinc deposits,

geologic work of ground water 334
Ground-penetrating radar, contaminant
hydrogeology, indirect detection
methods 370
Ground water:
basin hydrologic cycle, surface
features
lake interactions 93
surface-water interactions 95
composition of, aqueous
geochemistry 248
contaminant hydrogeology 344
See also Contaminant
hydrogeology
historical research on 2
Ground-water flow 33
basin hydrologic cycle 75
See also Basin hydrologic cycle
Darcy’s law 33
equations 58
See also Equations
fractured rock 48
geologic work of 326
See also Geologic work of ground water
heat transport in 191
See also Heat transport
hydraulic conductivity 37
anisotropicity and heterogeneity 39
aquifer classification 41
average values 42
Darcy’s law for anisotropic
material 43
distributions 38
measurement of 44
observed range 37

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Ground-water flow (Cont.)

mapping 45
hydrogeological cross sections 46
potentiometric surface 47
stimulation of, ground water
resources 142
theory of 3
unsaturated zone 51
fractured rock 54
hydraulic and pressure heads 51
variably saturated flow 54
water-retention curves 53
Ground water resources 136
development of 136
aquifer response 136
case study 137
management strategies 139
yield analysis 137
finite-difference equation 145
flow stimulation 142
MODFLOW codes 147
boundary conditions 153
case study 154
example 148
finite difference equations 148
modular program structure 148
operational issues 152
water-table conditions 153
Grouting, containment 418
Groveland Wells, Massachusetts,
contaminant removal case study 441
Gypsum, precipitation and dissolution
of natural ground-water systems
transport, unsaturated zone
chemical reactions 309

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Half-reactions, oxidation-reduction
reactions 272
Halogenated aliphatic compounds:
biotransformation 294
contaminant hydrogeology 350
Halogenated aromatic compounds:
biotransformation 294
contaminant hydrogeology 350
Halogenated hydrocarbons, aqueous
geochemistry 247
Hantush-Jacob leaky aquifer method,
hydraulic testing 110
Health effects, contaminant
hydrogeology, organic compounds 350
Health-risk assessment 454
Heat conduction equation 195
Heat transport 191
conduction, convection, and
equations
convective transport 193
equations 194
Fourier’s law 192
conduction and convection 191
energy resources 209
equations 194
forced convection 197
active depositional environments 203
mountainous terrain 205
regional flows 199
temperature and velocity 197
free convection 207
geological implications 208
onset of 207
sloping layers 208

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Heat transport (Cont.)

nuclear waste storage 210
program for 210
rock types 210
thermohydrochemical effects 212
thermomechanical effects 213
water flow and 3
Heterogeneity:
anisotropicity and, ground-water
movement 39
aquifer classification, ground-water
movement 41
Heterogeneous kinetic reactions, mass
transport equations, with reaction 299
Horizontal extension, abnormal fluid
pressures, isothermal basin loading
and tectonic strain 178
Hubert’s force potential, Darcy’s law,
ground-water movement 34
Hyde Park, Niagara Falls, New York
landfill:
contaminant removal case study 440
multiphase fluid systems, case
studies 404
Hydraulic conductivity. See also
Ground-water flow
fractured rock 50
ground-waler movement 37
Hydraulic head
Darcy’s law, ground-water
movement 34
ground-water movement, unsaturated
zone 51
mapping of ground-water movement 45

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Hydraulic testing 103

applications 124
dewatering problem 125
screen diameter and pumping
rates 125
step-drawdown test 125
water supply problem 127
bounded aquifers 120
computer calculations 128
conventional 105
Hantush-Jacob leaky aquifer
method 110
nonequilibrium modifications 108
steady-state behavior 109
Theis nonequilibrium pumping
method 105
water table aquifers 112
fractured or low-permeability rocks 122
partial penetration 118
prototype geologic models in 103
single-borehole tests 114
drill stem test 114
recovery in pumped well 114
slug injection (withdrawal) tests 115
well capacity 116
superposition principle 118
Hydrocarbons:
biotransformation 293
migration and entrapment, geologic
work of ground water 337
petroleum hydrocarbons,
contaminant hydrogeology 349
Hydrodynamic controls, containment 419
Hydrogeological cross sections, ground-
water movement, mapping 46
Hydrogeologic systems, geologic work
of ground water, self-organization 341

This page has been reformatted by Knovel to provide easier navigation.

