LBL-38383

UC-C14

TOUGH2
TOUGH2
TOUGH2
TOUGH2
TOUGH2 ^ . - ^ . _ , m —
SOFTWARE | %% I 1^ Z §—§ 2

K . PRUESS
A. SIMMONS
Y.S. W u
G. M O R I O I S

FEBRUARY 1 9 9 6

ERNEST O R L A N D O LAWRENCE
BERKELEY N A T I O N A L LABORATORY
EARTH SCIENCES D I V I S I O N
O N E CYCLOTRON ROAD
BERKELEY, C A 9 4 7 2 0

THIS WORK WAS PREPARED UNDER TUB O m C G <>t C I V I L I A N RAniOACTfVt

W A S T E M A N A . EMENT, U.S. OEPARTMEN) or EN*KC;V C O N T R A C T No. DE-
A C O 3 - 7 6 S F 0 0 0 9 B , AND D E > A ! 0 d - 7 8 E T 4 4 a O 2 AtlMINtSTEREO 9V THE
N E V A D A OPERATIONS Office I N COORITRATION W I T H T H E U.S. CCOLOGICAL
SURVEY, D E N V E R .

DISTRIBUTION OF THIS DOCUMENT IS UNLIMITED IU i i * ^ | t l i

 DISCLAIMER

This document was prepared as an account of wcri; sponsored by toe

United States Government. White this document is believed to coMiain
coneel information, neither Ihe Untied Stoles Government nor any
agency lhrreof, nor Tiie Regents of the University of California, nor any
of their employees, nukes say warranty, express cr Implied, or assumes
any legal responsibility for tho accuracy, completeness, or usefulness of
any information, apparatus, product, or process disclosed, or
represents thit its use would not infringe privately owned rights.
Reference herein to any specific commercial prodaet, process, or
service by its trade name, trademaik. manufacturer, ur otherwise, does
not necessarily constitute or imply its endorsement, recommendation,
or favoring by the United Slates Governmeirt cr any ageccy thereof, or
The Resents of the University of California. TUc views and opinions of
authors expressed herein do not necessarily stale or reflect those of ihe
United Stales Government or Any agency ihereof. cr The Regents of the
University of California.

Available to DOE and DOE Coatf&ciors

from the Office of Scientific and Technical lofarmstion
P.O. Box G2, Oak Ridge. TO 37831
Prices available from (615) 576-840!

Available (o the public from the

National Technical ] nfor(nation Service
U.S. Department of Commerce
5285 Port Royal Road. Springfield. V'A 22161

Ernest Orlando Lawrence Berkeley Nations* Labotatory

is an equal opportunity employer.
 DISCLAIMER

Portions of this document may be illegible

in electronic image products. Images are
produced from the best available original
document.
TABLE OF CONTENTS

1. Introduction 1

2. Requirements Specification 3

2.1 Requirements 3
2.2 Examples of Verification of Meeting Requirements 4

3. Design Description 7

3.1 Background on MULKOMTOUGH Family of Codes 7

3.2 Structure and Architecture 7
3.3 Physical Processes and Approximations II
3.4 Mathematical and Numerical Methods 12
3.5 Array Structure and Handling 14
3.6 Linear Equation Setup. 16
3.7 Dimensioning of Major Arrays 17
3.S Equation-of-State Modules 18
3.9 Specification of How Geometry 24
3.10 Initial Conditions and Restarting 23
3.11 Boundary Conditions 27
3.12 Software and Hardware Considerations 28
3.13 Performance Requirements and Design Constraints 29
3.14 Future Developments 30
3.15 Verification of Design Description 30

4. Software Validation Test Pian and Report 31

4.1 Validation Test Plan 31

4.2 Validation Tests 34
4.3 Summary 6!

TOUGH2 SOFTWARE QUALIFICATION - FEBRUARV 1996

 5. Software User Documentation ....65

5.1 Installation Procedures 65

5.2 Hardware and Software Operating Environments 65
5.3 Input and Output 65
5.4 User Features 70
5.5 Summary 70
5.6 Version History 71

Acknowledgements 72
References 73
Appendices 81
Appendix A. Mass and Energy Balances 81
Appendix B. Space and Time Discretization 83
Appendix C. Description of Flow in Fracrared Media 86
Appendix D. Nomenclature 89
Appendix E.TOUGH2 Selected Bibliography 9i
Appendix F. README File 120
Appendix G. Sample Problems for Code Demonstration 127

T O U C H 2 SOFTVMRE QUAIIFICATION - FEBRUARY 1996 ii

LiST OF FIGURES

Figure 1. Relationship ofTOUGH and TOUGH2 to MULKOM architecture .'. 1

Figure 2. Modular architecture of MULKOM and TOUGH2 8
Figure 3. TOUGH2 code structure 9-10
Figure4. Summary of program units and version history 11
Figures. Structure of thermophysical property arrays..'. 15
Figure 6. Linear equation structure 15
Figure?. Diustrafons of subregions on &eP-T diagram. 19
Figure 8. User options for supplying geometric data 25
Figure 9. Liquid saturation profiles for validation problem 1 35
Figure 10. Wellblock pressures for validation problem 2...-. 36
Figure 11. Simulated steady-state conditions in validation problem 3 37
Figure 12. Comparison of the analytical solution to the TOUGH2 solution
in validation problem 4 38
Figure 13. Comparison ofthe analytical and TOUGH2 solutions to the radial heat
transport problem in validation problem 5 39
Figure 14. Comparison of the analytical and the TOUGH2 solutions to the
Theis problem in validation problem 6 at t = t„„=10 days 40
Figure 55. Comparison of TOUGH2 simulation to analytical solution for pressure
buildups for non-isothermal injection into a fractured reservoir for
validation problem 7 41
Figure 16. TOUGHS predictions and experimental measurements at
x = 1.39 m in validation problem 8 42
Figure 17. a) Comparison of TOUGH2 predictions with experimental data
at the top and bottom of the heating element in validation problem 9.
b) Comparison of TOUGH2 predictions with experimental data
for greater distances from beater. 44
Figure 18. Comparison of TOUGH2 results and experimental data for validation
problem 10 45
Figure 19. Comparison of TOUGH2 to Warren-Root double-porosity solution
for validation problem 11 46
Figure 20. Temperature profiles for the analytical solution and the
TOUGH2 simulation for validation problem 12 47

TOUGH2 SOFTWARE QUALIFICATION - FMHUARV 1996 in

 Figure 21. Comparison of analytical and TOUGH2 solutions to problem
of 1-D flow through a heterogeneous formation for validation
problem 15a 51
Figure 22. Comparison of the analytical and TOUGH2 solutions to the
problem of 2-D flow to a single well in an anisotropic aquifer
for validation problem 15b 52
Figure 23. Irregular grid constructed for validation problem 16 53
Figure 24. Comparison of TOUGH2 simulation with the Theis solution for
an irregular grid as described in validation problem 16." 54
Figure 25. Comparison of analytical solution and TOUGH2 results for
heat conduction problem in validation problem 17a. 55
Figure 26. Comparison of air mass fraction profiles showing TOUGH2
results and analytical solution for validation problem 17b 56
Figure 27. Comparison of the pressure profile along the rock column
from the TOUGH2 simulation and the analytical
solution for validation problem 17c 57
Figure 28. Horizontal well model for validation problem 18 58
Figure 29. Comparison of analytical and TOUGH2 solutions showing pressure
drawdown for the 3-D validation problem 18. 59
Figure 30. Analytical and TOUGH2 predictions of pressure drawdown
distribution in the gas reservoir for validation problem 19 60
Figure 31. Comparison of TOUGH and semi-analytical solution showing
normalized cumulative liquid flux for slab absorption in validation
problem 20 61
Figure 32. TOUGH input formats with TOUGH2 extensions 66
Figure 33. Input formats for the new TOUGH2 data blocks 67
Figure 34. Input formats for MESHMAKER module 69
Figure 35. Space discretization and geometry data in the integral finite
difference method S3
Figure 36. Idealized "double porosity" model of afracturedporous medium 86
Figure 37. Subgridding in the method of "multiple interacting conrinua" (MINQ.... 87
Figure 38. Flow connections in the "dual permeability" model 88
Figure 39. Input file for sample problem 1 - code demonstration 128
Figure40. Selected output for sample problem 1 129

TOUGH2 SOFTWARE QUALIFICATION - FEBRUARY 1996 i7

Figure 41. Produced fluid temperature vs. time for vertical fracture problem
in sample problem 3 130
Figure 42. Temperature profiles for problem 4 along a line of injection to
production well after 36.5 years 131

TOUCH2 SOFTWARE QUAUFICAIiON - FEBRUARY 1 9 9 6 ; V

LIST OF TABLES

Table 1. TOUGH2 fluid property modules 8

Table 2. Summary of major common blocks IS
Table3. Ranges ofthermophysical parametersforphase regions in Figure 7. 20
Table4. Summary of EOS 1 21
Tables. Summary of EOS2 21
Table 6. Summary of EOS3 22
Tabic 7. Summary of EOS4 23
Tables. Summary of EOS5 24
Table 9. Summary of validation problems 32-34
Table 10. Comparison of TOUGH2 to thermophysicai property data from
steam tables in validation problem 13 48-49
Table 11. Comparison of TOUGH2 results to Kelvin's law's handling
of vapor pressure lowering in validation problem 14 49
Table 12. Reservoir properties, geometry, and discretization in validation
problem 15a 50
Table 13. Requirements validation cross-check 62-64
Table 14. New optional data blocks in TOUGK2 67
Table 15. Specifications for generation types 6S
Table 16. TOUGH2 disk files 70
Table 17. Elements and features ofsubproblems in sample problem 1 127

TOUGH2 SonwARe QUALIFICATION - FEBRUARY 1996 vi

1. INTRODUCTION
T0UGH2 is a numerical simulation code for multi-dimensional coupled fluid and
heat flow of multiphase, multicomponent fluid mixtuies in poious and fractured media. It
belongs to the MULKOM ("JfiJLti-KQMponent") family of codes (Pruess, 1983b, 1988)
and is a more general version of the TOUGH simulator (Pruess, 1987). The MULKOM
family of codes was originally developed with a focus on geoihermal reservoir simulation.
They ate suited to modeling systems which contain different fluid mixtures, with
applications to flow problems arising in the context of nigh-level nuclear waste isolation,
oil and gas recovery and storage, and groundwater resource protection. TOUGH2 is
essentially a subset of MULKOM, consisting of a selection of the better tested and
documented MULKOM program modules. 'iTierelationshipof TOUGH and TOUGH2 to
die MULKOM architecture is shown in Figure 1.

Figure 1. Relationship of TOUCH and TOUGH2 to the MULKOM arcluucturc

The purpose of this package of reports is to provide all software baseline

documents necessary for the software qualification of TOUGII2. In accordance with
Lawrence Berkeley National Laboratory (LBNL) Quality Implementing Procedure (Q1P)
SI.O, all components of the specified baseline documentation are provided. The software is
not acquired, therefore no parts of (he documentation are exempt from software verification
and validation. This report contains the following sections: 1) Requirements Specification.
2) Design Description, 3) Software Validation Test Plan and Report, 4) Software User
Documentation, and 5) Appendices. These sections comprise sequential parts of the

TOUCH2 SOFTWARE QUALIFICATION - FEBRUARY 1996

Software Life Cycle. They could have been written as a series of reports, but we chose to
combine them into one report. The report components are not intended to stand alone but
should be used in conjunction with the TOUGH User's Guide (Pruess, 1987) and with
TOUGH2 - A General Purpose Numerical Simulatorfor Multiphase Fluid and Heat Flow
(Pruess, 1991). The qualification package is complete with the attached Software
Identification Form, executable source code, list of users, and bibliography of selected
publications which utilized TOUGH or TOUGH2. No attempt was made to provide a
conclusive bibliography; rather, the better known applications in geologic disposal of
nuclear waste and geothermal reservoir processes are listed. Included in the package are
several additional key publications that provide full details of (he verification and validation
test problems described herein and additional examples of TOUGH2 applications. They
include the following: Proceedings of the TOUGH Vfortehop (Pruess, 1990); TOUGH
Simulations of Updegraff's Set of Fluid and Heat Flow 'Problems (Moridis and Pruess,
1992); Proceedings of the TOUGH Workshop '95 (Pruess, 1995); and Flow and
Transport Simulations Using T2CG1, a Package of Conjugate Gradient Solvers for the
TOUGH2 Family of Codes (Moridis and Pruess, 1995).
The version of TOUGH2 that is herein being qualified is the November 1994
Standard Version 1.11 housed at the Department of Energy's Energy Science and
Technology Software Center (ESTSQ in Oak Ridge, Tennessee.

TOUGH2 SOFTWARE QUALIFICATION • FEBRUARY 199C •>

2. REQUIREMENTS SPECIFICATION

TOUGH2 is used for a multitude of applications, including oil and gas reservoir
engineering, environmental problems involving contaminants, geothermal reservoir
processes, and geologic disposal of nuclear waste. A list of requirements for representing
conditions applicable to all of these settings would be long. Because Version l . i l of
TOUGH2 is being qualified for applications to the latter two modeling situations, a list of
requirements was drawn up that is specific to representing the dominant fluid flow and heat
transfer processes in geothermal reservoirs and in geologic disposal systems for nuclear
waste. In order to simulate these processes, TOXJGH2 must represent correctly the
physical processes of multi-phase flow in saturated and unsaturated heterogeneous media,
it therefore must provide an accurate description of constitutive physical laws and
properties affecting each phase. It also must be capable of handling different geometries in
one, two, and three dimensions. To handle the problems efficiently and smoothly,
TOUGH2 must meet requirements for flexibility and timing. The multiple requirements for
TOUGH2 with respect to geothermal and nuclear waste applications are listed below for
ease of referral. The code underwent a long period of testing tbroughout development of
the MULKOM code family and with a variety of settings and specifications, for the express
purpose of determining whether the code accurately handles the processes of interest.
Examples of some of the situations to which the code was applied are found in Appendix
E, Selected Bibliography, and in application examples that follow. Furthermore,
verification and validation tests have been performed that collectively address all of the
listed requirements. These are described in Section 4, Software Validation Test Plan and
Report.

2.1 Requirements

A. Flow processes driven by pressure, viscous, capillary, and gravity forces

- single-phase fluid flow
- single-phase gas flow
- two-phase flow of liquid and gas
B. Constitutive relations
- accurate description of thermophysical properties of water, vapor, and air
- dissolution of air in water
- vapor adsorption
- vapor pressure lowering due to suction pressures
C Phase change and interaction
- appearance and disappearance of phases
- interference between liquid and gas phases
- enhancement of gas phase permeability from Knudsen (slip flow) effects
- binary diffusion in gas phase
- vapor-liquid phase change
D. Heat transport
- heat conduction
- heat convection
- sensible and latent heat change.;
- conductive heat exchange with impermeable strata
- coupled fluid and heat flow
E. Dimensionality
- 1-D, 2-D, ard 3-D flows

TOUCH2 SOFTWARE QUAIIHOTION - FF8RUARY 1996 3

 F. Geometry
- Cartesian, cylindrical, irregular
G. Grid generation
H. Heterogeneities
- heterogeneous formations
- flow in porous and fractured media
I. Flexible specification of initial and boundary conditions, sources and sinks
J. Other
- automatic time-stepping
-restart capability
- compatibility with TOUGH
- user documentation
- sample problems to illustrate code use and verify accuracy
- internal version control
- modularized structure for modifying or adding capabilities

2.2 Examples of Verification of Meeting Requirements

Hundreds of applications of the MULKOM family of codes over more than fifteen
years have demonstrated mat the codes, including TOTJGH2, are applicable to complex
problems in nuclear waste disposal, geothermal reservoir processes, environmental
restoration, and that they obtain excellent results. Several successful simulations of field
observations and laboratory experiments, as well as code verifications, were presented at
the 1990 TOUGH Workshop (Pruess, 1990) and at the 1995 TOUGH Workshop (Pruess,
1995). Proceedings of both are included in this package. Many of the significant
applications, particularly those in nuclear waste, are listed below in roughly chronological
order.
An early comparative evaluation of geothermal reservoir simulation codes
established important benchmarks for subsequent code developments (Stanford, 1980).
MULKOM was extensively cross-checked against the SHAFT79 solutions developed for
this DOE code comparison study. An analytical solution obtained by Lauwerier (1955) for
non-isothermal flow in a homogeneous reservoir with conductive heat exchange to caprock
and baserock was closely matched by a MULKOM simulation (Pruess and Bodvarsson,
1984).
Using the MINC (Multiple Interacting gontinua) method, Pruess and Narasimhan
(1985) demonstrated excellent agreement between MULKOM simulations and the Warren
and Root (1963) double porosity solution for isothermal single-phase flow in a naturally
fractured reservoir. Prucss ct al. (1937) developed an analytical solution for the
propagation of a boiling front during water injection into a depleted vapor zone in one-
dimensional radial flow geometry. Numerical simulations with MULKOM closely matched
the analytically predicted front location. Vcrma ct al. (1985) numerically simulated a
laboratory two-phase flow experiment for measuring steam-water relative permeability,
obtaining good agreement with experimental data. Lam et al. (1988) successfully used
MULKOM and the MINC method to model laboratory heat sweep experiments in a
pressure vessel containing a configuration of regularly-shaped granite blocks.

McKibbin and Pruess (1989) presented analytical solutions for steady-state

geothermal heat pipes with C 0 , which closely agreed with results of MULKOM
2

simulations. Wu et al. (1990) developed a Buc;;ley-Lcvcrett type solution for immiscible

displacement in composite porous media. Satisfactory agreement was obtained with

T O U G H 2 SOFTWARE QUALIFICATION - FEBRUARY 1 9 9 6

 numerical simulations using MULKOM-GWF, a specialized version of MULKOM for
three-phase flow of gas, water, and foam (Pruess and Wu, 1988). Doughty and Pruess
(1990) investigated raermohydrologic conditions near heat-generating high-level nuclear
waste packages. For the idealized problem of an infinite linear string of waste packages in
a homogeneous porous medium with water as the single component, they obtained a semi-
analytical similarity solution which was closely matched by numerical simulations with
TOUGH.
Bodvarsson et al. (1990) reported three-dimensional weli-by-well numerical
simulations with MULKOM of the Olkaria, Kenya geotheimal reservoir. Using actual
flow rates and enthalpies from well data for the period from 1977 to 1983 to calibrate the
reservoir model, they developed performance predictions that matched observed behavior
of most wells from 1984 to 1987.

Falta and Pruess (1991) and Falta et al. (1992a. b) developed an enhanced version
of TOUGH for three-phase component flow of water, air, and volatile organic compounds,
known as STMVOC. They modeled steam injection experiments in one-dimensional
laboratory cores performed by Hunt et al. (1988) and successfully predicted the migration
of the seam condensation front, as well as profiles of temperature and pressure, and the
removal of rrichloroethylene from the column over time. Doughty and Pruess (1992)
extended their earlier work on a similarity solution for fluid and heat flow near a linear heat
source to a pore fluid consisting of water and air. The only simplifications made in their
treatment involved flow geometry. The process complexities of highly non-linear, fully
transient, two-phase, two-component fluid and heat flow with phase change were
rigorously accounted for, making this the most difficult solvable two-phase fluid and heat
flow problem presently available. Numerical simulations with TOUGH2 showed excellent
agreement with the semi-analytical similarity solution.

Based on the observation that vapor pressures in vapor-auininated geotherma!

reservoirs under undisturbed conditions are close to saturated vapor pressure for measured
temperatures (and thus contain water as both vapor and liquid phases), Pnicss and
O'Sullivan (1992) performed numerical simulations to investigate the nature and strength of
fluid retention from capillary suction, vapor adsorption, and vapor pressure lowering.
They ran the calculations with TOUGH2, which models vapor pressure lowering effects by
means of the Kelvin equation.

2.2.1 Applications of TOtJGH2 to Nuclear Waste Disposal

One motivation in developing the TO0GK2 code (Pruess, 1991) was from the need
of performance assessment in the high-level nuclear waste disposal project. TOUGH2 has
been used in many projects worldwide related to high-level nuclear waste projects from
basic research to applications (Pruess, 1990 and 1995). The TOUGH/TOUGH2 code was
used in (he early research efforts in studies of unsaturated zone flow in Yucca Mountain
(Rulon et al., 1986). In recent years, TOUGH2 has been used to develop the site-scale
unsaturated zone model for moisture, gas and beat flow in Yucca Mountain (Witrwer et at.,
1992,1993,1994, and 1995; and Bodvarsson et al., 1994). Furthermore. TOUGH2 has
also been used to perform sensitivity studies of Yucca Mountain to understand effects of
the ambient natural conditions on the mountain behavior (Wu et al., 1995; Ahlers ct al.,
1995a, 1995b). Associated with an inverse modeling tool (1TOUGH2, Finsterle, 1993),
TOUGH2 has been used to analyze ventilation experiments at the Grimsel Rock
Laboratory, Switzerland (Finsterle and Pruess, 1995) and fo estimate the nydrogeological
parameters, infiltration rate, water table, surface and lateral boundary conditions (Wu et a!.,
1995).

TOUCH2 SonwARE QUALIFICATION - KBRUARY 1996

 The general capability of TOUGH in handling fracture/matrix flow (Fruess, 1991)
makes it useful in simulating groundwater infiltration and percolation within the welded
unit at Yucca Mountain. It has been used for groundwater (ravel time calculations for the
Yucca Mountain Site Characterization Project (Ho et al., 1994; and Arnold ct ai., 1995).
Niemi and Bodvarsson (1988) addressed the question of how hystcrctic (history-
dependent) capillary pressure-liquid saturation relations may affect flow and liquid
saturation in a fractured rock system such as at Yucca Mountain. They used a hysteresis
model to simulate a system consisting of discrete fractures and rock matrix under periodic
infiltration pulses. They used material property values for densely welded tuffs at Yucca
Mountain. Their mode! showed strongly hysterctic behavior in the uppermost layer of the
matrix, which generated higher fracture flows and a more "smeared" matrix liquid
saturation versus depth distribution for the hysteretic case.

TOTJGH2 has also found wide applications in studies of thermobydrologic

phenomena associated with thermal loading issues in high-level nuclear waste disposal
(Witherspoon et a!., 1996). TOUGH2 has been extensively used in thermal loading
studies and performance assessment of high-level nuclear repositories (Prucss et al., 1990;
Pruess and Tsang, 1993 and 1994; and Wu et a!., 1995).
TOUGH2 has performed well on a series of fluid and heat flow problems that
involved one, two, and three-dimensional flows, with varying degrees of non-linearity,
coupling between fluid and heat flows, and complexity of boundary conditions. The
verification process at this stage of the Software Life Cycle consisted of testing the
MJJLKOM family of codes, including TOUGH2, over a period of years on the types of
problems described above. These results, as well as verification and validation problems
for the MULKOM/TGUGH codes, substantiate the accuracy of the process description
employed in the code, and the mathematical and numerical approaches used. The latter two
topics are discussed in the following section. Validation tests are discussed in Section 4.

TOUGH2 SOFTWARE QUALIFICATION . FEBRUARY 1996 6

3. DESIGN DESCRIPTION
This section describes the formal structure and architecture of TOUGH2. It details
the functional requirements of the software and how they are implemented. It explains how
TOUGH2 handles the physics of the processes modeled. It provides a description of
numerical models and methods and the five fluid property (equation-of-state or EOS)
modules. This section also describes the software structure, logic, and data structure and
flow. It includes performance requirements and design constraints.

3.1 Background on MULKOM/TOUGH Family of Codes

TOUGH2's predecessor is the code MULKOM f'MJLti-KQMponent"; Pruess,
1983b). MULKOM was built on the recognition that the governing equations for non-
isothermal, multi-phase, multi-component flow axe the same, regardless of the nature and
number of fluid components present. MULKOM featured a modular architecture, in which
a central module for flow and transport could be interfaced with several different fluid
property modules to model systems with different fluid mixtures. The coding of
MULKOM was never finalized but instead evolved into a collection of program modules
for specialized applications. Nuclear waste-oriented applications spurred an effort to
finalize and document a version of MULKOM for non-isothermal two-phase flows of
water and air, which became TOUGH (Transport Q$ Unsaturated groundwater and Heat)
(Pruess, 1987). The methodological choices made in the development of MULKOM and
their relationship to an earlier code, SHAFT79, were discussed by Pruess (198S).
TOUGH2 is a successor to TOUGH (Pruess, 1991). It is subroutine for
subroutine very similar to TOUGH, but offers additional features such as internal mesb
generation, an interna! version control system, more convenient means of specifying
boundary conditions, and an embedded fiactured-media capability with the MMC method
of'Multiple Interacting £ontinua". Unlike TOUGH, TOUGH2 provides the full multi-
component, multi-phase flexibility previously available only in the undocumented
MULKOM collection of modules. The name MULKOM is now used to refer to the
particular modular architecture of the codes, while actual coding implementations are
referred to as TOUGH and TOUGH2 (Moridis and Pruess, 1992).

The TOUGH2 family of codes includes: I) TOUGH2 for two-phase multi-

component fluid and heat flow with phase change; 2) T2DM, a TOUGH2 module for
Fickian hydrodynamic dispersion (Oldenburg and Pruess, 1993); and 3) T2VOC (Falta et
al., 1995) and M2NOTS (Adenekan, 1992; Adenekan et al., 1993) for three-phase multi-
component fluid and heat flow (see Figure 1).

3.2 Structure and Architecture

TOUGH2 implements a flexible, general-purpose architecture (see Figure 2) for
simulating fluid and heat flow in systems in which any number of components or species
can be distributed among several coexisting phases. A key feature of the code architecture
is an array structure that allows for flexible interfacing between the module that sets up and
solves the fluid flow equations and the equation-of-state (EOS) modules, which represent
fluid mixtures with different numbers of components and phases (see Table 1). Figure 3
provides a more detailed flow diagram of the code structure.

T O U G H 2 SOFIVKAR6 QUALIFICATION - FEBRUARY 1 9 9 6

 Baalnput t
and J
InltMation

«tr*r ;
SoluSono! Assvmblingand VSUOM ', Equation •
Linear Itcrathra Solution of of :
Suasions Flaw Eifliaft-ns fooondaiy" Stale :
PmmMr*

j Printed
] Output

!., ,.J
•BSSVUMC
m> H M I U

Figure 2. Modular architecture ofMULKOM and TOUCH2.

Table I. T0UGH2 Fluid Property Modules

MODULE CAPABILITIES
EOS1 water, water with tracer*
EOS2 water, C02
T
EOS3 water, air*
EOS4 water, air, with vapor pressure lowering
EOS5 water, hydrogen*
•optional constant-temperature capability
'similar to the EOS-raodule of TOUGH

User features includeflexibledimensioning of major arrays, capabilities for internal

processing offlowgeometry through automatic mesh generation (via MESHMAKER), and
enhanced abilities for specifying initial and boundary conditions. TOUGH2 input formats
are compatible with those of TOUGH, allowing TOUGH input files to be usee! wit'-.
TOUGH2. However, some default parameter settings are different in TOUGH2, so thai
minor adjustments in TOUGH input files may be in order.
Much of what program units do is spelled out in internal comments and in printed
output. TOUGH2 provides a tight version control system for meeting stringent demands
on reliability and referenceability of code applications. Each program unit, when first
called during a TOUGH2 simulation run, writes a one-line message specifying its name,
version number and date, and purpose. All version messages are optionally printed to
OUTPUT at the cud of a simulation run. An example of an OUTPUT file for the end of a
run is shown in Figure 4. This lists all of the program units used in the simulation, their
version number and date, and what they do. Users who wish to modify the code can

TOUCH2 SOFTWARE QUALIFICATION - FEBRUARY 1996 8

therefore maintain an easily referenceable record of code changes and applications by
updating version messages for each program unit they change for internal use. The coding
complies with the ANSIX3.9-I978 (FORTRAN 77) standard.

Simplified How Chart of TOUGH2

-define parameters
TOTJGK2 -set up common blocks
(main program) •serve as main executive routine
I -open and assign
* 1 IO disk files

i
1_

-read input data

MESHM: -mesh generation
RZ2D,WKZ3X>,
PRZ2D
(optional)

initialize arrays for mesh, generation, and

initial condition data

EOS satellite
routines:
SAT.COWAT,
SUPST.VEW,
VKCO.COVIS,
VJSS.RELP.FCAP

f» 1
,-•—*-•>. /«TER«fl \ . set iteration and time step
S T O r
V^ -> < KCYC=KCYCri> counter and convergence flafi

. KCftfrL.
new iteration

Figure 3. Structure o/TOUGffi.

TOUCH2 SOFTWARE QUALIFICATION - FEBRUARY 1996

 (eootiaiKd from previous pase)

1
j-selupsinkano
QU | source terms
-set up
accumulation ami tiow tenns;
-calculate Jacobian matrix
-check on convergence

YES

linear equation
solver (direct or
preconditioned
conjugate gradients)
increment primary variables

-detailed printout of
thermodynamic
conditions and flow
rates
-mass-mul volume
averages

Figure 3 (conl.). Structure of TOUGH2.

TOUGH2 SOFTWARf QUAIIFICATION . FEBRUARY 199fi 10

 flHIIIIHHIH«ltll(wlUtllt1HlHltiMIIIIMIItlHIIIHIlMIMtMMItMIII(IIIIIHIII*lt*lf»H*fll*Mtf4f*HIIMIHIHll4ltfMMIt

« s m r cr ruewi airs ISCB •

* •
•milfllllHMI "<ltHllllUllllllllHIII|l1IHHIH||,||| | ||, Hllli iMHllllllllltlilltllllltlllMHH4lrtHllttHll
MI t ll a

KIT \-asai Vui comais

is i.o !5 inn. mi ora m o ivctsi, H O I , iwso. una; IIMEI, MKSI, m inxn

n i.o »KSK mi MIK ISKJAH
is»«t i.o - II IHR mi HUD tu MU mvi» mtau FIIE .IUVI.
res« i.o ii nr im EstsmwnniicnituicwiieaciianiBi
«rs i.g tutu im o u s t s 2-1 M i aratiminin
nia i.o ijwsa mi umHUFaMi-isaonnuiiEsi
Km i.o nuioi ini lucsHtiuammwiof H I - ! I C »
ru» I.O iiiren im otouit WQE» of timintwi m m m mumranutmnic
«f lit i.o a W«IL mi iimtiuE Mn m FUES «tsw M MUC., <toc», w <i«n>
era i,s saira mi nraiire ran* r» munis :> rat
EOS 1,0 78, *HH mi MEM ... KHOFStJIOL nSKUKS UaOE FM HOT/HI
at 1,0 annum mo trowIUIEOBIIIK urauiianasm«raciianf TWEUWE
cout i,« a wa»i im UHII WTE* m m m m. new u mcuot OFramuwtU> MCSUE
OKI i.< »;«*ur mo v«f«Kxsm™imiw<.£Hiii««f»etio<ofrareuiUEHtrcsaK
via 1,9 swum mo tisetnoruoninTDiisrjcnoiOfitTOuiwsMmtssK
TIKI i.o inrai mo aiaurE mceirr sr wows snira
cvis i.o inustti »» OEFfictEir tat us WBE nscsm s w u m
sin i,o aamr mo usaunwv»«uniciiao.-mt£uiti:«ioresaiz
tat i,« swan mo iiii/ii«»U5«i!fi«Eun«KW«uma»!H>ni«<f aiuunw
tat i.» muta mi cviiua WSSUUFWCIIMIFUUMTIW
UIU I.I SWJ mi KiFKasiattriKMeEt rants, MS, no DOST
TSET i.o *i»a mi iKSMin trtn TO mere tiTi raj-ic:*: :mtn lira
mil i.o sua mi KSOXI <ii « a t u t t « IU n o CTa
m i.t BJSBUT mo USEOUEIU.towsuismuts
UKO i.i sjimr mo i«Et»tE nt K ui«> EMIIM soivtt hu<
Kl« KM«U.»l«ro£fMlttll«liirui
BKVM t,« lira mt mmmtmiauaBmaeaaoaaauao
n l.o ifotra itto cauauicnrnmsnc, wesrf.ir. m j a B r i w i
ssi i.o sun mi n m KSXIS ran iiaan, tatBiias, msiattauoss
nam i.o mow mo • • « o»inio« OF • nwc «*, mm KUMUT VUIUUS I« FIU tsnn

Figured. Summary of program units andversion history (from Pruess, 1991; Fig. 14)

3 . 3 Physical Processes a n d Approximations

Development o f the TOUGH simulator was motivated by problems involving
strongly heat-driven flow, for which conventional approaches to describing unsaturated
flow are not applicable (Narasimhan, 1982). A s temperatures approach or exceed the
boiling point of water, vaporization takes place with associated increases in vapor partial
pressure and strong forced convection of the gas phase. To describe these phenomena it is
necessary to employ a multiphase approach to fluid and heat flow, which fully accounts for
the movement of gaseous and fluid phases, their transport of latent and sensible heat, and
phase transitions between liquid and vapor. The gas phase in general will consist of a
mixture of water vapor and air, and both components must be trackcii separately.

TOUCH2 SOnVMRC QUAUFICATION - FEBRUARV 1996 11

 The TOUGH2 simulator takes account of the following physical processes with the
governing equations provided in Appendix A. Fluid flow in both liquid and gaseous
phases occurs under pressure, visco.-s, and gravity forces according to Darcy's law, with
interference between the phaser re. jseited by means of relative permeability functions.
Mass- and energy-balance equa. MIS are written in integral form for an arbitrary flow
domain. In addition, TOUGH2 considers binary diffusion in the gas phase. It accounts
for a simplified description of Kr jdsen diffusion by means of a Klinkenberg factor for
permeability (Knudsen, 1909; Klinkcnberg, 1941; Hadley, 1982). Capillary pressure,
vapor adsorption, and vapor pressure lowering effects are taken into account for the liquid
phase.

The transport equations are complemented with constitutive relationships, which

express all parameters as functions of a set of thermodynamic variables. All
themiophysical properties of liquid water and vapor are obtained within experimental
accuracy from steam table equations (International Formulation Committee, 1967). Air is
treated as an ideal gas, and additivity of partial pressures is assumed for air/vapor mixtures.
The viscosity of air/vapor mixtures is computed from a formulation given by Birschfeider
et a!. (1954), using steam table values for vapor viscosity. Air solubility in liquid water is
represented by Henry's law. However, because air solubility in water is very small, the
temperature dependence of Henry's constant is neglected. The value implemented in
TOUGH2, K„=10'°Pa, is accurate to within ± 1095- in the temperature range from 40°C to
I0O"C (Loomis, 192S). Data for air solubility in water at higher temperatures has not been
published and is not well known.

Heat transport occurs by means of conduction, wifli thermal conductivity dependent

on water saturation; convection; and binary diffusion, which includes both sensible and
latent heat. The description of Uieimodynamic conditions is based on assumed local
equilibrium of all phases. (MULKOM has also been used to model non-equilibrium
conditions, such as chemical reactions).
The governing equations used in TOUGH2 and their numerical implementation, are
applicable to one-, two-, or three-dimensional anisotropic porous or fractured media.
TOTJGH2 does not perform deformation analysis of the host medium and its effects on the
multiphase flow behavior, but it allows for porosity changes in response to changes in pore
pressure (compressibility) and temperature (expansivity).

3.4 Mathematical and Numerical Methods

Continuum equations for mass- and energy-balance are discrctized in space using
the "integral finite difference" method (Edwards, 1972; Narasimban and Witherspoon.
1970). (See Appendix B for a full derivation of methodology). Introducing appropriate
volume averages, we have

J*Mdv = V„M»
V.

Here M is a volume-normalized extensive quantity, and M„ is the average value of

M over V„. Surface integrals are approximated as a discrete sum of averages over surface
segments. Time is discretized fully implicitly as a first order finite difference to obtain the
numerical stability needed for an efficient calculation of multi-phase flow. Five different
weighting schemes for mobility averaging options are available for selection by the user.

TOUOK2 SOFTWARE QUALIFICATION - FEBRUARY 1996 12

 The five weighting schemes include: (1) upstream weighting of both mobility and
permeability; (2) upstream weighting of permeability and averaging of mobility between
adjacent elements; (3) upstream weighting of mobility and harmonic weighting of
permeability; (4) arithmetic averaging of mobility and harmonic weighting of permeability;
and (5) harmonic weighting of the product of mobility and permeability. For each volume
element (grid block N) there are a number of primary variables, the choice of which
depends upon the phase composition needed to define the thermodynamic state of the flow
system and therefore on the choice of EOS module. A coupled set of algebraic equations is
obtained (see Pruess, 1987). The total number of mass- and energy-balance equations is
equal to the product of the number of primary variables times the number of grid blocks.
The equations are strongly coupled because of interdependence of mass and heat flow.
They are highly non-linear, because of order-of magnitude changes in parameters during
phase transitions, and because of non-linear material properties such as relative
permeabilititcs and capillary pressures. Because of these features of the equation system,
TOUGH2 performs a simultaneous solution of the discretized mass- and energy-balance
equations, taking all coupling terms into account. Non-linearities are bandied by
performing the Newton/Raphson iteration. The unknowns are the independent primary
variables, which completely define the thermodynamic state of the flow system at time level
t '*'. Clarification is provided in die discussion of EOS modules, to follow.

Upon application of the Newtort/Raphson iteration (see Pruess, 1987), a set of

coupled linear equations is obtained. These can be solved with the direct solver MA23
(Duff, 1977) or with a preconditioned conjugate gradient solver, as discussed below.
Iteration is continued until the residuals are reduced to a small fraction of the accumulation
terms. See Appendix B for convergence criteria. Convergence is usually obtained in three
to four iterations. If convergence cannot be achieved within a certain number of iterations,
the time step size Ai is reduced and a new iteration process is started. AU derivatives
needed in the coefficient matrix are obtained by numerical differentiation. The resulting
Jacobian matrix is solved to yield the changes in the primary variables in each element of
die discretized domain from the previous iteration.

The entire geometric information of the space discretization is provided in the form
of a list of grid block volumes, interface areas, nodal distances, and components of
gravitational acceleration along nodal lines. The discretized equations are valid for arbitrary
irregular discretizations in one, two, or three dimensions, and for porous as well as for
fractured media. This flexibility should be used with caution, however, because the
accuracy of solutions depends upon the accuracy with which the various interface
parameters can be expressed in terms of average conditions in grid blocks (sec Pruess,
1987). A necessary condition for this is approximate thermodynamic equilibrium in almost
all grid blocks at almost all times (Pruess and Narasimhan, 1985). For systems of regular
grid blocks referenced to global coordinates, the equations reduce to a conventional finite
difference formulation (Peaceman, 1977).

In the original TOUGH2, the Jacobian was solved using MA28 (Duff, 1977) a
direct solver using sparse matrix storage techniques. However, storage and execution time
requirements of MA28 limited the use of TOUGHS to a maximum of a few thousand
equations. To address this limitation, Moridis and Pruess (1995) added T2CG1, a package
of preconditioned conjugate gradient solvers to complement TOUGH2's direct solver and
significantly increase the size of tractable problems. The conjugate gradient solvers
decrease execution time and memory requirements substantially, and make possible the
simulation of three-dimensional flow problems with tens of thousands of grid blocks on
workstations and PCs.

TOUCH2 SOFTWARE QUALIFICATION - FEBRUARY 1 9 9 6 13

 3.5 Array Structure and Handling
The number of primary thermodynamic variables that is needed to specify
completely the thermodynamic state of a flow system consisting of a number of NK
components which are distributed according to local thermodynamic equilibrium among a
number of NPH phases is NK+1. This is equal to the total number of balance equations
per grid block, namely NK mass balance and one energy balance equation. The
thermodynamic state of a discretized flow system containing NEL (number of elements)
volume elements or grid blocks is then completely specified by a set of NEL*NK1 primary
thermodynamic variables, to which correspond an equal number of mass and energy
balance equations. For transient flow systems, these primary variables are time-dependent,
and they represent the unknowns to be calculated in each time step. In TOUGH2 all of the
NEL*NK1 variables are stored sequentially in a one-dimensional array X: first (lie NK1
variables for grid block #1, then the NK1 variables for grid block #2, and so on (see
Figure 5). The starting location for primary variables for grid block N is NLOC+1, where
NLOC = (N-1)*NK1. More detail on handling primary thermodynamic variables is found
on pages 9-11 in Pruess (1991).