 Next Page

Index Terms Links

Hydrogeology:
contaminant 344
See also Contaminant hydrogeology
defined 1
historical perspective
chemical 3
physical 2
post-1960 4
related fields 4
Hydrologic cycle 5
base flow 8
basin 75
See also Basin hydrologic cycle
components of 5
equation 10
evapotranspiration 7
infiltration and recharge 8
Hydrolysis:
chemical reactions 277
dissolved contaminant transport
modeling, analytical approaches 376
Hydrophobic sorption, of organic
compounds, surface chemical
reactions 267
Hydrostratigraphic unit, defined 16

Infiltration, hydrologic cycle 8

Inhibition, multiphase fluid systems 394
Initial conditions, mass transport
equations 300
Injection-recovery systems. See Pump-
and-treat systems
Inorganic species, contaminant
hydrogeology 349

This page has been reformatted by Knovel to provide easier navigation.

Previous Page

Index Terms Links

In situ destruction 443

abiotic chemical destruction 449
bioremediation, intrinsic 443
bioventing and bioslurping 445
applicability 446
progress in 448
requirements 446
testing 447
Instantaneous point source model,
dissolved contaminant transport
modeling, analytical approaches 380
Interceptor systems, contaminant
removal 421
Interfacial tension, multiphase fluid
systems 394
Intergranular porous rock, ground-
water movement, fractured rock 49
Interstitial porosity, defined 13
Intrinsic bioremediation, in situ
destruction 443
Ion complexation, equilibrium
calculations and, complexation
reactions 264
Ions, aqueous geochemistry 238
See also Aqueous geochemistry
Isothermal basin loading, abnormal
fluid pressures 175
See also Abnormal fluid pressures
Isotropic processes 277
deuterium and oxygen-18 279
isotropic reactions 278
radioactive decay 277

Karst systems, contaminant

hydrogeology, solute plumes 357
Kd-based approaches, metals, sorption
of, chemical reactions 269

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Keytones, aqueous geochemistry 247

Kinetic reactions:
aqueous geochemistry 244
equilibrium models versus, aqueous
geochemistry 240
first-order, mass transport equations,
with reaction 298
heterogeneous, mass transport
equations, with reaction 299
oxidation-reduction reactions 275

Laboratory tests, hydraulic conductivity,

measurement of 44
Lacustrine deposits, porosity and
permeability 20
Lagoonal environment, porosity and
permeability 21
Lake—ground-water interactions, basin
hydrologic cycle, surface features 93
Landfills:
contaminant removal case study 440
multiphase fluid systems, case
studies 404
Landslides, basin hydrologic cycle 101
Land subsidence, one-dimensional
drained response 163
defomiable porous media
stimulation of 169
time rate of 167
stimulation of 169
time rate of 167
verticnl compression 165
Lateral succession of strata, porosity
and permeability, marine
environments 21
Lead-zinc deposits, geologic work of
ground water 332

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Leaky aquifer, Hantush jacob method,

hydraulic testing 110
Light nonaqueous phase liquids
(LNAPLs) 398
See also Contaminant removal; Multiphase
fluid systems
LNAPLs. See Light nonaqueous phase
liquids
Loading source, contaminant
hydrogeology 344
Localization, contaminant
hydrogeology 344
Longitudinal dispersion, dissolved
contaminant transport modeling,
analytical approaches 372
Low-permeability rocks, hydraulic
testing 122

Macrofracture, porosity and

permeability 30
Macroscopic scale, solute transport,
dispersion coefficient 221
Magnetometry, contaminant
hydrogeology, indirect detection
methods 371
Management strategies:
ground water resource development 139
remediation 420
Mapping:
of aquifers 15
of ground-water movement 45
Marginal marine environment, porosity
and permeability 20
Marine environments, porosity and
permeability 21
See also Porosity and permeability
Mass, in water, aqueous geochemistry 240