The choice of primary variables that define the thermodynamic state of the system
when phase change is involved in fluid and beat flow processes is an important
consideration. When a phase appears or disappears, the set of appropriate thermodynamic
variables may change. This problem can be dealt with in two ways. One way is to use a
set of "persistent" variables such as pressure and enthalpy, or density and internal energy,
which remain independent even as phase conditions change, so that they can be used
throughout the single and two-phase regions. The other possibility is to use the variables
pressure and temperature only for single-phase conditions and to switch to variables such
as pressure and saturation when a transition to two-phase conditions occurs. Experience
has proven this variable-switching approach to be a robust method for treating multiphase
systems and it has been implemented in the MULKOM, TOUGH, and TOUGH2 codes.
There are two additional arrays DX and DELX with structure identical to X. While
X holds the primary variables corresponding to ±c last successful (converged) time step,
DX holds the latest increments calculated during the Newton-Raphson iteration process.
Thus the latest updated primary variables are the quantities X+DX. The array DELX holds
7
small increments of the X themselves (typically on the order 10" *X) which arc used to
calculate incremented parameters needed for the numerical calculation of the derivatives in
the Jacobian matrix J = -3R/<)X, (see Figure 6; x, denotes the collection of all primary
independent thermodynamic variables). At the conclusion of a converged rime step, the
primary variables X arc updated, X —>X+DX.

TOUGH2 SOFTWARE QUALiriCATION - f£8RU/«V 1996 14

 now EOS
Modal:
' Mk

VOUMS MUMLY ttcotrejmrMiuMFrau

ELEMENT VAWA1LEJ

1 s nam
i. MM
xNcn i> IfflW
*1 p WW
« nn
p. PMW

j X" PJWOO+1)

Kt XJtOQCO)

X040ONK1}
MHOSEO

IKEi.
XOffilWXO

»KgiBq*irtgBO

figure 5, Structure ofihermopkysical property arrays in MULKOM and TOVGH2.

/ Ax = R
•1
* !
NKJ M»
n MQ*>

«0 21«
1
!
j 0MTWO1
Mi *...•-*.
Rft

Fiffw 6. Linear equation structure.

TOUCII2 SOFTWARE QUALIFICATION . FEBRUARY 1996

 As stated above, the number of mass- and energy-balance equations per grid block
is the same as the number of NK1 primary thermodynamic variables. In many
applications, however, the heat effects may be so* small that temperature changes would be
insignificant, and it would be sufficient to consider just the mass balance equations. The
simplest way of forcing temperature changes to zero would be to assign the solid matrix an
overwhelmingly large heat capacity, so that finite rates of heat exchange will cause
negligible temperature change. This approach is possible with TOUGH2, but it has the
disadvantage that the full number of balance equations must be solved, even though the
energy balance reduces to the trivial statement oT/3t =0. For certain EOS modules in which
temperature is a primary variable, TOUGH2 offers a more elegant way of running
problems without temperature changes which saves considerable computing time. In this
method, TOXJGH2 uses a parameter NEQ, distinct from NK, to number the balance
equations per grid block, with the normal case being NEQ= NK+1. However, the user can
choose to assign NEQ=NK in the INPUT file; then no energy equations are set up or
solved, and the number of coupled equations per grid block is reduced from NK+1 to NK.

The EOS module calculates all thermophysical properties (secondary parameters)

needed to assemble the mass- and energy-balance equations for the latest updated primary
variables X+DX. These parameters are then stored sequentially in a large array "PAR" (see
Hguie 5). The number of secondary parameters other than component mass fractions is
NB (usually NB=6); in addition there are NK mass fractions so that the total number of
secondary parameters per fluid phase is NBK=NB+NK. The PAR array structure is
shown in Figure S for the case of two fluid phases; however the coding permits any
number of phases, as specified by the parameter NPH. The NPH*NBK phase-specific
parameters are followed by temperature T and a void (unused) array member, so that the
total number of secondary parameters is NSEC=NPH*NBK+2.

Note that the thermophysical properties are needed not only for calculating the
residuals of the mass- and energy-balance equations, (Appendix B, Equation B.6) but also
for calculating their derivatives in the Jacobian matrix (Eqs. B.7 and B.8) Thus, secondary
parameters are required not only at the "state point" (latest X+DX), but also for the NEQ
additional sets of primary variables in which one of the primary variables at a lime is
incremented by DEjLX. Therefore, the total number of secondary parameters per grid block
is (NEQ+1)*NSEC. Secondary parameters for grid block #N start after location
#NLOC2=(N-l)*(NEQ+l)*NSEC of the PAR array. More detail on thermophysical
property array structure is found on pages 12-14 in Pruess (19°1).

3.6 Linear Equation Setup

The data provided by the FAR-array are used in the flow module of TOOGH2 fo
assemble the linear equations (Eq. B.8) that ate solved at each step of the Newton-Raphson
iteration procedure. These equations are arranged and numbered sequentially, as shown in
Figure 6, with the first NK equations per grid block representing component mass
balances, while the last equation (#NK1) represents the energy balance. The row indices
of the Jacobian matrix correspond to the sequence of primary variables in array X. If the
option NEQ^NK is chosen, only NK mass balance equations will be set up per grid block,
la this case only the first NK primary variables per grid block will contribute matrix
columns, while variable #NKI, which must be temperature, remains passive and is not
engaged or altered in the linear equation handling. However, all thermophysical parameters
will be calculated at the temperature values specified in variable #NK1.

TOUCH2 SOFTWARE QUAliriCATtON - rtBRUARY 1996 16

 Note that the accumulation terms of the balance equations depend only on primary
variables for one grid block, so that they will generate non-zero derivative terms only in an
NEQ*NEQ submatrix that is lor-ated on the diagonal of the Jacobian J. The flow terms,
being dependent on primary variables of two grid blocks, will generate two non-2ero
NEQ*NEQ submalrices of derivatives, which are located in the off-diagonal matrix
locations corresponding to the two grid blocks.

In TOUGH2 all Jacobian matrix elements as well as the entries in the vector R of
residuals are calculated in subroutine "MULT1". The calculation first assigns all matrix
elements arising from the accumulation terms, of which there are NEQ*NEQ. These arc
stored sequentially in a one-dimensional array CO ("coefficients"); matrix elements for grid
block N begin after location (N-1)*NEQ*NEQ in CO. The corresponding row and column
indices are stored separately in arrays IRN and ICN, respectively. Calculation of the
derivatives demands that each accumulation term is calculated NEQf-1 times; once for the
state point (X+DX), and NEQ limes for each of the NEQ primary variables incremented
(X+DX+DELX). Additional contributions to diagonal terms in the Jacobian J may arise
from sink and source terms, if present; these are assigned in subroutine QU called from
MULTI. Subsequently all flux terms are evaluated. These depend in general on the
2*NEQ primary variables of the two connected grid blocks, so that a total of 2*NEQ+1
flux terms need to be evaluated for calculation of the state point as well as of all derivative
terms.
After all matrix elements and members of the right-hand side vector of residuals
have been preconditioned, the subroutine package MA2S (Duff, 1977) or tbe package of
preconditioned conjugate gradient solvers (Moridis and Piuess, 1995) is called to solve the
linear equations (Eq. B.8). The resulting increments in the primary variables are added to
the array DX, and the process of linear equation setup and solution is repeated for the
primary variables X+DX. This process continues until the residuals are reduced below a
preset convergence tolerance. If convergence is not achieved within a specified maximum
number of iterations (default 8), the time step is repeated with reduced time, increment.

3.7 Dimensioning of Major Arrays

The major problem-size dependent arrays reside in COMMON blocks, which are
dimensioned by means of a PARAMETER statement in the main TOUGH2 program. An
informational statement on permissible problem size (number of grid blocks, etc.) is
provided in the printed output of a TOUGH2 run. When problem specifications exceed
array dimensions the execution stops with a diagnostic printout. The user must then
increase PARAMETER assignments, recompile the main program, and relink. A lis! of
major arrays used in TOUGH2 with their dimensions is given in Table 2.

TOUGII2 SOFTWARE QuAitriCATiON - FEBRUARY 1 9 9 6 17

 Table 2. Summary of Major Common Blocks
REFERENCE TO COMMON BLOCKS LENGTH
Elements E1-E6 NEL (=aumber of elements)
YTNWES, AHTRAN
Primal}' variables P1-P7 NEL'NKl'
Connections ci-cn NCON (^number of connections)
(interfaces) COMPS. PORVEL NCON'NPH
Linear equations LI SNZ=(NEL+2*NCON)*NEQ**2
L2JL3 (2-4) *NZ
L4 NEL'NEQ
L5 NEL*5*NEQ
L6 NEL*8*NEQ
1.7 NZ+5*NEQ»NEL+32
L8 NEL*5*NEQ
COARA6 NZ+32«NEQ»NEL+1000
Secondary parameters SECPAR NEL*(NEO+n*NSEC
*NKfcNK+!

3.8 Equntion-of-State Modules

The tberrnophysical properties of fluid mixtures needed in assembling the
governing mass- and energy-balance equations are provided by equation-of-state (EOS)
modules. The MULKOM (and hence TOUGH2) formulation in the general sense
accommodates any number of fluid components and phases that may fee present, but the
number of phases and components modeled is determined by the EOS module that is
selected.
Besides providing values for all secondary (thermophysical) parameters as
functions of the primary variables, the EOS module must fulfill three additional functions:
1) the phase conditions pertaining to a given set of primary variables must be
recognized (element-for-element),
2) the appearance and disappearance of phases must be diagnosed as primary
variables change during the Newton-Raphson iteration process, and
3) primary variables must be switched in response to a change of phase.
The primary variables/secondary parameters concept as implemented in MULKOM
and TOUG1-I2 eliminates any direct connection between the choice of primary variables and
the secondary parameters that are used to set up the flow equations. This provides
maximumflexibilityin the choice of primary variables because only secondary parameters
are used in the flow equations. Only one exception to this separation exists; pressure of a
reference phase is by convention always the first variable and it is used directly in the flow
equations. The choice of all other primary variables is open. Note that only one EOS
module at a time should be liniced with other TOUGJH2 modules.
3.8.1 EOS! (Water. Water with Tracer): This is the most basic EOS module. It
provides a description of pure water in its liquid, vapor, and two-phase states, and has a
capability of representing two waters of identical physical properties which contain
different trace elements. The default parameter settings for a single water component are
(NK, NEQ, NPH, NB) - (1, 2, 2, 6). The option NEQ = 1 is available for running
problems that involve only liquid water, or only superheated steam, under constant

TOIJCH2 SOfTWARE QUAI IFICATiON - fFRRIIARY 1 996 IS

 pressure conditions. The primary variables are (P, T) for single-phase points, and (P , S )
c c

for two-phase points. For the convenience of the user, it is possible to initialize two-phase
points as (T, S ); a numerical value of the first primary variable less than 374.15 will
t

automatically be taken to mean temperature (in °C) instead of gas pressure, and will cause
variables to be internally converted from (T, S ) to (P„.(T), S.) prior to execution.
5

The two-waters capability can be invoked by specifying (NK, NEQ, NPH, NB) =
(2, 3, 2, 6) in data block "MULTT* (see below). With tiiis option, two water mass
balances will be set up, allowing separate tracking of the two components. The primary
variables in this case are (P, T, X) for single-phase points, and (P , S , X) for two-phase
e g

points, where X is the mass fraction of Water 2 present. All thcrmophysical properties
(density, specific enthalpy, viscosity) are assumed independent of the component mixture:
i.e., indepsndent of the mass fraction X. This approximation is applicable for problems in
which the identity of different waters is distinguished by the presence of different trace
constituents, which occur in concentrations low enough to have no effect on the
thermophysical properties.
All water properties (density, specific enthalpy, viscosity, saturated vapor pressure)
are calculated from (he steam table equations as given by die Internationa! Formulation
Committee (1967). See Figure 7 and Table 3 for ranges. The formulation includes
subragion 1 (subcooied water below T * 350"C, subregion 2 (superheated steam) and
subregion 6 (saturation lice up to T = 350°C). Within these regions, density and internal
energy are represented within experimental accuracy. Viscosity of liquid water and steam
are represented to widain 2.5% by correlations given in the same reference. (Refer to the
original publication for details).

Figure 7. Illustration ofSubrcgions on the Pressure-Temperature Diagram

(from International Formulation Committee^ J967).

TOUCH2 SOFTWARE QUAllflCATION - FEBRUARV 1 9 9 6 19

 Table 5. Ranges oJThermophysical Parameters for Phase Regions in Figure 7.
Subrcgion Temp (°C) Sat Vapor Pressure
1 0.01-3S0 0.0061-1000
2 350-374.15 165.4-221.2
3 0.01-374.15 0-165.4

The phase diagnostic procedures are as follows. When initializing a problem, each
grid block has two primary variables (XI, X2). Whether X2 means gas saturation (two-
phase) or temperature (single phase) is decided from the numerical value: for X2 > i.5, X2
is taken to be temperature in °C, otherwise it is gas saturation. (Although physically
saturation is restricted to the range 0 < S < 1, it is necessary to allow saturations to exceed
1 during the Newton-Raphson iteration). If X2 is temperature, single phase conditions
exist; specifically, for P (=X1) > P (T) there is single phase liquid; otherwise there is
SJt

single phase steam. After initialization, the phase condition is ideiiliried simply bised on
the value for S , as stored in the array PAR. S = 0: single phase liquid; S = 1: single
e e c

phase vapor, 0 < S. < 1: two-phase.

Phase change is recognized as follows. For single phase points the temperature
(second primary variable) is monitored, and the corresponding saturation pressure is
compared with P. For a vapor (liquid) point to remain vapor (liquid), it is required that P <
P ^ O ^ P J ; if this condition is violatei a transition to two-phase conditions takes place.
The primaiy variables are then switched to (P., S.) and these are initialized as P = P„,(T),
e

S = 0.999999 if the point was in the vapor region, and S, = 0.000001 if it was in the
t

liquid region. For two-phase points S is monitored; it is required that 0 < S. <l for a
£

point to remain two-phase. If S < 0 this indicates disappearance of the gas phase; the
s

primary variables are then switched to (P, T) and the point is initialized as single phase
liquid, with T taken from the last Newton/Raphson iteration, and P = 1.000001 * P„(T).
For S > 1 the liquid phase disappears; again the primary variables are switched to (P, T)
and the point is initialized as single phase vapor, with T taken from the iast
Newton/Raphson iteration, and P = 0.999999 * P^,(T). In these transitions temperature is
preserved, rather than pressure, from the last iteration. A summary of EOS1 specifications
and parameters is given in Tabic 4 below.

T O U O H 2 SOFTWARE QUALIFICATION . FEBRUARY 199G 20

 Tabic 4. Summary ef EOS J
COMPONENTS #1: water
#2: "water 2" (optional)
PARAMETER CHOJCES

(NK,NEQ,NPH,NB)= (1,2,2,6) one waier component, non-isotbennal (default)

(1,1,2,6) only liquid, or only vapor, isothermal
(2,3.2.6) w o waters, non-isothermal*
PRIMARY VARIABLES
single phase conditions
(P, T, PQ)-(pressure, temperature, [mass fraction of water 2]*)
two-phase conditions
(P,S,,EX])-(gas phase pressure, gas saturation, (mass traction of water 2!')
t

*two waters cannot be run in isothermal mode, because in this case temperature is not the last primajy
variable
'optional, for NK=2 only

3,8.2 EOS2 (Water. CO-): This fluid property module was developed by
O'Sullivan et al. < 1985) for describing fluids in gas-rich geothennal reservoirs, which often
contain CO, mass fiactions from a few percent to occasionally 80% or more (Atkinson et
al., 1980). "It accounts for non-ideal behavior of gaseous C 0 , and dissolution of CO, in
2

the aqueous phase according to Henry's law with heat-of-solution effects. The
thermophysicai property correlations are based on the model of Sutton and McNabb
(1977); a formulation from Pritchett et al. (1981) is used for the viscosity of vapor-C0 2

mixtures. Specifications and psaf; iters of EOS2 are summarized in Table 5.

TableS. Summary of EOS2

COMPONENTS *1: water
*2: CO,
PARAMETER CHOICES
(NK,KEQ,NPH,NB)- (2,3.2.6) no other options are available
PRIMARY VARIABLES
single phase conditions
(P. T, PCOjHprcssure, temperature, C0 , partial pressure)
2

twe-phase conditions
(P..S.,PCOV( gas phase pressure, eas saturation. CO, partial pressure)
?

3.8.3 BOS3 (Water. Airt: This module is an adaptation of She EOS module of
TOUGH for the TOUGH2 program, and implements the same thcnuophysical properties
model (see Pruess, 1987). All water properties are represented by the steam tabic
equations (International Formulation Committee, 1967). Air is approximated as an idea!
gas, and additivity is assumed for air and vapor partial pressures in the gas phase, P = P,
+ P . The viscosity of air-vapor mixtures is computed from a formulation given by
v

Hixschfelder et al. (1954). The solubility of air in liquid water is represented by Henry's

SOUQH'i SOFIVWUte QlMUFiGWOfV . F£BRUAfCV 1996 21

 law; i.e., dissolved air mole fraction x/*"' is proportional to air partial pressure in the gas
phase.
EOS3 differs from the EOS module of TOUGH in the choice of primary
thermodynamic variables. In TOUGH the variables are (P, T, X) for single phase
conditions and (P , S , T) for two-phase conditions. The choice made in EOS3 is (P, X,
t e

T) for single phase and (P , S +10, T) for two-phase. The rationale for the choice of S,
e s

+10 as a primary variable is asYoilows. As an option, we wish to be able to run isothermal

two-phase flow problems with the specification NEQ = NK, so that (he then superfluous
heat balance equation needs not be engaged. This requires that temperature (T) be the third
primary variable. The logical choice of primary variables would then appear to be (P, X,
T) for single phase and (P , S , T) for two-phase conditions. However, both X and S
e f

vary over the range (0, 1) so that this would not allow a distinction of single phase from
two-phase conditions solely from the range of primary variables. By taking the second
primary variable for two-phase conditions to be X2 = S +10, the range of that variable is
£

shifted to (10,11) and a distinction between single and two-phase conditions can bs made.
Primary variabies can optionally be initialized identical to TOUGH specifications by setting
MOP(19) = 1. A summary of EOS3 specifications is given in Table 6.

Table 6. Summary ofEOSS

COMPONENTS #1: water
#2: air
PARAMETER CHOICES
(2,3,2.6) water and air, nonisothermal (default)
(NK,NEQ,NPH,NB)= (2.Z2.6) water and air, isothermal
PRIMARY VARIABLES'
single phase conditions
(P. X, THpressure, air mass fraction, temperature)
two-phase conditions
(P„S,+10,TKgas phase pressure, gas saturation plus 10, temperature)

*By setting MOP (19>»1, initialization can be made with TOUGH-style variables (P. T, X) fcr
single phase. (P.,S„ T) for two-phase.

3.8.4 EOS4 (Water. Air, with Vaoor Pressure Lowering Capability): This EOS
differs from EOS3 in that provision is made for vapor pressure lowering effects. Vapor
pressure is expressed by Kelvin's equation (Eq. A.9); it is a function not only of
temperature, but depends also on capillary pressure, which is a function of saturation. The
primary variables are (P, T, P„) for single-phase conditions and (P , S , PJfortwo-phase
t c

conditions. Temperature is not among the primary variables for two-phase conditions but
is determined from the relationship P - P, = ? , with P„ = P, (T, S,) as given in (Eq. A.9).
£ v

Other sets of primary variables, in particular temperature, could be used in two-

phase conditions. However, test calculations indicated that the choice (P , S , l\) usually
t

ied to better convergence behavior than the choice (P., S., T). Willi the variables (!»,, S . s

T), the amount of air present in a grid block becomes cohuolled by the difference between
total gas pressure, P , and effective vapor pressure, P, = P„,(T) • f {T,S,), which can be
Ei WL

subject to severe numerical cancellation. From the applications viewpoint, however,

initialization of theflowproblem with the set (P , S , T) may be more convenient. EOS4
c t

allows one to initialize two-phase points as (P , S , T); this capability can be selected by
r t

T O U G H 2 SOfTWARfc QUALIFICATION . FEBRUARV 1 <(9r>

 specifying M0P(19) = 3 in the JNPUT file. The default option for MOP(19) = 0 is (P ,
Sj., P,). The choice MOP(19) = 2 allows EOS4 to be initialized with EOS3 variables of (P,
JC T) for single phase, (P , JL + 10, T) for two-phase. In this way, continuation runs
t

with EOS4 can be made fiom EOS3-generated conditions.

When using MOP(19) * 0 options, data block or file INCON must terminate on a
blank record ( ' ' ) . If'+++' is encountered in INCON, it is assumed that primary variables
arc provided in agreement with internal usage; MOP{19) is then reset to zero and a message
to this effect is printed.
The ability to handle vapor pressure lowering effects makes it possible for a liquid
phase to be present under conditions where vapor partial pressure and gas phase total
pressure are less than the saturation pressure. In EOS4 the pressure at which liquid phase
density, enthalpy, and viscosity are evaluated is taken as P, = max (P , P„,). A difficulty
5

here is that temperature is not among the variables in two-phase conditions, so that P is 1B

only implicitly known; moreover, vapor pressure lowering effects are functionally
dependent on liquid phase density, which is also a function of temperature. This leads to a
potentiaUy unstable situation with regard to the choice of liquid phase pressure under
conditions where P. = P „ , which is common in boiling regions. To avoid this problem
liquid water density in the Kelvin equation is evaluated for vapor pressure lowering (Eq.
A.9) at P, = P^, which is a good approximation due to the small compressibility of water.
In all accumulation and Sow terms, the density of liquid water is evaluated at P, = max (P„
P„,>. Vapor pressure lowering can be optionally suppressed by setting MOP(20) = 1 . A
summary of EOS4 specifications is given in Table 7.

Table 7. Summary of EOS4

COMPONENTS m: water
#2: air
PARAMETER CHOICES

(NK, N"EQ, N?H. NB)= (2,3,2.6) water and air, nonisoihermal

(no other choices available)
VlCS? (70)='.: optionally suppress vapor pressure lowerisc effects
PRIMARY VARIABLES"

tift£[e phase conditions

(P. T, PiXpressurc, temperature, air partial pressure)
two-phase conditions
(P,,S„ P,)-gas phase pressure, gas saturation, air partial pressure)

'By setting MOP (19)=?, initialization of two-phase conditions can be made with ( f „ S>„ T).
'By setting MOP (19)=7, initialization can be made with EOS3-stylc variables of (P, X, T) tor
«in»ie nha«e. •?., S. * 10, T) for two-phase.

3.8.5 EOSS (Water. Hydrogen): This property module was developed to study the
behavior of groundwater systems in which hydrogen release is taking place. It is related to
EOSS, the main difference being that the air component is replaced by hydrogen, with
different taermopbysical properties (see Table 8). The assignment and handling of primary
thermodynamic variables in EOS5 is identical to EOS3 (see Table 6). The main differences
in the assignment of secondary parameters are as follows. Density of hydrogen gas is
computed from the idea! gas law. Viscosity and water solubility of hydrogen are

TOUGH2 SOFTWARE QlMlIf ICATION - FEBRUARY 1 9 9 6 23

 interpolated from the data given in Table S. For temperatures in excess of 25°C, the
solubility at 25°C is used. The parameter specifications of EOS5 are identical to those of
EOS3 as given in Table 6, with "air" replaced by "hydrogen".

Tabic S, Tliermophysical Properties of Hydrogen

Density at P=l bar Experimental* Ideal Gas Law'
5
T=2S0K .086546 kg/m .08660 kg/m 5

5
T=300K .080776 kg/m .0S0S2 kg/m 3

Viscosity*
T=0=C T=100°C
P=lbar 8.40xlO'Pas 10.33xlO*Pas
Ps-ieOter Z.5K\0*bxS !0.44xlO«Pas
Solubility in water ot P=l bar'
T=f/C 1.92xl(T'giygH,0
T=25°C 1^4X10-'BHJEH,O
•from Vargaflifc (1975), p. 39.
'universal jas constant R=S3t-S.56J/moVC; molecular weight of hydrogen 2.0160.
•after Dean (i9S5).
Solubility at different pressures is computed from Henry's law.

3.9 Specification of Flow Geometry

Handling of flow geometry in TOUGH2 is compatible with TOUGH input formats
and data handling. As in other "integral finite difference" codes (Edwards, 1972;
Narasimhan and Witherspoon, 1976), flow geometry is handled by means of a list of
volume elements (grid blocks) and a list of flow connections between them. This
formulation can handle regular and irregular flow geometries in one, two, and three
dimensions. Single- and multiple-porosity systems (porous and fractured media) can be
specified, and higher order methods, such as seven- and nine-point differencing, can be
implemented by means of aj.i. priate specification of geometric data (Pruess and
Bodvarsson, 1983).
Volume elements in TOUGH2 arc identified by five-character names, such as
'ELE10'. Flow connections are specified as ordered pairs of elements, such as
"(ELE10.ELE11)". A variety of options and facilities is available for entering and
processing the corresponding geometric data (sec Figure 8). As in TOUGH, element
volumes and domain identification can be provided by means of a data block "ELEME" in
the fNPUT file, while a data block "CONNE" can be used to supply connection data,
including interface area, nodal distances from the interface, and orientation of the nodal line
relative to the vertical. These data are internally written to a disk file MESH, which in turn
initializes the geometry data arrays used during the flow simulation. The data formats on
the file MESH are identical with the format specifications for data blocks ELEMB and
CONNE.

TOUGH2 SOFTWAHF QUALIFICATION - FEBRUARY 1996 24

 Y
/
geometry tfata in 1 geometry data for
INPUT lite
j internal mesh generation
{Modes ElEME, CONNE) (MoclMESHMAKER)
i
disk lite
•MESH-

Figure 8. User options for supplying geometric date

TOUGH2 offers additional avenues for defining flow system geometry. By means
of the keyword 'MESHMAKER' in the INPUT file, a special program module can be
invoked to perform a number of mesh generation and processing operations. The
MESHMAICER module itself has a modular structure; present sub-modules include
"RZ2D" for generating two-dimensional radially symmetric (R-Z) meshes, and "JCYZ" for
one-, two-, and three-dimensional Cartesian grids. Multiple-porosity processing for
simulation of flow in naturally fractured reservoirs can be invoked by means of the
keyword 'JflNC, {Pruess and Narasimhan, J 982, 1985; Pruess. J 983a; see Appendix Q .
The MINC process operates on the data of the "primary" (porous medium) mesh as
provided on disk file "MESH", and generates a "secondary" mesh containing fracture and
matrix elements with identical data formats on file "MINC". (ThefileMESH used in this
process can be either directly supplied by the user, or it can have been internally generated
either from data in INPUT blocks ELEME and CONNE, or from RZ2D or XYZ mesh-
making; see Figure 8). The internal mesh generation process will also write nodal point
coordinates on file MESH for graphical display. These data are written in 3E10.4 format
into columns 51-80 of each grid block entry in data block ELEME. At present, no internal
use is Iliad's of nodal point coordinates in TOUGH2.

In TOUGH2 elements are referenced by names consisting of a suing of five

characters, '12345'. These are arbitrary, except that the last two chararr us (#4 and 5) must

TOUCH2 SorrwAM QUALIFICATION - f EMUMXY I 996 25

 be numbers. Specific naming conventions have been adopted in the internal mesh
generation process. For RZ2D, the last two characters directly number the radial grid
blocks, from ! through 99. Character #3 is blank for the first 99 radial blocks, and then
runs through the sequence 1, 2, ...9, A, B, .... Z for a maximum of 3599 radial blocks.
The second character counts up to 35 grid layers as 1, 2, .... 9, A, B, ..., Z. The first
characleris 'A' for the first 35 layers, and is incremented to B, C, .... A, 1, 1, ..., 9 for
subsequent groups of 35 layers.

For rectilinear meshes generated by XYZ, characters 4 and 5 together number the
grid blocks in X-direction, while character #3 = 1, 2 9, A, B, ..., Z numbers Y-
direction grid blocks, and character #2, running through the same sequence as #3, numbers
grid blocks in Z direction. Overflows with more than 99 X-biocks, or more than 35 Y- or
Z-blocks advance character #1 through the sequence A, B. C,.... Z. Both RZ2D and XYZ
assign all grid blocks to domain #1 (first entry in block "ROCKS"); a user desiring changes
in domain assignments must do so by hand, either through editing of the MESH file, or by
appropriate source code changes in subroutines WRZ2D and GXYZ. TOUGH2 runs that
involve RZ2E> or XYZ mesh generation will produce a special printout, showing element
names arranged in their actual geometric pattern.

The naming conventions for the MINC process are somewhat different from those
originally adopted in the GMINC program (Pruess, 1983a) and are as follows. For a
primary grid block with name '12345', the corresponding fracture subelement in the
secondary mesh is named '2345' (character #1 replaced with a "Wank"). The successive
matrix continua are labeled by running character #i through 2, .... 9, A, B, .... Z. The
domain assignment is incremented by 1 for the fracture grid blocks, and by 2 for the matrix
grid blocks. Thus, domain assignments in data block "ROCKS" should be provided in the
following order: the first entry is the single (effective) porous medium (POMED), then
follows the effective fracture continuum (FRACT), and then the rock matrix (MATRX).
An example is given in Sample Problem 3 in Appendix G that follows.
Mesh generation and/or MTNC processing can be performed as pan of a simulation
run. Alternatively, by closing the INPUT file with the keyword 'ENDFT (instead of
'ENDCY'), it is possible to skip the flow simulation and only execute the MESHMAKER
module to generate a MESH or MTNC file. These files can then be used, with additional
user-modifications by hand if desired, in a subsequent flow simulation. MESHMAKJER
input formats are described in the Software User Documentation Report under "Preparation
of Input Data", and examples of practical applications are given in the sample problems in
Appendix G. Execution of MESHMAKER produces printed output which is self-
explanatory.

3.10 initial Conditions and Restarting

As in the TOUGH code, initial thermodynamic conditions for the volume elements
in the flow domain can be assigned to identical default values for all elements, or they can
be prescribed for each element individually by means of a data block "INCON". A file
"INCOif" written to the same specifications as data block "INCON" should also be used
for initialization.
A simulation problem can be run in several segments. At the end of a simulation
run TOUGH2 writes the primary diermodynaruic variables of all elements on a disk file
"SAVE" with format specifications identical to "INCON". For a subsequent continuation
run. file SAVE can be merged into the INPUT file as data block INCON or it can be

TOUGH2 SOFTWARE QUALIFICATION - FmmiARY 1996 26

 renamed as file INCON. In the latter case, no data block INCON should be present in the
INPUT file, as this would cause the INCON file to be overwritten.
TOUGH2 also offers the facility of assigning initial conditions uniformly
throughout selected zones of the simulation grid. This is invoked by means of a data block
INDOM, which provides information on the thermodynamic conditions in user-defined
domains. The format specifications for block INDOM are similar to those used in INCON.
Thermodynamic conditions given in block INDOM take precedence over default
assignments for the entire flow domain; specifications for individual grid blocks in INCON
supersede all other assignments.
The normal way of defining initial conditions is by directly providing the primary
thermodynamic variables. Note that these variables are generally different for different
EOS modules. The thermodynamic state variables that are used internally in TOUGH2 as
primary dependent variables may not always be the most convenient variables for a user to
initialize a flow problem. The parameter MOP(19) offers a variety of choices, which
permit initialization with variables different from tie internally used primary variables.
These choices are different for different EOS modules, and they are documented in the
printout produced by each EOS.
At the end of a simulation run, file SAVE will always be written with the internal
primary variables of the EOS module used. When modifying an INPUT file for a
continuation run, MOP(19) might therefore have to be changed to its default value
MOP(19)=0 for proper initialization. To minimirft the possibility of user error, an
automatic switch has been implemented in TOUGH2, as follows. The file SAVE as
internally written by TOUGH2 terminates on a record with '+++* in the first three
columns, followed by one record with restart information. When the data block INCON
or file INCON terminates on'+++' rather than a blank line, it is assumed that this INCON
was internally generated in a previous TOUGH2 run and that it is therefore written with the
internally used set of primary variables. Accordingly, when '+++' is encountered in
INCON the switch MOP(19) is reset to zero, and a message to that effect is printed.

3.11 Boundary Conditions

Boundary conditions are generally specified by means of appropriately chosen
volume elements, flow connections, and sinks and sources. Boundary conditions can be
of two basic types. Dirichlet conditions prescribe thermodynamic conditions, such as
pressure or temperature on the boundary, while Neumann conditions prescribe flaxes of
mass or heat crossing boundary surfaces. A special case of Neumann boundary conditions
is "no flux", which in the integral finite difference framework is handled by not specifying
any flow conditions across the boundary. More genera! flux conditions are prescribed by
introducing sinks or sources of appropriate strengths into the elements adjacent to the
boundary. Dirichlet-type boundary conditions, such as constant pressures or temperatures,
can be specified by introducing appropriate boundary elements and connections. A
connection consists of an interface area, and a pair of adjacent nodes at some distance from
the interface. Assigning very large volumes to such boundary elements will ensure that
their thermodynamic state remains unchanged in a simulation. It is also possible to fix
temperature and to allow pressure to vary. This can be done by means of assigning a very
large heat capacity to an element with "normal" volume. The only feature distinguishing
boundary elements from the "normal" grid blocks forming the flow domain is their large
volume (and/or heat capacity); in the calculations they are treated on an equal footing with
all other elements.

TOUCH2 SOFTWARE QUAUHCAIIOM - FEBRUARY 1996

 TOUGH2 implements Dirichlet conditions with the simple device of "active" and
"inactive" elements. By convention, elements encountered in data block ELBME (or files
MESH or MINC) are taken to be "active" until the first element entry with a zero or
negative volume is encountered. The first element with volume less than or equal to zero,
and all subsequent elements, are taken to be inactive. For die inactive elements no mass or
energy balance equations are set up, and their primary thermodynamic variables are not
included in the list of unknowns. Otherwise, inactive elements can appear in flow
connections and initial condition specifications like all other elements. This feature can be
used to specify Dirichlet boundary conditions by gathering all elements beyond the desired
flow domain boundary at the end of the ELEME block and inserting a dummy volume
element of zero volume in front of them. Thermodynamic conditions for the inactive
elements are maintained as initially prescribed during a simulation run.

The specification of inactive elements can ?Iso be used in the MESHMAKER

module to steer the MINC-process of subgridding volume elements. By convention,
inactive elements will remain unpartitioned, i.e., they will be treated as a single porous
medium. Users should beware that the MINC process may lead to ambiguous element
names when the inactive element device is used to keep a portion of the primary mesh as
unprocessed porous medium.

3.12 Software and Hardware Considerations (Efficiency, Portability, etc.)

TOUGH2 is portable to platforms that have an ANSI (FORTRAN 77) compder. It
can run on a PC, a Macintosh, a UNIX workstation (Sun, etc.), or a mainframe. It is
easily maintained because of the flexibility of its modularization and through its internal
version control, which allows the user to note version updates and to document
personalized changes to the program. Numerous tests have shown that it produces
consistent results on various platforms. TOUGH2's efficiency has been demonstrated
clearly in Moridis and Pruess (1995). The authors chose sixteen test problems of variable
complexity and size and compared the efficiency of the direct solver method using MA28 to
methods using three different conjugate gradient solvers. For each solver they tabulated the
grid size, number of equations, number of time steps, simulation time, number of
Newtonian iterations, and CPU time. The conjugate gradient solvers were invariably
faster, often by orders of magnitude, than the MA28 direct solver in all 16 problems.
MA28 could not solve most of the large and medium size problems; it was limited in size to
less than 2000 grid blocks for 2-D problems, and less man 400 for 3-D problems.
Performance of the solvers was specific lo the problem, compiler, and machine. The
conjugate gradient solvers had well-defined memory requirements, which were lower than
requirements for the MA2S. These low memory requirements and the speed of the routine
made possible the solution of 20,000 equation problems needed for these exercises on
personal computers. TOUGH2, in fact, can handle problems with as many as 100,000
equations using the conjugate gradient solver approach. Theoretically, there is no upper
bound to the number of equations it is capable of solving. Occasionally a specific solver
failed for a particular problem, but in all test problems, one or more solvers produced a
fast, efficient and accurate solution.

TOUGH2 SOFTWARE QUALIFICATION • FFRHUARY I99fi 20

 3.13 Performance Requirements and Design Constraints
Version 1.11 of TOUGH2 uses no pre- or post-processors and requires no
interfaces with external hardware or software. TOUGH2 has no prescribed requirements
for performance beyond its ability to model accurately the processes listed in the
Requirements Specification, Section 2. Its flexible modularization and choice of solvers
remove constraints on the size, dimensions, and complexity of design of a problem that can
be handled by the code. Therefore, this section applies primarily to caveats in problem
specification and choice of solver.
As with any code, proper set-up of the problem is necessary to obtaining reliable
results. Key to successful application of TOUGH2 is a careful consideration of the
physical processes that are involved in a given flow problem. In particular, space
discretization, time-stepping, and interface-weighting procedures need to be selected
carefully so that accurate results may be obtained. Without attention to these issues,
application of TOUGH2 may result in both inefficient performance and inaccurate results.

In selecting the conjugate gradient solvers over the direct solver, the user trades the
robustness and slow performance of a direct solver for the speed and reduced memory
requirements of an iterative solver. However, because they are case-, problem-, and
machine-specific, iterative solvers may be less reliable if they are not carefully applied.
Because of the case-specificity of the solvers, the user must watch carefully both the
convergence and the evolution of residuals over the time increments of the simulation. In
cases of unsatisfactory convergence, the user should examine the conjugate gradient
performance statistics in the file LINEQ generated by TOUGH2 runs, paying close
attention to the parameters ERR and ERR (see Moridis and Pruess, 1995 for more details).

Because a wide variety of options is available in a complex code such as TOUGH2,

the user must give careful consideration to peculiar features of a given problem when
setting it up if a reasonably accurate and efficient solution is to be obtained. TOUGH2 is
intended to be a "general-purpose" simulator. For example, there is no single weighting
scheme for general transient two-phase flows in composite media that simultaneously
preserves optimal accuracy for single-phase or steady two-phase flows. Another problem
arises in the weighting scheme for interface densities. For certain flow problems spatial
interpolation of densities may provide more accurate answers than a weighting scheme.
Issues of interface weighting arid associated discretization errors are especially important
when non-uniform or irregular grids are used. Additional complications related to interface
weighting arise in flow problems that involve hydrodynamic instabilities. Discussion of
examples of the nuances and pitfalls of modeling multiphase flows is provided in Pruess
(1991). In general, in the design and implementation of numerical schemes for multi-phase
flows, a trade-off occurs between accuracy and efficiency on the one hand, and flexibility
and robustness on the other. For a particular problem, small modifications in the source
code may produce substantial gains in accuracy and efficiency.

When making changes in the code, it is essential to preserve a continuous

dependence of all secondary parameters on die primary thermodynamic variables. True
numerical discontinuities, such as a non-zero capillary air entry pressure, are inadmissible.
They may lead to an unstable situation in which the residuals in the governing equations
(Eq. B.6) become discontinuous functions of die primary variables, so that it may be
impossible to reduce them to small values. A finite transition region for continuous
variation of parameters must be provided.

TOUCH2 SOFIWAM QUALIFICATION • FEBRUARY 1996 .29

 3.14 Future Developments
As stated previously, the Version 1,11 of TOUGH2 that is being qualified here
includes the better tested MULKOM program modules. LBNL has developed additional
modules, however, and these are expected to become part of future releases of TOUGH2.
They include EOS modules for multiphase fluid mixtures containing hydrocarbon and non-
Newtonian fluids, capabilities for modeling rock-fluid interactions with dissolution and
precipitation processes and associated porosity and permeability change, coupled
radionuclide transport in porous fractured media, modules for production and injection well
operations and scheduling, preprocessor programs for higher-order differencing schemes
to minimize space discretization errors, and specialized routines for the study of multiphase
flow processes in fractured media, including hysteresis effects.

3.15 Verification of Design Description

Verification that TOUGH2 meets the design description and structure depicted in
Figure 3 was conducted during testing and debugging of the code. Sample runs were set
up to test the ability of the EOS modules to handle primary variables in the manner
described above. Once the runs were completed, sample problems were constructed to
demonstrate the code. These are found in Appendix G.

TOUGH2 SOFrwASfQlMIIFIC-ATION - rfiBIMRY 199C 30

4. SOFTWARE VALIDATION TEST PLAN AND REPORT

4.1 Validation Test Plan

The Validation Test Plan for TOUGH2 is quite straight-forward. It calls for
implementing the code in the variety of applications specified by requirements, testing and
debugging the code for errors, and subsequently verifying and validating the code. The
terms "verification" and "validation" are used differently in quality assurance software
terminology than in scientific research. In quality assurance terms, "software verification"
means checking to ensure that a following stage (e.g. implementation) fulfills the
requirements of apreceding stage (e.g. design), where the Software Life Cycle is built on a
"waterfall approach" in which requirements are established, followed by code design, then
code implementation, and finally validation. Validation means that the operational code
satisfies all the requirements specified for it in the requirements stage (Mangold, 1993).
The verification process has been described in the two preceding sections, in which the
requirements are specified and the code design is explained. Verification testing and
debugging took place as each module of the code was created. Some of the verification
tests performed to ensure that the code is able to model the processes specified by the
requirements are given as sampie problems in Appendix G.
4.T.1 Acceptance Criteria: A number of additional verification problems were run
to compare the results obtained by TOUGH2 against known analytical solutions to such
formalized problems as the Theis solution, for example. The acceptance criterion for
establishing that the TOUGH2 results showed sufficient agreement to the analytical
solution was demonstration that the obtained results for TOUGH2 and the analytical
solution differed by five percent or less (95% or better agreement). The five percent choice
is somewhat arbitrary; depending upon the nature of the problem, it may be possible (and is
always desirable) to achieve much closer agreement. The validation problems are
summarized in Table 9. A short summary of each, along with a graphical comparison of
TOUGH2 and (he analytical solution, is found in the paragraphs following. For some
applications, comparison could not be made to an analytical solution but could be made to
experimental results in the laboratory or field. In the world of scientific conceptual models,
this type of comparison is called "model validation". Comparisons of TOUGH2 solutions
with experimental results are also included in Table 9 and described in the following section
along with a graphical comparison. Acceptance criteria are difficult to quantity for
comparisons against experimental or field data because the closeness of fit will depend
partly on the goodness of original data, experimental artifacts, and heterogeneity in the
natural system. Where possible, we try to obtain an agreement of 90% or better. The
reader is referred to references listed in Column 6 of Table 9 for further details of the
problem set up and specifications.

In the terminology of software engineering, "validation" tests are those that ensure
that the software meets the requirements specified for it. The "verification" tests also fulfill
the definition of "validation", because in order to obtain good results when compared to
analytical solutions, the software must be able to model the processes of interest in the
problem. However, additional validation problems were run for this qualification effort to
cover specified requirements of TOUGH2 that are not part of the problem descriptions for
tests listed in Table 9. A summary of the validation test problems and results is provided
below. Full descriptions of these validation tests will be found in a future report.

Throughout the development of the MULKQMATOUGH codes an effort was made

to maintain continuity in numerical performance, i.e., different code variations were

JOUGH2 SOFTWARE QUALIFICATION • FESKIMRY 1996 31

 exercised on several earlier benchmark problems to guard against errors in coding.
Therefore, verification or validation exercises performed with any member of the
MULXOM/TOUGH family are significant in establishing credibility for ali other members
as well.
Table 9. Summary of Validation Problems
# PROBLEM DIMENSIONS FEATURES ISSUES REFERENCES
TITLE
1 infiltration 1-D, linear isothermal code verification against #2 in Praess
horizontal known semi-analytical (1987); also #1 in
solution (Philip, 1955; Moridis and Praess
Ross eta!., 19S2) (1992)
2 How to £ 1-D, radial water and steam only. phase transitions; #4 in Pmr -
geothermai no air; sensible and propagating boiling (1987), also #4 in
well I atcnt heat effects; from; code verification Moridis and Pruess
coupled fluid and heat against known semi- (1992)
flow analytical and numerical
solutions (Garg, 1973,
I9S0)
3 transient heat 1-D, linear coupled fluid and heat code verification against Pruess (1991);
pipe flow with air; liquid-gas similarity solution Doughty and
counter-flow with very (Doughty and Prucss, Pruess (1992)
strong binary diffusion; 1991,1992)
sensible and latent heat
effects
4 coupled heat i-D, linear non-isothermal code verification It 2 in Moridis and
and mass convection, diffusion (Avdonin, 1954, Boss, Prucss (1992)
transport 19S2)

5 heat transport 1-D, radial single phase non- code verification #3 in Moridis and
isothermal, convection, (Avdonin. 1964, Ross, Pruess (1992)
diffusion, sensible and 19S2)
latent heal effects #2 in Moridis and
Pnicss<1995)
6 Theis problem i-D, radial single-phase. validation against rrl in Moridis and
(flow toward a isotbeimal, viscous analytical solution Prucss (1995)
well) forces (Theis, 1935)

7 Coupled fluid 1-D, radial heat conduction, M7NC, verification Pruess and VVu
and heat flow frocluie-mntiix flow, (1993)
in fracture single-phase, non-
isothermal

8 infiltration 2-D, carteb'ian isothermal, two-phase validation against #6 in Moridis and

heterogeneous medium, experimental data Pruess (1995)
seepage face mixed (Vauulin el at., 1979)
boundary condition,
interference between
liquid and gas, gravity
effect.

TOUCH2 SOFIWARE QUALIFICATION - TiBRIMRV I 99fi 32

Table 9 (tout)- Summary of Validation Problems
it PROBLEM DIMENSIONS FEATURES ISSUES REFERENCES
TITLE
9 convection cell 2-D, cylindrical single phase, non- validation against lab #7 in Moridis and
isothermal, experiment (Keda, iTuess(1992)
heterogeneous soil, no 1984)
mass flow boundary, 3 in Moridis and
flow channeling, Pruess(I995)
sensible and latent heat
effects

10 iwo-phase flow 2-D. cylindrical simultaneous heat and validation against lab #S in Moridis and
massflow,phase experiment (Krugcr and Pruess, 1992
change, time-variant Rumey, 1974; Faust
pressure boundary, and Mercer, 1979)
interference between
liquid and gas phase

11 Warren-Root 1-D, radial transientflow,double- code verificatfon against diis paper

Solution pcrosity medium analytical solution
(Warren and Root,
1963)

12 Lauwerier heat 2-D, cartesian conductive and code verification against this paper
transfer convective heat transfer analytical solution
solution in porous media (Lauwerier, 1955)

13 handling of NA water, water vapor, air TOUGH2 calculated this paper

thermophysical properties compared to
properties steam tables (CRC.
1993)

14 vapor pressure NA coupling between comparison with this paper

lowering capillary and vapor predictions from
adsorption effects, arid Kelvin* s equation
vapor piessure

15a heterogeneous 1-D single-phase slightly code verification against this paper
compressible liquid analytical solution of
Moridis (1995)

15b flow to single 2-D single-phase, slightly code verification against this papci
well with compressible fluid, analytical solution
anisotropic infinite anisotropir (Papadapoulns, 1965)
formation aquifer

16 single-phase 2-D, cartesian transient flow in verification of irregular this paper

transient flow horizontal, isotropic, grid capability using
with irregular isothermal aquifer Theis solution
sird

TOUGH2 SOFTWARE QUAUFICATION - FEBRUAKV 1996 33

 Tjbte 9 (com). Summary of Validation Problems
PROBLEM DIMENSIONS FEATURES ISSUES REFERENCES
TITLE
I7a hut conduction 1-D, linear semi-infinite rock code verification against this paper
column, step change in analytical solution of
T: ignores flow effects Carsiaw and Jaeger
(1959)

lTo binary vapor 1-D vapor and air diffusion code verification against this paper
diffusion in gas phase, semi- analytical solution of
infinite rock column; Carlsaw and Jaeger
ignores vapor (1959)
adsorption

17c gas Row with 1-D steady single-phase gas code verification of gas this paper
Klinkenberg flow across linear rock permeability
effects column, isothermal enhancement a: tow
pressures (Klinkenberg,
1941)

horizontal well 3-D anisotropic reservoir code verification against this paper
with analytical solution of
compressible Goodeand
liquid Thambynayagarn (1987)

19 single-phase 1-D, radial flow to single vertical code verification against this paper
gas flow well at center of analytical solution of
bounded cylindrical Kabir and Hasen (1986)
formation

20 water 1-t), radial fracturd-matrix code verification agains; Zimmerman et a!,

absorption into interaction in two-phase semi-analytical solution (1990)
porous matrix How

4.2 Validation Tests (Keyed to Table 9)

1. Infiltration - 1-D. Linear (Pruess, 1987; Moridis and Pruess, 1992)

This ..> a one-dimensional problem that considers infiltration into a semi-infinite tube of
partially saturated soil. Philip (1955) obtained a semi-analytical solution for this problem,
using a similarity transformation method. Detailed specifications were given in a report by
Ross et al. (1982), who proposed this problem "as a benchmark case for numerical
simulators. The infiltration boundary in the problem is represented by means of a very
large element (LBO), with a nodal distance set to a small non-zero value, to avoid relative
permeability at the boundary to cv taken from the downstream clement designated as F l .
The solution obtained by Philip (1955) treats the gas phase as a passive spectator at
constant pressure. This approximation is enforced by surrounding the soil tube with a ring
of very large volume, which is assigned a pressure of P = 1 bar. Different interface
weighting procedures were used, labeled 0, Ml, and M2 in Figure 9. The simulated

TOUGH2 SOFTWARE QUAIIFICATION • rraxuARV 199C 34

 0.00 0.02 0.04 Q.OS O.OS 0.10 0.12
Distance {m}

Figure 9. Liquid saturation profilesfor verification problem 1 (from Moridis and Pruess, 1995),

results are seen to agree well with the semi-analytical solution given by Ross et ai. (1982)
and meet the acceptance criterion of 95% or better agreement.
2. Flow to a Gcpthermal Well (Pruess. 1987; Moridis and Pruess, 1992)
This problem deals with radial flow to a geothermal well. Garg (I97S) developed a semi-
analytical theory for radial flow to a geolhermal well, which accounts for phase transitions
and propagating boiling fronts. He presented simulated results for production at a constant
rate of 14 icg/s from a 100 m thick reservoir that is initially in single phase liquid conditions
of T = 300'C, P = 90 bars. In response to production, pressures drop to the saturated
vapor pressure, and a boiling front moves out into the reservoir. The computational mesh
consists of 10 grid blocks with AR = 1 m, and an additional 40 grid blocks with AR*«,, =
OAR,- out to an outer radius of 2000 m. Simulated pressures in the weiiblock (element
AAI) are plotted versus time in Figure 10. Comparison of TOUGH with Garg's results is
excellent and meets the acceptance criterion of 95% or better agreement.

TOUGH2 Sonw/uw QUALIFICATION - FEBRUARY J 996 35

 Figure 10. Wcttblock pressures for verification problem 2 (from Pruess. 1937).

5. Heat Pipe (Pruess. 1991; Doughty and Pruess, 1992J

Validation Problem 3 simulates a heat pipe with cylindrical geometry. It considers a case in
which a heat pipe (a heat transfer system which operates via a liquid-vapor counterfiow
process) is artificially induced by nuclear heat-generating waste packages in a partially
saturated medium. The TOUGH2 input file uses the EOS3 fluid properties module. The
problem places a constant strength, linear heat source in an infinite homogeneous medium
with uniform initial conditions. The MESHMAKER module is used to generate a one-
dimensional radial grid extending to a large radius, which for practical purposes can be
considered infinite. Boundary conditions are constant. Most of the formation parameters
are identical to data used in previous modeling studies at Yucca Mountain (Pruess et a!.,
1990). Because fracture effects are not included in the simulation, beat pipe effects would
be very weak at the low rock matrix permeabilities (on tiw order of one microdarcy)
encountered at Yucca Mountain. To obtain a more pronounced heat pipe effect, absolute
permeability was arbitrarily increased to 20 millidarcy and capillary pressures were reduced
by a factor of 100 compared to that at Yucca Mouutaia To examine space discretization
effects, runs were made with a coarse, medium, and Cue mesh using MESHMAKER. As
shown in Figure 11, excellent agreement is obtained by comparing the fine mesh results
with those of an exact similarity solution (Doughty and Pruess, 1991, 1992). Under
constant initial conditions the governing partial differential equations in radial distance r and
time t reduce to ordinary differential equations through the introduction of a similarity
2
variable Tj^r/t" . Tile resulting equations are coupled and nonlinear, necessitating a
numerical integration (Doughty and Pruess, 1991,1992).
The input file for problem 3 can also be executed with the EOS4 Quid property
module, which includes vapor pressure lowering effects. This can serve as a
benchmarking reference for the EOS4 module. The results are quite similar to those
obtained with EOS3 when the problem is set up with a constant rate of heat generation,
except that because of very strong vapor pressure lowering effects, diyoul near the heater is
slowed. Comparison of this problem with the similarity solution affords the most
comprehensive code verification available, because ail of the non-linearities of two-phase

TOUGH2 SOFTWARE QUALIFICATION - TFIIRIJARV 1996 36

 200 — " ' ' I—— -' i' r—-^pT

160 0.8
3-
c S
f t -...TOUGH2 o S
0.6
! 120 f | similarity 11
a-
a 2- L f solution ra u>
™ X
T3 3
0.4 =i =
_ p_ _ _ sr—
a<
40 F \ 0.2

0--'
k
-10 - 8 - 7 - 6
Z=tog(r/4f)
Figure 11. Comparison of TOUGHS with simitariiy solwion forproblem 3 (from Pruess, 1991).

flow behavior and of fluid and heatflowcoupling are rigorously described by the similarity
solution.
4. Coupied Heat and Mass Transport - Avrfonin Problem fMoridis and Pruess, 1992)
The analytical solution to this problem of one-dimensional heat and mass transport was
developed by Avdonin (1964). Ross et al. (1982) described the problem and presented a
solution. Cold water is injected into a semi-infinite, high-temperature aquifer at a specified
constant mass flow rate. The overburden and underburden are impermeable to mass and
heatflow,acting as no-flow and adiabatic boundaries and reducing the equation governing
heat transport to that of convection-diffusion. Updegraff (1989) modeled tin's problem in
his comparison study of three simulation codes that model strongly coupled mass and heat
flow in unsaturated porous media (TOUGH, NOKIA, and PETROS). Moridis and Pruess
(1992) discuss their simulation of Updegraff" s set of fluid and heat flow problems and the
difference between their results using TOUGH and those of Updegraff.

The space discretization used by Updegraff(1989) consisted of 500 equally spaced

gridbiocks and two boundary gridblocks. The space and time discretization were based on
criteria established by Reeves et al. (1986). A very large volume was assigned to the
boundary grid blocks, thus ensuring constant boundary pressures and temperatures during
the simulation. An initial pressure distribution varyingfromthe kit-hand to the right-hand
boundary was distributed across the grid blocks. The prescribed pressure differential on
the boundaries created an influx which resulted in an equivalent system. For this flux to
equal the desired injection rate, Updegraff specified an appropriate permeability.

It turned out that Updegraff used erroneous water property values. The differences
between his values and those in the steam tables had a significant impact on the solution.
Therefore, Moridis and Pruess (1992) computed the analytical solution using values from
the steam tables. They also corrected Updegraff s error in relative permeability. Moridis
and Pruess (1992) created three modified data sets. The first used an upstream weighting

TOUCH2 SOfTWAKE QuAllflCATION - FEBRUARY 1 9 9 6 37

 of mobilities, permeabilities, and thermal properties. The second data set used a midpoint
weighting scheme. The second and third data sets differed in boundary conditions and in
the treatment of sources and sinks. The use of direct injection into the fust grid block in
these two data sets reduced the size of the input data file by 90%. Strictly speaking,
constant rate injection into a flow system initially at uniform pressure will result in a
transient change offlowrates away from the injection point, rather than giving steady-state
mass flow throughout. However, because of the large permeability and the small
compressibility of liquid water, the pressure diffusivity at these conditions is very large, so
that the region with practically steady flow expands rapidly and runs well ahead of the
thermal front.

The corrected analytical and the TOUGH solutions obtained with the modified input
files are shown in Figure 12. The runs required 1300 time steps to reach the maximum
simulation time of *„„, = 130,000 sec because of the time-step constraint of AT < 100 sec.
Moridis and Pruess (1992) concluded that with upstream weighting TOUGH properly
predicts the midpoint of the front, but there is a certain amount of numerical dispersion
which results in a broadening of the front. On the other hand, it is a more robust numerical
weighting scheme, being more stable in difficult situations with long time steps. The
midpoint weighting scheme is more accurate and in excellent agreement with the analytical
solution. This test meets the acceptance criteria of 95% or better agreement.

Distance (m)

Figure 12. Comparison of the analytical solution to the TOUGH solutions in validation problem 4 (from
Moridis and Pruess, 1992).

5. Radial Heat Transport (Moridis and Pruess, 1992; Moridis and Pruess, 1995)
The radial heat transport in this problem was originally solved analytically by Avdonin
(1964) and was later described by Ross et al. (1982). Cold water is injected into a semi-

TOUGH2 SOFTWARE QUALIFICATION - FEBRUARY 1 9 9 6 38

 infinite, high-temperature aquifer. This problem is very similar to the one-dimensional heat
transport problem described in validation problem 4, except that it uses radial rather than
Cartesian coordinates. The overburden and underburden are impermeable to mass and heat
flow, acting as no-flow and adiabatic boundaries and reducing me equation governing heat
transport to that of convection-diffusion.
The radius of the aquifer is sufficient to approximate serru-infinite behavior (1000
in), at which a constant temperature (equal to the initial temperature of 170%) is imposed.
TOUGH2 predictions of the temperature distribution in Figure 13 at f = t^ a 1C? sec after
initiation of the cold water injection are shown for comparison to the analytical solution.
This simulation was performed using the EOS1 fluid property module with NK = 1
and NEQ = 2. The number of gridbloc'us used was 127, resulting in a total of. N = 254
equations. The grid was generated with the MESHMAKER submodule. A very large
M 3
volume (10 m } was assigned to the single boundary grid block in the radial direction
(127th), thus ensuring constant boundary pressures and temperatures during the duration
of the simulation. With flow rate directly specified through a mass source, the problem
becomes insensitive to the precise value of permeability.
Figure 13 shows excellent agreement between the analytical and the TOUGH2
solutions al r = 1 0 ' sec. This is well within the acceptance criterion of 95% or better
agreement. This problem has a "similarity solution" in terms of the variable i*/t
(O'Sullivan, 1981; Doughty and Pruess, 1990, 1992). The TOUGH2 results are
consistent with the rVr invariance diat exists in the problem so thai at i =10* sec. the
TOUGH2 solution virtually coincides with the analytical solution.

T i 1 1 = r
0 tOD 200 300 400 S00
OIstairc» (m)

Figure 13. Comparison of the analytical and TOUGH2 solutions to tlie radial heat transport problem in
validation problera 5 (from Moridis and Prucu. WJ2).

T<"><' 'Ml SOFTWARE QUALIFICATION - FEBRUARY 7996 39

 6. Theis solution fMoridis and Pruess, 1995)
This problem represents the classical Theis (1935) problem of one-dimensional radial flow
toward a well of radius r„ -» 0 in a homogeneous circular aquifer with infinite-acting
boundaries. Input parameter values are given in Moridis and Pruess (1995). The EOS1
fluid property module with NK = 1 and NEQ = 1 was used for the TOUGH2 simulation.
s
The outer boundary extends to 10 m, which is sufficient for the aquifer to act as infinite.
A total of N = 104 equations are solved, as the domain is subdivided into 104 grid blocks.
Figure 14 shows the analytical solution and the TOUGH2 solution obtained for an aquifer
temperature of 20°C. The two solutions are virtually identical. Therefore, the acceptance
criterion for validation is met. Note that in uns and similar problems of single-phase flow
in confined aquifers (where the compressibility is small and the only variable changing is
pressure) the convergence criterion for the relative error in die Newtonian iterations must be
set to a sufficiently small number. More specifically, the parameter RE1 in Record
s
PARAM.3 of the PARAM data block must be reset from its default value of 1 0 to a
number several orders of magnitude smaller. The maximum RE1 value should not be
s
larger than 30 .

n
—j——j i t i "i t'
-1 ! 4 s
10-* 10 10° 10' 10 *.0' W 10
Distance from well (m)

Figure 14. Comparison of ike analytical and the TOVGH2 solutions to the THEIS problem
aft = t „ , = 10 days (front Moridis and Pruess, 199$).

7. O3npledjluidjmi.heat_fi0^inJracwre_(Pniess and Wu, 1993)

Pruess and Wu (1993) developed a new method for modeling fluid and heat flow in
fractured reservoirs which is an extension of a technique developed by Vinsome and
Westerveld (1980) for calculating heat exchange between permeable layers and
impermeable semi-infinite confining beds during thermally enhanced oil recovery. Piuess
and Wu's method combined afinite-differencedescription of global flow in the fractures
with an analytical representation of interporosiry flow by means of trial functions for fluid
pressures and temperatures in the matrix blocks. The trial functions contain adjustable

TOUCH2 SOFTWARE QlMllFICATION - FtlRUARY 1996 40

 parameters which are calculated for each time step in a fully coupled way based on matrix
block shapes and dimensions. They incorporated the method into MULKOM and verified
it by comparison with exact analytical solutions for fluid and heat exchange with individual
matrix blocks. Applications were made to geothermal well test and production-injection .
problems with inlerporosity fluid and heat flow. A shown in Figure 15, the MTJLKOM
simulation using multiple-interacting-continua (MTNC) and numerical solution show
excellent agreement.

1 5
10* 10' TO TO Iff* «> K?

Figure 25. Comparison ofMULKOM to semi analytical solution for pressure buildups for non-
isothermal injection into a fractured .'eservoirfor validation problem 7 (from Pruess and Wu, 1993).

Test problems 8 through 10 are verification problems for the purposes of software
engineering, but are validation problems in the sense of scientific model validation. Instead
of comparisons to analytical solutions, the TOUGH2 simulated results are compared to
results obtained from field or laboratory experiments to ascertain the "validity" of the
conceptualization of the model presented in the TOUGH2 problem.
8, Infiltration - Vauclin Problem fMoridis and Pruess, 1992; 1995)
Test problem 8 describes a two-dimensional infiltration laboratory experiment conducted by
Vauclin el al. (1979), who provided measurement data and a numerical solution (therefore
this is both a verification and a "validation" problem). Water infiltrates al a rate of
4.111x10* m/scc over a length of 0.5 m of a vertical slab of soil. Due to symmetry only
half of the problem needs to be modeled. The bottom boundary and the left boundary (line
of symmetry) are considered impermeable to flow. Below z m 0.65 m the right boundary is
a constant pressure boundary with a water saturation of 1 at tbe bottom; above z = 0.65 m it
is a seepage surface, i.e. a mixed type of boundary condition which sets the water flux
il to zero when the medium is unsaturated and has a bead equal to the hydraulic head
X rt the medium is saturated.
For this simulation cite domain was subdivided into 378 grid blocks. The
MESHMAKER facility in TOUGH2 was used to generate the grid. A very large volume
3
(1(P m ) was assigned to the right-hand boundary grid blocks, thus ensuring constant
boundary pressures, saturations, and temperatures throughout the simulation. The

TOUCH2 SOFTWARE QUMIFICATION . rfumiARY 1996 41

 simulation was performed with the EOS3 fluid property module, using NK = HEQ = 2
(isothermal conditions) in block MULT! to solve just two equations per grid block. This
results in a total of N = 756 equations. A more complete discussion of the data input is
provided by Moridis and Pruess (1992).

An upstream weighting scheme for mobilities (MOP (11) = 2) was used in this
simulation. A no-flow top boundary was used because previous simulations had indicated
that the mass transfer through this block was insignificant. Water was injected directly into
four specific grid blocks by specifying appropriate sources.
Initial pressures and saturations were hand-calculated and assigned using the
hydrostatic pressure distribution below the water table and the capillary pressure vs. water
content relationship in Vauclin et al. (1979). Coding the capillary pressure function into
OUGH2 is a small effort; the FORTRAN code for the Vauclin et al. (1979) capillary
pressure function is provided in Moridis and Pruess (1995).
Figure 16 shows the TOUGH2 predictions and the experimental observations at the
desired locations and times. The agreement between experimental data and numerical
simulation results is good, but measurable local deviations are observed. Therefore, not all
areas of the curve meet an acceptance of 90% or better agreement to experimental data. The
reason for the deviations is due to the heterogeneity of the soil slab; this is discussed
extensively in Moridis and Pruess (1992).

WBaimsruin -T .

'| I I i i I i • • i I • • i i I " i « ! • • " I " ' ' I »

0.0 0.1 0.2 0.3

WaMr cwitinl (nfVin')

Figure 16. TOUGH2 predictions and experimental infiltration measurements atx = J.jfimin lest 8 (from
Moridis and Pruess. 199S).

TOUCH2 SOFTWARE QUALIFICATION - romuARY i <JS6 42

 9. Convection Cell fMoridis and Pruess. 1992; 1995)
This problem involves a laboratory convection cell experiment described, performed, and
modeled by Rcda (19S4). A porous medium consisting of glass beads with an average
diameter of 0.65 mm fills the annular region- between two vertical concentric cylinders.
Application of heat generates a thermal buoyancy force, giving rise to the development of
convection cells. This problem was used to test the ability of TOUGH2 to simulate
transient two-dimensional simultaneous heat and mass flow, and was discussed in detail by
Moridis and Pruess (1992).
To evaluate TOUGH2, numerical predictions were compared to measurements of
the thermal response of the surface of the inner cyliner at the elevations z - 0.4192,
0.62S8, 0.7336, and 0.7860 m (i.e. z/Ar = 2, 3, 3.5, and 3.75.). These comparisons
s
were made over a period of / = f„„ = 10 sec for a power level of 278.3 W/m. The strong
variation of temperature in the immediate vicinity of the heater, combined with the need to
include permeability enhancement effects in the simulations, required a grid with
sufficiently fine discretization near the beater. The MESHMAKER facility- of TOUGH2
was used to generate the grid. Finer spatial discretization was used near the heater with
successively larger blocks at increasing distance from the heater.

Two simulations were run. The first did not consider permeability enhancement.
The second simulation used nice different sub-domains of porous media. Initial pressure
and temperature conditions were set. No separate step of gravity equilibration was needed
because the process occurs very fast. The EOSi fluid property module was used for this
simulation with NK = I and NEQ = 2. The 416 grid Modes resulted in a iota! of N = 832
equations.

Simulation results were compared with the measurements made by Reda (1984) at
various distances from the heater over time. Figures 17 a and b compare the experimental
measurements and numerical predictions. Numerical results with and without permeability
enhancement are shown. A very good agreement between experiment and prediction is
observed for the period of transient convection, as well as for the eventual steady state.
Most parts of the curves show 90% or better agreement and therefore meet the acceptance
criterion for comparison to experimental data. Figure 17 a shows an extreme sensitivity of
temperature to radial distance from the beater in the immediate vicinity of the inner radius.
A less dramatic dependence of temperature on radial distance r above the heater is shown
in Figure 17 b.

Results of the simulation show that permeability enhancement has a significant

impact on the temperature distribution. Temperatures predicted without permeability
enhancement ate consistently higher above and below the heater. The temperature
differential is small initially, but keeps increasing during the transient period and stabilizes
as steady-slate is approached. Furthermore, the temperature differential appears earlier and
is more pronounced at the top of the heater. This indicates that without flow channeling
initiation of convection is slower, and a weaker convection process occurs at later times
when steady-state is approached. Neglecting channeling effects does not produce mote
accurate results despite its apparent better agreement with measurements near the steady-
state because of very steep temperature gradients near the heater. A more detailed
discussion is found in Moridis and Pruess (1992; 1995).

TOUCH2 SOmVARC QlMliriCVtlON - FEBRUARY VJ96 43

17a) 17b)

Figzire 17.
c Comparison o(TOUCH2 predictions \ii\h experimental data at the top and bottom f__r ~ S and 2
respectively) of the heating element in lest problem 9.
b. Comparison qfTOVCHl predictions with experimental data a z/Ar = 3.5 and z/A r = 3.75 (from
Moridis andPruess, 1995).

10. Two-phase flow (Moridis and Pruess, 1992)

This problem modeled flashing (vaporizing) flow from a synthetic sandstone core. The
experiment was performed and modeled by Kroger and Ramey (1974). Faust and Mercer
(1979) later independently modeled the experiment. This problem tests TOUGH2's ability
to simulate simultaneous heat and mass flow, as well as vaporization of water (phase
change). The saturated core cylinder is placed in an oven and heated to 19S.9"C. The left
boundary is insulated to prevent mass and energy flux, while the right end of (lie core is a
''rime-variable-pressure'' boundary with a pressure decline described by Updegraff (1989).
Prior to emplacement in the oven, the saturated core is brought to an initial pressure and
heated so as to affect a linear temperature variation between the left and right end. The
relative permeability curves in the experiment were described by Corey's equations (Kruger
and Ramey, 1974). Numerical predictions were compared to the measured temperature
distribution along the length of the core cylinder at a time of r = 300 sec.

Moridis and Pruess (1992) performed numerical simulations using TOUGH and
compared them to the results obtained by Updegraff (1989). The boundary conditions of
Updegraff s approach did not approximate the ones described in the Kruger and Ramey
(1974) experiment Updegraff asserted the; TOUGH could not handle the type of
boundary conditions called for by the experiment, but in fact it can handle them easily,
these being the constant temperature boundary at the outside edge of the core, zero
temperature gradient boundary, and the transient pressure boundary. Moridis and Pruess
(1992) modified Updegraff s approach and data inputs by creating two new data sets, but
kept his space discretization. To simulate the constant temperature boundary, they added a
second set of grid blocks, which formed a ring surrounding the cylindrical core and

TOUGHS! SOFTWARE QUAI If IGkTION - FFMIMW 199fi 44

 assigned a very large volume to each of the grid blocks of this domain. The addition of the
ring boundary created essentially a two-dimensional problem.

Moridis and Pruess (1992) introduced a very general transient-pressure boundary

condition by treating air as an ideal gas and submitting it to suitable time-dependent rates of
air injection and withdrawal. They created two data sets, one with the ring boundary and
one without, and made TOUGH runs using both files. By treating the boundary conditions
in the manner described in full detail in Moridis and Pruess (1992), they were able to
eliminate all of the problems encountered by Updegraff. Figure 18 shows numerical
results compared to experimental data of Kruger and Ramey (1974). The agreement is
satisfactory in meeting the acceptance criterion for comparison to experimental data;
observed discrepancies may reflect that experimentally achieved boundary conditions at the
core outlet may be more complex than presumed in the idealized one-dimensional model.

0.0 0.2 0.4 0.6 0.0 1.0

Rttttlv* dl*tane»

Figure IS. Comparison of TOUGH data and experimental data jor problem 10
(tram Moridis and Proas, 1992).

11. Warren-Root double-pornsitv solution fthis paper):

This problem is designed to examine the capability and accuracy of TOUGH2 in simulating
transient flow in a double-porosity medium. The problem concerns transient behavior of
water injection into liquid in a horizontal, uniformly fractured, infinite and isothermal
formation. Warren and Root (1963) presented an analytical solution for this problem
using a double-porosity approach.

A one-dimensional radial grid was generated for the TOUGH2 simulation. The
Infinite radial fractured/matrix domain is represented by 50 elements with an outer radius of
1,000 m and thickness of 10 m. The three-dimensional fracture network and cubic matrix
system are used in the TOUGH2 discretization. The matrix blocks are represented by 1 x 1
x 1 m cubes. The fracture permeability and aperture ate correlated by the cubic law.

The frxture/rnatrix properties and fluid parameters used are: fracture porosity <j», =
O.00Q6; matrix porosity $„ = 0.3; fracture permeability fc, = 1.0 x 10''W; matrix

TOLIGH2 SOFTWARE QUALIFICATION - FfSXIMRY 1996 4S

 6 a
permeability k, = 1.0 x I0"' m ; temperature T = 2S°C; rock compressibility C, = 0; fluid
10 -1 3
compressibility C, = 4.48 x 10" Pa ; and fluid viscosity u. = .898 x 10" Pa»s. The
double-porosity parameters as defined by Warren and Root (1963) are: a = 60; co = 2.0 x
3 5
10" ; X - 6 x 10" . The injection rate at the well is 0.01 nrVs, welliore radius is 0.1 m, and
the outer boundaries are treated as a constant pressure condition in the TOUGH2 simulation
instead of an infinite system. The early time solution will not be affected by the boundary.
A comparison of the TOUGH2 simulation and the Warren-Root solution is shov/n
in Figure 18 for the well pressure response. Figure 19 indicates that the TOUGH2
simulated result is in excellent agreement with the analytical solution for this problem, and
meets the acceptance criteria of 95% or better agreement, indicating typical doublc-porosity
behavior of two-parallel semi-log straight lines on a pressure plot

30 ""-•i ""••! " " " . , m

H """I '

S.
s
Ui
S SO
a.
a
o
£
15

10
1
10** 10"' 10' 10 10' 1ft* 10' 10* 10* 10* 10* 10* 10'*
DlmenslonlesaTime
Figure 19. Comparison at TOUOH2 lo Wurren-Rooidouble-porosity solution.

12. Lauwcrier heat transfer solution (this paper):

This problem is designed to examine the capability and accuracy of the TOUGH2 code in
simulating two-dimensional conductive and convectjvc heat transfer in porous media. The
problem concerns hot water injection into a linear water layer with a constant injection rate,
displacing the in-situ "cold" water, in which thermal conduction is igao;ed. At the same
time, heat transfer occurs perpendicularly into impermeable layers, in which only the heat
conduction in the direction perpendicular to the linear water layer is considered. Lauwcricr
(1955) derived an analytical solution for this two-dimensional heat transfer problem,
against which results of a TOUGH2 simulation were compared.

A two-dimensional rectilinear grid of 4,400 elements was generated using the

TOUGH2 MESHMA1CER for a grid of 10 x 60 m. In order to reduce the effects of spatial
discretization on the numerical .solution, a very fine grid was used; the 10 m length,
parallel to the linear water layer, was divided into 200 uniformly spaced gridblocks. Along
(lis direction perpendicular to the water layer, the 60 ni width was divided into 22 rows
noH-uuiformly, with the finest spacing next to the water layer. Also the mesh connections

TOUr,H2 SOFTWARF QuAUFICATiON • fFRRUARV 199ft 4G

 in the direction parallel to the water layer were deleted in the input file for consistency with
the analytical treatment which includes transverse heat conduction only.

The formation properties and fluid parameters used are: water layer porosity <•», =
5
1.0; impermeable layer porosity § - 0.3; water layer permeability k, = .89 x 1 0 " m ;
t

half width of water layer b = 0.05 m; rock compressibility Cr = 0; fluid compressibility C,

1
= 4.48X10" * Pa"'; fluid viscosity u\ = .89x 10"' Pa»s; water specific heat c„ = 4.17965
3
Mcg»°C; rock specific heat c = 1,000 J/kg»°C; water density p„= 1,000 kg/m ; and rock
t

3
grain dens/ty p, = 1,600 kg/m . The boundary and initial conditions are: pore velocity in
water layer i o m injection V = 5.0035 m/s; injection temperature T = 26 °C; and initial
w 0

temperature Ti = 25 ° C. The outlet end conditions for both water layer and impermeable
layer were treated as constant temperature boundaries.
A comparison of temperature profiles along the water layer &om the TOUGH2
simulation and the analytical solution is shown in Kgure 20 for three different times.
Figure 20 indicates that the TOUGH2 simulated results are in excellent agreement with the
analytical solution for diis problem and meet the acceptance criteria for having 5% or less
discrepancy, except at the thermal front where some slight numerical dispersion effects
exist.

Distance (m)

Figure 20. Temperature profiles for the analytical solution and the TOVGH2 simulation for the
Luuwerier heat transfer solution in Problem 12.

13. Correct handling of thermophvsical properties (this papert:

This is a trivial problem designed to confirm the validity of the theruiophysical properties of
the reservoir fluids in TOUGH2 simulations. The fluids considered in TOUGH2 include

TOUCH2 SOFTWARE QlMllfiCATION • FEBRUARr 1 9 9 6 47

 (a) water, (b) water vapor, (c), air, (d) C0 , and H . In this verification exercise only the
2 2

first three are considered.

TOUGK2 was run with a small number of gridblocks and with no communication
between them, thus affecting invariable conditions. Tae fluid properties at the different
conditions were then printed out, and were tabulated. In Table 10 the TOUGH2-caiculated
properties are compared to values obtained from CRC (1993) for a variety of pressure and
temperature conditions. The two sets of property values practically coincide, thus
confirming the correctness of thermophysical properties calculated in TOUGH2 and
meeting the acceptance criteria for being within 5% of actual values. In this case one would
want the TOUGH2 and actual values to match more closely than by 95%, and in fact, they
fall within less than 1% discrepancy.

Table W. Comparison ofTOUGH2 to tkermophysicai property datafrom steam tables in validation

problem 13.

TOTJGH2 RESECTS TABLES

Substance P(?a) TO PJ-PL frst'"*)

H ,K (J/k )
s L s
Pg.PL "*&'">''> Hg.f'L (Jlkg't

Air 1.0i3x:0 5 20 1.2036 ( p ) g 9.8825x10* <H ) 1.2036 (f»g) 9.8824x10* (Hg)

S

3.0x10 s !0 3.6903 ( ) P g 8.8624x10* (Kg) 3.6903 ( p ) g 8.8624x10* (Kg)

fi.OxlO 5 50 6.4671 (p ) 5 6.4671 ( p ) 3

!.2944x:0 ,'U-)
g
S.2943X10 (H ) g
g

Water/Vapor ;.oi3x1o s 5 1000.00 (p > L 2.1107x10* <K ) IOOOCPL)

2.1107x10* ai )
L L

1.013xlO S 20 998.32 (PO 8.3955x10* (HO 998.32 (p,J 8.3955x10* (H ) L

i.onxio 5
90 965.13 ifiO S
3.7696xlO ( H L ) 965.13 ipQ 3.7696x10 (H ) 5
L

1.013x10 s 100 0.59755 ( p ) g 2.676x10 (H )6

L
0.5*758 (p;) 6
2 676xi0 (!1[)

i.onxio 5
150 0.52323 ( p ) c 2 7762x10 (Hj,) s
0.52323 (p,j 2.7162x10 (H >6
L

3.0xl0 5 10 999.89 (p,,) 4 22«7xlfi*(H._) 999.89 (Pi) 4.2267x10* (H ) L

5.065xi0 5 10 999.99 <p ) L 4 24S8xl0 (H^) 6

999.99 (P]) 4.7488x10* (K ) L

5.065x10 s 30 995.93 ( p )L 12612X10 (HL) 5 995.93 ( p )L 1.7612xi0 (H >5

L

5.065x10 s 150 916.79 ( p )L 6.32I7X10 <H )5

916.79 ( P I J 6.3J17xl0 (K )5
L
L

4.6709x10 S 149.3 2.5032 <P;J 2.7445x10* (h„) 2.5032 (p|_) 2.7445x10* (H| j

917.43 (PL) S
6.291 l x l O ( H [ ) 917.43 (p^) s
6.2912x10 (HL)

TOUGH2 SOFIWAM QlMUHCATION - FtBKlMRY 1 9 9 6 48

 Table 10 (cont). Comparison ofTOUGH2 to tkermophysical property data from steam tables in
validation problem 13.

TOUGH2 RESULTS TABLES

1.013X10 6 100 958.56 (p i L

4.1975xl0 (H )5 S5836 (p ) L
3
4.197SxI0 (H ) L
L

1.013x10 s 150 917.9 (Pi) 6.324SX10 (H >5

917.9 ( p j 6.3248x10* (H )
L L

5.065x10 s 150 919.44 (p ) L

5
6.35X10 (HL) 919.44 (fii) S
6.3S01X10 (HL)

11.437 29.84 8.1787xlO~ (p )

S
2.5572xlQ (H )6
g S.1SX10" 5
2.557xi0 6
g

14. Vapor Pressure Lowering (this paper):

This problem examined the implementation of vapor pressure lowering as a function of
decreased matrix saturation (increased capillary suction). An analytical relationship for this
problem, -mown as the Kelvin equation, was presented by Edlefsen and Anderson (1943).
A one-dimensional grid composed of eight gridblocks was generated for this
simulation. Each gridblock was assigned a different liquid saturation, while all other
properties remained constant from block to block. The saturations tested varied fiom 0.5 to
0.001 (fully saturated = 1.0). Temperature was kept constant for all times at 26.85 "C
(300K), and all the properties of water were evaluated at this temperature. Other relevant
properties, which were used to calculate both the analytical solution and the numerical
(TOUGH2) solution were: vapor pressure of free water = 0.03531 bar, atmospheric
pressure = 1.01325 bar, molecular (formula) weight of water = 18.016 kg/kmole, and the
3
density of water = 997.0 kg/m . Capillary suction was evaluated using a standard van
Genuchtea (1980) model, and values for the van Genuchten parameters were chosen
arbitrarily to be representative of a tuff matrix. The parameters chosen were a = 0.001
and m = 0.412.

Agreement between the TOUGH2 calculations and the analytically calculated values
is excellent, as shown in Table 11. None of the numerically calculated values differed from
the analytical solution by greater than one-one hundred thousandths (0.00001) of a bar. In
no case did the relative error exceed three-hundredlhs (0.03) of one percent. Therefore, the
validation acceptance criterion was easily met for ihis test case.
Table 11. Comparison of vapor pressure lowering between TOUGH2 and Analytical
solution (Edlefsen and Anderson, 1943).
Saturation PvCTOUGIK) l'v (Analytical) Delta (bar) 9b error
0.S 0.03532 0.03531 o.oooot 0.03
0.1 0.03531 O.O3530 0.00001 0.03
0.05 0.03530 0.03529 o.ooooi 0.03
0.01 0.03514 0.03513 0.00001 0.03
0.005 0.03483 0.034S2 0.00001 0.03
0.003 0.0343! 0.03430 0.00001 0.03
0.002 0.03354 0.03354 0.00000 0.00
0.001 0.03075 0.03074 0.00001 0.03

TOUCH2 SOFTWARE QUALIFICATION - FEBRUARY' 1996 49

15a. Flow in a Heterogeneous Formation with Multiple Wells ("this paper):
Single-pliase, Slightly Compressible Liquid, 1-Dimensional, Heterogeneous
This problem describes flow in a horizontal heterogeneous system with multiple wells and
finite boundaries. The analytical solution to this problem was developed by Moridis (1995)
using the Transformational Decomposition method, which first defines the conditions at the
boundaries of the heterogeneous subdomains and then describes the pressure distribution
using analytical sub-solutions within each subdomain.
The problem has five subdomains. The various rocks in the subdomains are
assumed incompressible. The reservoir geometry, dimensions, and properties (porosity <j>
and permeability k), as well as the well rates and the well locations, ate presented in Tabic
12. The initial pressure is p = 5000 psi and the water compressibility is considered
3 1
constantat c = 3.134x10"° psH(2.200xlO- Pa" ). For the TOUGH2 simulation, the
w

domain was discretized into a total of 178 gridblocks.