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Mass distribution, equilibrium

calculations, multiphase fluid
systems, unsaturated zone 409
Massively instrumented field tracer
tests, solute transport 233
Mass transport:
in ground-water flows, geologic
systems 326
See also Geologic work of ground water
in natural ground-water systems 303
See also Natural ground-water systems transport
Mass transport equations 296
advection-diffusion equation 297
advection-dispersion equation 297
boundary and initial conditions 300
diffusion equation 296
with reaction 298
equilibrium sorption reactions 299
first-order kinetic reactions 298
heterogeneous kinetic reactions 299
Mathematical models, basin hydrologic
cycle, topographic driving forces 76
Mechanical dispersion, solute transport,
dispersion concepts 219
Mesozoic rock group, porosity and
permeability 24
Metals:
enhancement of mobility,
complexation reactions 264
mobility of
control, oxidation-reduction
reactions 276
enhancement, complexation
reactions 264
sorption of, Kd-based approaches to
modeling, chemical reactions 269
trace metals, contaminant
hydrogeology 347

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Metamorphic processes, fluid phase

and 5
Meteoric waters, natural ground-water
systems transport, mixing as
chemical change agent 303
Microfracture, porosity and
permeability 30
Microorganisms 285
See also Colloids
biofilnis 287
biotransformation 293
complex transformation pathways 295
halogenated aliphatic compounds 294
halogenated aromatic compounds 294
hydrocarbons 293
polychlorinated biphenyls (PCBs) 295
overview 285
processes 291
biodegradation 292
biofilm kinetics 292
reaction rates 288
sampling and enumeration 287
subsurface ecology 290
Microscopic scale, solute transport,
dispersion coefficient 220
Milk River aquifer. natural ground-water
systems transport. case study 316
Mineralization, geologic work of
ground water 330
See also Geologic work of ground water
Mineralogy, hydrogeology and 4
Miscellaneous nonvolatile compounds,
aqueous geochemistry 246
Mississippi Valley-type lead-zinc
deposits, geologic work of ground
water 332

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Mobility, of metals:
control of, oxidation-reduction
reactions 276
enhancement of, complexation
reactions 264
Modeling, dissolved contaminant
transport modeling 372
See also Dissolved contaminant
transport modeling
MODFLOW codes 147
boundary conditions 153
case study 154
example 148
finite-difference equations 148
modular program structure 148
operational issues 152
water-table conditions 153
Mohr-Coulomb theory, reservoirs,
tectonic processes, pore fluids 188
Molecules, aqueous geochemistry 238
See also Aqueous geochemistry
Mountainous terrain:
basin hydrologic cycle, topographic
driving forces 83
heat transport, forced convection 205
Mudstones, diagenesis, porosity and
permeability 27
Multidimensional transport models,
dissolved contaminant transport
modeling, analytical approaches 378
Multilevel samplers, contaminant
hydrogeology, sampling methods 363
Multiparameter equilibrium models,
surface chemical reactions 269

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Multiphase fluid systems 393

air-permeability testing 412
concepts 393
inhibition and drainage 394
interfacial tension and capillary
forces 394
relative permeability 395
saturation and wettability 393
solubility and effective solubility 397
DNAPL site recognition 413
LNAPLs and DNAPLs 398
case studies 404
occurrence of DNAPLs 399
Occurrence of LNAPL 399
quantitative methods 403
secondary contamination 401
partitioning 405
saturated zone 411
Unsaturated zone 406
gas transport by diffusion 408
mass distribution equilibrium
calculations 409
volatilization 406
Multiple-borehole tests, fractured or
low-permeability rocks 123

Natural gradient test, solute transport 232

Natural ground-water systems transport 303
age dating of ground water 319
direct methods 319
indirect methods 322
case study, Milk River aquifer 316
mixing as chemical change agent 303
diffusion in deep sedimentary
environments 305
meteoric and original formation-
waters 303

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Natural ground-water systems transport (Cont.)