Figure 20 shows the pressure drawdown at t = 200 days obtained from the analytical
method and the TOUGK2 simulation. It is obvjous that the two solutions literally coincide
and therefore that the acceptance criterion for the validation ate met.

Table 12. Reservoir Properties, Geometry, and Discretization in Test Problem 15a
Dimensions in (x,y,z): 5000 x 300 x 50 (ft)
Discretization: M s 17S Dx's of variable size, My = M = 1
x z

Sub- k(md) Subdomain # of Wells Well Rales Well position

domain
* length (ft) (obl/D) (local x, ft)

1 50 0.2 800 2 -30,20 100. 560

2 20 0.18 1200 3 -20 (all) 300, 770. 910

3 8 0.12 1400 1 -25 490

4 6 0.08 800 1 -25 650

5 2 0.07 SCO 0 - -

TOUCH2 SOFTWARE QUALIFICATION - FEBRUARY 1996 SO

 I.... I.... 1.... I li m i l , n l i i lit.

450.

£ 400-

350-
'I "•'»
C 1000 2000 3000 4009 5000
x(ft)
Figure 21. Comparison of the analytical and the TOVGH2 solutions to the problem of I'D flow through a
heterogeneous formation with multiple wells for problem 15a.

15b. Flow to a Single Well at the Center of an Infinite Reservoir (this paper):
Single-phase, Slightly Compressible Liquid, 2-Dimensional, Anisotropic
In this problem we compare the TOUGH2 and the analytical solutions of flow of
water to a vertical well located in the center (x = 0, y = 0) of an aquifer which is infinite in
area! extent, homogeneous, anisotropic, and of uniform thickness. The two-dimensional
analytical solution to this problem was developed by Papadopulos (1965) using integral
transforms in an approach very similar to the Theis (1935) solution, which assumes a fully
penetrating well and constant water and formation properties.

The analytical and the TOUGH2 solutions at t = 20 days were computed using the
following reservoir and fluid properties: T = 1000 m /day, Ty = 100 m'/day (directional
x

iransmissivities), 0 = 0.3 (formation porosity), S= 1.0x10 (storage coefficient), h = 50 z

m (formation thickness), q = 1000 m /day. For the TOUGH2 simulation, the domain was
discrelized into a total of 3,600 gridblocks (60x60 ia x,y). Due to symmetry, only one
quarter of the domain was considered (from 0 to infinity in both x and y).
Figure 22 shows the pressure drawdown along the x axis at y = 0 m, as well as along the
x=y axis. The agreement between the analytical and the TOUGH2 soiutions is excellent
and falls well within the acceptance criterion of having less than 5% discrepancy.

TOUGHS SOmVAJtr QUAtiriCATtON - fEERUARV 1936 Si

 Figure 22. Comparison of the analytical and tlie TOUGII2 solutions to tin problem of2-Dflow to a
single well in an anisotropic aquifer for problem 15b.

16. Single-phase transient flow simulation using an irregular grid (this paner'i:
This problem is designed to examine the capability and accuracy of the TOUGH2 code in
simulating transient flow using an irregular grid. The problem concerns transient
flow o f a slightly compressible liquid in a horizontal, uniform, infinite, and isothermal
aquifer, for which the analytical solution of Theis (1935) is available. The system is
initially at static conditions and is fully saturated with water. Water is injected through a
fully penetrating well with constant volumetric rate from t = 0.
The two-dimensional grid of the TOUGH2 simulation is shown in Figure 23.
Because of the symmetry of the problem, only a quarter of the flow geometry is discretized
and simulated. The infinite radial flow domain is represented by a rectangular domain of
2,000 x 2,000 m with thickness of 10 m. The irregular integral finite difference grid, as
shown in Figure 22, was generated based on randomly selected block centers, with a total
of 1,000 elements. Three locations of A (x=9.67ro, y=8.33m), B (x=26.90m,
y=73.02m), and C(x=305.12m, y=118.12m) are chosen at which pressures from the
analytical and numerical solutions were compared.

TOUCH2 SOfTWARt QlMIIPir-MlON - rmmiARV 1 9 M 52

 0 300 400 COD SCi 1000
X Coordinate (m)

Figure 23. Irregular grid constructed for problem 16.

The rock and iiuid parameters used are: porosity <J> = 0.2; permeability k= 1.0 x
s
10""m ; temperature T = 25°C; rock compressibility C, = 0; fluid compressibility C, =
,0 -3
4.48 x 10" Pa"'; and fluid viscosity u, = .898 x 10 Pa«s. The injection rate at the well is
0.001 m Vs, and the outer boundaries in the TOUGK2 simulation are treated as a Crst-type.
constant pressure condition. Therefore, the early time numerical solution can be compared
with the analytical solution before the finite boundary effects take place.

Comparison of the TOUGH2 simulation and the Theis solution is shown in Figure
24, for the three observations points, indicating overall excellent agreement between the
two solutions for all the three locations and demonstrating that me comparison meets the
acceptance criteria of having 5% or less discrepancy. It should be mentioned mar a well is
represented as a line source in the Theis solution; however, the well is approximated in the
TOUGH2 simulation by a finite soil column with a top area of 17.8 m \ This explains why
a small difference exists between the rwo solutions at location A, which is close to the well.

10UCH2 SorrwARE QUALIFICATION . FEBKUARV 1996 53

 Figure 24. Comparison of TOUCH2 simulation with Theis solution for an irregular grid,
as described in problem 16. ____

1
17a. Qns-djrocnstpnai heat condueppn fthjs papa );
This problem is designed to examine the accuracy of the TOUGH2 code in simulating
conductive heat transfer in porous media. Toe problem concerns heat conduction into a
semi-infinite linear rock column. Initially, the system is at uniform temperature, and a step
change in temperature is imposed on the boundary from t=0. Then heat starts to conduct
into the column. When effects of water flow on the heat transfer can be ignored, the heat
transfer problem becomes one of heat conduction in solids, for which many analytical
solutions are available. For the test problem of interest, an analytical solution from
Carslaw and Jaeger (1959) is used.
In the TOUGH2 simulation, a one-dimensional linear grid of 1,000 elements was
generated using the TOUGH2 MESHMAKER for a ten meter domain of unit cross area. In
order to eliminate the effects of fluid flow, rock porosity and permeability were set to zero.
The formation properties and thermal parameters used are: permeability k = 0.0 nr; rock
compressibility C, = 0; rock specific heat c, = 1,000 J/kg °C; thermal conductivity K=2.0
5
W/m°C; and rock grain density p , = 1,600 kg/m . The boundary and initial conditions are:
boundary temperature T = 50 • C; and initial temperature T,"= 25 " C. The outlet end
0

condition for the rock column was treated as a constant temperature boundary, the same as
the initial condition.
A comparison of the temperature profiles along the rock column from the TOUGH?,
simulation and the analytical solution is shown in Figure 25 for three different times, 1, 10,
and 50 days. Figure 25 indicates that the TOUGH2 simulated temperanire results are in
excellent agreement with the analytical solution for this problem and meet the acceptance
criteria of having 5% or less discrepancy from the analytical solution. In fact, it is difficult
to delect any discrepancy between the two.

TOUf.UJ SOFTWAHF QUALIFICATION . ffBRUARV 19<>6 54

 0 2 4 fi 8 10
Dtstanoo{m)

Figure 25. Comparison of analytical solution, and TOVGH2 results for heat conduction problem
in problem 17a.

17b. One-dimensional binary vapor diffusion (thispapert:

This problem is designed to examine the accuracy of the TOUGH2 code in simulating
vapor and air diffusion in the gas phase. The problem concerns air component diffusion
into a semi-infinite linear rock column. The system initially contains single-phase gas at
isothermal conditions of 2S°C. The gas phase consists of air and water vapor only, and
initially the mass fraction of vapor is 0.990 throughout. The adsorption of air or vapor on
the rock solids is ignored. At t = 0, the air mass fraction on the boundary is increased to 1.
Then the air component starts to diffuse into the column by binary diffusion. For this test
problem, the governing partial differential equation, boundary and initial conditions arc
identical to a heat conduction problem (pure diffusion). Therefore, an analytical solution of
heat conduction from Carsiaw and Jaeger (1959) is used in this comparison study.

In the T0UGH2 simulation, a one-dimensional linear grid of 1,000 elements was

generated using the TOUGH2 MESHMAKER for a 10 m domain of unit cross area. The
EOS3 fluid property module was used. IK. order to eliminate effects of advective gas flow,
the water saturation and die rock-permeability were set to zero. The formation properties
and diffusion parameters used are: permeability k = 0.0; rock compressibility Ci = 0;
formation temperature T « 25 "C; porosity <!> » 1; tortuosity t = 1; effective diffusivity
4 J s
<J>S,TD = 4.0975x;0' m /s and binary diffusion coefficient, D = 1.1707x10 m'/s. The
boundary and initial conditions are: boundary air mass fraction Xair = 1.0; and initial air
mass fraction Xair = 0.99. The outlet end condition for the rock column was treated as a
constant mass fraction boundary, the same as the initial condition.

A comparison of the air mass fraction profiles along the rock column from the
TOUGH2 simulation and the analytical solution is shown in Figure 26 for three different
times, 1, 5, and 10 days. Figure 26 indicates that the TOUGH2 simulated results arc in
excellent agreement with the analytical solution for this problem and meet the acceptance
criteria of having 5% or less discrepancy from the analytical solution.

TOURl 12 SOFTWARE QUAIMGATION • FCBRUARY 1996 55

 0 2 * 6 8 10
Distance (m)

Figure 26. Comparison air mass fraction profiles showing TOUCH! results and analytical solution for
problem 17b.

17c. One-dimensional s;as flow with Klinkenherp effects (this panerl:

This problem is designed to examine the capability and accuracy of the TOUGH2 code in
simulating porous medium gas flow with the Klinfcenberg effect (1941). The problem
concerns a steady gas flow problem across a linear rock column. The system contains
single-phase gas at isothermal conditions, and a constant gas mass injection rate is imposed
on the inlet of the rock column. The outlet end of the rock column is kept at a constant
pressure. Klinkenberg(1941) gave the following relation:

K«-ic(i+i) (1)

where J^. is the gas-phase permeability; and K^is the gas permeability at infinite pressure;
b is the fUinkcnbcrg coefficient (Pa); and P is the gas-phase pressure.
Under the steady state flow condition stated above, an analytical solution can be derived for
the gas pressure distribution along the rock column,

2
P(x)~->+{6 +/'|+24ft + 2? fi(i-.r)/PK-j "
m (2)

where P is the outlet boundary pressure; q is the gas mass injection rate per uuit area; n is
b m

the gas viscosity; L is the length of the rock column; and [3 = Mj/RT is the gas
compressibility factor.

TOUCH2 SOHWAM QUALIFICATION . FMKUARY 1996 56

 In the TOUGH2 simulation, a one-dimensional linear grid of 1,000 elements was
generated using the TOUGH2 MESHMAKER for a 10 m domain of unit cross sectional
area. In order to eliminate the effects of liquid, the water saturation was set to zero. The
formation and Klinkenberg parameters were selected from a laboratory study of welded tuff
atYuccaMountain(Reda,1987), The parameters used are: porosity <> = 0.3; permeability
2 s
k = 5x10"" m ; Klinkenberg coefficient b = 7 . 6 x 1 0 Pa; rock compressibility Cr = 0;
5 3
formation temperature T = 2 5 "C; and compressibility factor jj = 1.1885X10' kg/Pa»m .
The boundary conditions are: air mass injection rate q = 1x10* kgfe; and outlet boundary
m

pressure P = 1 bar.
b

A comparison of the pressure profile along the rock column from the TOUGH2
simulation and the analytical solution is shown in Figure 27. Figure 27 indicates that the
TOUGK2 simulated pressure distribution is in excellent agreement with the analytical
solution for this problem and meets the acceptance criterion of having less than 5%
discrepancy.

Distance (m)

Flgun 27. Comparison of the pressure profile along she rock column from the 'WUGH2 simulation and
the analytical solution for problem 17c

38. 3 - » How to a Horizontal Well ("this paper):

Single-phase, Slightly Compressible liquid, 3-Dimensianal, Anisotropic
This problem involves verification of TOUGH2 against the analytical solution tor flow of a
single-phase fluid to an infinite-permeability horizontal well located in a semi-infinite
homogeneous and anisotropic reservoir of uniform thickness and width. The three-
dimensional analytical solution to this problem was developed by Goode and
Thambynayagam (1987) using successive integral transforms. The solution involves a
slightly compressible liquid with constant compressibility, assumes constant liquid and
formation compressibilities, and neglects gravitational effects.

TOUCH2 SOmVAKE QlMUFIWION • FEBRUARY 1996 57

 Figure 2S shows a schematic of the horizontal well model and identifies a number
of important parameters. In the problem used for the comparison between the analytical
and the TOUGH2 solutions the following reservoir and fluid properties were used: k = x

100md, ky =50md,k = 10md (formation permeabilities), $=0.1 (formation porosity),

z
3 1
ct = 3.134x10"* psf' (= 2.200xl0" Pa" , total system compressibility), h - 220 ftz

(formation thickness), h = 2200 ft (formation width), L b = U0.5 ft, i

x z M= 109.5 ft,
s 3
Lxd = 800 ft, L = 500 ft. q = 2000 STB/D (3.68x10 m /S, well flow rate), m = 1.0 cp
w

(viscosity) and Bo = 1.0025 (fonnation volume factor, corresponding to a fluid density of

water (998.3 kg/hr"). The initial reservoir pressure is p = 5000 psi.

For the TOUGH2 simulation, the domain was discretized into a total of 9,996
gridblocks (17x2Sx21 in x,y,z) of non-uniform size. A very high permeability (5.0x10
md) was assigned to the k permeability of the wellbore to simulate the infinite permeability
x

assumption in the analytical solution. The well flow rate was distributed uniformly over
the iength of the well.
Figure 29 shows the pressure drawdown at the well (x = 1100 ft, y = 0 ft, z = 110
ft) over time. An excellent agreement between the analytical and the TOUG112 solutions is
observed, well within the acceptance criterion of less than 5% discrepancy, thus confirming
the validity of the TOUGII2 solution for this type of problem.

x=o

Figure 28. Schematic of the horizontal well system.

TOUCH2 SOFJWAMQUAIIFICATION - FHWUARY 1996 ss

 I

'I ' " " " 1 T

0.01 0.1 1 10 100
n e w time, In hr* 1000

Figure 29. Comporitou of the analytical and the TOUGH2 solutions (pressure drawdown at the welt) to
the problem of 3-D flow to a horizontal well in an anisotropic aquifer for problem IS.

Analytical solutions arc generally not available for three-dimensional problems

because their full complexity cannot be verified. However, the problem described above
with a 3-D horizontal well validates TOUGH2's ability to model three-dimensional
problems accurately. It is also worth noting the three-dimensional simulations described
by Moridis and Pruess (1995) because they obtained good results using data from nuclear
waste and geothermal settings. Their Test 10featuredsingle- and two-phase Sow under
non-iiothermai conditions in a three-dimensional geothermal reservoir model, as described
by Antunez et al. (1994). Moridis and Pruess (1995} employed an irregular grid in Test 11
to a three-dimensional model of the Cerro Prieto geothermal field featuring non-isotheimai
multi-phase flow with phase changes and fracture-matrix interactions. A similar test was
their Test 13 which depicted channelized two-phase flow in fractured media in three
dimensions at Yucca Mountain, as described by Pruess and Tsang (1994).
! 9. l.D Radial Flow of a Compressible Gas fthis nanert:
Single-phase, Compressible Gas, I-Dimensional, Homogeneous
This problem involves radial flow of a gas (compressible fluid) to a single vertical well
located at tne center of a bounded cylindrical formation. The analytical solution to this
problem was developed by Kabir and Hasan (1986) using perturbation techniques.
In the problem the reservoir is finite but infinite-acting due to its very large radius.
_ 2
The following reservoir properties were used: k = 2.367xl0 l2 m (formation
r

permeability), 0*0.35 (formation porosity), and h m 50 m (formation thickness). In both

z

TOUGH2 and the analytical solution the gas properties were taken to be those of air, which
was considered an ideal gas. The initial reservoir pressure was /> = 6x10 Pa, and the

TC.:"H2 sonwAM Quwt ICATION - FMRUARV 1996 59

 initial temperature was T = 50 °C. Gas was removed from the system at a rate of 1.15573
kg/sec. For the TOUGH2 simulation, the domain was discrerized into a total of 114
gridblocks of non-uniform size.
Figure 30 shows the distribution of the pressure drawdown along the r-axis at a
time t = 6.89 days from the beginning of gas removal. Under the assumption of ideal gas
properties, there is an excellent agreement between the analytica! and the TOUGH2
solutions, well within the acceptance criterion of having less than 5% discrepancy.

? to"

Distance from tlw well (m)

Figure 30. Analytical and TOVGH2 predictions of the pressure drawdown distribution in the gas
reservoir for problem 19.

20. Water absorption into porous matrix

This is a problem designed to examine the capability and accuracy of the TOUGH code in
simulating transient interaction between fractures and matrix in two phase flow conditions.
Zimmerman et al. (1990) developed an approximate analytica! solution for calculatiug water
absorption into porous spherical, cylindrical and slablike matrix blocks whose characteristic
curves are of van Genuchten-Mualem type. Here, the analytical solution is used to check
the numerical results from the TOUGH code.
A slab shape of blocks was chosen in this study, and a one-dimensional radial grid
was generated for the TOUGH simulation, the half thickness of die slab is 0.20 m, and is
subdivided into 20 grid blocks of equal volume. The block is initially at uniform saturation
of 0.6765. Then, the outer boundary is saturated with water at zero water potential from
H), so that water begins imbibing into the block.
The fracture/matrix properties used arc those that have been estimated for (he
Topopah Spring member ot the Paintbrush Tuff of Yucca Mountain: matrix porosity

TOUCH2 SOFTWAM QUALIFICATION - fwmMKV ivib 60

 ls 5
<j>„=O.K; matrix permeability k„ = 3.°xlO' m ; and van Genuchtcn parameters, n=3.04;
5
m=0.67I; S =0.984; S,=0.31S; and OJ=1.147xlO' ; water viscosity u=0.001 Fa.s.
s

A comparison of the TOUGH simulation and the analytical solution is shown in

Figure 31 for the relative water influx versus dimensionless time. Rgure 31 indicates that
the TOUGH2 simulated result is in reasonable agreement with the analytical solution for
this problem, in terms of estimation of water imbibition flux into the slab block.

Emensiontess TBTK. kt/a?#a*

Figure 32. normalized cumulative liquidflux for TOUGH end semi'Onalytical solution in slab
absorption problem of test problem 20.

4.3 Summary

Verification of design of TOUGH2 was met exclusively through numerical tests

and did not require the employment of additional procedures or confirmatory methods.
Software validation of TOUGH2 was accomplished by testing in the 20 test cases
described above. As a demonstration that the specified requirements for TOUGH2 have
been met, Table 13 provides a cross-check of the requirements against the validation
problems described above tiiat satisfy those requirements. As shown in Table 13, the
validation tests that were run for qualification purposes collectively satisfy the
Requirements Specification and all meet the acceptance criteria.

As additional examples of TOUGH2 applications that fit the software engineering

definition of "validation" because they demonstrate that the code can model the processes it
purports to model, the TOUGH2 Selected Bibliography (Appendix E) lists papers that
applied TOUGH2 to solve problems involving the processes listed under Requirements
Specification (e.g.. heat pipe, infiltration, convection cell).

TOUCH2 SonwARr QUALIFICATION - I-'EMUARV (996 61

 Table 13. Requirements Validation Cross-Check
Requirement Test Case I Acceptance ; Reference
Criteria !
' A. flow processes
gravity effects i 8.9 :
met i Pruess (1987)
capillary forces ! 3.8 ; Moridis and Pruess
viscous forces 1 6,8,11,15a, : (1992,1995)
> 15b, IS. 19
1
single-phase gas flow J, 6, 9, 11, ; met : Pruess < 1987)
: 15a, 55b, ; Prucss and Wu( 1993)
• 16, 18. 19 this paper, Moridis and
: Pruess (1992.1995)
two-phase flow of ' 3. 8. 10 '. met i Moridis and Pruess (1992,
liquid and gas '• 1995)

. Constitutive Relations
accurate description of thermophy; 13' this paper
properties :

dissolution of air in water ' this paper

: Pruess (1987)
i Mcridis and Pruess (1992)
vapor adsorption this paper
; Pruess (1987)
Moridis and Pniess (1S92)
' vapor pressure lowering this paper; Pruess (1987)
due to suction pressure

; C. Phase Change
/Interaction
phase (dis)appearance 2,10 < Morjdis and Pruess (1992)
I
phase interference - 3, 8, 10 j Moriois and Pruess
(liquid and gas) ; | (1992,1995)

permeability enhancement 17c this paper

from slip flow effects \
(Knudsen)
binary diirusion in gas 3,17b ! litis paper
phase •• Pnjcss(I9S7)
Muridis and Pniess (1992)
vapor-liquid phase change 2, 3, 10 Pruess(1987)
Moridis and Pruess (1992)

TOUGH2 SonrwARC QuAiiriCATiON - FEBRUARY 1996

 Table 13 (conO. Requirements Validation Cross-Check
Requirement \ Test Case j Acceptance [ Reference
i Criteria
1
-y
conduction 7, i2, 17a ; met [ this paper

4,5,9 , met : Moridis and Pmess (1992,1995)

sensible and latent 2, 3, 5, 9, 10 ! met Prucss(1987)

heat changes : Moridis and Prucss
(1992.1995)
conductive heat 7,12 ; Pruess and Wu (1993),
exchange with \ this paper
impermeable strata
coupledfluidand 2, 3, 4, 7, 10, : met j Pruess and Wu (1993)
heat flow 12 ; • Moridis and Proess (1992)
i Pruess(1987)
£. iBimenswaalHy .••<<,•- •SSX--Y
:
1-D I, 3,4, 6, 7, met 1
this paper, Pruess (1937)
II. 15a, 17a, ; ! Proess and Wu (1993)
17c, 19 : Moridis and Pruess (1995)

2-D 8, 9, 10, 12, , met Moridis and Pruess (1992)

15b, 16 this paper
Moridis and Pruess (1995)
this paper

~r-"=>*j
:
g. Geometry £ '§£&
•"-"r- •} -••«-• * - • -

Unear 1,3,4,17a, ! met i this paper, Moridis and

: 17b,17c i : Pruess(1992)
:
Ptness/1937)
Cancsian ,12 | this paper

cylindrical 9. SO ! Moridis and Pruess (1995)

Moridis and Pruess (1992)

radial , 2. 6. 7, 11, 19 I met ; this paper. Pruess and Wu

i (1993), Pruess (1987)
:
Moridis and Pruess (3995)
irregular '16 j met : this paper

1
^PaStlsteJi .fr; r -." 7•*'"t_C
i-K^if.*!
j - ^ ^ ^ j ^ ^ j •,_ „,
:.93fe85°fiSB '^^mL^^T
heterogeneous fornuiions : S, 15a, ISb, 18 ;' met this paper
flow in porous m.dia . 4, 6, 15a, i5b. i met Moridis and Pniess
18.20 • 0992, 1995)

TOUGI12 SOFTWARE QUAltrlCATION - FIBRIMRY 1996 63

 Table 13 (canl.). Requirements Validation Cross-Check
Requirement ! Test Case i Acceptance ; Reference
; Criteria
ilow in fiactuted ! 7, 8. 11 i met : this paper
: Pruess and Wu (1993)
Moridi5andPmess(1995)
I. flexible handling .10 ' Moridis and Prucss (1992)
of boundary
conditions, e t c
J. Other
!
automatic time- all cases met by its lhispaper; Pruess (1991)
steppin" existence

restart capability iNA met by its this paper; Prusss(1991)

, existence

compatible with 1,2,3 'demonstration • this paper: Pruess (1987)

TOUGH :
I that TOUGH anc Moridis and Pruess (1992)
TOUGH2 obtain
same results I
user documentation '• all cases ' met by its this paper; Pruess (1991)
i • existence
illustrative sample ! sample problem! met by their this paper; Pruess (1991)
problems-code 1-4 Sect. S.5 . existence
demonstration and ' this paper
accuracy
internal version all cases . met by its this paper; Prucss (1991)
control ! existence;
; see Fig. 4
'. in this paper

TOUGH2 SomvARt QUALIFICATION - FEBRUARY 199G 64

5. SOFTWARE USER DOCUMENTATION

5.1 Installation Procedures

The distribution media for TOUGH2 varies. For installation options on a worfc
station or mainframe, information should be obtained from the ESTSC distribution center at
Oak Ridge*. For a PC, the program is installed by inserting the diskette into drive B and
typing "install". Ir tractions for operating TOUGB2 will appear in the "README" file,
which is included a* Appendix F to this document

5.2 Hardware and Software Operating Environments

As stated previously, TOUGH2 can be operated in any type of environment: on a
mainframe computer, at a workstation, and on a PC or a Macintosh.

5.3 Input and Output

This section describes input and output files, file formats, options, default
parameters data files, commands, and execution. TOTJGH2 docs not specify acceptable
ranges of inputs and outputs; these are determined exclusively by whether the results make
reasonable sense and require knowledge on the user's part of operative processes.
Likewise, any errors that result will come about primarily due to the user's choice of
options, parameter values, and problem setup. The MULKOM family of codes, including
T"OUGH2, has been tested and debugged for over a decade with scores of applications.
Only one coding error is known to exist. This occurs in the naming convention for very
large grids in the meshmaking RZ2D submodule. The error will be corrected in an updated
version of TOUGH2. For code nomenclature see Appendix D.

5.3.1 Preparation of Input Data: TOUGH2 input is provided through a file INPUT,
organized into data blocks which are labeled by five-character keywords. With a few
exceptions, the order of data blocks is arbitrary. TOUGH2 input formats are compatible
with those of TOUGH. Figure 32 gives a listing of TOUGH input formats and indicates a
number of optional additional parameters that in TOUGH2 are provided through the same
data blocks. TOUGH2 also has a number of new, optional data blocks. These are listed in
Table 14, and the corresponding input formats are shown in Figure 33.

'Energy Scicence Technology Software Center

P.O. Box 1020, Oak Rldje, Tennessee 37831, U.S.A.
PJione: <<2J) 57CZ«06, Pax: (423)576.2865

TOUGH2 SOFTWARE QUAltriCATION - fCDRUARV 1996 65

 TOWH.Input « U h TflttCH2 Extcnalan*

UiTUu.l.lU I l.t L,UJ t U U I U . U 1-l.ltl 11.1 JJiJ f 111J U.I.LiJ-UJ.1 J-tLU.1 I I t) 11 U l U L t U 11 l.t.

r
i i I. <• . l . t . n i t i . I t I . > i ' H . . t l "'"~"''l n . m m l I ...
i i i i i t M l | | l . M i , . i . n . . . . .
| l n . . . , |
| r , • ,,.,,
, i i i | M M,
j^ jiija.'iu<. ia'l»ai'l " a''''|'''H- ''i iai r'Wtf''[ ""
• . t . l . t t . ! • . • . ! . • • • ! . . • • t... it.. ( I • ,.t t

Figure 32. TOUGH input formats with T0U0H2 extensions

TOUGH2 SOfTWAKC QIMIIHCAIION . FMRUARY 1996 66

 N « ( o p t i o n a l ) Dole Stocks Ir> TCUGH2

n.fiMw»««iE.»i f t ' " " " M " " " " " ' t i.T.i t \ f i t n m f t t

Um-li-UJLiJ.
l.t.t l t , M I I 1 l ^ 1 t l , H
P f »' > » n i m m f f f i i f m t i t t r m m i

®P " " " r i m m i m

' " " " " ' "
u m i n m u
• m n t i m n n i i i m m t t t M H m
" " " " M "

» " * f «»t n l m H M i i i m m . . m u m i m m m i
'""""""""
M l " 1 1
" " " '"•'• i i yti i H I I I H I 11 m i n m . 1.1 IJUU.1.X,*.U.TJJL

t^P.f.w.lrt* >... " " " " i t

,u .7^:
• • • " • " ' " " " " . • • m i n i m u m M m i
Figure 33. Input formats for new TOVGUi Jala Mocks

Table 14. New data blocks in T0VCH2 (optional)

Keyword Function
MESHM invokes internal mesh generation and processing facilities

MULTI allows to select number of fluid components and balance equations per grid block-
applicable only with certain EOS modules that offer different options

INDOM permits dotuain-speeifie initialization of thermodynamic conditions

NOVER if present, optionally suppresses a printout of versions and dates of the program units
executed in a T0UGH2 run
ENDH alternative to "ENDCY" for closing a T01TGH2 input file: will cause flow simulation to
be skipped; useful if only mesh generation is desired

The following discussion first summarizes the new parameters and options of
TOUGH2 in the TOUGH data blocks. It then discusses the input formats and choices
available through the new data blocks. For a discussion of input variables that are identical
to the ones used in TOUGH, one should refer to the TOUGH User's Guide (Prucss,
1987).

5.3.2 enhancements in TOUGH Blocks: Comments or text can be inserted between

data blocks anywhere in a T0UGH2 input file. Such records will generate a one-line
printed output that says "Have read unknown block label '(first five characters)' • ignore
this and continue reading input data", but will otherwise be ignored. (In TOUCH,
execution simply stopped when an unknown block label was encountered).

Several of the MOP parameters (first record in block 'PARAM') that control
optional printout and some ealculational choices, have different options and settings than in
TOUGH. Each TOUGH2 run will produce a one-page informative printout of availahle
selections and options chosen. Additional parameters provided through TOUGH data
blocks are as follows (see Figure 32). The second (optional) record in block "ROCKS' hat

TOUCH2 SOFTUMW QlMlWCATION . FOMMIrr 139b «7

a parameter GK which is the Kiinkenberg parameter b in the gas phase permeability
relationship k = k^l+WP). In partially saturated media, vapor diffusion can be
considerably enhanced in comparison to the expression given in Eq. A.7, due to phase
change effects (condensation/evaporation) at the pore level. Such enhanced diffusion can
be modeled by specifying a suitable value, typically or order 1, for the parameter B= <j>S.i.
This is to be entered as parameter BE in the first record in block PARAM. In block ELEM,
AHTX is the contact area of a grid block with the top or bottom boundary of Ihe flow
system. This can be used with certain EOS modules for a heat exchange calculation with
semi-infinite half spaces that represent the confining beds of a flow system (as in Sample
Problem 3, Appendix G). The X, Y, Z data in clement records are nodal point coordinates.
These are not used at all in TOUGH2, but can be optionally provided in the EUEME block
to facilitate plotting.

When working with different EOS modules, there is a need to be able to specify
injection of different fluid components (or heat). Trble 15 lists the TYPE specifications
that can be used in data block GENER in the input file.

Table IS. Specifications for Generation Types

Component
Codewords
Component (variable "Type"
Index in block GENER) EOS1 EOS2 EOS3 EOS4 EOS5
#1 COM1, MASS, WATE waterl water water water water
#2 COM2, AIR. WATR water2 CO :
air air H,
«•
*4*
COM3 - - •
- -
*NK! r
COM4
HEAT
-heat -heat .heal -
Ileal
-heal
-not used in EOS1 through EOS5
TJK1=NK+1

Thus, a user working with ti\e "two waters" option of EOS module EOS1 would
specify TYPE=COMI (or MASS, or WATE) to inject "water 1", while specification of
TYPE=COM2 (or AIR, or WATR) would allow injection of "water 2".
5.3.3 Input Formats for MESHMAKKR: At present there arc three sub-modules
available in MESHMAKER (see Figure 34): keywords 'RZ2D' or •RZZDL' invoke
generation of one- or two-dimensional radially symmetric R-Z meshes; 'XYZ' initiates
generation of a one, two, or three-dimensional Cartesian X-Y-Z mesh; and 'MTNC calls a
f
modified versif" " the 'GMDJC program (Pruess, 1983a) to sub-partition a "primary"
porous iimliuw inesb into * secondary mesb for fractured media, using the method of
"multiple interacting conlixiua" (Pruess and Narasimhan, 1982, 1985). The meshes
generated under keyword 'RZ2D' or 'XYZ' options, or assignment of 'ELEME' and
'CONNE' blocks in (he INPUT file must precede the MESHMAKER/M1NC data. Sec
Pruess (1991) lor preparation of input data for the three MESHMAKER sub-modules.
Complicated geometric settings may require application of external mesh generators.

TOUOH2 SomvAKC QlMllflCATrON - FWRIMftv 1996 CS

 HESHNNCER - T«-rfUtnal«n.l R-2 Crtdn

\**.W**,1* itm " " " . M . t , ! " " " " " i " ) 1
" " " " ) " " " 1

1 l * f l • ! • • .1 M i l l H I H .1 l i t .

• • " • • " " " t,li.LLLl.UJJ.i.t1i.t " " ' If

"tt | i ii.j,H .M,,i

1 , M
i h V m m n m r m ^ i m m m m m
k I i ! j V l

MMi^iUMtilimiiMilHitiinitniiiiliiltiiiiin.imuimli.1 i
rcLMial! 111111111111II niitin.il.
i M i i . i i i | i i n m i i
l

M
•***^it lnM 1 •" "' I.'
,
' '* "
"1 "' V. '' 1
i M H 1
imiiiiimiiii
HIM I il " .mi 1i " "«' . " ) i " " "
•LLU m z L m i u j L i J.i 11 n i I.I.LI.
""""" MfM.t *" H"I M f J . U J J M f T T t "l H" Ii InI1UJJJU
i|,f,[J,'

" : "• H »
'""""""""
U l . ' " ' I " " , U ' • " ? '•H'i-un 11111 t.u.ixi 11 m 111 n ) t tj-Ljm.LUJLu.i,
"m
" " " I ' " "

Figure 34. Input formats for MESHMAKER module.

T O U C H 2 SOITWAMQUAUHCATION • FttRUAKY 1 9 9 6 69
 5.4 User Features
Much of the data handling in TOUGH2 is accomplished through disk files which
are written in a format of 80 characters per record, so that code users can edit and modify
them with any norma] text editor. Table 16 summarizes the disk files other than (default)
INPUT and OUTPUT used in TOUGH2. Most of these are also used in TOUGH, and
files with the same names in both codes have identical formats. The use and function of
these files is described in the following sections and in the Appendix G Sample Problems
(Pruess, 1991). Further information is available in the TOUGH User's Guide {Pruess,
1987).

•table 16. TOVGH2 Disk Files

File Use
MESH written in subroutine INPUT from ELEME Mid CONNE data, or in moduie MESHMAKER from
mesh specification data
read in REHLE to initialize ail geometry data arrays used to define the discrelized flow problem
GENER written in subroutine INPUT fromGENER data
read in RF2LE to define nature, strength, and time-dependence of sinks and sources
INCON written in subroutine INPUT from INCONdau
read in KHLE to provide a complete specification of thermodynamic conditions
SAVE written in subroutine WRIFI to record thermodynamic conditions at the end of a TOUGH2
simulation run
compatible with formats of file or data block INCON for initializing a continuation run
MINC written in module MESHMAKER with MESH-compauble specifications, to provide all geometry
data for a rractured-porous medium mesh (double porosity, dual permeability, etc J
read (optionally) in subroutine RFILE to initialize geometry data for a fractured-porous system
UNEQ written in linear equation solver "MA28", to provide informative messages on linear equation
solution
TABLE (optional; available only with certain EOS modules) written in subroutine QLOSS to record data
on heat exchange with impermeable confining layers, or heat and fluid exchange with embedded
matrix blocks in afcaetured-porousmedium
read in QLOSS in a continuation run
VERS written in all TOUGH2 program units with informational message on version number, dale, and
function
read in main program "TOUGH2" and printed to default OUTPUT at the conclusion of a TOUGH2
-
simulation run; printing of version information is suppressed when keytvork 'NOVER is present
in INPUT file

5.5 Summary
Verification of this phase of the Software Life Cycle which relates to installation.
user documentation, and provision of code demonstration cases was completed by
providing with this package the report TOUGH2 - A General Purpose Numerical
Simulator for Multipliase Fluid and Heat Flow (Pnicss, 1991) and the sample problems
described in Appendix G which demonstrate how the user sets up certain types of problems
and manipulates various aspects of the code.

TQUGH2 SorTWAtt QUALIFICATION . rtuRUARv 3 9 % 70

5.6 Version History
TOUGH2 was released in May 1991 and transferred to the National Energy
Software Center, which performed some tests on it before the Center was disbanded. The
code was then transferred to DOE'S Energy Science and Technology Software Center
(ESTSC). TOUGH2 remained unchanged until November 1994 when the T2CG1 package
of conjugate gradient solvers was added. The November 1994 Version 1.11 of TOUGH2
is the version described in this software qualification package. An adaptation of this
version for PC's was transferred to ESTSC in January 1995. It inchdes only very minor
modifications, and is not considered a separate, distinct version of the code. Additional
TOUGH2 modules are in use, e.g. a module with hydrodynamic dispersion. These will be
released in due course, following extensive testing, documentation, and some packaging.
When they are ready for release, a new version of TOUGH2 will be senr to ESTSC.

TUUUH2 iOmvAKE QU/UINCMION • FttKIMKV 1996 7T

ACKNOWLEDGMENTS
The authors are grateful to Donald Mangold and Bo Bodvarsson for helpful
discussions during this work and to Robeit Zimmerman and Mark Bandurraga for
contributing to die bibliography. Thanks are due to Jerry Fairley for providing Test Case
14. Xahan Noorishad, Donald Mangold, Kenzi Karasaki, and Bo Bodvarsson provided
thoughtful reviews which improved the manuscript greatly. We appreciate die efforts of
MariaFink, Vicki Franco, and Carol Taliaferro in processing the document. Funding for
this quality assurance effort was provided through the USGS. This work was supported
fay [DOE-OCRWM-YMP-M&O] under DOE contract no. DE-AC03-765F00098.

TOIJCH2 SOFTWARE QUAIIFICATION • rFwuiARY 1911". 72

REFERENCES
Adenekan, A.E., Numerical Modeling of Multiphase Transport of Multicompoaent Organic
Contaminants and Heat in the Subsurface, Ph.D. Dissertation, Univ. of California
Berkeley, 1992.
Adenekan, A.E., Patzek, T.W., and Pruess, K., Modeling of Multiphase Transport of
Multicomponent Organic Contaminants and Heat in the Subsurface: Numerical Model
Formulation, Water Resources Research, 29{ll), pp. 3727-3740.1993.
Ahiers, C.R, Bandurraga, T.M., Chen, G., Finsterle, S., Wu, Y.S., Bodvarsson, G.S.,
Kwicklis, E., Rousseau, I., and Flint, L., Performance analysis of the LBNL/USGS
three-dimensional unsaturated zone site-scale model, (in press, a).

Ahiers, C.G., Bandurraga, T.M., Chen, G., Finsterle, S., Wu, Y.S., and Bodvarsson,
G.S., Summary of model calibration and sensitivity studies using the LBNL/USGS
three-dimensional unsaturated zone site-scale model (in press, b).
Antuncz, A. Moridis, G., and Pruess, K., Large-Scale Geothermai Reservoir Simulation
on PCs, Report LBL-35192, Lawrence Berkeley National Laboratory, Berkeley,
California, 1994.
Arnold, B.W., Altaian, S. J., Robey, T.H.. Barnard, R.W., and Brown, T.J.,
Unsaturated Zone Fast-Path Flow Calculations for Yucca Mountain Groundwater
Travel Time Analyses (GWTT-94). Report SAND95-0857, Saudia National
Laboratories, Albuquerque, New Mexico, September 1995.

' Atkinson, P.G., Celati, R., Corsi, R., and Kucuk, F., Behavior of the Bagnore
Stearn/COj Geothermai Reservoir, Italy, Soc. Pet. Eng. J., pp. 228-238. August 1980.
Avdonin, N.A., Some Formulas for Calculating the Temperature Field of a Stratum
Subject to Thermal Injection, Neft'i Gas, 3. pp. 37-41.1964.

Aziz, K., and Settari, A., Petroleum Reservoir Simulation, Applied Science Publishers,
Ltd., London, 1979.
Bodvarsson, G.S., Pruess, K., Hfliifcwa, C , and Ojiambo, S.B., Evaluation of Reservoir
Model Predictions for the Olkaria Geothermai Field, Kenya, Geothermics, 19(5), pp.
399-414,1990.
Bodvarsson, G.S., Chen, G., and Wittwer. C , Preliminary Analysis of Thrce-
Dimensional Moisture Flow within Yucca Mountain, Nevada, Proceedings of the Fifth
Annual High-Level Radioactive Waste Management Conference, American Nuclear
Society and American Society of Civil Engineers, Las Vegas, Nevada, pp. Xxxxx,
May 1994.

Bodvarsson, G.S., Chen, G., Zimmerman, R.W., and Kwicklis, E.M., Accuracy and
Efficiency of Semi-analytical Dual-Porosiry Simulator for Row in Unsaturated
Fractured Rock Masses, Radioactive Waste Management and Environmental
Restoration, 19. pp. 193-208, 1994, and Report LBL-34323, Lawrence Berkeley
Laboratory, Berkeley, California, 1994.

JOUCH2 SOFTWARE QUAUHCATION . FCIKUAKV 1996 73

 Caislaw , H. S., and Jaeger, J. C , Conduction of Heat in Solids , Oxford, Clarendon
Press, pp.60, 1959.
Doughty, C , and Pruess, K„ A Similarity Solution for Two-Phase Fluid and Heat Flow
Near High-Lcvel Nuclear Waste Packages in Porous Media, Int. J. Heat and Mass
Transfer, 33, pp. 1205-1222, 1990.
Doughty, C , and Pruess, K., A Similarity Solution tor Two-Phase Water, Air, and Heat
Flow Near a Linear Heat Source in a Porous Medium, Report LBL-30051, Lawrence
Berkeley National Laboratory, Berkeley, California, 1991.
Doughty, C , and Pruess, K., A Similarity Solution for Two-Phase Water, Air, and Heat
How Near a Linear Heat Source in a Porous Medium, J. Geophys. Res. 97(B2), pp.
182101838, 1992.
Duff, I.S., MA28 - A Set of FORTRAN Subroutines for Sparse Unsynunetric Linear
Equations, AERE Harwell Report R 8730, July 1977.
Eastman, G.Y., The Heat Pipe, Scientific American, 218(5), pp. 38-46, May 1968.

Ediefsen, N.E., and Anderson, A.B.C., Thermodynamics of Soil Moisture, Hilgardia,

15, pp. 31-298, 1943.
Edwards, AX., TRUMP: A Computer Program for Transient and Steady State
Temperature Distributions in Multidimensional Systems, National Technical
Information Service, National Bureau of Standards, Springfield, Virginia, 1972.
Falta, R.W., and Pruess, K., STMVOC User's Guide, Report VBL-30758, Lawrence
Berkeley National Laboratory, Berkeley, California, June 1991.
ralta, R.W., Pruess, K., Finsterle, S„ and Battistelli, A. T2VOC User's Guide, Report
LBL-36400, Lawrence Berkeley National Laboratory, Berkeley, California, March
1995.
Falta, R. W., Pruess. K., Javandal, I., and Wiiherspoon, P.A.. Numerical Modeling of
Steam Injection for the Removal of Nonaqueous Phase Liquids from the Subsurface. 1.
Numerical Formulation, Water Resources Researrch, 28(2), pp. 433-449,1992.
Falta, R. W., Pruess, K., Javandal, I., and Withcrspoon, P.A., Numerical Modeling of
Sream Injection for the Removal of Nonaqueous Phase Liquids from the Subsurface. 2.
Code Validation and Application, Water Resources Researrch, 2S(2), pp. 45 J-564,
1992.
Faust, C.R., and Mercer, J.W., Geothermal Reservoir Simulation. 2. Numerical Solution
Techniques for Liquid- and Vapor-Dominaled Hydrothennal Systems, Water
Resources Research, 15, pp. 31-46,1979.
Forsyth, P.A., Wu, Y.S.. and Prucss, K., Robust numerical methods for saturated-
unsaturated Dow with dry initial conditions in heterogeneous media, Advances in Water
Resources, IS, pp. 25-38,1995.

TOIJGHJ SOFTWARE Q I M I IHMTION • FFRMMRV 1996 74

 Finsterlc, S., 1T0UGH2 User's Guide Version 2.2, Report LBL-34581, Lawrence
Berkeley Laboratory, Berkeley, California, 1993.
Finsterlc, S., and Pruess K., "3TOUGH2: Solving TOUGH Inverse Problems,"
Proceddings of the TOUGH Workshop 1995, Report LBL-37200, Lawrence Berkeley
Laboratory, Berkeley, California, pp. 287-292, March 20-22,1995.

Garg, S. K., Pressure Transient Analysis for Two-Phase (Liquid Water/Steam)

Geoihermal Reservoirs, Paper SPE-7479, presented at 53rd Annual Fall Technical
Conference and Exhibition of we SPE, Houston, Texas, October 197S.
Goode, P.Q., and Thambynayagam, R.K.M., Pressure Drawdown and Buildup Analysis
of Horizontal Wells in Anisotropic Media, SPE Formation Evaluation, pp. 683-697,
1987.
Hadiey, R.G., Theoreiical Treatment of Evaporation Front Drying, Int. J. Heat Mass
Transfer, 25(10), pp. 1511-1522, 1982.
liayden, N.K., Johnstone, J.K., and Peters, R.R., Parameters and Material Properties for
Hydrologic Modeling of the Topopah Spring Unit, Memo to Distribution, Sandia
National Laboratories, Albuquerque, New .Mexico, September 27,1983.
Hirschfelder, J.O., Curtiss, C.F., and Bird, R.B., Molecular Theory of Gases and
Liquids, John Wiley and Sons, Inc., New York, 1954.
Ho, C , and Eaton, R., Studies of Tnerniohydrologic Flow Processes Using TOUGH2,
Report SAND94-2011, Sandia National Laboratories, Albuquerque, New Mexico,
December 1994.
Hunt, J.R.. Sitar, N., and Udell, K.S., Nonaqueous Phase liquid Transport and Cleanup,
2, Experimental Studies, Water Resources Research, 24(8), pp. 1259-1269,1988.
International Formulation Committee, A Formulation of the Thermodynamic Properties of
Ordinary Water Substance, IFC Secretariat, DAsseldorf, Germany, 1967.
Kabir, C.S., and Hasan, A.R., Pre&acture Testing in Tight Gas Reservoirs, SPEFE, pp.
128-138, April 1986.

Kiinkenberg, L.J., The Permeabiliiry of Porous Media to Liquids and Gases, 'm.APl
Drilling and Production Practice, pp. 200-213, 1941.
Knudsen, M. Die Gesetze der Molckularstrocmung und der Inneren Reibungsslroeuiung
der Gase durch Roehren, Armelen der Physlk, 28, pp. 75-131,1909.
Kruger, P., and Ramey, H.J., Stimulation and Reservoir Engineering of Geothermal
Resources, Stanford Geothermal Program Report SGR-TR-1, Stanford University,
Stanford, California, June 1974.
Lam, S.T., Hunsbcdt, A., Kruger, P., and Pruess, K., Analysis of the Stanford
C-eotncrmal Reservoir Model Experiments Using Uje LBL Reservoir Simulator,
Geothermics, 17, pp. 595-605, 1988.

TOUCU2 SomVASE QUAllflCATION - FtCMMftr 1 9 9 6

rs
 Lauwerier, H.A., The Transport of Heat in an Oil Layer Caused by the Injection of Hot
Fluid, Applied Science Research, 5, Section A(2-3), pp. 145-150, Martinus Nijhoff,
Publisher, The Hague, 1955.
Loomis, A.G., Solubilities of Gases in Water, in E.W. Washburn (ed.). International
Critical Tables, Vol. m, pp. 255-257, McGraw-Hill, New York, 1928.
Mangold, D., Qualifying Codes Under Software Quality Assurance: Two Examples as
Guidelines for Codes That Are Existing or Under Development, Report LBL-34423,
Lawrence Berkeley National Laboratory, Berkeley, California, May 1993.
McKibbm, R., and Pruess, K., Some Effects of Non-Condensible Gas in Gcotherma!
Reservoirs with Steam-Water Counterflow, Geotkermics, 18(3), pp. 367-375,1989.
Moridis, G., The Transformational Decomposition Method for Compressible Fluid Flow
Simulations, SPE Advanced Technology Series, 3, pp.163-172,1995.
Moridis, G-, and Pruess, K., TOUGH Simulations of Updegraff s Set of Fluid and Heat
Flow Problems, Report LBL-32611, Lawrence Berkeley National Laboratory,
Berkeley, California, 1992.
Moridis, G., and Pruess, K., Flow and Transport Simulations Using T2CG1, a Package
of Conjugate Gradient Solvers for the TOUGH2 Family of Codes, Report LSL-
36235, Lawrence Berkeley National Laboratory, Berkeley, California, April 1995.
Narasimhan, T.N., and Witherspoon, P.A., An Integrated Finite Difference Method for
Analyzing Fluid Flow in Porous Media, Water Resources Research, 12(1), po. 57-64.
1976.
Niemi, A., and Bodvarsson, G.S., Preliminary Capillary Hysteresis Simulations in
Fractured Rocks, Yucca Mountain, Nevada, in Rapid and Far-reaching Hydrologic
Processes in the VadoseZon, J. Contain. Hydrology, 3, pp. 277-291,1988.
Oldenburg, CM., and Pruess, K., A Two-Dimensional Dispersion Module for the
TOUGH2 Simulator, Report LBL32505. Lawrence Berkeley National Laboratory.
Berkeley, California, 1993.
O'Sullivan, M.J., A Similarity Method for Gcothermal Well Test Analysis, Water
Resources Research, 17(2), pp. 390-398,1981.
O'Sullivan, M.J., Bodvarsson, G.S., Pruess, K., and Blakely, M.R., Fluid and Heat
Flow in Gas-rich Geolhermai Reservoirs, Soc. Pet. Eng. J., 25, pp. 215-226, April
1985.
Papadopulos, I.S., Nonsteady Flow to a Well in an Infinite Anisotropic Aquifer, Symposia
Intern. Assn. Sci. Hydrology, Dubrovnik, 1965.
Peaceman, D.W., Fundamentals of Numerical Reservoir Simulation, Elsevier,
Amsterdam, 1977.
Philip, J.R., Numerical Solution of Equations of the Diffusion Type with Diffusivity
Concentration Dependent, Tram. Faraday Sac, 51, pp. 885-892, 1955.

TOUCH2 SOFTWARE QlMUHCATION . FMKUARY 1 9 9 6 7fi

 Pritchett, J. W., Rice, M. H., and Riney, T. D., Equation-of-state for Water-Carbon
Dioxide Mixtures: Implications for Baca Reservoir, Report DOE/ET727163-8,
Systems, Science and Software, La Jolla, California, February 1981.
Pruess, K., GMINC - A Mesh Generator for Flow Simulations in Fractured Reservoirs,
Report tSL-15227, Lawrence Berkeley National Laboratory, Berkeley, California,
1983a.
Prucss, K., Development of the General Purpose Simulator MULKOM, Report LBL-
15500, Annual Report for Earth Sciences Division - 1982, Lawrrence Berkeley Lab,
Lawrence Berkeley National Laboratory, Berkeley, California, 1983b.

Pruess, K., TOUGH User's Guide, Report NUREG/CR-4645 (Nuclear Regulatory

Commission, Washington, D.C.), and Report LBL-20700, Lawrence Berkeley
National Laboratory, Berkeley, California, 1987.
Pniess, K., SHAFT, MULKOM, TOUGH: A Setof Numerical Simulators for Multiphase
Fluid and Heat Flow, Geotermia, Rev. Mex. Geoenergia, 4, pp. 185-202; also Report
LBL-24430, Lawrence Berkeley National Laboratory, Berkeley, California, 1988.
Pruess, K., (editor). Proceedings of the TOUGH Workshop, Report LBL-29710,
Lawrence Berkeley National Laboratory, Berkeley, California, September 1990.
Pruess, K., TOTJGH2 - A General-Pnrpose Numerical Simulator for Multiphass Fluid and
Heat Flow, Report LBL-29400, Lawrence Berkeley National Laboratory, Berkeley,
California, 1991.
Pruess, K., (editor). Proceedings of the TOUGH Workshop 1995, Report LBL-37200,
Lawrence Berkeley National Laboratory, Berkeley, California, April 1995.
Pruess, K., and Bodvarsson, G. S., A Seven-point Finite Difference Method for Improved
Grid Orientation Performance in Pattern Steam Floods, Paper SPE-12252, presented at
the Seventh Society of Petroleum Engineers Symposium on Reservoir Simulation, San
Francisco, California, 19S3.
Pruess, K., and Bodvarsson, G. S., Thermal Effects of Reinjection in Geothermal
Reservoirs with Major Vertical Fractures, J. Pet. Tech., 36, pp. 1567-1578,1984.
Pruess, K. Calori, C, Celati, R., and Wu, Y. S., An Analytical Solution for Heat Transfer
at a Boiling Front Moving Through a Porous Medium, Int. J. Heat and Mass Transfer,
30, pp. 2595-2602,1987.
Prucss, K., and O'Sullivan, M., Effects of Capillarity and Vapor Adsorption in the
Depletion of Vapor-Dominated Geothermai Reservoirs, paper presented at 17th
Workshop on Geothermal Reservoir Engineering, Stanford University, Palo Alto,
California, January 1992.
Prucss, K., and Narasimhan, T. N„ On Fluid reserves and the Production of Superheated
Steam from Fractured Vapor-dominated Thermal Reservoirs, J. Geophys. Res., 87,
(Bll),pp. 9329-9339, 1982.

TOUCH;: SOFTWARE QUALIFICATION - FEBRUARY 1996 77

 Pruess, K., and Narasimhan, T. N., A Practical Method for Modeling Fluid and Heat
Flow in Fractured Porous Media, Soc. Pet. Eng. J., 25, pp. 14-26,1985.
Pruess, K., and Tsang, Y., Modeling Strongly Heat-Driven Flow Processes at a Potential
High-Lcvcl Nuclear Waste Repository at Yucca Mountain, Nevada, Proceedings of the
Fourth Annual International High Level Radioactive Waste Management Conference,
American Nuclear Society and American Society of Civil Engineers, Eds., April 1993,
and Report LBL-33597, Lawrence Berkeley Laboratory, Berkeley, California, 1993.
Pruess, K„ and Tsang, Y„ Thermal Modeling for a Potential High-Level Nuclear Waste
Repository at Yucca Mountain, Nevada, Report LBL-3538I, Lawrence Berkeley
National Laboratory, Berkeley, California, 1994.
Pruess, K., Wang, J.S.Y., and Tsang, Y.W., On Thermohydrological Conditions Near
High-Lcvel Nuclear Wstes Emplaced in Partially Saturated Fractured Tuff. Part 1:
Simulation Studies with Explicit Consideration of Fracture Effects, Water Resources
Res., 26, pp. 1235-1248, 1988.
Pruess, K., and Wu, Y.S., Ou. PVT-Data, Well Treatment, and the Preparation of Input
Data for an Isothermal Gas-Water-Foam Version of MULKOM, Report LSL-25783,
Lawrence Berkeley National Laboratory, Berkeley, California, August 1988.
Pruess, K„ and Wu, Y.S., A New Semianalytical Method for Numerical Simulation of
Fluid and Heat Flow in Fractured Reservoirs, Paper SPE-18426, presented at Tenth
SPE Symposium on Reservoir Simulation, Houston, Texas, Februaray 1989.
Prucss, K„ and Wu, Y.S., A New Semianalytical Method for Numerical Simulation of
Fluid and Heat Flow in Fractured Reservoirs, Soc. Pet. Engineers Adv. Teclu Ser., i.
pp. 63-72, 1993.
Reda, D.C., Natural Convection Experiments About a Finite-Length Cylindrical Heat
Source in a Liquid-Saturated Porous Medium, Report SAND83-2209C, Sandia
National Laboratories, Albuquerque, New Mexico, 1984.
Reda D. C , Slip-flow experiments in welded tuff: the Knudson diffusion problem.
Coupled Processes Associated with Nuclear Waste Repositories, Tsang, C. F., ed.,
pp. 4S5-493. 1987.
Reeves, M., Ward, D.S., Johns. N.D., and Cranweil. R.M., Theory and Implementation
for SWIFTO, The Sandia Waste Isolation Flow and Transporr Model for Fractured
Media, Release 4.84, NUREG/CR-3328, and SAND83-U59, Sandia National
Laboratories, Albuquerque, New Mexico, August 1986.
Ross, 8., Mercer, J.W., Thomas, S.D., and Lester, B.H., Benchmark Problems for
Repository Siting Models, Report NUREG/CR-3097, Nuclear Regulatory
Commission, Washington, D.C., December 1982.
Rulon, J., Bodvarsson, O.S., and Montazer, P., Preliminary Numerical Simulations of
Ijrouodwater Flow in the Unsaturated Zone, Yucca Mountain, Nevada, Report LBL-
20553, Lawrence Berkeley National Laboratory. Berkeley, California, 1986.

TOUGH2 SOFTWARE QUALIFICATION - FEUMMRY 1996 7«

 Stanford Geothermal Program (ed.). Proceedings Special Panel on Geothermal Model
Intercomparison Study, Report SGP-TR-2, presented at Sixth Workshop on
Geothermal Reservoir Siting, Stanford University, Palo Alto, California, December
1980.

Sutton, F.M., and McNabb, Boiling Curves at Broadlands Field, New Zealand, N. Z. J.
Sci. 20, pp. 333-337.
Thcis, C. V., The relation between the lowering of the piezometric surface and the rate and
duration of discharge of a well using ground-water storage, Trans., AGU , pp. 519-
524, 1935.
Udell, K.S., and Fitch, J.S., Heat and Mass Transfer in Capillary Porous Media
Considering Evaporation, Condensation, and Non-Condensible Gas Effects, paper
presented at the 23rd ASME/AIChE National Heat Transfer Conference, Denver,
Colorado, pp. 103-110, American Society of Mechanical Engineers, New York,
August 1985.
Updegraff, CD., Comparison of Strongly Heat-Driven Flow Codes for Unsaturated
Media, , Report SAND 88-7145, Sandia National Laboratories, Albuquerque, New
Mexico, August 1989.
VanGenuchten, M. Th., A Closed-Form Equation for Predicting the Hydraulic
Conductivity of Unsaturated Soils, Soil Sci. Soc. Am. J., 44, pp. 892-898,1980.
Vauclin, M., Khanji, D., and Vacchaud, G., Experiemntal and Numerical Study of a
Transient, Two-dimensional Unsaturaled-Saturated Water Table Recharge Problem,
Water Resources Research, 15(5), pp. 1089-1101,1979.
Vcrma, A.K., Pruess, K., Tsang, C.F., and Witherspoon, P.A., A Study of Two-Phase
Concurrent Flow of Steam and Water in an Unconsolidated Porous Medium, paper
presented at 23rd ASME/AIChE National Heat Transfer Conference, Denver,
Colorado, August 1985.
Vlnsome, P.K.W., and Westcrveld, J., A Simple Method for Predicting Cap and Base
Rock Losses in Thermal Reservoir Simulators, J. Canadian Pet. Tech., pp. 87-90,
July-September 1980.

Warren, J.E., and Root, P.J., The Behavior of Naturally Fractured Reservoirs, Soc. Pet.
Eng. J„ pp. 245-255, Transactions, AIME, 228,1963.
Witherspoon, P.A.,« al., Thermohydrologic Modeling and Testing Program: Peer Review
Record Memorandum, January 1996.
Wittwcr, C.S., Bodvarsson, G.S.. Chomack, M.P., Flint, A.L., Flint, L.E., Lewis. B.
B., Spcngler, R.W., and Rautman, C.A., Design of a Three-Dimensiosal Site-Scale
Model for the Unsaturated Zone at Yucca Mountain, Nevada, Proceedings of the Third
Annual International Conference on High Level Radioactive Waste Management,
American Nuclear Society and American Society of Civil Engineers, eds.. Las Vegas,
Nevada, April, 1992.

TOUCH2 SOFTWAM QlMliriCATSON - FORIMRY 1 9 9 6

»
 Wittwer, C.S., Bodvarsson, G.S., Chornack, M.P., Flint, A.L., Hint, L.E., Lewis, B.
£., Spengler, R.W., and Rautnian, C.A., Design of a Three-Dimensional Site-Scale
Model for the Unsaturated Zone at Yucca Mountain, Nevada, Proceedings of the Third
Annual International Conference on High Level Radioactive Waste Management,
American Nuclear Society and American Society of Civil Engineers, eds., Las Vegas,
Nevada, April, 1992.

Wittwer, C.S., Chen, G., and Bodvarsson, G.S., Studies of the Role of Fault Zones on
Fluid Flow Using the Site-Scale Numerical Mo.del of Yucca Mountain, Proceedings of
the Fourth Annual International Conference on High Level Radioactive Waste
Management, American Nuclear Society and American Society of Civil Engineers,
eds.. Las Vegas, Nevada, April, 1993.
Wittwer, C.S., Bodvarsson, G.S., Chornack, M.P., Flint, A.L., Flint, L.E., Lewis, B.
B., Spongier, R.W., and Rautman, C.A., Development of a Thrce-Dimensionai Site-
Scale Model for the Unsaturated Zone at Yucca Mountain, Nevada, Radioactive V/aste
Management and Environmental Restoration, Harwood Academic Publishers, G0ibH,
U.S., Vol. 19, pp. 147-167,1994.

Wittwer, C.S., Chen, G., Bodvarsson, G.S., Chornack, M., Hint, A., Hint, L.,
Kwickiis, E., and Spengler, R.. Preliminary Development of die LBL/USGS Three-
Dimensional Site-Scale Model of Yucca Mountain, Nevada, Report LBL-37356,
Lawrence Berkeley Laboratory, Berkeley, California, 199S.
Wu, Y.S., Pruess, K., and Chen, Z.X., Buckley-Leverett How in Composite Media,
Report LSL-28937, Lawrence Berkeley National Laboratory, Berkeley, California,
1990.
Wu, Y.S., Bandurraga, T.M., Ahlers, C.K, Finsterle, S., Chen, G., Haukwa, C ,
Bodvatsson, G.S., Kwicklis, E., Rousseau, J., and Flint, L., Calibration and
Extension of the LBL/USGS Three-Dimensional Site-Scale Model of Yucca Mountain,
Nevada, Report LBL-___, Lawrence Berkeley Laboratory, Berkeley, California, in
press.
Zimmerman, R.W., Bodvarsson, G.S., and Kwicklis, E.M., Absorption of Water into
Porous Blocks in Various Shapes and Sizes, Water Rcso. Res.,26, pp. 2797-2806,
1990.

TOUGH2 SOFTWARE QUAUHCATIQN - FEURWARY 1996 no

APPENDIX A. MASS A N D ENERGY BALANCES
The basic mass- and energy-balance equations solved by MULKOM, TOUGH and
TOUGH2 can all be written in the following general form:
K W M
4- J M' 'dV= J F ' « . d T J q dV+ (A.l)
d t
v n r„ v„

The integration here is over an arbitrary subdomain Vn of the flow system undt. study,
which is bounded by the closed surface r„. The quantity M appearing in the accumulation
term denotes mass or energy per unit volume, with K = 1, . . ., NK labeling the mass
components, and K = N K + 1 for the heat "component."

The general form of the mass accumulation term is

M^oTspppXff' (A.2)

The totai mass of component K is obtained by summing over all fluid phases p = 1, . . . ,
NPH. Sp is the saturation (volume fraction) of phase P, p is density of phase p, andxff'
p

is the mass fraction of component K present in phase p. Similarly, the heat accumulation
term in a multi-phase system is
NPH
M(NK+D = 4. £Sj5Ppup + (1 - <W PR C T , R

1
P- (A.3)
where u,, denotes internal energy of fluid phase p.

The mass flux term is a sum over phases

NPH , ,

= I
P (A.4)
for K = 1, . . . , NK. Individual phase fluxes are given by a multi-phase version of
Darcy's lav/:

F>—«Sjpp(Vi^pp ) B (A.5)

Here k is absolute permeability, k^, is relative permeability of phase p. up is viscosity, and

F& = P+Pc.p (A.<5)
is the pressure in phase P, which is the sum of the pressure P of a reference phase, and the
capillary pressure of phase p relative to the reference phase, g denotes the vector of
gravitational acceleration. Gas phase permeability can be specified to depend on pressure,
according to the Klinkcnberg relationship k = ko(I + b/P), where ko is absolute

TOUCHi iOTTWfJte QUAUflC/lTlON • FitXUAHY 1996 81

 DATA:
Table 8.1: RADIATION IN HOME
Background cpm Scale = x1 time = 60 seconds

MATERIALS COUNTS PER

MINUTE
LANTERN MANTLE
POTTERY GLAZE
SMOKE DETECTOR
WATCH
GLASS CRYSTAL

Table 8.2: RADIATION DECAY

Radioactive Isotopes Radiation(s) Gamma Ray Notes
Source: Energy
Lantern 232-rh alpha rays & beta 0.240 MeV natural
mantle & gamma rays
Pottery 235 (J, 238 u alpha rays & beta 0.063 MeV natural
Glazed & gamma rays
Smoke 241 m
A
alpha rays & beta 0.059 MeV man-made
Detector & gamma rays
Painted 226Ra alpha rays & beta 0.0510 MeV natural
Watch & gamma rays
Glass ''Pb ? ?MeV ?
Crystal

QUESTIONS:
1. Which of the house-hold materials had the greatest radiation
intensity? The least radiation intensity?
2. What radioactive element is present in a Coleman Mantle? in a smote
detector? In a panted watch?
3. What unit is used to describe the magnitude of radioactive decay
energies?
4. An isotope of the element calcium has a half-life of 12 years, how
long will it take for 1/4 of the calcium to decay?
a. 4 year b. 8 years c. 12 years d. 16 years e. 24 years

28
APPENDIX B. SPACE AND TIME DISCRETIZATION*
The continuum equations (B.l) are discretized in space using the "integral finite
difference" method (Edwards, 1972; Narasimhan and Withcrspoon, 1976). Introducing
appropriate volume averages, we have

/MdV . V M„
n (B.l)

where M is a volume-normalized extensive quantity, and M„ is the average value of M over

V , Surface integrals are approximated as a discrete sum of averages over surface segments
n

Anm-
K A
jF .ndr = X rmFr», (B.2)
r„ »>
Here F is the average value of the (inward) normal component of F over the surface
n m

segment Anm between volume elements V„ and V . The discretization approach used in the
m

integral finite difference method and the definition of the geometric parameters are
illustrated in Fig. 35.

Figure 35. Space discretization and teomeiry data w the integral finite difference method.

* Aiiptee horn TOUCH2 repwi,Proe» (1991).

TOUUH2 SOfTWAH QuAMHCMION - FtMllAiV 1990 "JJ

 The discretized flux is expressed in terms of averages over parameters for elements V„ and
V . For the basic Darcy flux term, Eq. (A.5), we have
m

Ujj.i)
P,nm p g n m

Dntn
•* -Inml- J
where the subscripts (nm) denote a suitable averaging at the interface between grid blocks n
and m (interpolation, harmonic weighting, upstream weighting). D is the distance
n m

between the nodal points n and m, and g is the component of gravitational acceleration in
n m

the direction from m to n.

The discretized form of the binary diffusive flux in the gas phase is

« - = - (* VsL teLiw ^ ^ (M.

For the group (<J> S T) geometric weighting is used at the interface, while D and p are
averaged between grid blocks n and m.
Substituting Eqs. (B.l) and (B.2) into the governing Eq. (A. 1), a set of first-order
ordinary differential equations in time is obtained.

-nmFL+q? CB.S)

Time is discretized as a first order finite difference, and the flux and sink and source
k+ K
terms on therighthand side of Eq. (B.5) are evaluated at the new time level, t ' = i + At,
to obtain the numerical stability needed for an efficient calculation of multiphase flow. This
treatment of flax tenns is known as "fully implicit," because the fluxes are expressed in
1:+1
terms of the unknown thermodynamic parameters at time level t , so that these unknowns
are only implicitly defined in the resulting equations; see e.g. Peaceman (1977). The lints
discretization results in the following set of coupled non-linear, algebraic equations

A + v
R„ _ M„ -M n - y ^ j i nraP m n n1„ i ( J J 6 )

,k
where we have introduced i-siduals R j j . For each volume element (grid block) V„
there arc NEQ equations < K = 1,2 NEQ; usually, NEQ = NK -i 1), so that tor a flow
systea- with NEL grid blocks (B.6) represents a total of NEL • NEQ coupled non-linear
equations. The unknowns are the NEL • NEQ independent primary variables I-M; i • 1
k
NEL * NEQ} which completely define the state of the flow system at time level t *'. Tnese
equations are solved by Newton/Raphson iteration, which is implemented as follows. We
k+1
introduce an iteration index p and expand the residuals RjJ' in Eq. (B.6) at iteration step
p + 1 in a Taylor series in terms of those at index p.

TOUCH? SOFIWAU QUAtlfICATKJN . FEMUARY 1 9 9 6 84

 g K.k+l
R

R ^( , ) - B ^ W t s a .
x p+1
x -x
( i.p+l i,pj
(B.7)
= 0

Retaining only terms up to first order, we obtain a set of NEL • NEQ linear equations
for the increments (xj.p+i - xj.p):

M
{«**-«*) = < h,)
K
V f n„ (B.8)
7 9xs
All terms dR /dxj in the Jacobian matrix are evaluated by numerical differentiation. Eq.
n

CB.8) is solved by sparse direct matrix methods (Duff, J977) or itcratively by means of
preconditioned conjugate gradients {Moridis and Pruess, 19951 Iteration is continued until
,k+l
the residuals R * are reduced below a preset convergence tolerance. The selection and
switching of primary variables in a TOUGH2 solution depend on the phase conditions.
The variable switching procedure affects the updating for secondary dependent variables
but does not affect the equation setup because the equations are still mass and energy
conservation equations for each block.
pK,k+l
K
n.p-t-l
(B.9)
MK.k+1
n,p+l
5
The default (relative) convergence criterion is Si *= 10" . When the accumulation terms ace
,k+1
smaller than £2 (default 62 = 1), an absolute convergence criterion is imposed, IR* l <,
£; > 62. Convergence is usually attained in 3 - 4 iterations. If convergence cannot be
achieved within a certain number of iterations (default 8}, the time step size At is reduced
and a new iteration process is started.
It is appropriate to add some comments about our space discretization technique.
The entire geometric information of the space discretization in Eq. (B.6) is provided in the
form of a list of grid block volumes V„, interface areas A„ , nodal distances X>nn> and
m

component* gun of gravitational acceleration along nodal lines. There is no reference

whatsoever to a global system of coordinates, or to the dimensionality of a particular flow
problem. The discretized equations are in fact valid for arbitrary irregular discretizations in
one. two or three dimensions, and for porous as well as for fractured media. This
flexibility should be used with caution, however, because the accuracy of solutions
depends upon the accuracy with which the various interface parameters in equations such
as (B.3, B.4) can be expressed in terms of average conditions in grid blocks. A general
requirement is that there exists approximate thermodynamic equilibrium in (almost) all grid
blocks at (almost) all times (Pruess and Nanuiimhua, 1985). For systems of regular grid
blocks referenced to global coordinates (such as r - z, x. - y - z\ Eq. (B.6) is identical to a
conventional finite difference formulation (e.%. Peacemau, 1977).

TOUCH2 sornvAKC QIMUHCATION • r»nu/wr 1996 85

APPENDIX C DESCRIPTION OF FLOW IN FRACTURED MEDIA*
Figure 36 illustrates the classical double-porosity concept for modeling flow in
fractured-porous media as developed by Warren and Root (1963). Matrix blocks of low
permeability are embedded in a network of inter-connected fractures. Global flow in the
reservoir occurs only through the fracture system, which is described as an effective
porous continuum. Rock matrix and fractures may exchange fluid (or heal) locally by
means of "interporosiry flow," which is driven by the difference in pressures (or
temperatures) between matrix and fractures. Warren and Root approximated the
interporosiry flow as being "quasi-steady," with rate of matrix-fracture interflow
proportional to the difference in (local) average pressures.

Pisure 36. Idealized "double peroshy" modtl cf a fractured porous medium.

The quasisteady approximation is applicable to isothermal single phase flow of

fluids with small compressibility, where pressure diffusivities are large, so that pressure
changes in the fractures penetrate quickly all the way into the matrix blocks. However, for
multiphase flows, or coupled fluid and heat flows, the transient periods for interporosiry
flow can be very Ions (tens of years). In order to accurately describe such flows it is
necessary to resolve the driving pressure, temperature and mass fraction gradients at the
matrix/fricturc interface In the method of "multiple interacting continua" (rvflKC; Pruess
and Narasimhan, 1982, 1985), resolution of these gradients is achieved by appropriate
subgridding of the matrix blocks, as shown in Fig. 37. The MINC concept is based on the

* MijMi ton TOUGH2 report, Proos (mi).

TOUCH2 SotlWMtc QtMUfKMiON • FiiKUARr 1996 nr.

 notion that changes in fluid pressures, temperatures, phase compositions, etc. due to the
presence of sinks and sources (production and injection wells) will propagate rapidly
through the fracture system, while invading the tight matrix blocks only slowly. Therefore,
changes in matrix conditions will (locally) be controlled by the distance from the fractures.
Fluid and heat flow from the fractures into the matrix blocks, or from the matrix blocks into
the fractures, can then be modeled by means of one-dimensional strings of nested grid
blocks, as shown in Fig. 37.

Fractures

' Matrix Blocks

Pipire 37. SubRriddinn in the method of "multiple interacting continua" (MINC).

In general it is not necessary to explicitly consider sufagrids in all of the matrix

blocks separately. Within a ceitain reservoir subdomain (corresponding to a finite
difference grid block), all fractures will be lumped into continuum # 1, all matrix material
within a certain distance from thefractureswill be lumped into continuum # 2, matrix
material « larger distance becomes continuum # 3, and so on. Quantitatively, the
subgridding is specified by means of a set of volume fractions VOL(j), j = i J, into
which the 'primary" porous medium grid blocks are partitioned. The MWC-process in the
MESHMAKER module of T2VOC operates on the element and connection data of a porous
medium mesh to calculate,forgiven data on volume fractions, the volumes, interface areas,
and nodal distances for a "secondary" fractured medium mesh. The information on
fracturing (spacing, number of sets, shape of matrix blocks) required for this is provided
by * "proximity function" PROX(x) which expresses, for a given reservoir domain V , the 0

total fraction of matrix material within a distance x from the fractures. If only two continua
are specified (one for fractures, one for matrix), the MINC approach reduces to the
conventional double-porosity method. Full details are given in a separate report (Pruess,
1983s).

The MINC-method as implemented in llw MESHMAKER module can also describe

global matrix-matrix flow. Figure 38 shows the most general approach, often referred to as
r
"dual permeability," in which global flow occurs in both fracture and matrix continua. It i
also possible to permit matrix-matrix flow only in the vertical direction. For any given

) O U C H 2 SOfTWAM QUAtiriCAIKIN - FtntUAKY 199S B7

 fractured reservoir flow problem, selection of the most appropriate gridding scheme must
be based on a careful consideration of the physical and geometric conditions of flow. The
MINC approach is not applicable to systems in which fracturing is so sparse that ihe
fractures cannot be approximated as a continuum.
•?m -M •^m ^3
®z <F <tr

tr\ &r\ or\ or \ ©- T

-GF- & ©=
Figure 38* Flow connections in the "dual permeability" model Global Jlow occit*s between Imth
fracture (F) and matrix (M) grid blocks. In addition there is F'M interporosity flow.

TOUGH2 SOFTWARE QlMllfJCATJON . friRUARY 199f» DS

 D. NOMENCLATURE
2
A area,m
b Klinkenbcrg parameter, Pa
B effective vapor diffusion strength parameter, replaces the group <J>ST g

in Eq. (A.7), dimensionless

C specific heal, J/kg-°C
d penetration depth for heat conduction, m
D 2
diffusion coefficient, m /s
D distance, m
DELX small increments of primary variables for computing numerical
derivatives
DX increments of primary variables during Kewton-Raphson iteration
2
f diffusive flux, kg/m -s
fvPL vapor pressure lowering factor, dimensionless (Eq. A.9)
2 2
F mass or heat flux, kg/m -s or W/m
2
gravity acceleration, m/s
h specific enthalpy, J/kg
i index of primary thermodynamic variable
J Jacobian matrix
2 _12 2
k intrinsic permeability, m ( 1 0 m = 1 darcy)
kr relative permeability, dimensionless
K thermal conductivity, W/m-°C
K H
Henry's constant. Pa
m index of volume element (grid block)
rriair molecular weight of air
mR o 2 molecular weight of water
mj molecular weight of liquid
3 3
M accumulation term in mass or energy balance equation, kg/m or J/m
n index of volume element (grid block)
N index of volume element (grid block)
NB number of secondary parameters other than mass fractions in PAR
array (usually NB = 6)
NBK NB + NK
NEL number of volume elements (grid blocks) in flow domain
NEQ number of balance equations per voluaie element; NEQ = NKI or
NEQ = NK
NK number of mass components present
NKI NK+1
NLOC storage location after which primary variables start for grid block N;
NLOC = ( N - l ) * N K l
NL0C2 storage location after which secondary parameters start for ^rid biock
N;NTOC2 = (N - 1*(NEQ + 1)*NSEC
NPH number of phases
NSEC number of secondary parameters per volume element; NSEC =
NPK*NBK + 2
D index in Newton-Raphson iteration
P pressure. Pa
?a.e. air entry pressure, Pa
Pc capillary pressure, Pa

TOUCH2 SOFTWARE QUALIFICATION - FEBRUARY 1996 89

 Psat saturated vapor pressure, Pa
3 3
q volumetric sink or source rate, kg/m -s or W/m
r radius, m
3 3
R residuals in mass or energy balance equations, kg/m or J/m
R universal gas constant, 8314-56 J/°C-mole
S saturation (voidfractionoccupied by afluidphase), dimensionless
t time, s
T temperature, °C
u specific internal energy, J/kg
V volume, m 3

X distance, m
x,X primary thermodynamic variable
(K)
V
mol fraction of component K «n phase (5

• xf mass fraction of component K in phase ji

z = iog[r/(t)^3 similaritj' variable for cylindrical flow geometry

Greek
P~ phase index (p = liquid, gas)
2
8 thermal diffusiviry, m /s
K component index
3
P density, kg/m
2

r area, m
porosity, dimensionless
-©•

T tortuosity factor, dimensionless

H viscosity, Pa-s
Subscripts
a air
i3 phase
c capillary
f fracture
CT
gas
: initial
i liquid
r relative
R rock
V vapor

T O U G H 2 SOFTWAflt QUALIFICATION - FEBRUARY 1 9 9 6 90

APPENDIX E- TOUGH2 SELECTED BIBLIOGRAPHY

Adenekaa, A.E., Patzek, T.W., and Pruess, K., "Removing Organic Contaminants from
the Subsurface by Steam Injection: Numerical Modeling," Paper presented at the
Symposium on Exploration, Characterization and Utilization of California Heavy
Fossil Fuel Resources American Chemical Society, San Francisco, CA, April 1992.
Adenekan, A.E., Patzek, T.W., and Pruess, K., Modeling of multiphase transport of
multicomrjonent contaminants and heat in the subsurface-numerical model
formulation. Water Resources Research, vol. 29, no. I I , pp. 3727-3740,1593.
Adenekaa, A.E., ftuess K., and Falls R.W., Removal of toetoLomeihyhsae <xjataninadoa
from the subsurface—a comparative evaluation of different remediation strategies
by means of numerical simulation, Report LBL-30272, Lawrence Berkeley
Laboratory, Berkeley, CA, December 1990.
Ahlers, C.F., Ba^durraga, T.M., Chen, G., Finsterle, S., Wu, Y.S., and Bodvarsson,
G.S., Summary of model calibration and sensitivity studies using the LBNL/USGS
three-dimensional unsaturated zone site-scale model, Submitted for publication in
September 1995.
Ahlers, C.F., Bandurraga, T.M., Chen, G., Finsteric, S., Wu, Y.S., Bodvarsson, G.S.,
Kwicklis, E., Rousseau, J., and Flint, L., Performance analysis of the
LBNL/USGS three-dimensional unsaturated zone site-scale model, Submitted for
publication in September 1995.
Amistoso, A.E., Aquino, E.G., Aunzo, Z., Jordan, O.T., Sta. Ana, F.X.M,, Doughty,
C , and Bodvarsson, G.S., Reservoir analysis of the Painpinon geothennal field,
Negros Oriental, Philippines, Geothermics, vol. 22, no. 5/6, Report tJSL-33922,
Lawrence Berkeley Laboratory, Berkeley, CA, 1933.
Andrews, B..^ Design/test calculations for two-phase flow in a dual-porosity, duai-
permeabflity medium using TOUGH, Memorandum to NAGRA, Baden,
Switzerland, September 1988.
Antunez, E., Lippoaann, M., Ali-Khan, M., and Boardman, T., "Simulalion of the Heber
Geothermal Field, A TOUGH/PC Application," Proceedings of the TOUGH
Workshop *95, Report LBL-37200, Lawrence Berkeley Laboratory, Berkeley, CA,
pp. 101 - 106, March 20-22, 1995.
Antunez, E., Moridis, G., and Pruess, K., "Large-Scale Thrcc-Dimcnsional Geothermal
Reservoir Simulation on Small Computer Systems," Proceedings of the World
Geothermal Congress '95, pp. 2977-2980, International Gcothcrmal Association,
May 1995.