saturated zone chemical reactions 310
cation exchange 315
redox reactions 313
soluble salts dissolution 312
weak acid-strong base reactions 310
unsaturated zone chemical reactions 306
cation exchange 309
gas dissolution and redistribution 306
gypsum precipitation and
dissolution 309
organic reactions 309
sulfide oxidation 309
weak acid-strong base reactions 307
Natural loading events:
drained response to, one-dimensional
consolidation 163
undrained response to, one-
dimensional consolidation 160
Natural weak acid-base systems 256
See also Weak acid
Nonaqueous phase liquids (NAPLs) 398
See also Contaminant removal; Multiphase fluid systems
Nonequilibrium equation,
modifications, hydraulic testing 108
Nonvolatile compounds, miscellaneous,
aqueous geochemistry 246
Normal compaction, diagenesis,
porosity and permeability 25
Nuclear waste storage 210
program for 210
rock types 210
thermohydrochemical effects 212
thermomechanical effects 213
Nutrients, contaminant hydrogeology 348

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Oil spills, contaminant removal case

Study 438
One-dimensional consolidation 159
deformable porous media 159
See also Deformable porous media
drained response to natural loading
events 163
flow equation development 159
undrained response to natural
loading events 160
One-dimensional drained response, land
subsidence 163
stimulation of 169
time rate of 167
vertical compression 165
One-dimensional loading, abnormal
fluid pressures, isothermal basin
loading and tectonic strain 176
Ore deposits, geologic work of ground
water 331
See also Geologic work of ground water
Organic complexation, complexation
reactions 265
Organic compounds:
aqueous geochemistry 245
biotransformation of, oxidation-
reduction reactions 276
hydrophobic sorption of, surface
chemical reactions 267
Organic contaminants, contaminant
hydrogeology 349 351
Organic reactions, natural ground-water
systems transport, unsaturated
zone chemical reactions 309
Organic solutes, chemical reactions 258
Organometallic compounds, aqueous
geochemistry 247
This page has been reformatted by Knovel to provide easier navigation.
Index Terms Links

Original formation waters, natural

ground-water systems transport,
mixing as chemical change agent 303
Orogenic belt expulsion, lead-zinc
deposits, geologic work of ground
water 335
Otis Air Base (Cape Cod,
Massachusetts), dissolved
contaminant transport modeling,
case study 386
Oxidation, sulfide, natural ground-water
systems transport, unsaturdted
zone 309
Oxidation numbers, oxidation-reduction
reactions 272
Oxidation-reduction reactions 272
kinetics and dominant couples 275
mobility of metals control 276
organic compounds,
biotransformation of 276
oxidation numbers, half-reactions,
electron activity, and redox
potential 272
Oxygen-18, isotropic processes 270

Paleozoic rock group, porosity and

permeability 23
Partial penetration, hydraulic testing 118
Partitioning, multiphase fluid systems 405
Percolation theory, tectonism and
fracture formation, porosity and
permeability 31
Permeability. See also Porosity and
permeability
Darcy's law, ground-water
movement 37
described 15

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Permeability (Cont.)
hydraulic conductivity relationship 44
relative, multiphase fluid systems 395
Petroleum hydrocarbons, contaminant
hydrogeology 349
Petrology, hydrogeology and 4
Phase transformations, abnormal fluid
pressures 183
Phenols, aqueous geochemistry 247
Phosphorous compounds, aqueous
geochemistry 247
Phreatic seismograph, earthquake,
tectonic processes, pore fluids 188
Physical hydrogeology, historical
perspective 2
See also Hydrogeology
Physical weathering, porosity and
permeability 16
Piezometers, contaminant
hydrogeology, sampling methods 362
Plate counts, microorganisms 288
Plumes. See Solute plumes
Polychlorinated biphenyls (PCBs):
biotransformation 295
contaminant hydrogeology 350
Polynuclear aromatic hydrocarbons,
aqueous geochemistry 248
Pore fluids. See Abnormal fluid
pressures; Deformable porous
media; Tectonic processes
Porosity and permeability 13
See also Permeability
continental environments 16
erosion, transportation, and
deposition 17
weathering 16
continental-marine boundary 20