Antunez, E., Pruass, K., and Moridis, G„ "Use of TOUGH2 on Small Computers,"
Proceedings of the TOUGH Workshop '95, Report LBL-37200, Lawienw
Berkeley Laboratory, Berkeley, CA, pp. 265-270. March 20-22, 1995.
Antunez, E., Walters, M. A., and Bodvarsson, G.S., Numerical study of tiiu northwest
geysers geothermal field-a case study of the Coldwater Creek SteamCeld,
Gectkermics, vol. 23, no. 2, 1994.

TOUGH2 SorrWARC QlMUHCATION - FEBRUARY 1 996

Arens, G., Hoglund, L., and Wiborgh, M., "Requirements on Sealing Measures Due To
nas Production," Proceedings of the TOUGH Workshop '95, Report LBL-37200,
Lawrence Berkeley Laboratory, Berkeley, CA, pp. 52-57, March 20-22,1995.
Aunzo, Z., Steingrimsson, B., Bodvarsson, G.S., Escobar, C , and Quintanilia, A.,
"Modeling Studies of the Ahuachapan Geothennal Field, El Salvador," Proceedings
of the Fourteenth Geothermal Reservoir EngineeriDg Workshop, Stanford
University, Stanford, CA, pp. 287-295, 1989.
Battistelli, A., Calore, C , and Pruess, K., "A Fluid Property Module for the TOUGH2
Simulator for Saline Brines with Non-Condensable Gas," Presented at the
Eighteenth Geothermal Reservoir Engineering Workshop, Stanford University,
Stanford, CA, January 26-28, 1993.
Battistelli, A., Calore, C. and Pruess, K., "Analysis of Salt Effects on the Depletion of
Fractured Reservoir Blocks," Proceedings of the World Geothermal Congress, pp.
1613-1618, International Geothermal Association, May 1995.
Battisteili, A., Calore, C„ and Pruess, K., "Vapor Pressure Lowering Effects Due To
Salinity and Suction Pressure in the Depiction of Vapor-Dominated Geothermal
Reservoirs," Proceedings of the TOUGH Workshop '95, pp. 77-83, Report LBL-
37200, Lawrence Berkeley Laboratory, Berkeley, CA, March 20-22,1995.
Betz, C , Emmet, M., Farber, A., Helmig, R., Kaleris, V., and Kobus, H., "Soil
Remediation by Heat Injection: Experiments and Numerical Modeling,"
Proceedings of the TOUGH Workshop '95, Report LBL-37200, Lawrence
Berkeley Laboratory, Berkeley, CA, pp. 187-192, March 20-22,1995.

Bodvarsson, G.S., Model predictions of the Svartsengi geothermai field, Iceland, Water
Resources, Res., vol. 24, No. 10, pp. 1740-1746, Report LBL-21253, Lawrence
Berkeley Laboratory, Berkeley, CA, 1988.
Bodvarsson, G.S., Numerical modeling of geothermal systems with applications to Krafla,
Iceland and Olkaria, Kenya, Geolhermal Reservoir Engineering, E. Okandan, also
in J. Geothermal Ener., 1988.
Bodvarsson, G.S., Aunzo, Z., Chen, G., and Haukwa. C , "Recent Development of the
LBL/USGS Site-Scale Model of Yucca Mountain, Nevada," Proceedings of the
Sixth Annual International Conference on High Level Radioactive Waste
Management, American Nuclear Society and American Society of Civil Engineers,
Eds.. Las Vegas, NV, May 1-4,1995.
Bodvarsson, G.S., and Bjbrnsson, S.. Analysis of pressure, enthalpy and CO. transients
in well BR21, Ohaaki, New Zealand, Trans., Geothermal Resources Council, vol.
11, pp. 503-507, Report LBL-23718, Lawrence Berkeley Laboratory, Berkeley,
CA, 1987.
Bodvarsson, G.S., BjSrosson, S., Gunnarsson, A., Gunnlaughsson, E., Sigurdssou, O..
Steffanson, V., and Steingrimsson, B., "A Summary of Modeling Studies of the
Nesjavellir Geothermal Held, Icelaud," Proceedings of [he Thirteenth Geolhenual
Reservoir Engineering Workshop, Stanford University, Stanford, CA, in press.
Report LBL-24675, Lawrence Berkeley Laboratory, Berkeley, CA, 1988.

TOUCH2 SOFTWARE QUAUFICATION - FEBRUARY 1 9 9 6 92

 Bodvarsson, G.S., Bjomsson, S., Gunnarsson, A., Gunnlaughsson, E., Sigurdsson,
O., Stefansson, V., and Steingrimsson, B., The Nesjavdlir geothermal field,
Iceland, 1. Field characteristics and development of a three-dimensional numerical
model, J. Geotlier. Sci. Tech. vol. 2, no. 3, pp. 189-228, Report LBL-28760,
Lawrence Berkeley Laboratory, Berkeley, CA, 1990.
Bodvarsson, G.S., BjSrnsson, S., Gunnarsson, A., Gunnlaugbsson, E., Sigurdsson, O.,
Stefansson, V., and Steingrimsson, B., The Nesjavellir geothermal field, Iceland,
2. Evaluation of the generating capacity of the system, J Geother. Sci. Tech., vol.
2, no. 4, pp. 229-261, Report LBL-28761, Lawrence Berkeley Laboratory,
Berkeley, CA, 1991.
Bodvarsson, G.S., Chen, G., Haukwa, C , and Kwickiis, £., "The Use of TOUGH2 for
the LBL/USGS Three-Dimensional Site-Scale Model of Yucca Mountain, Nevada,"
Proceedings of the TOUGH Workshop '95, pp. 84-90, Report LBL-37200,
Lawrence Berkeley Laboratory, Berkeley, CA, pp. 14-20, March 20-22,1995.
Bodvarsson, G.S., Chen, G., and Wittwer, C , "Preliminary Analysis of Three-
Dimensional Moisture Flow within Yucca Mountain, Nevada," Proceedings of the
Fifth Annua] International Conference High Level Radioactive Waste Management,
American Nuclear Society and American Society of Civil Engineers, Eds., Las
Vegas, NV, May 22-26,1994.
Bodvaisson, G.S., Chen, G., Zimmerman, R.W., and Kwickiis, EM., Accuracy and
efficiency of semi-analytical dual-porosity simulator for flow in unsaturated
fractured rock masses. Radioactive Waste Management and Environmental
Restoration, vol. 19, no. 1, pp. 193-20S, Report LBL-34323, Lawrence Berkeley
Laboratory, Berkeley, CA, 3994.
Bodvarsson, G.S., and Cox, B.L., Numerical studies of gravity effects in two-phase
reservoirs. Tram:, Geolhermal Resources Council, vol. 10, pp. 429-436. Report
LBL-21935, Lawrence Berkeley Laboratory. Berkeley, CA, 1986.
Bodvarsson, G.S., Cox, B.L., and Ripperda, M.A., "Effects of Steam-Liquid
Counterfiow on Pressure Transient Data from Two-Phase Geothermal Reservoirs,"
Presented at the Sixty-Second Annual Technical Conference and Exhibition of the
SPE, Paper SPE-16775, Dallas, TX, September 19S7.
Bodvarsson, G.S., Cox, B.L., and Rippcrda, M.A., Effects of steam-liquid counterflow
on pressure transient data from two-phase geothsrmal reservoirs, SPE Journal on
Reservoir Engineering, vol. 4, No. 2, pp. 187-193, Report LBL-237I9,
Lawrence Berkeley Laboratory, Berkeley, CA, 1989.
Bodvarsson, G.S., and Gauike, S., Effects on non-condensable gases on fluid recovery in
fractured geothermal reservoirs, SPE Reservoir Engineering, vol. 2 No. 3, pp.
335-342, Report LBL-21JI2, Lawrence Berkeley Laboratory, Berkeley, CA,
1987.

TOUGH2 SOFTWARE QUALIFICATION - FEBRUARY 1996 93

 Bodvarsson, G.S., Gislason, G., Gunnlaughsson, E., Sigurdsson, O., Stefansson, V.,
and Stcingrimsson, B., "Accuracy of Reservoir Predictions for the Nesjavellir
Geothermal Field, Iceland," Paper presented at trie Eighteenth Geothermal
Reservoir Engineering Workshop, Stanford University, Stanford, CA, Report
LBL-33921, Lawrence Berkeiey Laboratory, Berkeley, CA, January 26-28,1993.
Bodvarsson, G.S., and Lippmann, M.S., A modeling study of the natural state of the
Heber geothermal field, Trans. Geothermal Resources Council, Vol. 7, no. 441-
448, 19S3.
Bodvarsson, G.S., and lippmann, MJ., Numerical studies of the heat and mass transport
in the Ceiro Prieto geothermal field. Water Resource Res., vol. 19, No. 3, pp.
753-767, 1983.
Bodvarsson, G.S., and Lippmann, M.J., Numerical sti'dies of enthalpy and C 0 transient
2

in two-phase wells, Trans. Geothermal Resources Council, vol. 8, pp. 289-294,

1984.
Bodvarsson, G.S., and Lippmann, M.J., "The Generating Capacity of die Heber
Geolhermal Field," Proceedings of the Ninth Geothermal Reservoir Engineering
Workshop, Stanford University, Stanford, CA, Report SGP-TR-74, pp. 157-166,
Report LBL-16677, Lawrence Berkeley Laboratory, Berkeley, CA, 1983.
Bodvarsson, G.S., Lippmann, M.J., and Gaulke, S.W., "Enthalpy Transients in
Fractured Two-Phase Geotheimal Systems," Proceedings of the Geotheraial
Resources Council 19S5 International Symposium on Geotfiermal Energy, Kailua
Kona, HI, Report LBL 193675, Lawrence Berkeley Laboratory, Berkeley, CA,
pp. 533-540, 1985.
Bodvarsson, G.S., Lippmann, MJ., and Pruess, K., "Geysers Reservoir Studies," Paper
presented at DOE Geothermal Program Review XI, Berkeley, CA, Report L8L-
34245, Lawrence Berkeley Laboratory, Berkeley, CA, April 27-28,1993.
Bodvarsson, G.S., McGraw, M.A., Flint, L.E., and Flint, A.L., Numerical modeling of
lateral infiltration into the Paintbrush Tuff at Yucca Mountain, Nevada, Report
LBL-35389, Lawrence Berkeley Laboratory. Berkeley, CA, 1994.
Bodvarsson, G.S., Mishra, S., and Attanayake, M J . , Estimating properties of
unsaturated fractured formations from injection and failoff tests, Report LUL-
24796B, Lawrence Berkeley Laboratory, Berkeley, CA, 1987.
Bodvaisson, G.S., Mishra, S., and Attanayake, M.P., Injection and failoff test analysis to
estimate properties of unsaturated fractures, AGU Monograph on Flow and
Transport through Unsaturated Fractured Rocks, vol. 42. pp. 142-156. 1987.
Bodvarsson, G.S., Niemi, A., Spencer, A., and Attanayake, M. P., Preliminary
calculations of the effects of air and liquid water drilling on moisture conditions in
unsaturated rocks, Report LBL-25073, Lawrence Berkeley Laboratory, Berkeley,
CA, 1988.

TOUGM2 SOFTWARE QUAHFiCATION - FEBRUARY 1996 94

Bodvarsson, G.S., O'Sullivan, MJ., Pruess, K., and Blakely, M.R., Fluid and heat flow
in gas-rich geothermal reservoirs, Soc. Pet. Eng. J„ vol. 25, No. 2, pp. 215-226,
Report LBL-]6329, Lawrence Berkeley Laboratory, Berkeley, CA, 1985.
Bodvarsson, G.S., and Pruess, K., "Modeling Studies of Geothermal Systems with a Free
Water Surface," Proceedings of the Ninth Geothermal Reservoir Engineering
Workshop, Stanford University, Stanford, CA, Report SGP-TR-74, pp. 351-356,
1983.
Bodvarsson, G.S., and Pruess, K., Thermal effects of reinjection in geothermal reservoirs
with major vertical fractures, J. Pet Tech., vol. 36, No. 10, pp. 1567-1S78,
Report LBJL-J6269, Lawrence Berkeley Laboratory, Berkeley, CA, 1985.
Bodvarsson, G.S., Pruess, K., Haukwa, C , and Ojiambo, S.B., Evaluation of reservoir
model predictions for the Olkaria East geothermal field, Kenya, Geothermics, vol.
19, no. 5, pp. 339-414, Report LBL-26771, Lawrence Berkeley Laboratory,
Berkeley, CA, 1991.

Bodvarsson, G.S., Pruess, K., and Lippmann, M.J., Modeling of geothermal systems, J.
Pet. Tech., vol. 38. no. 10, pp. 1007-1021, September 1986.
Bodvarsson, G.S., Pruess, iC, and Lippmann, MJ., Numerical models for the evaluation
of geothermal systems, Revista Brasileira de Gecfisica, vol. 5, no. 2, pp. 311-
318, Report LBL-22047, Lawrence Berkeley Laboratory, Berkeley, CA, 1989.
Bodvarsson, G.S., Pruess, K., and O'Sullivan, M.J., Injection and energy recovery in
fractured geothermal reservoirs. Society of Petroleum Eng. J., vol. 25, no. 2, pp.
303-312, Report LBL-15344, Lawrence Berkeley Laboratory, Berkeley, CA,
April 1985.
Bodvarsson, G.S., Pruess, K_, Stefansson, V., Bjirnsson, S., and Ojiambo, S.B., "A
Summary of Modeling Studies of the East Olkaria Geothermal Field, Kenya,"
Gcothermal Resources Council 1985 Symposium on Geothermal Energy, Kaiiua
Kona, HI, Report LBL-19367, Lawrence Berkeley Laboratory, Berkeley, CA, pp.
295-302, 1985.
Bodvarsson, G.S., Pruess, K., Stefansson, V., Bjimsson, S., and Ojiamho, S.B., The
East Olkaria geotherrnal field, Kenya, 1. History match with production and
pressure decline data, Geophys. Res., vol. 92, no. B2, pp. 521-539, Report L8L-
20098, Lawrence Berkeley Laboratory, Berkeley, CA, 1987.
Bodvarsson, G.S., Pruess, K., Steransson, V., Bjirnsson, S., and Ojiarribo, S.B., Tne
East Olkaria geothermal field, Kenya, 2. Predictions of well performance and
reservoir depletion, J. Geophys. Res., vol. 92, No. B2, pp. 541-554, Report LBL-
20099, Lawrence Berkeley Laboratory, Berkeley, CA, 1987.
Bodvarsson, G.S., Pruess, K., Stefansson, V., and Eliasson, E.T., A summary of
modeling studies of the Krafla geothermal field, Trans., Geothermal Resources
Council, Vol. 7, pp. 391-396, 1983.

TOUGH2 SOFTWARE QUALIFICATION - FEBRUARY 199C ; 95

Bodvaisson, G.S., Pruess, K., Stefansson, V., and Eliasson, E.T., The Krafla
geothermal field, Water Resources Research, Vol. 20, No. I I , pp. 1531-1544,
Report LBL-16203, Lawrence Berkeley Laboratory, Berkeley, CA, 1984.
Bodvarsson, G.S., Pruess, K., Stefansson, V., and Eliasson, E.T., The Krafla
geothermal field, Iceland: 3. The generating capacity of the field, Water Resources
Res., vol. 20, pp. 1545-1559, Report LBL-16203, Lawrence Berkeley
Laboratory, Berkeley, CA, 1984.
Bodvarsson, G.S., Stefansson, V., Steingrimsson, B., Bjfrnsson, Gunnarsson, A., and
Gunnlaugsson, E., "Natural State Model of the Nesjavellir Geothermal Field,
Iceland," Proceedings of the Eleventh Geothermal Reservoir Engineering
Workshop, Stanford University, Stanford, CA, Report SGP-TR-84, pp. 109-116",
1986.
Bodvarsson, G.S., and Witherspoon, P.A., "How Rate Decline of Steam Wells in
Fractured Geothennal Reservoirs," Proceedings of (he Tenth Annua! Geothermal
Reservoir Engineering Workshop, Stanford University, Stanford, CA, Report
SGP-TR-84, Report LBL-18439, Lawrence Berkeley Laboratory, Berkeley, CA,
pp. 105-112, 1985.
Brown, T.P., Lehman, L.L., and Nieber, J.L., 'Testing Conceptual Unsaturated Zone
Flow Models for Yucca Mountain," Proceedings of the Fifth Annual International
Conference on High Level Radioactive Waste Management, American Nuclear
Society and American Society of Civil Engineers, Eds., Las Vegas, NV, May 22-
26, 1994.
Buliivant, D.P., "Making MULKOM/TOUGH Faster and Easier to Use," Proceedings of
the TOTJGH Workshop, Report LBL-29710, Lawrence Berkeley Laboratory,
Berkeley, CA, pp. 139-142, September 13-14,1990.
BuIIivant, D.P., O' Sullivan, MJ., and Yang, Z., 'Experiences Using MuWgrid for
Geothermal Simulation," Proceedings of the TOUGH Workshop '95, Report LBL-
37200, Lawrence Berkeley Laboratory, Berkeley, CA, pp. 281-286, March 20-22,
1995.
Buschcck, T., and Nitao, J.J., "Modeling Hydrothcrmal Flow in Variably Saturated,
Fractured, Welded Tuff During the Prototype Engineered Barrier System Field Test
of the Yucca Mountain Project," Proceedings of the TOUGH Workshop, Report
LBL-29710, Lawrence Berkeley Laboratory, Berkeley, CA, pp. 9-14, September
13-14, 1990.
Buscheck, T.A., and Nitao, J.J., "The Impact of Thermal Loading on Repository
Performance at Yucca Mountain," Proceedings of the Third High Level Radioactive
Waste Management International Conference, Las Vegas, NV, pp. 1003-1017,
April 12-16,1992.
Calore, C , GianslU G., and, Pruess, K., "Water-CO, Version of MULKOM Code: A
Tool for Studying the Origin and Trends of CO, In Geothermal Reservoirs,"
Proceedings of the TOUGH Workshop, Report LBL-29710. Lawrence Berkeley
Laboratory, Berkeley, CA, pp. 53-60, September 13-14, 1990.

TOUCH2 SOnWARE QUAIIFICATION • FIDSUARV 1 9 9 6 9G

 Christian-Fear, TJL., and Webb, S.W., 'Modification and Application of TOUGH2 as a
Variable-Density, Saturated Flow Code and Comparison to SWIFT H Results,"
Proceedings of the TOUGH Workshop '95, Report LBL-37200, Lawrence
Berkeley Laboratory, Berkeley, CA, pp. 2833, March 20-22,1995.
Cox., B.L., and Bodvarsson, G.S., "Preliminary Studies of Two-Phase Effects on
Pressure Transient Data," Proceedings of the Eleventh Geothermal Reservoir
Engineering Workshop, Stanford University, Stanford, CA, SGP-TR-84, pp. 7-
14, Report LBL-21249, Lawrence Berkeley Laboratory, Berkeley, CA 1986.
Cox, B i . , Finsterle, S., and Yang, J.S.Y., "Experimental and Numerical Aqueous Flow
Through a Partially Saturated Fracture," Proceedings of the Sixth Annual
International Conference on High Level Radioactive Waste Management, American
Nuclear Society and American Society of Civil Engineers, Eds.. Las Vegas, NV,
May .M, 1995.
Cox, B.L., and Pruess, K., Numerical experiments on convcctsve heat transfer in water-
saturated porous media at near-critical conditions, Transport in Porous Media,
vol. 5, no. 3, pp. 299-323, 1990.
Cox, B.L., Pruess, K., and McKibbin, R., "Mathematical Modeling of Near-Critical
Convection," Presented at the Thirteenth Geothermal Reservoir Engineering
Workshop, Stanford University, Stanford, CA, Report LBL-24774, Lawrence
Berkeley Laboratory, Berkeley, CA, January 19-21,1988.
Ds la Torre, E. and Suarez, M.C., "DIOMRESKQcm): An Efficient Method Based on
Kryiov Subspaces to Solve Big, Dispersed, Unsymmetricai Linear Systems,"
Proceedings of the TOUGH Workshon '95, Report LBL-37200, Lawrence
Berkeley Laboratory, Berkeley, CA, pp. 2*99-304, March 20-22, 1995.
Dodge, F.T., and Green, R.T., "Pressure-Driven Gas Flow in Heated, Partially-Saturated
Porous Media," Proceedings of the Fifth Annual International Conference on High
Level Radioactive Waste Management, American Nuclear Society and American
Society of Civil Engineers, Eds., Las Vegas, NV, May 22-26,1994.
Doughty, C , "Flow Reduction Due To Degassing and Redissolution Phenomena,"
Proceedings of the TOUGH Workshop '95, Report LBL-37200, Lawrence
Berkeley Laboratory, Berkeley, CA, pp. 227-232, March 20-22,1995.
Doughty, C , and Pruess, K., A semi-analytical solution for heat pipe effects near high
level nuclear waste packages buried in partial!;' saturated geologic media.
International Journal of Heat and Mass Transfer, vol. 31, no. 1, pp. 79-90.
1988.
Doughty, C , and Pruess, K., A similarity solution for two-phase fluid and heat flow near
high-level nuclear waste packages empiaced in porous media, International Journal
of Heat and Mass Transfer, vol. 33, no. 6, pp. 1205-1222. 1990.
Doughty, C , and Pruess, K., "Verification of TOUGH2 Against a Seaiianalyiical Solution
for Transient Two-Phase Fluid and Heat Flow in Porous Media," Proceedings of
the TOUGH Workshop, Report LBL-29710, Lawrence Berkeley Laboratory,
Berkeley, CA, pp. 113-120, September 13-14,1990.

TOUGH2 SOFTWARE QUAIIFICATION - FEBRIMRV 1 9 9 6 97

Doughty, C , and Pruess, K., "A Mathematical Model for Two-Phase Water, Air and Heat
- Flow Around a Linear Heat Source Emplaced in a Permeable Medium," Presented
at the ASMEWICHE National Heat Transfer Conference, Minnepolis, MN, July
1991.
Doughty, C , and Pruess, K., A similarity solution for two-phase water, air, and heat flow
near a linear heat source in a porous medium, Journal of Geophysical Research,
vol. 97, B2, pp. 1S21-1838, 1992.
Eugster, S.M., and Senger, R.K., 'Investigation of Potential Water Inflow into a
Ventilated Tunnel of the Proposed Low/Intermediate Level Waste Repository in
Switzerland," Proceedings of the TOUGH Workshop '95, Report LBL-37200,
Lawrence Berkeley Laboratory, Berkeley, CA, pp. 7-13, March 20-22,1995.
Falta, R.W., and Brame, S., "Numerical Simulation of In-Situ DNAPL Remediation by
Alcohol Flooding," Proceedings of the TOUGH Workshop '95, Report LBL-
37200, Lawrence Berkeley Laboratory, Berkeley, CA, pp. 193-198, March 20-22,
1995.
Falta, R.W., Javandel, I., Pruess, K., and Witherspoon, P. A., Density-driven flow of
gas in the unsaturated zone due to the evaporation of volatile organic compounds,
Water Resources Research, vol. 25, no. 10, pp. 2159-2169,1989..
Falta. R.W., and Pruess, K., "STMVOC: A Numerical Simulator for Three Phase
Contaminant Transport in Nonisolhennal System," Proceedings of the TOUGH
Workshop, Report LBL-297 10, Lawrence Berkeley Laboratory, Berkeley, CA, pp.
29-36, September 13-14,1990.
Faita, R.W., Pruess, K., and Chesnut, D.A., Modeling adveetive contaminant transport
during soil vapor extraction, Groundwater, Vol. 31, no. 6, pp. 1011-1020, 1993.
Faita, R.W., Pruess, K., ST2VOC user's guide. Report LBL-30758, Lawrence Berkeley
Laboratory, Berkeley, CA, June 1991.
Falta, R.W., Pruess, K., Finsierie, S., and Battistelli, A., T2VOC user's guide. Report
LBL-36400. Lawrence Berkeley Laboratory, Berkeley, CA, 1995.
Falta, R.W., Prucss, K., Javandcl, I., and Witherspoon. P.A., Numerical modeling of
steam injection for the removal of nonaqueous phase liquids from the subsurface.
1. Numerical formulation, Water Resources Research, vol. 28, no. 2, pp. 433-
449, 1992.
Falta, R.W., Pruess, K., Javandel, I., and Witherspoon, P.A., Numerical modeling of
steam injection for the removal of nonaqueous phase liquids from the subsurface.
2. Code validation and application. Water Resources Research, vol. 28, no. 2, pp.
451-564, 1992.
Fair, J.M., "Gaseous Transport of Volatile Organics in Porous Media: Comparison of
Mathematical Models," Proceedings of the TOUGH Workshop, Report L8L-
20710, Lawrence Berkeley Laboratory, Berkeley, CA, pp. 21-28, September 13-
14, 1990.

IOUCJH2 SOFTWARE QlMliriCATION - FtCMMRV 1 3 9 6 98

 Finsterle, S., FLG: Design calculation for a gas test at Grimscl Rack Laboratory, Interner
Berichl, pp. 89-62, Nagra, July 1989.

Finsterle, S., ITOUGH2 user's guide version 2.2, Report LBL-34581, Lawrence Berkeley
Laboratory, Berkeley, CA., 1993.
Finsterle, S., Inverse modeling of test SB4-VM2/216.7 at Wellenberg, Report LBL-
354S4, Lawrence Berkeley Laboratory, Berkeley, CA, March 1994.
Finsterie, S., Bodvarsson, G.S., and Chen, G., "Inverse Modeling as a Step in the
Calibration of the LBL/USGS Site-Scale Model of Yucca Mountain," Proceedings of
the Sixth Annual International Conference on High Level Radioactive Waste
Management, American Nuclear Society and American Society of Civil Engineers,
Eds., Las Vegas, NV, May 1-4,1995.
Finsterie, S. and Mishra, S., "Interpretation of Hydraulic Tests in a Two-Phase Flow
System Using TOUGH," Proceedings of the TOUGH Workshop, Report LB-L-
29710, Lawrence Berkeley Laboratory, Berkeley, CA, pp. 133-138, September 13-
14, 1990.
Finsterle, S., Moridis, G.J. and Pruess, K., A TOUGH2 equation-of-state module for the
simulation of two-phase flow of air, water, and a miscible gelling liquid, Report
LBL-36086, Lawrence Berkeley Laboratory, Berkeley, CA, May 1994.
Finsterle, S., Moridis, G., Pruess, K., aad Persoff, P., Physical barriers formed irom
gelling liquids: Numerical design of laboratory and field experiments, EOS,
Transactions of the American Geophysical Union, 1994 Spring Meeting Supplement,
vol. 75, no. 16: 151, Report LBL-35315, Lawrence Berkeley Laboratory, Berkeley,
CA, 1994.
Finsterle, S., and Pruess, K., Design calculations for a combined ventilation and brine
injection experiment at the Grimsel Rock Laboratory, Switzerland, Report LBL-
34460, Lawrence Berkeley Laboratory, Berkeley, CA, October 1993.
Finsterlc, S., and Prucss, K., Solving the estimation-identification problem in mo-phase
flow modeling, Water Resources Research, vol, 31, no. 4, pp. 913-924, 1995,
Report VBL-34853, Lawrence Berkeley Laboratory, Berkeley, CA, 1993.
Finsterle, S., and Pruess, K., "Estimating Two-Phase Hydraulic Properties by Inverse
Modeling " Proceedings of the Fifth Annual International Conference on High Level
Radioactive Waste Management, American Nuclear Society and American Society of
Civil Engineers, Eds., Las Vegas, NV, May 22-26, 1994.
Finslerle, S., and Pruess, K., Optimizing multiphase aquifer remedialion using
ITOUGH2, Report U3L-36088, Lawrence Berkeley Laboratory, Berkeley, CA.,
1994.
Finsterie, S., and Pruess, K., "1TOUGH2: Solving TOUGH Inverse Problems,"
Proceedings of the TOUGH Workshop '95, Report LBL-37200, Lawrence
Berkeley Laboratory, Berkeley, CA, pp. 287-292, March 20-22,1995.

TOUGH2 SOFTWARE QUALIFICATION - FCBRUARV I 996 99

 Finsterlc, S., and Pruess, K., "Using Simulation-Optimization Techniques to Improve
Multiphase Aquifer Remediation," Proceedings of the TOUGH Workshop '95,
Report LBL-37200, Lawrence Berkeley Laboratory, Berkeley, CA, pp. 181-186,
March 20-22,1995.
Finsterle, S., Schiueter, E., and Pruess, K., Exploratory simulations of multiphase effects
in gas injection and ventilation tests in an underground rock laboratory, Report LBL-
2SS10, Lawrence Berkeley Laboratory, Berkeley, CA, 1990.
Finsterle, S., and Vomvoris, S., Inflow to Stripa validation drift under two-phase
conditions: scoping calculations, NAGRA International Report NN1B 91-40,
Wettingen, Switzerland, 1991.
Flint, A.L., Flint, L. E., and Hevesi, J.A., "The Influence of Long-Term Climate Change
on Net Infiltration at Yucca Mountain, Nevada," Proceedings of the Fourth Annual
International Conference on High Level Radioactive Waste Management, American
Nuclear Society and American Society of Civil Engineers, Eds., Las Vegas, NV,
April 1993.
Freeze, G.A., Larson, K. W., Davies, P.B., and Webb, S.W., "Using TOUGH2 to
Model the Coupled Effects of Gas Generation, Repository Consolidation, and
Multiphase Brine and Gas Flow at the Waste Isolation Pilot Plant," Proceedings of
the TOUGH Workshop '95, Report LBI^37200, Lawrence Berkeley Laboratory,
Berkeley, CA, pp. 34-39, March 20-22,1995.
Gaulkc, S., Bodvarsson, G.S., Aquino, B., Doughty, C , Okusu, N., Halfman, S., and
Ripperda, M., Reservoir analysis of The Geysers geothermal field, Report LBL-
1558, Lawrence Berkeley Laboratory, Berkeley, CA, 1988.
Gelier, J., and Pruess, K., "On Water Infiltration in Rough-Walled Fractures,"
Proceedings of the Sixth Annua! International Conference on High Level
Radioactive Waste Management, American Nuclear Society and American Society
of Civil Engineers, Eds., Las Vegas, NV, May 1-4,1995.
Glass, R.J., Tidwell, V.C., Flint, A.L., Peplinski, W., and Castro, Y., "Fracture-Matrix
Interaction in Topopah Spring Tuff: Experimental and Numerical Analysis,"
Proceedings of the Fifth Annual International Conference on High Level
Radioactive Waste Management, American Nuclear Society and American Society
of Civil Engineers, Eds., Las Vegas. NV, May 22-24, 1994.
Green, S.T., Martin, R.H., and Svedeman, S., "Use of TGUGH Computer Code to
Simulate a Laboratory-Scale Experiment," Proceedings of the TOUGH Workshop,
Report LBls-29710, Lawrence Berkeley Laboratory, Berkelev, CA, Dp. 121126,
September 13-14, 1990.
Green, R.T., and Rice, G., "Numerical Analysis of a Proposed Percolation Experiment at
the Pena Blanca Natural Analog Site," Proceedings of the Sixth Annua!
International Conference on High Level Radioactive Waste Management, American
Nuclear Society and American Society of Civil Engineers, Las Vegas, NV, May
1995.

TOUGH2 SOFTWARE QlMUHCATION - FEBRUARY 1 9 9 6 100

Hadgu, T„ Zimmerman, R.W., and Bodvarsson, G.S., Coupled reservoir-wellbore
simulation of geomsnnal reservoir behavior, Geothermics, vol. 24, no. 2, pp.
145-166, Report LBL-36141, Lawrence Berkeley Laboratory, Berkeley, CA,
1995.
Hadgo, T., Zimmerman, R.W., and Bodvarsson, G.S., "Coupling of the Reservoir
Simulator TOUGH and the Wellbore Simulator WFSA," Proceedings of the
TOUGH Workshop '95, Report LBL-37200, Lawrence Berkeley Laboratory,
Berkeley, CA, pp. 84-89, March 20-22,1995.
Halfman, S., Lippmann, M.J., and Bodvarsson, G.S., "Numerical Model of the Cerro
Prieto Field," Proceedings of the Eleventh Geotheimal Reservoir Engineering
Workshop, Stanford University, Stanford,, CA, SGP-TR-84, pp. 127-134,
Report LBL-20523, Lawrence Berkeley Laboratory, Berkeley, CA, 1986.
Ho, C.K., "Modeling Infiltration into a Tuff Matrix from a Saturated Vertical Fracture,"
Proceedings of the Fifth Annua! International Conference on High Level
Radioactive Waste Management, American Nuclear Society and American Society
of Civil Engineers, Eds., Las Vegas, NV, May 22-26,1994.
Ho, C.K., "Studies of Non-Isothermal Flow in Saturated and Partially Saturated Porous
Media," Proceedings of the Fifth Annual International Conference on High Level
Radioactive Waste Management, American Nuclear Society and American Society
of Civil Engineers, Eds., Las Vegas, NV, May 22-26, 1994.
Ho, C.K., "Assessing Alternative Conceptual Models of Fracture Flow," Proceedings of
the TOUGH Workshop '95, Report LBL-37200, Lanrence Berkeley Laboratory,
Berkeley, CA, pp. 323-330, March 20-22, 1995.
Ho, C.K., "Numerical Simulation of Multicomponent Evaporation and Gas-Phase
Transport Using M2NOTS2," Proceedings of the TOUGH Workshop '95. Report
LBL-37200, Lawrence Berkeley Laboratory, Berkeley, CA, pp. 221-225, March
20-22, 1995.
Ho, C.K., Dunn, E„ and Sobolik, S.R., "Ventilation and Vapor-Phase Transport Near the •
ESF Tunnel," Proceedings of the Sixth Annua! International Conference on High
Level Radioactive Waste Management, American Nuclear Society and American
Society of Civil Engineers, Eds., Las Vegas, NV.May 1-4,1995.
Ho, C.K., and Eaton, R.R., "TOUGH2 Model of the G-Tunne! Heater Test," Proceedings
of me Sixth Annual International Conference on High Level Radioactive Waste
Management, American Nuclear Society and American Society of Civil Engineers,
Eds., Las Vegas, NV, May 1-4.1995.
Ho, C.K., and Wilson, M.L., "Assessing Alternative Conceptual Models of Fracture
Flow," Proceedings of the Sixth Annual International Conference on High Level
Radioactive Waste Management, American Nuclear Society and American Society
of Civii Engine'rs, Eds., Las Vegas, NV, May 1-4, 1995.
Holfold, D.J., and Payer M.J., "Unsaturated Zone Moisture and Vapor Movement Induced
By Temperature Variations in Asphalt Barrier Field Lysimeters," Proceedings of the
TOUGH Workshop, Report LBL-29710, Lawrence Berkeley Laboratory.
Berkeley, CA, pp. 15-20 September 13-14, 1990.

TOUGII2 SOFTWAREQUAUFICATION - FEBHUARV 199G ToT

Hudson, D. B., Flint, A.L., and Gucrtal, W.R., "Modeling a Ponded Infiltration
Bxperiment at Yucca Mountain, Nevada," Proceedings of the Fifth Annual
International Conference on High Level Radioactive Waste Management, American
Nuclear Society and American Society of Civil Engineers, Eds., Las Vegas, NV,
May 22-26,1994.

Javeri, V., "Analysis of Nuclide Transport Under Natural Convection and Time Dependent
Boundary Condition Using TOUGH2," Proceedings of the TOUGK Workshop
'95, Report LBL-37200, Lawrence Berkeley Laboratory, Berkeley, CA, pp. 1-6,
March 20-22,1995.
Kaiasaki, K., Segan, S„ Pruess, K., and Vomvoris, S., A study of two-phase flow in
fracture networks, Report LBL-34706, Lawrence Berkeley Laboratory, Berkeley,
CA, 1994.
Kidd, E., "Energy Science and Technology Software Center," Proceedings of the TOUGH
Workshop "95, Report LBL-37200, Lawrence Berkeley Laboratory, Berkeley, CA.
pp. 277-280, March 20-22, 1995.
Kissiing, W.M., White, S.P., O'SulUvan, M.J., Bullivant, D.P., and Bro\ , K.L.,
"Modeling Chloride and CO, Chemistry at the Wairatcci Geothermal Field, New
Zealand," Proceedings of "the Twenty-First Annual Geothcrmal Reservoir
Engineering Workshop, Stanford University, Stanford,, CA, January 1996.
Kovscek, A.R., Patzek, T.W., and Radke, C.J., "FOAM3D: A Numerical Simulator for
Mechanistic Prediction of Foam Displacement in Multidimcnsions," Proceedings of
the TOUGH Workshop '95, Report LBL-37200, Lawrence Beikelev Laboratory,
Berkeley, CA, pp. 131-136, March 20-22, 1995.
Kwicklis, E.M., and Healy, R.W., Numerical investigation of steady liquid water flow in
a variably saturated fracture network, Water Resources Research, vol.29, no. 12,
pp. 4091-4102,1993.
Kwicklis, E., Kealy, R. W., Lu, N., and Bodvarsson, G.S., "Application of TOUGH to
Hydroiogic Problems Related to the Unsaturated Zone Site Investigation at Yucca
Mountain, Nevada," Proceedings of the TOUGH Workshop "95, Report LBL-
37200, Lawrence Berkeley Laboratory, Berkeley, CA, pp. 335-343, March 20-22,
1995.
Lai, C.H., and Bodvarsson, G.S. "Numerical Studies of Cold Water Injection into Vayor-
Domiaated Geothermal Systems," Paper presented at Society of Petroleum Engineers
Western National Meeting, Report SPB-21788, Long Beach, CA, March 1991.
Lai, C.H., Bodvarsson, G.S., and Truesdeil, A. H., Modeling studies of heat transfer ana
phase distribution in two-ahase geothermal reservoirs, Geothermics, vol. 23, no.
1, pp. 3-20, 1994.
Lai, C.K., Pruess, K., and Bodvarsson, G.S., On the accuracy of the M1NC
approximation. Report LBL-21025, Lawrence Berkeley Laboratory, Berkeley, CA
February 1986.

IOUGH2 SOFTWARE QUAliriCATION - FEBRUARY 1 9 9 6

Lam, S., Hunsoedt, A., and Kruger, P., Analysis of the Stanford geothermal reservoir
model experiments using the LBL reservoir simulator, Report SGF-TR-85, Stanford
Geothermai Program, Stanford University, Stanford, CA, April 1985.

Lam, S.T., Hunsbcdt, A., Kruger, P., and Pruess K., Analysis of the Stanford
geothermrJ reservoir model experiments using the LBL reservoir simulator,
Geothermics, vol. 17 no. 4, pp. 595-605, Report LBL-25957, Lawrence Berkeley
Laboratory, Berkeley, CA, 1988.
Lefebvre, R., Modeling acid mine drainage in waste rock dumps, Proceedings of the
TOUGH Workshop '95, Report LBL-37200, Lawrence Berkeley Laboratory,
Berkeley, CA, pp. 239-245, March 20-22,1995.
Lippmann, MJL, Aunzo, Z., Laky, C , Steingrimsson, B., Bodvarsson, G.S., Truesdeli,
A. H., Halfman-Dooley, S.E., and Cucllar, G., Prc-expioitation state of the
Ahuachapan geothermal field, El Salvador, Geothermics, vol. 20, no. 1, pp. 1-22,
1991.
Lippmann, M.J., and Bodvarsson, G.S., A modeling study of the natural state of the
Heber geothennal field, California, Geothermal Resources Council Trans, vol. 7,
pp. 441-447, 1983.
Lippmann, M.J., and Bodvarsson, G.S., Numerical studies of the heat and mass transport
in the Cerro Prieto geotherraal field, Mexico, Water Resources Research, vol. 19,
pp. 753-767, 1983.
Lippmann, M.J., and Bodvarsson, G.S., "The Generating Capacity of the Heber
Geolhermal Field, California," Proceedings of the Ninth Annual Geoiherroa!
Reservoir Engineering Workshop, Stanford University, Stanford,, CA, Report
SGP-TR-74,^. 157-166, December 13-15, 1983.
Lippmann, M J., and Bodvarsson, G.S., The Heber geothermal field, California: Natural
state and exportation modeling studies, J. Geophys. Res. vol. 90, no. B1, pp. 745-
758, 1985."
Lippmann, M.J., Halfman, S.E., and Bodvarsson, G.S., "Quantitative Mode! of the Cerro
Prieto Field," Proceedings of the Eleventh Geothermal Reservoir Engineering
Workshop Stanford University, Stanford,, CA, Report SGP-TR-93. pp. 127-134,
January 21-23, 1986.

Lippmann, M.J., Haukwa, C , Bodvarsson, G.S., and Mainieri, A., "Preliminary

Reservoir Engineering Studies of theMkavalles Geothennal Held," Proceedings of
the Seventeenth Geothermal Reservoir Engineering Workshop Stanford University,
Stanford, CA, Report SGP-TR-M1, pp. 127-137, January 29-31, 1992.
Lippmann, M.J., Lacky, C , Bodvarsson, G.S. Retana, M., and Cueilar, G.,
"Hydrogeoiogic Model of the Ahuachan Geoiherma! Field, El Salvador,"
Proceedings of the Fourteenth Geothermal Reservoir Engineering Workshop
Stanford University, Stanford,, CA, January 24-26, Report SGP-TR-122, pp.
265-272, 1989.

TOUGH2 SOFTWARE QUALIFICATION - FEBRUARY 1996 103

 Lippmann, MJ., Oldenburg, CM., and Pruess, K., "Double-Diffusive Convection in
Liquid-Dominated Geothcrmal Systems with High-Salinity Brines," Proceedings of
the Nineteenth Geothermal Reservoir Engineering Workshop Stanford University,
Stanford,, CA, Report SGP-TR-147, pp. 209-216, Report LBL-35039, Lawrence
Berkeley Laboratory, Berkeley, CA, 1994.
Lippmann, M.J., Oldenburg, CM. and Pruess, K., "Heat and Mass Transfer in
Kypersaline Geothermai Systems," Proceedings of the World Geothermal
Congress, Florence, Italy, pp. 1933-1938, May 18-31, 1995.
Lippmann, M.J., Ripperda, M., Bodvarsson, G.S., Cuellar, G., and Escobar, C , An
exploitation mode! and performance predictions for the Ahuacbap& geothermal
field, El Salvador, Geotkermics, vol. 20, no. 4, pp. 181-196,1991.
Lippmann, M.J., Ripperda, M., Bodvarsson, G.S., Goranson, C. and Witherspoon, P.
A., The Ahuachapan geothermal field, El Salvador: Exploitation model,
performance predictions and economic analysis. Report LBL-31J32. Lawrence
Berkeley Laboratory, Berkeley, CA, 1991.
Lippmann, M.J., andTraesdcll, A.H., Beneficial effects of groundwater entry into liquid-
dominated gcotherma! systems, Geothermal Resources Council Trans., vol. 14,
pp. 721-727, 1990.
Lippmann, M J., and Truesdell, A.H., Interaction of cold-water aquifers with exploited
reservoirs of the Cerro Prieto geothermal systems, Geothermal Resources Council
Trans., vol. 14, pp. 735-741,1990.
Lippmann, M.J., and Truesdell, A.H., "Reservoir Simulation and Geochcmica! Study of
Cerro Prieto I Wells," Proceedings of the 15th Geothermal Reservoir Engineering
Workshop Stanford University, Stanford, CA, Report SGP-TR-130, pp. 211-220,
January 23-25, 1990.
Lippmann, M.J., Truesdell, A.H., Manon, A.M., and Halfman, S.E., "The
Hydrogeoiogic-Geochemical Mode! of Cerro Prieto Revisited," Proceedings of the
Fourteenth Geothermal Reservoir Engineering Workshop Stanford University,
Stanford, CA, Report SGP-TR-122, pp. 163-172, January 24-26, 1989.
Lippmann, MJ., Truesdell, A.H., Marion, A.M., Quijano, L., and Coplea, T.. "Boiling
and Condensation Processes in the Cerro Prieto Beta Reservoir Under
Exploitation," Proceedings of the Seventeenth Geotherrnal Reservoir Engineering
Workshop Stanford University, Stanford, CA, Report SGP-TR-I41, pp. 205-214,
January 29-31, 1992.
Lippmann, MJ., Tsang, C.F., Mangold, D. C , and Doughty, C , Prediction of
reinjection effects in the Cerro Pricto gcothermal system, Geothemiics, vol. 13, pp.
141-162, 1983.
Lupo, MJ. and Moridis, G.J., "Predicting the Distribution of Contamination from a
Chlorinated Hydrocarbon Release," Proceedings of the TOUGH Workshop '95,
Report LBL-3T2GQ, Lawrence Berkeley Laboratory, Berkeley, CA, pp. 163-168
March 20-22,19Sj.

TOUGH2 SOFTWARE QUALIFICATION - FEBRUARV I 996 104

Manai, T., "EVEGAS (European Validation Exercise of gas. Migration Model) Project,"
Proceedings of tie TOUGH Workshop '95, Report LBL-37200, Lawrence
Berkeley Laboratory, Berkeley, CA, pp. 305316, March 20-22,1995.
McCartin T., Codell, R., and Nicholson, T., 'Two-Phase Row and Solute Transport
Simulations in a Tuff Drilkore," Proceedings of the TOUGH Workshop, Report
LBL-29710. Lawrence 3erke!ey Laboratory, Berkeley, CA, pp. 127-132,
September 13-14,1990.
McCraw, L.E., Flint, L.E., Bodvarsson, G.S. and Flint, A.L., Numerical modeling of
lateral infiltration into the Paintbrush Tuff at Yucca Mountain, Nevada, Report
L3L-35389, Lawrence Berkeley Laboratory, Berkeley, CA, 1994.
McCray, J.E., and Falta, R.W., "Air Sparging for Subsurface Remediation: Numerical
Analysis Using T2VOC," Proceedings of the TOUGH Workshop '95, Report
LBL-37200, Lawrence Berkeley Laboratory, Berkelev, CA, pp. 211-217, March
20-22, 1995.
McGuinness, M.J., Blafceiey, M., Pruess, K., and O'Sullivan M.3., Geothermal beat pipe
stability: solution selection by upstreaming and boundary conditions, Transport in
Porous Media, vol. 11, pp. 71 - 100, 1993.
McKibbin, R., and Pruess, K-, Some effects of non-condensable gas in geothermal
reservoirs with steam-water counterflow, Geothermics, vol. IS, no. 3, pp. 367-
375, 19S9.
McPherson, B., and Brcdehoeft, J., "Analysis of Over-Pressure Mechanisms in the Uinta
Basin, Utah," Proceedings of the TOUGH Workshop '95, Report LBL-37200,
Lawrence Berkeley Laboratory, Berkeley, CA, pp. 157-162, March 20-22-1995.
Menzies, A.J., and Forth, J.L., "Modeling the Response of the Geothermal System at
Lihir Island, Papua, New Guinea, to Mine Dewatering," Proceedings of the
TOUGH Workshop '95, Report LBL-37200, Lawrence Berkeley Laboratory,
Berkeley, CA, pp. 245-251, March 20-22,1995.
Misbxa, S., Attanayake, M.P., and Bodvarsson, G.S., Estimating properties of
unsaturated fractured formations from injection and falloff tests, Report LBL-
24796, Lawrence Berkeley Laboratory, Berkeley, CA, 1937.
Mishra, S., Attanayake, M.P., and Bodvarsson, G.S., Injection and falloff test analysis to
estimate properties of unsaturated fractures, AGU Monograph on Flow and
Transport through Unsaturated Fractured Rocks, vol. 42, .pp. 149-156. 1987.
Montazer, P., "Air-TOUGH, a Fully Three-Dimensional Linking of Atmosphere with Soil
Using Eddy Diffusivity Concept and V-TOUGH," Proceedings of the TOUGH
Workshop '95, Report LBL-37200, Lawrence Berkeley Laboratory, Berkeley, CA,
pp. 331- 334, March 20-22,1995.
Moridis, G.J., "A New Set of Direct and Iterative Solvers for the TOUGH2 Family of
Codes," Proceedings of the TOUGH Workshop '95, Report LBL-37200,
Lawrence Berkeley Laboratory, Berkeley, CA, pp. 293-298, March 20-22,1995.

TOUCH2 SOFTWARE QUALIFICATION - FEBRUARY 1996 105

 Moridis, G.J., and Pruess, K., TOUGH simulations of Updegraffs set of fluid and heat
flow problem. Report LBL-326U, Lawrence Berkeley Laboratory, Berkeley, CA,
1992.
Moridis, G.J., and Pruess, K., "Air Barriers for Waste Containment in the Subsurface,"
Proceedings of the TOUGH Workshop '95, Report LBL-37200, Lawrence
Berkeley Laboratory, Berkeley. CA, pp. 175-180, March 20-22, 1995,
Moridis, G., and Pruess, K., T2CG1: a package of preconditioned conjugate gradient
solvers for the TOUGH2 family of codes, Report LBL-36235, Lawrence Berkeley
Laboratory, Berkeley, CA, 1995.
Moya, S.L., and Iglesias, E.R., "Numerical Simulation of Carbon Dioxide Effects in
Geothennal Reservoirs," Proceedings of the TOUGH Workshop '95, Report LBL-
37200, Lawrence Berkeley Laboratory, Berkeley, CA, Dp. 119-130, March 20-22,
1995.
Niemi, A., and Bodvarsson, G.S., Preliminary capillary hysteresis simulations in fractured
rocks, Yucca Mountain, Nevada. In: P.F. German (editor). Rapid and far-reaching
hydrologic processes in the vadose zone, J. Contain. Hydro!., vol. 3, pp. 277-291
1988.
Niemi, A., and Bodvarsson, G.S., Preliminary capillary hysteresis simulalions for
fractured rocks—model development and results of simulations. Report LBL-
23593, Lawrence Berkeley Laboratory, Berkeley, CA, 1991.
Niemi, A., Pruess, K. and Bodvarsson, G.S., Incorporation of the capillary hysteresis
model HYSTR into the numerical code TOUGH, Report LBL-23S92, Lawrence
Berkeley Laboratory, Berkeley, CA, 1991.
Niemi, A., Spencer, A., Attanayake, M.P. and Bodvarsson, G.S., Preliminary
calculations of the effects of air and liquid water drilling on moisture Conditions in
unsaturated rocks, Report LBL-25073, Lawrence Berkeley Laboratory, Berkeley,
CA, 1988 s

Nitao, J.J., V-TOUGH - an enhanced version of the TOUGH code for the thermal and
hydroiogio simulation of large-scale problems in nuclear waste isolation, UCID-
2J954, Lawrence Livermore National Laboratory, Livcrmorc, CA, 1989.
Nitao, J.J., "Increasing the Efficiency of the TOUGH Code for Running Large-Scaie
Problems in Nuclear Waste Isolation," Proceedings of the TOUGH Workshop,
Report LBL-29710, Lawrence Berkeley Laboratory, Berkeley, CA, pp 143-148
September 13-14-, 1990.
Okabe, T., Osato, K., and Takasugi, S., "Case Study Using Faster TOUGH,"
Proceedings of the TOUGH Workshop, Report LBL-29710, Lawrence Berkeley
Laboratory, Berkeley, CA, pp. 61-66, September 13-14,1990.
Oldenburg, CM., Benson, S., Pruess, K., Daiscy, J., Brown, N., Gold, L., and
MacFarlane J., "The SELECT Environmental Remedy Selection Tool: A Platform for
T2VOC Multiphase Transport Modeling," Presented at the 1995 ASME/AIChE
National Hew Transfer Conference, Portland, OR, August i995.

TOUGH2 SOFIWARS QUALIFICATION - FEBRUARY 1996 106

 Oldcnberg, CM., Hinkins, R.L., and Pruess, K., "On the Development of MP-
TOUGH2," Proceedings of the TOUGH Workshop '95, Report LBL-37200,
Lawrence Berkeley Laboratory, Berkeley, CA, pp. 252-258, March 20-22, 1995. i
Oldenburg, CM., and Pruess, K., A two-dimensional dispersion module tor the
TOUGH2 simulator, Report LBL-32505, Lawrence Berkeley Laboratory,
Berkeley, CA, September 1993.

Oldenburg, CM., and Pruess, K., On numerical modeling of capillary barriers, Water
Resources Research, vol. 29, no. 4, pp. 1045-1056, 3993.
Oldenburg, CM., and Pruess, K., Numerical simulation of coupled flow and transport
with TCUGH2: a verification study, Report LSL-35273, Lawrence Berkeley
Laboratory, Berkeley, CA, 1994.

Oldenburg, CM., and Pruess, K., T2DNM: radionuclide transport forTOTJGH2, Report
LBL-34868, Lawrence Berkeley Laboratory, Berkeley, CA, 1994.
Oldenburg, CM., and Pruess, K., Dispersive transport dynamics in a strongly coupled
groundwater-brine flow system, Water Resources Research, vol. 31, no. 2, pp.
289-302, 1995.
Oldenburg, CM., and Pruess K. Mixing witfi first-order decay in variable velocity porous
' media flow, in press in Transport in Porous Media, 1995.
Oldenberg, CM., and Pruess, K., "Strongly Coupled Single-Phase Flow Problems:
Effects of Density Variation, Hydrodynamic Dispersion, and First Order Decay,"
Proceedings of the TOUGH Workshop '95,* Report LBL-37200, Lawrence
Berkeley Laboratory, Berkeley, CA, pp. 205-210, March 20-22,1995.
Oldenburg, CM., K. Pruess, K-, and Lippmann, M.J., '"Double-Diffusive Convection in
Liquid-Dominated Geothermal Systems with High-Salinity Brines," Presented at
the 19th Geothermal Reservoir Engineering Workshop Stanford University,
Stanford,, CA, Report LBL-35039, Lawrence Berkeley Laboratory, Berkeley, CA,
January 1994.
Oldenburg, CM., Pruess, K., and Lippmann M., "Heat and Mass Transfer in Hypersaiine
Geothcrmal Systems," Proceedings of the World Geothcrmal Congress 1995, pp.
1647-1652, International Gcotacrmai Association, May 1995.
O'Sullivan, M.L, "A Simple Model of a Vapor-Dominated Geothermal Reservoir,"
Proceedings of the TOUGH Workshop, Report LBL-29710, Lawrence Berkeley
Laboratory, Berkeley, CA, pp. 37-44, September 13-14,1990.
O'Sullivan, M.J., Bodvarssoc, G.S., Pruess, K., and Blakeley M. R. Fluid and heat
(

flow in gas-rich geothermal reservoirs. Society of Petroleum Engineers Journal,

vol. 25, no. 2, pp. 215-226, April 1985.

TOUGH2 SoFTwwse QUALIFICATION - HBHIIARY 1996 107

O" Sullivan, MJ., and Bullivant, D.P., "A Graphical interface to the TOUGH Family of
How Simulators," Proceedings of die TOUGH Workshop '95, Report LBL-
37200, Lawrence Berkeley Laboratory, Berkeley, CA, pp. 90-95, March 20-22,
1995.
Pantazidou M., "LNAPL Infiltration in the Vadose Zone: Comparisons of Physical and
Numerical Simulations," Proceedings of the TOUGH Workshop '95, Report LBL-
37200, Lawrence Berkeley Laboratory, Berkeley, CA, pp. 169-174, March- 20-22,
1995.
Persoff, P., Pruess, K., Benson, S., Wu, Y.S., Radke, C.J., Witherspoon, P-A., and
Shikari, Y.A., Aqueous foams for control of gas migration and water coning in
aquifer gas storage, Energy Sources, vol. 12, pp. 479-497,1990.
Piepho, M.G., "Preliminary Analysis of Tank 24i-C-!06 Diyout Due to Large Postulated
Leak and Vaporization," Proceedings of the TOUGH Workshop '95, Report LBL-
37200, Lawrence Berkeley Laboratory, Berkeley, CA, pp. 40-45, March 20-22,
1995.
Poppei, J., and Fischer, D., "Prognostic Simulation of Reinjection - Research Project
Geotherma! Site Neustadt-GIewe, Germany," Proceedings of the TOUGH Workshop
'95, Report LBL-37200, Lawrence Berkeley Laboratory, Berkeley, CA, pp. 96-100,
March 20-22,1995.
Pruess, K., Development of the general purpose simulator MULKOM, Annual Repon
1982, Earth Sciences Division, Report LBL-15S00, Lawrence Berkeley Laboratory,
Berkeley, CA, 1983.
Pruess, K., GM1NC - a mesh generator for flow simulations in fractured reservoirs,
Report LBL-J5227, Lawrence Berkeley Laboratory, Berkeley, CA, March 1983.
Pruess, K., Heal transfer in fractured geothertnal reservoirs wim boiling. Water
Resources Research, voi. 19, no. 1, pp. 201-208, February 1983.
Pruess, K., A quantitative model of vapor dominated geothermal reservoirs as heat pipes
in fractured porous rock, Geothermal Resources Council, Transactions, vol. 9, Part
B, pp. 353-361, August 1985.
Pruess, K., TOUGH user's guide, Report LBL-20700, Lawrence Berkeley Laboratory,
Berkeley, CA, and Nuclear Regulatory Commission Report NUREG/CR<I645,
Washington, D.C., 1987.
Pruess, JC, 'Modeling of Fluid and Heat Flow in Fractured Geothermai Reservoirs,"
Proceedings of the International Symposium on Geothermal Energy, Kumamoto and
Beppuk, Japan, pp. 205-209, November 1988.
Pruess, K., SHAFT, MULKOM, TOUGH: a set of numerical simulators for multiphase
fluid and heat flow. Report LBL-24430, Lawrence Berkeley Laboratory, Berkeley,

TOUGH2 SOFTWARE QvAimcAnoN - FEMIMRV 1996 108

 Pruess, K-, Modeling studies of multiphase fluid and heat flow processes in nuclear waste
isolation, W. Lutze and R. C. Ewing, (eds.;, Scientific Basis for Nuclear Waste
Management XII, Materials Research Society Symposium Proceedings, Pittsburgh,
PA, pp. 793-803, 1989.
Pruess, K., Modeling of geotherma' reservoirs: fundamental processes, computer
simulations and field applications, Geotkermics, vol. 19 , no. 1, pp. 3-15,1990.
Pruess, K., Numerical modeling of gas migration at a proposed repository for low and
intermediate level nuclear wastes at Oberbavenstock, Switzerland, Report LBL-
25413, Lawrence Berkeley Laboratory, Berkeley, CA, 1990.
Pruess, K., "Overview of TOUGH2, A General-Purpose Numerical Simulator for
Multiphase Nonisothermal Flow," Proceedings of the TOUGH Workshop, Report
LBL 29710, Lawrence Berkeley Laboratory, Berkeley, CA, pp. 89-9S, September
13-14, 1990.

Pruess, K., (editor). Proceedings of the TOUGH workshop, Report LBL-297I0,

Lawrence Berkeley Laboratory, Berkeley, CA, September 1990.
Pruess, K., EOS7, an equation-of-state module for the TOUGH2 simulator for two-phase
flow of saline water and air. Report LBL-31114, Lawrence Berkeley Laboratory,
Berkeley, CA, 1991.
Pruess, K., Grid orientation and capillary pressure effects in the simulation of water
injection into depicted vapor zones, Geothermics, vol. 20, no. 5/6, pp. 257-277,
1991.
Pruess, K., "Grid Orientation Effects in the Simulation of Cold Water Injection into
Depleted Vapor Zones," Presented at the Sixteenth Gccthcrmal Reservoir Engineering
Workshop, Stanford University, Stanford, CA, Report LBL-30117, Lawrence
Berkeley Laboratory, Berkeley, CA, 1991.

Pruess, K., TOUGH2, a general-purpose numerical simulator for multiphase fluid and heat
flow, Report LBL-29400, Lawrence Berkeley Laboratory, Berkeley, CA, May 1991.
Prucss, K., Analysis of flow processes during TCE infiltration in heterogeneous soils at
the Savannah River Site, Aiken, South Carolina, Report LBL-32418. Lawrence
Berkeley Laboratory, Berkeley, CA, June 1992.
Pruess, K., Brief guide to the MINC-mcthod for modeling flow and transport in fractured
media, Report LBL-32195, Lawrence Berkeley Laboratory, Berkeley, CA, May
1992.
Pruess, K., Dispersion module for TOUGH2, Report LBL-32505, Lawrence Berkeley
Laboratory, Berkeley, CA, 2993.
Pruess, K., "Geysers Injection Modeling," Presented at the Geothermal Program Review
XH, U.S. Department of Energy, Report LBL-35500, Lawrence Berkeley
Laboratory, Berkeley, CA, April 1994.

TOUCH2 SOFTWAKE QUALIFICATION - fEsm/AHY J9SC 109

 Pruess, K., "Liquid-Phase Dispersion During Injection into Vapor-Dominated Reservoirs,"
Presented at the 19th Annual Geothermal Reservoir Engineering Workshop Stanford
University, Stanford, CA., Report LBL-35059, Lawrence Berkeley Laboratory,
Berkeley, CA, January 1994.
Pruess, K., On the validity of a Fickian diffusion model for the spreading of liquid
infiltration plumes in partially saturated heterogeneous media. Computational
Methods in Water Resources X, vol. 3, pp. 537-544, Kluwer Academic Publishers,
Dordrecht, 1994.
Pruess, K., A Fickian diffusion model for the spreading of liquid plumes infiltrating in
heterogeneous media, Transport in Porous Media, in press 1995.
Pruess, K., "Effective Parameters, Effective Processes: From Porous Flow Physics to In-
Situ Remediation Technology," Paper presented at the VEGAS-Symposium,
University of Stuttgart/Germany, Report LBL-37414, Lawrence Berkeley
Laboratory, Berkeley, CA, September 25-27,1995.
Pruess, K., "Numerical Simulation of Water Injection into Vapor-Dominated Reservoirs,"
Proceedings of the World Geothermal Congress, pp. 1673-1680, Internationa!
Geothermal Association, May 1995.
Pruess, K., (editor). Proceedings of the TOUGH Workshop '95, Report LBL-37200,
Lawrence Berkeley Laboratory, Berkeley, CA, March 1995.
Pruess, K., "Injection Plume Behavior in Fractured, Vapor-Dominated Reservoirs," Paper
presented at the Twenty-First Annual Geothemia! Reservoir Engineering
Workshop, Stanford, University, Stanford, CA, January 1996.
Pruess, K., and Antunez, E., "Applications of TOUGH2 to Infiltration of Liquids in Media
with Strong Heterogeneity," Proceedings of the TOUGH Workshop '95, Report
LBL-37200, Lawrence Berkeley Laboratory, Berkeley, CA, pp. 6976, March 20-
22, 1995.
Pruess, K., and Bodvarsson, G.S., "A Seven-Point Finite Difference Method for
Improved Grid Orientation Performance in Pattern Steam Floods," Proceedings of
the Seventh Society of Petroleum Engineers Symposium on Reservoir Simulation,
San Francisco, CA, Paper SPE-12252, pp. 175 -184,1983.

Pruess, K., and Bodvarsson, G.S., Thermal effects of reinjection in geothermal reservoirs
with major vertical fractures, J. Pet. Tech., vol. 36, no. 10, pp. 1567-1578.1984.
Pruess, K., Bodvarsson, G.S., Schroeder, R.C., and Witherspoon, P.A., Modei studies
of the depletion of two-phase gcothermal reservoirs. Society of Petroleum
Engineers Journal, vol. 22 , no. 2, pp. 2S0-290, April 1982.
Pruess, K., Bodvarsson, G.S., Stefansson, V. and Eliasson, E X , The Krafla
geolheima! field, Iceland: 4. History match and prediction of individual wel!
performance. Water Resources, Res., vol. 20, no. 11, pp. 1561-1584, Report
LBL-J6203, Lawrence Berkeley Laboratory, Berkeley, CA, 1984.

TOUGH2 SOFTWARE QUALIFICATION - FEBRUARY 1996 110

 Pruess, iL, Caiore, C , Celati, R., and Wu, Y.S. An analytical solution for heat transfer
at a boiling front moving through a porous medium, Int. J. of Heat and Mass
Transfer, vol. 30, no. 12, pp. 2595-2602, 1987.
M
Pniess, K., Celati, R., Caiore, C , and D'Amore, F., C 0 Trends in the Depletion of the
2

Larderello Vapor-Dominated Reservoir," Presented at the Tenth Gcothermal

Reservoir Engineering Workshop, Stanford University, Stanford, CA, Report
LBL-I9092, Lawrence Berkeley Laboratory, Berkeley, CA 1985.

Pruess, IC, Celati, R., Caiore, C , and Cappetti, G., "On Fluid and Heat Flow in Deep
Zones of Vapor-Dominated Gcothermal Reservoirs," Paper presented at the Twelfth
Annual Geothermal Reservoir Engineering Workshop, Stanford University,
Stanford, CA, Report LBL-22SI0, Lawrence Berkeley Laboratory, Berkeley, CA,
pp. 89-96. 3987.
Pruess, SC, and Enedy, S., "Numerical Modeling of Injection Experiments at The
Geysers," Presented at the Eighteenth Geothermal Reservoir Engineering
Workshop, Stanford University, Stanford, CA, Report LBL-33423, Lawrence
Berkeley Laboratory, Berkeley, CA January 26-28,1993.
Pruess, K., Finsterle, S., Persoff, P., and Oldenburg, C , "Phenomecological Studies of
Two-Phase Processes for Nuclear Waste Isolation," Proceedings of the Fifth
Annual International Conference on High Level Radioactive Waste Management,
American Nuclear Society and American Society of Civil Engineers, Eds., May 22-
26, 1994.
Prccss, K., and Karasaki, K., "Proximity Functions for Modeling Fluid and Heat Flow in
Reservoirs with Stochastic Fracture Distributions," Presented at the 8th Geotherma!
Reservoir Engineering Workshop, Stanford University, Stanford, CA, December
1982.
Pruess, K., and Narasimhan, T.N., On fluid reserves and the production of superheated
steam from fractured, vapor-dominated geothermal reservoirs, J. Ceophys. Res.,
vol. 87, no. B l l , pp. 9329-9339,1982.
Pruess, K., and Narasimhan, T.N., A practical method for modeling fluid and heat flow
in fractured porous media. Society of Petroleum Engineers Journal, vol. 25, pp.
14-26, 1985.
Pruess, K., and Narasimhan, T.N., "Numerical Modeling of Multiphase and
Nonisothermal Flow in Fractured Media," Proceedings of the International
Conference on Fluid Flow in Fractured Rocks, Atlanta, GA, May 1988.
Pruess, K., and O'Sullivan, M., "Effects of Capillarity and Vapor Adsorption in the
Depletion of Vapor-Dominated Geothermai Reservoirs," Presented at the
Seventeenth Geothermal Reservoir Engineering Workshop, Stanford Univeisity,
Stanford, CA, Report LBIJ-31692, Lawrence Berkeley Laboratory, Berkeley, CA,
1992.
Prucss, K., and Tsang, Y.W., On two-phase relative permeability and capillary pressure of
rough-walled rock fractures, Water Resources Research, vol. 26, no. 9, pp. 1915-
3926, 1990.

TOUGII2 SOFTWARE QUALIFICATION . FEBRUARY 1996 in

 Pruess, K., and Tsang, Y., "Modeling of Strongly Heat-Driven Flow Processes at a
Potential High-Level Nuclear Waste Repository at Yucca Mountain, Nevada,"
Proceedings of the Fourth Annual International High Level Radioactive Waste
Management Conference, American Nuclear Society and American Society of Civil
Engineers, Eds., Report IBL-33597, Lawrence Berkeley Laboratory, Berkeley,
CA, April 26-30,1993.
Pruess, K., and Tsang, Y., 1990, Thermal modeling for a potential high-level nuclear
waste repository at Yucca Mountain, Nevada, Repon LBL-35381, Lawrence
Berkeley Laboratory, Berkeley, CA, 1994.
Pruess, K., Tsang, Y.W. and Wang, J.S.Y., Numerical studies of fluid and heat flow
near high-level nuclear waste packages emplaced in partially saturated fractured
tuff, Report LBL-18552, Lawrence Berkeley Laboratory, Berkeley, CA,
November 1984.
Pruess, K., Tsang, Y.W., and Wang, J.S.Y., "Modeling of Strongly Heat-Driven Flow in
Partially Saturated Fractured Porous Media," Proceedings of the Seventeenth
International Congress on the hydrogeology of Rocks of Low Permeability,
University of Arizona, Tucson, Arizona, pp. 486-497,1985.
Pruess, K., and Wang, J.S.Y., TOUGH- a numerical model for nonisothermal unsaturated
Slow to study waste canister heating effects, Mat. Res. Soc. Symp. Proc, vol. 26,
Scientific Basis for Nuclear Waste Management, G. L. McVay, ed.. North
Holland, New York, pp. 10131-1038, 1984.
Pruess, K. and Wang, J.S.Y., Numerical modeling of isothermal and nonisothermal flow
in unsaturated fractured rock - a review, Flow and Transport Tlirough
Unsaturated Fracture Rock, D. Evans and T. Nicholson, (ed.), American
Geophysical Union, Geophysical Monograph, vol. 42, pp. 11-21,1987.
Pruess, K., Wang, J.S.Y., and Tsang, Y.W., Effective continuum approximation for
modeling fluid and heat flow in fractured porous tuff, Report LBL-20778,
Lawrence Berkeley Laboratory, Berkeley, CA, 1988.
Pruess, K., Wang, Jf.S.Y., and Tsang, Y.W., On thermohydrologic conditions near high-
level nuclear wastes emplaced in partially saturated fractured tuff. 1. simulation
studies with explicit consideration of fracture effects, Water Resources Research,
1990, vol. 26, no. 6, pp. 1235-1248, 1990.
Pruess. K., Wang, J.S.Y., and Tsang, Y.W., On thermohydrologic conditions near high-
level nuclear wastes emplaced in partially saturated fractured tuff. 2. effective
continuum approximation, Water Resources Research, 1990, vol. 26, no. 6, pp.
1249-1261, 1990.
Pruess, K., Wercs, O., and Schroeder, R.C., Distributed parameter modeling of a
producing vapor-dominated geotliennal reservoir — Serrazzaao, Italy, Water
Resources Research, vol. 19 , no. 5, pp. 1219-1230, October 1983.
Pruess, K., Wilt, M., Bodvarsson, G.S., and Goldstein N.E., Simulation and resistivity
modeling of a geothermal reservoir with waters of different salinitv, Geothennics,
vol. 12 no. 4, pp. 291-307,1983.

TOUCK2 SOFTWARE QUALIFICATION - TEBRUARY 1396 T12

 Pruess, K., and Wu, Y.S., "A Semi-Analytical Method for Heat Sweop Calculations in
Fractured Reservoirs," Proceedings of the Thirteenth Geothermal Reservoir
Engineering Workshop, Stanford University, Stanford, CA, Report SGP-TR-113,
pp. 229-223, Report LBL-24463, Lawrence Berkeley Laboratory, Berkeley, CA,
19SS.
Praess, K., and Wu, Y.S., On PVT-data, well treatment and the preparafion of input data
for an isothermal gas-water-foam version of mulkom, Report LBL-25783,
Lawrence Berkeley Laboratory, Berkeley, CA, August 1988.
Pruess, K., and Wu, Y.S., A new semianalytical method for numerical simulation of fluid
and heat flow in fractured reservoirs. Society of Petroleum Engineers Adv. Tech.
• Ser., vol. 1, pp. 63-72, 1993.
Reeves, M. and Lingineni, S., "Drift-Scale Thermohydrologic Analyses for the Proposed
Repository at Yucca Mountain, Nevada," Proceedings of the TOUGH Workshop
'95, Report LBL-37200, Lawrence Berkeley Laboratory, Berkeley. CA, pp. 21-
22, March 20-22,1995.
Rulon, J., Bodvarsson, C.S., and Montazer, P., Preliminary numerical simulations of
groundwater flow in the unsaturated zone. Yucca Mountain, Nevada, Report LBL-
20553, Lawrence Berkeley Laboratory, Berkeley, CA, 1986.
Sato, T., Okabe, T., Osato, K., and Tafcasugi, S., "Graphical User Interface for
TOUGH/TOUGH2 - Develcpment of Database, Pre-Processor, and Post-Processor,:
Proceedings of the TOUGH Workshop '95, Report LBL-37200, Lawrence Berkeley
Laboratory, Berkeley, CA, pp. 271-276, March 20-22,1995.
Schiueter, E. and Pruess, K., Sensitivity studies on parameters affecting gas release from
an underground rock cavcm, Report LBL-28818, Lawrence Berkeley Laboratory,
Berkeley, CA, 1990.
Scneviratne, A., and Findikakis, A.N., "The Effect of Vadose Zone Heterogeneities on
Vapor Phase Migration and Aquifer Contamination by Volatile Organics,"
Proceedings of the TOUGH Workshop '95, Report LBL-37200, Lawrence
Berkeley Laboratory, Berkeley, CA, pp. 226-227, March 20-22,1995.
Senger, R. K., Zuidema, P., Kngi, M., Mislira, S., and Jaquet, O., "Investigating the
Effects of Near-Field Characteristics on Gas and Water Transport from a LflLW
Repository," Presented at the SMiRT Post Conference Seminar No. 10. Report
NIB 94-44, NAGRA, Wettingen, Switzerland, May 1994.
Scnger, R.K., Investigation of two phase flow associated with gas generation from a
low/Intcrincdiatc-levcl waste repository, Report NIB 95-10, NAGRA Wettingen,
Switzerland, February 1995.
Shaikh, M., Ych, T.-CJ., and Rasmussen, T.C., "Two-Phase Nonisothermal Ilydrologic
Transport Simulations at the Apache Leap Tuff Held Site," Proceedings of the
TOUGH Workshop, Report LBL-29710, Lawrence Berkeley Laboratory,
Berkeley, CA, pp. 1-8, September 13-14,3990.

TOUGH2 SOFTWARE QUALIFICATION - FMRUARY 1 9 9 5 113

 Shan, C , "Verification of T2VOC Using an Analytical Solution for VOC Transport in
Vadose Zone," Proceedings of (he TOUGH Workshop '95, Report LBL-37200,
Lawrence Berkeley Laboratory, Berkeley, CA, pp. 218-220, March 20-22,1995.
Suarez M.C., and Manon A., "Injection of Cold Water and Air into a Two-Phase Volcanic
Hydrothermal System," Proceedings of the TOUGH Workshop, Report LBL-
29710, Lawrence Berkeley Laboratory, Berkeley, CA, pp. 45-52, September 13-
14, 1990.
Suarez, M.C., and Samaniego, F., 'Triple Porosity/Permeability Flow in Faulted
Geoihermal Reservoirs: Two-Dimensional Effects," Proceedings of the TOUGH
Workshop "95, Report LBL-37200, Lawrence Berkeley Laboratory, Berkeley, CA,
pp. 113-118, March 20-22,1995.
Todesco, M., "Modeling of Hydrothermal Circulation Applied to Active Volcanic Areas-
The Case of Vulcano (Italy)," Proceedings of the TOUGH Workshop '95, Report
LBL-37200, Lawrence Berkeley Laboratory, Berkeley, CA, pp. 107-112, March
20-22, 1995.
Tsang, Y.W., and Pruess, K., A study of thermally induced convection near a high-level
nuclear waste repository in oartially saturated fractured tuff. Water Resources
Research, 23 (10), pp. 1958-1967,1987.
Tsang, Y.W., and Pruess, K., Preliminary studies of gas phase flow effects and moisture
migration at Yucca Mountain. Report LBL-28319, Lawrence Berkeley Laboratory,
Berkeley, CA, 1989.
Tsang, Y.W., and Pruess, K., Further modeling studies of gas movement and moisture
migration at Yucca Mountain, Nevada, Report LBL-29127, Lawrence Berkeley
Laboratory, Berkeley, CA, 1990.
Tsang, Y.W., and Pruess, K., "Modeling Studies of Gas Movement and Moisture
Migration at Yucca Mountain, Nevada." Proceedings of the TOUGH Workshop,
Report LBL-29710, Lawrence Berkeley Laboratory, Berkeley, CA, pn. 99-106,
September 13-14,1990.
Tsang, Y.W., Pruess, K., and Wang, J.S.Y., 'The Role of Fault Zones in Affecting
Multiphase Flow at Yucca Mountain," Proceedings of the Fourth International
Conference on High Level Radioactive Waste Management, American Nuclear
Society and American Society of Civil Engineers, Eds., Las Vegas, NV, Aprii
1993.
Updegraff, CD., Comparison of strongly heat-driven flow codes for unsaturated media.
Report SAND-88-7145, Sandia National Laboratories, Albuquerque, NM, August
1989.
Updegraff, CD., and Bonano, E.J., "Comparison of Strongly Heat-Driven Flow Codes
for Unsaturated Media," Presented at the International Conference and Workshop
on the Validation of Flow and Transport Models for the Unsaturated Zone,
Ruicoso, NM, May 1988.

TOUGH2 SOrTWARE QlMlirtCATION - FEBRUARY 1 9 9 6 114

 Verma, A.K., Effects of phase transformation on steam-water relative permeabilities,
Doctoral Dissertation, University of California, Berkeley, CA, Report LBL-20594,
Lawrence Berkeley Laboratory , Berkeley, CA, March 1986.
Verma, A.K., and Pruess, K., "Enhancement of Steam Phase Relative Permeability Due
To Phase Transformation Effects in Porous Media," Presented at the Eleventh
Geothermal Reservoir Engineering Workshop, Stanford University, Stanford, CA.,
Report LBL-21027, Lawrence Berkeley Laboratory, Berkeley, Ca 1986.

Verma, A„ and Pniess, K., Effects of silica redistribution on performance of high-level

nuclear waste repositories in saturated geologic formations, in C.F. Tsang (ed.),
Coupled Processes Associated with Nuclear Waste Repositories, Academic Press,
NY, 1987.
Verma, A. and Pruess, K., Thermohydrologic conditions and silica redistribution near
high-level nuclear wastes emplaced in saturated geological formations, Journal of
Geophysical Res., vol. 93, no. B2, pp. 1159-1173,19SS.
Verma, A.K., Pruess, K., Bodvarsson, G.S., Tsang, C.F., Witherspoon P.A., "Design
and Development of a Test Facility to Study Two-Phase Steam/Water How in Porous
Media," Paper presented at the Ninth Geothermal Reservoir Engineering Workshop,
Stanford University, Stanford, CA., December 1983.
Veima, A.K., Pruess, K., Tsang, C.E. and Witherspoon P.A., "A Study of Two-Phase
Concurrent Flow of Steam and Water in an Unconsolidated Porous Medium,"
Proceedings of the Twenty-Third National Heat Transfer Conference, American
Society of Mechanical Engineers, Denver, CO, pp. 135-143,19S5.
Wang, J.S.Y., Cook, N.G.W., Wollenberg, H.A., Carnahan, C.L., Javandel, I., and
Tsang, C.F., "Geohydrologic Data and Models of Rainier Mesa and Their
Implications to Yucca Mountain," Proceedings of the Fourth Annual International
Higvel Radioactive Waste Management Conference, American Nuclear Society and
American Society of Civil Engineers, Eds., Las Vegas, NV, April 1993.

Wang, J.S.Y., and Narasimhan, T.N., Hydrological mechanisms governing fluid flow in
fractured welded units and porous non-welded units ai Yucca Mountain. Report
LBL-21022, Lawrence Berkeley Laboratory, Berkeley, CA, 1986.
Wang, J.S.Y., and Narasimhan, Hydrologic modeling of vertical and lateral movement of
partially saturated fluid flow near a fault zone at Yucca Mountain, Report LBL-
23510, Lawrence Berkeley Laboratory, Berkeley, CA, June 1987.
Webb, S.W., TOUGH2 simulations of the TEVES projeci inducing the behavior of a
single-component NAPL, SAND94-1639, Sandia National Laboratories,
Albuquerque, NM, (in review).

Webb, S.W., and Chen, J.C., "Phasic Pressure Difference Effects in Two-Phase Flow for
Dissolved Gas Exsolution," Proceedings of the TOUGH Workshop, Report LBL-
29710, Lawrence Berkeley Laboratory, Berkeley, CA, pp. 107-112, September
13-14, 1990.