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Porosity and permeability (Cont.)

fluid pressure and, porosity and
permeability 3
marine environments 21
ancestral seas 22
diagenesis 24
lateral and vertical strata
succession 21
paleozoic rock group 23
permedbility 15
porosity and effective porosity 13
tectonism and fracture formation 29
uplift, diagenesis. and erosion 27
Porous media, storage properties of,
ground-water flow equations 62
Potential evapotranspiration, hydrologic
cycle 7
Potentiometric surface, ground-water
movement, mapping 47
Precipitation:
evapotranspiration, hydrologic cycle 7
hydrogeologic systems, self-
organization, geologic work of
ground water 341
of solids, chemical reactions 262
Prediction, ground water resources 144
Pressure head, ground-water
movement, unsaturated zone 51
Pressure solution, diagenesis, porosity
and permeability 25
Primary porosity, defined 13
Proportionality constant, Darcy's law,
ground-water movement 36
Prototype geologic models, hydraulic
testing 103
Protozoa, microorganisms 285
See also Microorganisms

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Pump and-treat systems 422

contaminant removal 421
design methods 425
capture zones 426 430
model-based approaches 429
pilot-scale system expansion 425
stimulation-optimization
techniques 429
problem of 422
technical considerations 423
Pumping rates, screen diameter and,
hydraulic testing 125

Quality assurance, sampling,

contaminant hydrogeology 362

Radar, ground-penetrating, contaminant

hydrogeology, indirect detection
methods 370
Radioactive contaminants, contaminant
hydrogeology 346
Radioactive decay:
dissolved contaminant transport
modeling, analytical approaches 376
isotropic processes 277
Reaction rates, aqueous geochemistry.
equilibrium versus kinetic
descriptions 241
Reactive barrier systems, abiotic
chemical destruction 450
Recession, base flow, hydrologic cycle 8
Recharge:
artificial, ground water resource
development 139
hydrologic cycle 8

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Recharge-discharge relations, basin

hydrologic cycle, surface features 91
Recovery in pumped well, single-
borehole tests 114
Redox potential, oxidation-reduction
reactions 272
Redox reactions, natural ground-water
systems transport, saturated zone
chemical reactions 313
Regional flows, heat transport, forced
convection 199
Relative permeability. multiphase fluid
systems 395
Remediation 417
containment 417
grouting 418
hydrodynamic controls 419
sheet pile walls 418
slurry walls 417
stabilization and solidifkition 420
surface seals and surface drainage 418
contaminant removal 420
See also Contaminant removal
management options 420
Reservoirs:
basin hydrologic cycle 97
tectonic processes, pore fluids 187
Resource Conservation and Recovery
Act (RCRA) 4
Retardation equation, dissolved
contaminant transport modeling,
analytical approaches 375
Risk assessment 450
case study 456
data collection and evaluation 451
environmental assessment 456
exposure assessment 451
health-risk assessment 454

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Risk assessment (Cont.)

toxicity assessment 453
types of 455
Rock fractures. See Fractured rock
Rock groups:
geologic age of 23
hydraulic conductivity 39
nuclear waste storage 210
porosity and permeability 23
Rock matrix, compressibility of,
ground-water flow equations 64
Roll-front uranium deposits, geologic
work of ground water 331

Saline soils, geologic work of ground

water, mineralization 337
Sampling:
contaminant hydrogeology 360
network designs 360
quality assurance 362
methods of, contaminant
hydrogeology 362
See also Contaminant hydrogeology
microorganisms 287
Sandstone:
marginal marine environment 21
secondary porosity in, diagenesis 26
Saturated, defined 5
Saturated zone, multiphase fluid
systems 411
Saturated zone chemical reactions,
natural ground-water systems
transport 310
See also Natural ground-water systems
tansport
Saturation, multiphase fluid systems 393