TOUGH2 SonwARe QUALIFICATION - FEBRUARY 1996 115

Webb, S.W., and Larson, K-W.," The Effect of Stratigraphic Dip on Multiphase How at
the Waste Isolation Pilot Plant," Proceedings of the TOUGH Workshop '95,
Report LBL-37200, Lawrence Berkeley Laboratory. Berkeley, CA. pp. 22-27,
March 20-22,1995.
Webb. S.W.. Larson, K.W.. and Christian-Fear, T.L., "Summary of Applications of
TOUGH2 to the Evaluation of Multiphase Flow Processes at the WTPP,"
Proceedings of the TOUGH Workshop '95, Report LBL-37200, Lawrence
Berkeley Laboratory, Berkeley, CA, pp. 46-51, March 2022,1995.
Webb, S.W., and Phelan, J.M., "Prediction of Single Component NAPL Behavior for the
TBVES Project Using TVOC," Proceedings of the TOUGH Workshop '95, Report
LBL-37200, Lawrence Berkeley Laboratory, Berkeley, CA, pp. 199-404, March,
20-22, 1995.
Webb. S.W., and Stormont, J.C., "Modeling of Capiiiary Barriers and Comparison to
Data," Proceedings of the TOUGH Workshop '95, Report LBL-37200, Lawrence
Berkeley Laboratory, Berkeley, CA, pp. 317322, March 20-22, 1995.
Weir, G.J., White, S.P., and Kissing, W.M., "Reservoir Storage and Containment of
Greenhouse Gases," Proceedings of the TOUGH Workshop '95, Report LBL-
37200, Lawrence Berkeley Laboratory, Berkeley, CA, pp. 233-23S, March 20-22,
1995.
White, S.P., "Speeding-Up TOUGH," Proceedings of the TOUGH Workshop, Report
LBL-29710, Lawrence Berkeley Laboratory, Berkeley, CA, pD. 73-88, September
13-14, 1990.
White, S.P., "Transport of Reacting Chemicals in a Two-Phase Reservoir," Proceedings
of the Sixteenth New Zealand Geotherrnal Workshoo, Auckland, New Zealand,
1994
White, S.P., Multiphase non-isothermal transport of systems of reacting chemicals, Water
Resources Res., vol. 32, no. 7, pp. 1761-1772, 1995.
Witherspoon, P., Fuller, P., and Finsterie, S., "Three-Dimensional Multiphase Effects in
Aquifer Gas Storage," Proceedings of the TOUGH Workshop '95, Report LBL-
37200, Lawrence Berkeley Laboratory, Berkeley, CA, pp. 137156, March 20-22,
1995.
Wittwer, C.S., Bodvarsson, G.S., Chornack, M.P., Flint, A.L., Flint, L.E., Lewis,
B.B., Spengler, R.W., and Rautman, C.A., "Design of a Three-Dimensiona! Site-
Scale Model for the Unsaturated Zone at Yucca Mountain, Nevada," Proceeding of
the Third Annual International Conference on High Level Radioactive Waste
Management, American Nuclear Society and American Society of Civil Engineers,
Eds., Las Vegas, NV, April 1992.
Wittwer, C.S., Bodvarsson, G.S., Chornack, M.P., Flint, A.L., Flint, L.E., Lewis,
B.B., Spengler, R.W., and Rautman, C.A., Development of a three-dimensional
site-scale mode! for the unsaturated zone at Yucca Mountain, Nevada, Radioactive
Waste Management and Environmental Restoration, Karwocd Academic
Publishers, GmbH, U.S., vol. 19, pp. 147-167, 1994.

T O U G H 2 SOFTWAR6 QUALIFICATION - FEBRUARY i 9 9 6

i«S
 Wittwer, C.S., Chen, G., and Bodvarsson, G.S., "Studies of the Role of Fault Zones on
Fluid Flow Using the Site-Scale Numerical Model of Yucca Mountain,"
Proceedings of the Fourth Annual International Conference on High Level
Radioactive Waste Management, American Nuclear Society and American Society
of Civil Engineers, Eds., Las Vegas, NV, April 1993.
Wittwer, C.S., Chen, G., Bodvarsson, G.S., Chornack, M., Flint, A., Flint, L.,
Kwicklis, E. and Spengler, R., Preliminary development of the LBL/USGS three-
dimensional site-scale model of Yucca Mountain, Nevada, Report LBL-37356,
Lawrence Berkeley Laboratory, Berkeley, CA, 1995.
Wu, Y.S., Bandurraga, T.M., Ahlers, C.F., Finsterle, S., Chen, G., Haukwa, C , and
Bodvarsson, G.S., "On Calibration of the TJZ Site-Scale Model of Yucca
Mountain," Submitted to the 7th International High-Level Nuclear Waste
Management Conference, 1996.
Wu, Y.S., Bandurraga, T.M., Ahlers, C.F., Finsterle, S., Chen, G., Haukwa, C ,
Bodvarsson, G.S., Kwicklis, E., Rousseau, J., and Flini. L., Calibration and
extension of the LBL/USGS three-dimensional site-scale model of Yucca
Mountain, Nevada, Lawrence Berkeley Laboratory, Berkeley, CA, in press, 1996.
Wu, Y.S., and Pruess, K., A multiple-porosity method for simulation of naturally
fractured petroleum reservoirs, SPE Reservoir Engineering, 3, pp. 327-336,1988.
Wu, Y.S., and Pruess, K., "On Verification, Use, and Treatment of Non-Newtonian
Behavior of the Numerical Simulator MULKOM-GWF," Proceedings of the
TOUGH Workshop, Report LBL-29710, Lawrence Berkeley Laboratory,
Berkeley, CA, pp. 73-88, September 13-14,1990.

Wu, Y.S., Pruess, K. and Chen, Z.X., Buckley-Leverett flow in composite media, SPE
Advanced Technology Series, vol. 1, No. 2, pp. 36-42. 1993.
Wu, Y.S., Pruess, K., Withcrspoon, P.A., Displacement of a Newtonian flaid by a bon-
Newlonian fluid in a porous medium, Transport in Porous Media, vol. 6, pp.
115-142, 1991.
Xiang, Y., Mlshra, S., and Dunlap, D., "Parametric Analysis of a Two-Dimensional
Groundwater Flow Model of the Unsaturated Zone at Yucca Mountain,"
Proceedings of the TOUGH Workshop '95, Report L3L-37200, Lawrence
Berkeley Laboratory, Berkeley, CA, March 20-22,1995.

Zimmerman, R.W., and Bodvarsson, G.S., An approximate solution for study of one-
dimensional absorption in unsaturated porous media. Water Resources Research,
vol. 25, no. 6, pp. 1422-1428, Report LBL-25629, Lawrence Berkeley
Laboratory, Berkeley, CA, 1989.
Zimmerman, R.W., and Bodvarsson, G.S., Semi-analytical solutions for flow problems in
unsaturated porous media, ASME Monograph on Multiphase Transport in Porous
Media, FED, vol. 82, HTD. vol. 27, pp. 23-28, 1989.

T O U G H 2 SOFTWARE QUALIFICATION - FEBRUARY 1 9 9 6 117

 Zimmerman, R.W., and Bodvarsson, G.S., "Combined Analytical/Numerical Approaches
to Solving Fluid Flow Problems in the Unsaturated Zone at Yucca Mountain,"
Proceeding of the International High-Level Radioactive Waste Management
Conference, American Nuclear Society, Las Vegas, NV, Report LBL- 28358,
Lawrence Berkeley Laboratory, Berkeley, CA, pp. 818-824, April 8-12,1990.
Zimmerman, R.W., and Bodvarsson, G.S., "Incorporation into TOUGH of an Analylical
Source/Sink Term for Fracture/Matrix Flow," Proceedings of the TOUGH
Workshop, Report LBL-29710, Lawrence Berkeley Laboratory, Berkeley, CA, pp.
67 72, September 13-14,1990.
Zimmerman, R.W., and Bodvarsson, G.S., "Combined Analytical/Numerical Approaches
to Solving Fluid Flow Problems in the Unsaturated Zone at Yucca Mountain."
Proceedings of the Second Annual International Conference on High Level
Radioactive Waste Management, American Nuclear Society and American Society
of Civil Engineers, Eds., Las Vegas, NV, 1991.
Zimmerman, R.W., and Bodvarsson, G.S., "Semi-Analytical Treatment of Fracture/Matrix
Flow in a Dual Porosity Simulator for Unsaturated Fractured Rock Masses,"
Proceedings of the Third Annual International Conference on Ifigh-Leve!
Radioactive Waste Management, American Nuclear Society and American Society
of Civil Engineers, Eds., Las Vegas, NV, April 1992.
Zimmerman, R.W., and Bodvarsson, G.S., "Approximating Ihe Imbibition and
Absorption Behavior of a Distribution of Matrix Blocks by an Equivalent Spherical
Block," Proceedings of the Fifth International High-Level Nuclear Waste
Management Conference, Las Vegas, NV, no 4, pp. 2030-2037, 1994.
Zimmerman, R.W., and Bodvarsson, G.S., "A Simple Procedure for Estimating the
Effective Hydraulic Conductivity of a Two-Dimensional Saturated or Partly-
Saturated Fracture Network. Proceedings of the Sixth Annual International
Conference on High Level Radioactive Waste M.w, cement, American Nuclear
Society and American Society of Civil Enginceti, ids., Las Vegas, NV, Mav
1995.
Zimmerman, R.W., and Bodvarsson, G.S., Effective block size for imbibition or
absorption in dual-porosity media. Geophysical Research Letters, vol. 22, no 11,
pp. 1461-1464, Report LBL-37074, Lawrence Berkeley Laboratory, Berkeley, Ca,
1995.

Zimmerman, R.W., and Bodvarsson, G.S., Effective transmissiviry of two dimensional

fracture networks, International Journal of Rock Mechanics, in press, 1996.
Zimmerman, R.W., Bodvarsson, G.S., Chen, G., and Hadgu, T.. A, Numerical dual-
porosily mode! with semi-analytical treatment of fracture/matrix flow, Water
Resources Research, vol. 29, pp. 2127-2i37, Report LBL-34148. Lawrence
Berkeley Laboratory, Berkeley, CA, 1993.
Zimmerman, R.W., Bodvarsson, G.S., Flint, A.L., and Film, L.E., "An Inverse
Procedure for Estimating the Unsaturated Hydraulic Conductivity of Volcanic
Tuff," Proceedings of the Fourth Annual International High Level Radioactive
Waste Management Conference, American Nuclear Society and American Society
of Nuclear Engineers, Eds., Las Vegas. NV, April 1993.

TOUGH2 SOFTWARE QUALIFICATION - FEDRUARY 1996 lie

 Zimmerman, R.W., Chen, G., Hadgu, T., and Bodvarsson, G.S., A numerical dual-
porosity model with semi-analytical treatment of fracture/matrix flow, Water
Resources Res., no. 29, pp. 2127-2137, Report LBL-34148, Lawrence Berkeley
Laboratory, Berkeley, CA, 1993.
Zimmennan, R.W., Chen, G., Hadgu, T., and Bodvarsson, G.S., Accuracy and
efficiency of a semi-analytical dual-porosity simulator for flow in unsaturated
fractured rock masses. Radioactive Waste Management and Environmental
Restoration, vol. 19, no. 1, pp. 193-208, LBL-34323, 1994.
Zimmennan, R.W., Hadgu, T., and Bodvarsson, G.S., "A New Lumped-Parameter
Approach to Simulating Flow Processes in Unsaturated Dual-Porosity Media,"
Proceedings of the TOUGH Workshop, Report LBL-37200, Lawrence Berkeley
Laboratory, Berkeley, CA, pp. 259-264, September 13-14,1990.
Zimmerman, R.W., Hadgu, and Bodvarsson, G.S., Coupling of a reservoir simulator and
a wellbore simulator for geothermal applications, Geothermal Resour.
Council, no. 17, pp. 499-505, Report LBL-34314, Lawrence Berkeley
Laboratory, Berkeley, CA, 1993.
Zimmerman, R.W., Hadgu, T., and Bodvarsson, G.S., Transient dual-porosity
simulations of unsaturated flow in fractured rocks. Report LBL-36S07, Lawrence
Berkeley Laboratory, Berkeley, CA, 1995.
Zimmerman, R.W., Kwicklis, E.M., and Bodvarsson, G.S., Absorption of water into
porous blocks of various shapes and sizes, Waster Resources Research, vol. 26,
pp. 2797-2806, Report LBL-27511, Lawrence Berkeley Laboratory, Berkeley,
CA, 1990.

T O U G H 2 SOFTWARE QUALIFICATION - FEBRUARY 1996 119

APPENDIX F. README FILE

TOUGH2 Version i.l 1 January 1994

with preconditioned conjugate gradient module T2CG1
(PC Version December 1994)
[email protected] RFM.ME READ.ME READ.ME
Update 12/94
This flyer contains brief instructions for installing and running TOUGH2 under
DOS or WINDOWS operating systems on an IBM PC or compatible. The TOUGH2-PC
code can be executed on 386 machines or better. With present dimensioning
(PARAMETER statements in main program) for up to 800 grid blocks and 2400
connections, memory requirements are approximately 4 Mb of RAM when one of the
conjugate gradient solvers is being used. When using the direct solver MA28 the same 4
MB of RAM will allow the execution of models with a maximum of S00 grid blocks and
1500 connections.
The TOUGH2 code requires 64-bit arithmetic, while the 386, 486 and 585
(Pentium) machines have 32-bit wordlength. Accordingly, in the TOUGH2-PC version all
floating point variables have been declared "real*83! (double precision), and all floating
point constants have been converted from Ew.d to Dw.d format. The only other changes
in TOUGH2-PC relative to the mainframe version are:
(1) Subroutines SECOND, FXDAYS, and ADDAYS have been appended to module t2cgi
for obtaining CPU times.
(2) FORMAT stalements with more than 127 "XI field spaces have been modified
according to 129X —> I20X.9X, to avoid a problem with the LAHEY compiler,
Version 5.XX.
No modifications in function names are needed, because TOUGH2 uses generic
FORTRAN77 function names throughout
TOUGK2-PC is distributed on a 3.513 high-density diskette, formatted under the
MS-DOS operating system for IBM PCs or compatibles. THOUGH2PC is set to be
compiled and linked using a LAHEY Fortran compiler version 5.00 or better, other
compilers that can access extended memory (memory beyond i Mb) may be used instead
with minoi changes to the CPU timing subroutines. A quick fix to the CPU timing routine
is explained in detail in the 'Additional notes' section in this flyer.
The distribution diskette includes the following 17 files:
(1) read.me - the file you're reading
(2) «2cgl.for - contains the main program of TOUGK2, with PARAMETER statements
for flexible dimensioning of all major arrays, and initialization of disk files; also
includes a revised version of subroutine LINEQ dial calls a suite of preconditioned
conjugate gradient routines; l2cg].for replaces this program module t2rn.for of the
pievious version of TOUGK2 (Version 1.0, April 1991);

TOUGH2 SOFTWARE QUALIFICATION - FEBRUARY 1996 120

 (3) rneshrn.for - module with internal mesh generation capabilities
(4) eosi.for - equation of-st@t6@;fc @,;water-or,@rwo-waters
(5) eos2.for - equation of state for water/C02 mixtures
(6) eos3.for - equationof state for water/air (similar to TOUGH)
(7) eos4.for - equation of state for water/air with vapor pressure lowering capability
(8) eos5.for - equation of state for water/hydrogen; otherwise similar to eosS.for
(9) t2f.for - the core module of TOUGH2; il reads input data, initializes arrays and
parameters, sets up the Newton-Raphson iteration, and performs time stepping; it also
contains the water property routines (steam table equations) and the relative
permeability and capillary pressure function which are Used in the equation-of-state
modules
(10) ma28.for linear equation soiver from the Harwell subroutine librar written by l.S.
Duff. (MA28 is subject to proprietary restrictions, and use outside of TOUGHS
requires proper license. Contact: The Harwell Subroutine Library Liaison officer, Mr.
S. Marlow, Building 8.9, Harwell Laboratory. Didcot, Oxon 0 X 8 QRA, United
Kingdom.)
(11) makefile file, that contains the required commands and instructions to compile, link and
create an executable file using the LAHEY Fortran compiler Ver. S.XX. Can be
modified to customize installation of TOUGH2, or to compile and link the code with
other equation of state (EOS) modules,
(!2) instatl.bat file to install and create the required directories to start using the program
according to the suggested structure in the "makefile" file
(13) delef.bat utility file Jo erase T0UGH2 generated output files between runs. This file
should be used with caution and customized to the specific needs of a run.
In addition, there are four input files for sample problems.
(14) sami - input file for sample problem 1 (code demonstration and comparison wish
TOUGH)
(3 5) fhp - input file for sample problem 2 (beat pipe in cylindrical geometry)
(16) rvf - input file for sample problem 3 (heat sweep in a vertical fracture)
(17) rfp - input file for sample problem 4 (five-spot geothenrta! production/injection)
Files (2) - (10) contain the TOUGH2 source code. The only change in comparison
to version 1.0, April 1991, is the replacement of file l2m.for with t2cgi.for, the
preconditioned conjugate gradient module. When compiling and linking TOUGH2,
l2cgl.for should come first, then meshm.for, then the desired EOS-module, then t2f.for,
and finally ma28.for. Note that only one of the EOS modules must be linked at a lime.
On an IBM PC, or compatible 386 machine or better with a DOS-based operating
system, using LAREY compiler Ver. 5.00 or later, small simulations may be tun from

10UCH2 SOJTW/UJE QUAUfJCATION - FtSRUARY 1 9 9 6 1 at

 DOS. The number of elements in a simulation model will depend on the amount of
extended memory (XMS) available in the machine. For big simulation jobs using the
LAHEY compiler Ver. S.XX, it is necessary that all extended memory (memory beyond 1
Mb) must be made available to the LAHEY compiler prior to compilation and linking. To
perform this operation all Terminate and Stay Resident Programs (TSR) must be removed.
it is recommended to use stripped-down versions of the CONFIG.SYS and
AUTOEXECBAT files specific for running TOUGH2. MS-DOS Version 6.0 or later
allow the use of multiple CONFIG.SYS and AUTOEXECBAT files, see the MS-DOS and
FORTRAN compiler manuals for details. Running TOUGH2 under WINDOWS using the
LAHEY Fortran compiler, has the advantage of accessing virtual memory (temporary space
on a hard drive). This allows large simulation jobs to run on machines with limited amount
of RAM. The trade-off is a slower execution due to data swapping between physical RAM
and virtual memory. On machines with 16 MB of RAM and more, it is possible to run
TOUGH2 under WINDOWS in the background and use other WINDOWS application in
the foreground.

To install and run TOUGH2 using the LAHEY Fortran compiler Ver. 5.XX proceed as
follows. (This procedure assumes that the compiler is properly installed)
(1) Insert the distribution diskette in drive B:, type INSTALL and press <enter>. The
install file will create a TOUGH2 directory, and copy TOUGH2-PC into this
directory. It will also create a DEVELOP subdirectory into which it will copy the
input files for the sample problems.
(2) Customize the "makefile" file as desired.
(3) Within the TOUGH2 directory, type the following command:
make
The code will men be compiled and linked and, if the "makefile" file provided on
the distribution diskette was not modified, an executable Ut2eosl.exell incorporating the
EOS 1fluidproperty moduie will be written to the TOUGH2\DEVELOP directory.
To check on proper code installation, within the TOUGH2\DEVELOP directory
execute sample problem No.4, five-spot geoicermal production and injection, with the
following command.
t2eosl <rfp >rfp.out
The first part of the command uploads the Ut2eosl.exeU file and runs it using the
'irfpll file as the input deck. The output is redirected to the 'Irfp.oul" file. The il>rfp.out"
part of the command is optional; if it is nor present, all output will be displayed on the
screen.
Additional notes.
An effort was made to have the TOUGH2 source code comply with the ANSI
X3.9-1978 (FORTRAN 77) standard, and on most machine and compiler combinations,
the code should compile and run without modifications. TOUGH2 makes several calls to
an external routine "SECOND" for obtaining elapsed CPU times, such as
CALL SECONDfTZERO).

TOt)nK2 SOFTWARE QUALIFICATION - FEBRUARY 1996 122

 The PC version of TOUGH2 includes the subroutines SECOND, ADDAYS and
ELDAYS that worktogetherto provide the elapsed CPU times, and are compatible with die
LAHEY Fortran compiler. Different compilers have different conventions for obtaining
CPU time, and appropriate adjustments must be made. Is some cases no facility for
assigning CPU time may be available; the user may then relinquish the CPU timing by
replacing the subroutine SECOND at the end of file t2m.for with a dummy routine, as
follows.

SUBROUTINE SECOND(T)
REAL*8T
CHARACTER ELTIME*U, FECHA*8
COMON /ELTJMZELTIME, EECRA
T=O.DO
EUmm=W:0Q:O0.00'
FECHA ='01/01/01'
RETURN
END
This dummy subroutine will avoid an "unsatisfied external,, error; ail execution
times and dates will be reported as 0.0 and 01, respectively, in the printout.
The file t2cgl.for includes the following routines:
(i) revised versions of the program units normally supplied in t2m.for,
(ii) a version of subroutine LINEQ that is appropriately modified for
interfacing with the conjugate gradient solvers, and
(iii) a set of preconditioned conjugate gradient routines.
The presence of a modified version of subroutine LINEQ in l2cgI.for will create a
situation of "duplicate names" during the linking process, as a subroutine LINEQ is also
present in standard TOUGH2 (file t2f.for). On most computers the linker wiil simply use
the first program unit with a given name, and will ignore subsequent program units with
the same name. On some computers the presence of duplicate names during linking will
create a fatal error. The simplest way to avoid this is to change the names of the unwanted
program units. Specifically, the name conflict can be avoided by renaming LINEQ in
t2f.for to LINEX, say, prior to compilation and linking. In the PC version of TOUGH2
LINEQ was removed from t2f.for.

The only user-definable input parameter associated with the conjugate gradient
module is MOP(21) in block PARAM, which selects different solvers as follows.
MOP(21) = 0: default; is set internally to MOP(21) = 3.
1:
2: direct solution with MA28 (as in TOUGH2, Version 1.0). routine
DSLUBC: biconjugate gradient solver with incomplete LU-
factorization.
3: routine DSLUCS: Lanczos-typc biconjugatc gradient solver with
incompiem LU-factorization.
4: routine DSLUGM: generalized minimum residual solver with

TOUGH2 SOFTWARE QUALIFICATION - FeeHOARv 1996 T 23

 incomplete LU-@tactori-z-a-tion.@:.-
Apart from the choice of the linear equation solver, there are no userdcfinable
parameters, and tbe conjugate gradient package would be ran with unmodified TOUGH2
input files.
The computational work for iterative solvers increases much less with problem size
and matrix bandwidth than is the case for direct solvers. Therefore, iterative solution is the
method of choice for virtually all 3-D problems, and for I-D and 2-D problems with more
than a few thousand simultaneous equations. Present dimensioning of the PC version of
t2cgi.for is for up to 800 grid blocks and up to 3 ecluations per grid block. Maximum
problem size can be easily changed in PARAMETER statements in the main program. If
the number of grid blocks exceeds 4 digits (9999), FORMAT statement « IS05 in
subroutine RPILE in t2f.for must be changed from 2014 to 16IS.
Iterative solvers such as conjugate gradients do not, generally speaking, have the
same robustness as direct solution techniques, and the performance of the different
algorithms is problem-dependent. The default solver option invokes the LancZOB-type
biconjugate gradient solver which often is die most computationally efficient. Our
simulations of a broad variety of flow problems indicate that each of the 3 conjugate
gradient solvers included in the T2CG1 package is optimal for certain cases, while failing
for others (Moridis et al„ 1994). Therefore, for any given flow problem, users are
encouraged to experiment and try the different solvers offered.
As mentioned before, T2CGI requires no user-definable parameters other than the
choice of linear equation solver. Occasionally users may wish to modify the default
settings for iteration and convergence parameters. These parameters are defined upon the
first call (ICAJLL = 1) in subroutine LINEQ in module T2CGI, as illustrated in the
following code fragment.
IF(ICALL.EQ.l) THEN
WRrTE(11,899)
899 F0RMAT(6X,1UNEQ 0.91 CG 31 JANUARY 19941.6X,
XT1NTERFACE FOR LINEAR EQUATION SOLVERS'/
X47X.ICAN CALL MA28 OR A PACKAGE OF CONTUGATE GRADIENT',
XI SOLVERS')
c
MATSLV=M0P(21)
IF(MATSLV.EQ.O.OR.MATSLV.GT.4)MATSLV=3
NMAXrr=MAX(20,NEL'NEQ/10)
ICLOSR=2
IF(MATSLV.EQ.4) 1CL0SR=0
CLOSUR=I.D-6
1SYM=0
1L7NTT=0
NVECTR=30
SRED=l.D-25
ENDIF
For very difficult problems, it may be necessary to tighten the convergence criterion
CLOSUR beyond the default of I.D-6, and to increase the maximum number of iterations
beyond the value specified in NMAXrr (NEL*NEQ is the order of the matrix of the linear
equation system). Information on convergence behavior for each Newtonian iteration is

TOUGH2 SOFTWARE QIMUFICATION - FEBRUARY 1996 124

 written onto a disk file ILINEQI.
Files (14) - (17) contain input files for the sample problems presented in die
TOUGH2 report. Users should run several of the sample p@o@iems@ to check on
proper code installation. Due to machine-dep<3>ndent roundoff, TOUGH2 may produce
slightly different results on different computers. For same-size rime steps, all primary
variables (pressure P, temperature T, saturation 9, etc.) and their changes (DX1, DX2....)
should agree to typically four digits or better. However, on different computers the
iteration sequence for a time step may be slightly different, and occasionally a different
number of iterations may be required for convergence. If automatic time stepping is used,
a different number of iterations for convergence may subsequently cause different-size time
steps to be taken; naturally this will then produce somewhat larger discrepancies in results
because of different time truncation errors.

Of all the numbers processed by TOUGH2, the most sensitive are the residuals,
i.e., the differences between left hand sides (accumulation terms) and right hand sides
(flow terms) of the governing equations. During the Newton/Raphson iteration process
these residuals are reduced to smaller and smaller values, until they drop below specified
convergence tolerances. As convergence is approached, the residuals are subject to
increasingly severe numerical cancellation, arising Horn subtracting two numbers with
diminishing difference. Maximum residuals are (optionally printed during the iteration
process as "MAX. RES.", and are also printed in the header of a full time step printout (as
"MAX. RES." or "RERMH). These numbers can serve as a convenient check when
evaluating reproducibility of code applications. Small numerical differences due to
roundoff etc will first show up in different values for "MAX. RES.", 2sas before giving
any visible changes in primary variables or their increments.

TOUGH2 is documented in:

K. Pruess, TOUGH2 - A General-Purpose Numerical Simulator for
Multiphase Fluid and Heat Flow, Lawrence Berkeley Laboratory Report
LBL- 29400, May 3991.
The TOUGH2 report is not a self-contained free standing document. For instructions on
preparing input data, users also need the following report:
K. Proess, TOUGH User's Guide, Lawrence Berkeley Laboratory Report
LBL - 20700, June 1987 (also available as Nuclear Regulatory
Commission Report NUREG/CR-4645).
Information on the conjugate gradient algorithms and applications to large 2-D and 3-D
flow problems are presented in:
Moridis, G., K. Pruess, B. Anfunez, and C. Oldenburg. T2CGI, A Package of
Preconditioned Conjugate Gradient Solvers for the TOUGH2 Family of Codes,
Lawrence Berkeley Laboratory Report LBL-35518, Lawrence Berkeley
Laboratory, Berkeley, CA, 1994.

Antunez, A., G. Moridis and K. Pruess. Large-Scale Geotliermal Reservoir

Simulation on PCs, Lawrence Berkeley Laboratory Report LEL-35192, presented
at 19th Workshop on Geolhermal Reservoir Engineering,
Stanford University, Stanford, CA, January 1994.

TOUCH2 SOFTWAM QUALIFICATION - FHRUARV 1 9 9 6 12S

The DSLUBC and DSLUCS routines used in the conjugate gradient package were written
by Anne Greenbaum (Courant Institute) and Marie K. Seager (Lawrence Livermore
National Laboratory). DSLUGM was written by Peter Brown, Alan Hindmarsh, and Mark
K. Seager (all of LLNL). These routines were adapted and improved for incorporation into
the TOUGH2 code by George Moridis (Lawrence Berkeley Laboratory). Interfacing with
TOUGH2 was done by George Moridis and Karsten Pruess. Emilio Antunez (LBL)
performed the adaptation to PC.
Distribution of the TOUGH2 code is handled by:
Energy Science and Technology Software Center
P.O. Box 1020
Oak Ridge, Tennessee 37831
U.S.A.
phone (615) 576-2606
fax (615)576-2865
email: estsc @ adonis.osci.gov
The address of the code developer is:
Karsten Pruess
Mail Stop 50E
Lawrence Berkeley Laboratory
Berkeley, CA 94720
U.S.A.
fax: (510)486-5686
email: k_pruess @ lbl.gov

TOUGH2 SOFTWAKB QUALIFICATION - FEBRUARY 1996 126

APPENDIX G. SAMPLE PROBLEMS

Pruess (3991) provides sample problems that are useful for checking the proper
installation of TOUGH2 and cross-referencing it to TOUGH and for handling common
applications of the code. The problems are summarized briefly hero; however, the user is
urged to consult Pruess (1991) for printouts of the input files, schematics of problem set­
up, and data plots.

G.l Sample Problem 1

Problem 1 is identical to sample problem 1 from the TOUGH User's Guide
(Pruess, 1987) and is a good check on proper installation of TOUGH2. It consists of a
number of one- and two-element subproblems which are independent of one another, but
which are run simultaneously and therefore must go through the same sequence of tiirte
steps. The lubproblems perform flow and/or injection and withdrawal of water, air, and
heat, with highly non-linear phase and component (disappearances that require subtle
numerical procedures. The input file for the EOS3 fluid properties module is almost
completely identical to that of TOUGH sample problem 1, except that a few MOP-
pararneters are set differently because of different defaults in TOUGH2. Table 17 lists the
elements and features of the subproblems in problem 1. As shown in Pruess (1991. Figure
9), the results are virtually identical to those generated by TOUGH. Minor differences
occur in the maximum residuals during the iteration process. After time step 2, TOUGH2
takes different time steps because of different default settings. Sample input and output are
shown in Figures 39 and 40 respectively.

Table 17. Summary of Sample Problem #] Features

Connection or Process Features

Element
(Ft, F2) How from single phase liquid (no air) phase transitions from liquid to two-
into single phase gas (no vapor) phase, gas to two-phase; appearance of
water component
(F3, F4) flow form hot two-phase conditions into vaporization and condensation; phase
coid two-phase conditions transitions from two-phase to liquid and

(F5, F6) flow of air into single phase liquid phase transition from liquid to two-
phase; appearance of air component
F7 injection of air into cold liquid phase transition from liquid to two-
phase; appearance of air component
l-'S production of fluid from single phase phase transition from liquid to two-
phase; vaporization
liquid
phase transition from two-phase to gas;
F9 iniection of heat into two-phase fluid
vaporisation
F10 withdrawal of heat from single phase phase transition from gas to two-phase;
vapor condensation
iho 1-sho !2 fluid production and injection demonstration of generation opdons

TOUCH2 Sorrwt.it QUAIIHCATION • rmRUARY 199&

 6
«5AU1- CWS 0EW«t1IAT2W( PHASE TRAKSITIOttS, COWKEWT (OW-lJ^KUfff, ?
THIS IH?UT « u i s uiCHiLt D I « £ « M T F W « THAT «** s * « n . e riOBtJu l
l
or TOUCH, w e s t ceuuoire AME WSCTTCD 7D NOTE W E o ^ « « ' * « i - • S " S K 5
ARE «RJLQM» »t A CGUAM COUKTM 10 MCIUTATt rWCtt ALICttCKT OP M W .
A FEWW-B'TIBKS AUK SCT MFFeWIVr [ME StCTIOH * .1 IN T0UW3 CUI0E>^
THAWS "nit, ** rwTti-U*"""*"" * J-JJ JH*.

i.ce
-B —-a
UM»MlMt4»lN7iI

_™l.„,.._.«—-

-1. •—•-?—
1 3
£3 --*.»-
F 1 9 iTn A H ; — - 1 - «•.—_4—
SW 1 lEHWE
SK011 *1 SSKCKS «.
t.e«

f IP a l G. c. t.
f W i fi. s. 1.
r tf 4 i K. s. t.

F 1
l . » ao. 1.
r 2
I.£6 170. 0.
F 3
l.CS r?.t
r * .*«
.•(9 31*.
t».W
F C e.
t.eo 1»«.
P C
1».G* IKS, i.
P T 0.
i.ci SB.
F 0
F »
I.P M),
«.
F 10
l.« .*» ».
4#.E6 2ia. c.
SttSll
E3.E5 ;i*.
SHOW
4s.es i»t.
COCK— . j „ —m-.—9'-"-
f 7 * 1W AM i.t1 &.*•:£«
rF inn. »U$5
F
«Nor
lera.
MEAT
HEAT
.!.«-*
-f:g
sua if HASS1 -I. s.e«
sw w 1 FINV -I.
-oo v 3 «.: * U,'
W5S
2.EZ
4.C3
e.j -f.3 •*.a -1.1
SX3 1 * 4 3 1 tuasi
z.u 4.LJ
t.i !»' -e.a •1.1
WO wr i «av l.t-13 i.a
SH9U* It * *ATE1 I.E3
i.i #.3
5M0I1 M L « « '
.(A J S 't 1.4E8
oav l.=-w
3.DE9
i.e:
EK31»EL«* BEL.V a.t-i* 1.6* i.ba
,„_«.. 1... . . . — e
VUttV
, „-.„,„- —..™e™,,

Figure 3P. Inputftlefrr sample problem 1 - fm/g demonstration.

TO'JCH2 SOFTWARE QUALIFICATION - FEBRUARY 1996 128

 iMHMtttUDWiai• I t. «<•»« OCX [ft. RW,II(- HI
iKwnumnrua
iii MRWPRMMI nam (.imMHtwi
uuiirucr
in tiej^wot tiBit utrBn«tti m *jum<nttt m HHUMIHI urn am ».tnn«*»u

...ItCURM... « ( l| It —MtTIJatlHWMU Mi, K • MnoiH*l U l U l l i t FMrM 1

nmmmn>rtwnM|Eqm«um«f » am ua*«.mutr4i I H M M N H «»*.UMtk««

tmiuftntnuiuMBtisxKittcanv r> am iia.i-HtufcH) »-i.SHer>w H>LIBMHM
„,lr»t»„. «I I, II —BUUXUMMMIl IU.«l>UtIMMlUUWMtUilW 1
,..nau««. nt 1,11- . - - . . -
f II i> <ni«t,mwiip
nwu>aititiMMMu Hwn«iiii».onoi UU-JMUMS
Bnmmktnai i, ti-nntnoi s t p c a *.tta<rei*a
HIHIWWUIWHtWHIIIMIIWIWIMIIiaW^
rca.HK mc u s las at itti i» K* w.m. to >t *cm
t.imwm t i 4 i *,iM»rfl «.iwwn «.Mtw*a MUttr*! n MMM**)

tan. i w i i * a. uut iua mi Kit- a a.

mi na-o an «u aMHti tivtMii
r i !VT!StwiJHN^«f.itmi^it.itirir44*.i)su>4riliffh-HtnneH«,
l Mia*MKUMI
1 » 1 MWMxit fclHIMI I, fclWInlH. L UMfeMl. L LtfS*»tl
r i > m w m fcwnmi u i w r c i.Tt!wi t.wrwrfi rmtani Mimt'm. *•«*•** Mawa-ai
i t *«.muwUUNiiMa(.«n)Mi«.nnHai.iuik-tti.o«rfii.fB(Mii. «na«<n IJWI«*I
f s i Migmw t.iwBn«i «.powa a,ttwt i . i w w i twaxa MCBHW t. «.HWWKI MBM.*!
i k ktMUiiaMWHimt.min'u«, «.tMMtii. (.IKSMI. lawirtJL
r T > «JrWMW i.wnii *.twi««M MWiM«M.wtt««n.tiuw«i ajEnwtt. aJMtru MMM1
f * t*,mn*aijms*at. uawiit. •. tWMtt. «. ira*va
i * f «.IUOM( ».M**W<O *.imtrm *.iw*ra IJIMMM i.m»«4i kunteM *, *.•&•<« t,iwj*u
f U » tJT*W« LBOMM 1.1—I H t 1. ». tWMti. UftFrHl K.
IMI ll «.«lffcnl UaHhMU.Oatr*! J H H t m u i l H W I ».1MgrM W W - j a M * * MfUhHl MUM**]
MI nfc«»aH;i.wtiniJ»wfciii u w t H ' W U a a i W M W ^ i T r o ^ ^
Ml U«48»MI LttMkXlIJWItl MWtrM|.LBh-WLUMrM U U t o N - ^ W I H I U S O M 1 MU1**«
MS O4.<2JMMU«lBMltJMIlMi U1VMI*.13Sb«aU1MH4 LJWV4J•vOtfkHtUSTIf« fcHMptl
H I n «,«mierr a.wtf tin *.***** M W I * M I N I M * » M I U B V M . . H M * * u n n — I uaaiv«>
8* J 11 t.tOHi« U l t l t « « IJHBMI UKTlKM L I B * * * LBSIHI UOCf*J -.M* B *JS5*fC ••MCli^l
HI 1tK«IWtfl»I.HVMHlUNaww«.11fMr^«.Wm^*.KlC^MM«AM-U)IM»«U}a»a>>IBmfWl
HI imstwo uaaMt a a m Lmamt.iimi-»matrw waihti •.«**« maafct* a-mumi
pan »«>W«OHKl.tX<MH>^g3W«««««APltriW^t»Mr^i;Mri.r^AHW^/-UWlW^AIJU*^«.mC*^J
xu s»,aci*HJtrntiHJi. t.i*H*ttt. *. UMM»«. «. (.nirm«i
•ea B ajinj*« a,tti»Mi*. «,im«Ml t. i. «.wUJr*»«. a. (.TMi^al

Figur* 40- Selected output for sample problem J.

G.2 Sample Problem 2

Sample Problem 2 simulates thermal interference along a preferential flow path by
modeling nomsctberraal injection into and production from a single vertical fracture. The
fracture is bounded by semi-infinite half-spaces of impermeable rock, which provide a
conductive heat supply. A special feature of the problem is that the semi-anaiyticat method
of Vinsome and Westerveld (1980) is used to describe heat conduction in the confining
layers, reducing the dimensionality of the problem from 3-D to 2-D. Water remains in
single-phase liquid conditions throughout so that capillary pressures and relative
permeability are not needed.
The problem uses the EOSI fluid property module and is run in tiuee separate
segments. The first run produces mesh generation only. Steps required for mesh
generation using MESHMAKER are explained in Pruess (1991). The second step
calculates a hydrostatic pressure equilibrium in the fracture under isothermal conditions.
This calculation uses the MESH file obtained in step one, and requires the following
modifications in the input file. The time step counter MCYC is reduced to 4, the generation
items are removed, convergence tolerance is decreased to achieve a tighter control on
gravity equilibrium, the 'ENDCY* statement is removed, and the semi-analytical heat
exchange calculation is disengaged by setting MOP(15)=0. The third step is the

TOUGH2 SOFTWARr QUALIFICATION - FEBRUARY 1 9 9 6 129

 production/injection run using an input file along with the MESH file as used in the gravity
equilibration. A plot of the transient temperature changes at the producing element is given
in Figure 42.
Tima (Days)
1 10 ioo
300 pp

JSO —

4- 250 —

5 240 -

220 -

6 r
io io IO 8

Time (Seconds)

Figure 41. Producedfluidtemperature versus time for vertical fracture problem

in sample problem 2.

G.3 Sample Problem 3

Sample Problem 3 is the five-spot problem of gcothermai production/injection. It
considers a large well field with wells arranged in a "five-spot" configuration. Because of
symmetry, only 1/8 of the problem needs to be modeled". The computational grid was
generated by means of a separate preprocessor program that has not yet been integrated into
the TOUGH2 package. The problem specifications correspond to conditions that may
typically be found in deeper zones of hot and fairly tight fractured two-phase reservoirs
(Prucss and Narasimhan, 1985). The input file for use with the EOS1 fluid property
module models the system as a fractured medium with embedded impermeable matrix
blocks in the shape of cubes. The matrix blocks arc assigned a small porosity so that they
will contain a small amount of water. This prevents the water mass balance equation from
degenerating to 0. The MESHMAKER module is used to perform MINC-partitioning of
the primary grid. Alternatively, the MINC process can bo disabled and the problem run as
an effective porous medium. Figure 38 shows temperature profiles along a line of injection
to production well for a porous medium and two different fracture spacings.

By small adjustments to input files in Problems 2 and 3, a number of problem

variations can be run, e.g. injection into vapor-dominated systems for Problem 2, and neat
exchange with confining beds for Problem 3.

T O U G H 2 SOFTWARE QUALIFICATION - FEBRUARY 1996 130

 J —i • t 1" 1
' c * * a
e o •
2S0 0 o
o
g 9
o
1 O
2 3 0
I 0
a
1 0 o Porous meaum
< KNCD-SOm
150 "injection
a 0 KNCD-2S0m
we3
1
e ftoduc£c»—»
'/-. j * i
100 200 300 <Q SCO £00 700
Distance (m)

Figure 42. Temperature profiles for problem 3 along a line of injection to production well after 36.5
years. MINC results for a SO m fracture spacing are virtually identical to porous medium
results while MINC results for a 250 m fracture spacing show lower temperatures.

TOUGH2 SOFTWARE QUALIFICATION - FEBRUARY 1996 131