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Screen diameter, pumping rates and,

hydraulic testing 125
Sea-level canal, basin hydrologic cycle 98
Seals, surface, containment 418
Secondary porosity:
carbonate rocks, uplift, diagnesis,
and erosion 29
defined 13
in sandstones, diagenesis 26
Sedimentary environments, deep,
diffusion in, natural ground-water
systems transport 305
Sedimentary rock, hydraulic
conductivity 39
Sediments, hydraulic conductivity 39
Seismicity-induced fluid injection,
tectonic processes, pore fluids 186
Seismic methods, contaminant
hydrogeology, indirect detection
methods 371
Sewage, spring monitoring, carbonate
terrain 87
Shales:
Dakota aquifer 22
diagenesis, porosity and permeability 26 27
Sheet pile walls, containment 418
Single-borehole tests 114
drill stem test 114
fractured or low-permeability rocks 123
recovery in pumped well 114
slug injection (withdrawal) tests 115
well capacity 116
Single well injection/withdrawal, with
multiple observation wells, solute
transport 233
Single well pulse test, solute transport 232
Slope stability, basin hydrologic cycle 101

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Slug injection (withdrawal) tests, single-

borehole tests 115
Slurry walls, containment 417
Soil-gas characterization, contaminant
hydrogeology, indirect detection
methods 367
Soil permeability estimates, soil-vapor
extraction 435
Soil-vapor extraction 431
heterogeneity and efficiency 436
indications for 433
overview 421
soil permeability estimates 435
system components 432
Solidification, containment 420
Solid phase, aqueous geochemistry 240
Solid sampling, contaminant
hydrogeology, sampling methods 364
Solid solubility, chemical reactions 262
Solubility:
multiphase fluid systems, concepts 397
of solids, chemical reactions 262
Soluble salts, dissolution of, natural
ground-water systems transport,
saturated zone chemical reactions 312
Solute plumes, contaminant
hydrogeology 352
See also Contaminant hydrogeology
Solute transport 215
advection 215
dispersion coefficient 220
dispersivity as medium property 221
macroscopic and larger levels 221
microscopic scale 220
dispersion concepts 216
diffusion 218
mechanical dispersion 219
Fickian model of dispersion 223

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Solute transport (Cont.)

fractud rock 226
geostatistical model of dispersion 228
autocovariance and autocorrelation
functions 229
correlated fields generation 230
dispersivity estimation 230
mean and variance 228
tracers and tracer tests 231
Contaminant plumes and
environmental tracers 233
field experiments 232
massively instrumented field tracer
tests 233
Solution, chemical reactions 257
Sorption:
equilibrium, mass transport
equations, with reaction 299
hydrophobic, of organic compounds,
chemical reactions 267
of metals, Kd based approaches to
modeling, chemical reactions 269
Sorption isotherms, surface chemical
reactions 266
Speciation modeling, stability and,
complexation reactions 263
Spring monitoring, sewage, carbonate
terrain 87
Stability, speciation modeling and,
complexation reactions 263
Stabilization, containment 420
Steady-state behavior, hydraulic testing 109
Step-drawdown test, hydraulic testing 125
Stimulation, of land subsidence, one-
dimensional drained response 169
Stimulation-optimization techniques,
pump-and-treat systems 429

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Stochastic model, of dispersion, Borden

tracer experiment 235
Storage properties, of porous media,
ground-water flow equations 62
Strain. See Abnormal fluid pressures;
Deformable porous media;
Tectonic processes
Stratigraphy, hydrogeology and 4
Stress and strain. See Abnormal fluid
pressures; Deformable porous
media; Tectonic processes
Structure, hydrogeology and 4
Subnormal pressure, abnormal fluid
pressures 184
Subsidence. See Land subsidence
Subsurface ecology, microorganisms 290
Sulfide oxidation, natural ground-water
systems transport, unsaturated
zone chemical reactions 309
Superposition principle, hydraulic
testing 118
Surface features, basin hydrologic
cycle 91
See also Basin hydrologic cycle
Surface Mining Act 4
Surface reactions, chemical reactions 266
See also Chemical reactions
Surface seals and drainage,
containment 418
Surface-water—ground-water
interactions, basin hydrologic
cycle, surface features 95

Tectonic processes. See aiso Abnormal

fluid pressures; Deformable porous
media

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

lead-zinc deposits, geologic work of

ground water 335
pore fluids 185
earthquake, phreatic seismograph 188
fluid injection, seismicity-induced 186
fluid pressures and thrust faulting 185
reservoirs, seismicity-induced 187
porosity and permeability 29
strain, abnormal fluid pressures 175
Tensiometer, ground-water movement,
unsaturated zone 52
Theis nonequilibrium pumping method,
hydraulic testing 105
Theis solution, heat flow and 3
Thermal expansion, of fluids, abnormal
fluid pressures 179
Thennohydrochemical effects, nuclear
waste storage 212
Thermomechanical effects, nuclear
waste storage 213
Three-dimensional consolidation 169
elastic properties 169
flow equations 171
Thrust faulting, fluid pressures and,
tectonic processes 185
Time-drawdown method,
nonequilibrium equation, hydraulic
testing 108
Time rate, of land subsidence, one-
dimensional drained response 167
Time-step size, MODFLOW codes,
operational issues 152
Topographic driving forces, basin
hydrologic cycle 75
See also Basin hydrologic cycle
Total porosity, effective porosity and 14
Toxicity assessment 453

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Trace metals, contaminant

hydrogeology 347
Tracers and tracer tests, solute
transport 231
contaminant plumes and
environmental tracers 233
field experiments 232
massively instrumented field tracer
tests 233
Transient response, steady-state
behavior, hydraulic testing 109
Transportation, porosity and
permeability 17
Transport processes, hydrogeology 4
Transverse dispersion, dissolved
contaminant transport modeling,
analytical approaches 377
Tritium, age dating of ground water,
natural ground-water systems
transport 319
Two-well tracer test, solute transport 233

Undersaturation problem, carbonate

rocks, geologic work of ground
water 329
Undrained response, to natural loading
events, one-dimensional
consolidation 160
Unsaturated, defined 5
Unsaturated zone:
chemical reactions in, natural
ground-water systems transport 306
See also Natural ground-water systems transport
contaminant hydrogeology, sampling
methods 367
ground-water movement 51
See also Ground-water flow

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Unsaturated zone: (Cont.)

multiphase fluid systems 406
See also Multiphase fluid systems
Uplift, porosity and permeability 27
Uranium deposits, roll-front, geologic
work of ground water 331

Velocity of flow:
Darcy's law, ground-water
movement 34
heat transport, forced convection 197
Vertical compression:
abnormal fluid pressures, isothernial
basin loading and tectonic strain 178
land subsidence, one-dimensional
drained response 165
Vertical succession of strata, porosity
and permeability, marine
environments 21
Volatilization:
chemical reactions 259
multiphase fluid systems, unsaturated
zone 406

Waste disposal, hydrogeology 4

See also Contaminant hydrogeology
Water, compressibility of, storage
properties of porous media,
ground-water flow equations 63
Water flow, heat flow and 3
Water-level fluctuation:
drained response to natural loading
events, one-dimensional
consolidation 163

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Water-level fluctuation: (Cont.)

undrained response to natural
loading events, one-dimensional
consolidation 160
Water-retention curves, ground-water
movement, unsaturated zone 53
Water structure, aqueous geochemistry 240
Water supply problem, hydraulic
testing 127
Water table aquifers, hydraulic testing,
conventional 112
Water table conditions, MODFLOW
codes 153
Weak acid-base systems, natural 256
Weak acid-strong base reactions,
natural ground-water systems
transport:
saturated zone chemical reactions 310
unsaturated zone chemical reactions 307
Weathering, porosity and permeability 16
Well capacity, single-borehole tests 116
Well hydraulics, hydrogeology and 3
Wells:
contaminant hydrogeology, sampling
methods 362
dewatering problem, hydraulic
testing 125
freshwater-saltwater interface, basin
hydrologic cycle 91
recovery in pumped well, single-
borehole tests 114
Well yield, step-drawdown test,
hydraulic testing 125
Wettability, multiphase fluid systems 393
Withdrawal (slug injection) tests, single-
borehole tests 115

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Yield analysis, ground water resource

development 137

This page has been reformatted by Knovel to provide easier navigation.