EFFECT OF D E L A Y E D B A C K F I L L O N OPEN STOPE MINING M E T H O D S

by

CRISTIAN ANDRES CACERES DOERNER

B.Sc, Universidad de Chile, 1997

A THESIS SUBMITTED IN PARTIAL FULFILMENT OF

THE REQUIREMENTS FOR THE DEGREE OF

MASTER OF APPLIED SCIENCE

in

THE FACULTY OF GRADUATE STUDIES

MINING ENGINEERING

THE UNIVERSITY OF BRITISH COLUMBIA

March 2005

© Cristian Andres Caceres Doerner, 2005

 Abstract ii

A B S T R A C T

Open stoping is comprised of large rectangular voids separated by intervening pillars so

as to minimize the size of the exposed surface and thereby reducing the potential for wall
slough and in turn external dilution.

These pillars provide support to the exposed wall; however, they result in ore loss and
increased costs such as having to establish slots for blasting. Longhole mining methods such
as Avoca or longhole retreat as practiced at the Musselwhite mine of Placer Dome employs
100% extraction with the use of fill walls to provide support to the adjacent stope.
Transverse open stoping also practiced at Musselwhite employs cemented rock fill
adjacent to a mined stope. The question is how to account for the overall stope wall stability
when the adjacent support is backfill. The backfill does not provide the same support as that
of a rock pillar, however, due to the increased use of fill abutments one has to develop a
methodology that accounts for this reduced overall support element as it does reduce the
overall stope surface exposed.
It has been shown in this thesis that the backfill wall does not provide the same overall
stability to an individual stope as would avail itself if the stope had rock abutments.
This is the focus of this study in order to establish design criteria to enable one to
employ existing methods for stope design such as the Stability Graph by augmenting input
parameters that have been calibrated through field measurements, analytical assessments,
numerical modeling and laboratory testing to evaluate the effect of mining adjacent to a
backfilled stope.

Sill pillars are employed at the Musselwhite mine to allow for multiple mining horizons
with unconsolidated backfill placed immediately above the intervening sill. These sills can be
comprised of unmined ore when the economics are such as to negate their mining or
alternatively they are replaced by a constructed sill mat to allow for mining underneath by
non-entry methods and thereby containing the overlying backfill.
Numerical studies were conducted to investigate modeled results of mining under a
cemented rockfill sill mat and to develop criteria for sill mat design. The results obtained
from this analysis can be extrapolated to other operations that utilize backfill as part of the
Abstract iii

mining sequence. Design curves were developed for the stability of sill mats for various
stope configurations and cemented rockfill strength properties

This, coupled with defining the effect of mining adjacent to backfill, forms the focus of

this thesis.
Table of Contents

T A B L E OF CONTENTS

ABSTRACT "
TABLE OF CONTENTS IV
LIST OF TABLES VII
TABLE OF FIGURES VIII
ACKNOWLEDGMENTS XI
1 INTRODUCTION 1
1.3 BACKGROUND 2
1.4 THESIS OVERVIEW 3

1.5 CONTRIBUTIONS MADE BY THESIS 3

2 THE MUSSELWHITE MINE 5

2.1 INTRODUCTION 5
2.2 GEOLOGY 6
2.2.1 Regional Geology 6
2.2.2 Mine Geology 7
2.3 UNDERGROUND MINING METHODS 9
2.3.1 Avoca 9
2.3.2 Transverse Retreat Open Stoping 10
2.3.3 Drilling and Blasting Methods 10
2.3.4 Ore Handling 11
2.3.5 Ground Support //
2.3.5.1 Development Support 12
2.3.5.2 Stope support - Stope backs and walls 12
2.3.6 Backfill 13
2.3.6.1 Mining Underneath Backfill - CRF Sill Mats 14
2.3.6.2 Mining Adjacent to Backfill 14
2.4 EQUIPMENT 16

3 REVIEW OF DESIGN METHODOLOGIES 17

3.1 BACKFILL 17
3.1.1 Purpose of Backfill. /7
3.1.2 Types of Backfill 19
3.1.2.1 Rockfill 20
3.1.2.2 Uncemented Rockfill - URF 20
3.1.2.3 Cemented Rockfill with Portland cement Slurry - CRF 21
3.1.3 Backfill Properties 21
3.1.4 Parameters Affecting Cemented Rockfill Strength 26
3.2 VERTICAL LOAD ON CEMENTED ROCKFILL SILL MA T PILLARS 28
3.2.1 Sill Mat Pillar Definition 28
3.2.2 Background - Loads Acting by Unconsolidated Material 29
3.2.3 Maximum Horizontal Pressure on the Walls of the Stope 31
3.2.4 Coefficient of Lateral Earth Pressure 33
3.2.4.1 Categories of Lateral Earth Pressure 33
3.2.4.2 Calculating Lateral Earth Pressure Coefficients 35
3.2.5 Maximum Vertical Load Exerted by Backfill on the Floor of the Stope 38
3.2.5.1 Terzaghi's Formulation 39
3.2.5.2 Blight's Formulation for Inclined Stopes 39
3.2.6 Vertical Load of Unconsolidated Material as a Function of the Material's Height 40
3.2.6.1 The Janssen Method - Silo Theory 40
3.2.6.2 The Reimbert Method - Silo Theory 41
3.3 STABILITY OF CEMENTED ROCKFILL SILL MATS-ANALYTICAL SOLUTION 42
Table of Contents v

3.3.1 Caving Failure 42

3.3.2 Flexural Failure 43
3.3.3 Sliding Failure : 44
3.3.4 Rotational Failure 44
3.3.5 Strength Properties 44
3.4 EMPIRICAL STABILITY DESIGN METHODS • 45
3.4.1 The Stability Graph Method 1
46
3.4.2 The Stability Graph Method - Radius Factor* 48
3.5 SOFTWARE AND DEVICE UTILIZED 49
3.5.1 Flac - Two Dimensional Explicit Finite Difference Method*'
2D
49
3.5.2 Map3D - Three Dimensional Boundary Element Method* 2
50
3.5.3 NeuroShell Predictor - Neural Networks fl
51
3.5.4 Unwedge - Underground Wedge Stability Analysis* 4
52
3.5.5 Dips - Graphical and Statistical Analysis of Orientation Data *4
53
3.5.6 Methods Ground Control Assessment - Cavity Monitoring System Surveys* 5
53

4 DATABASE 55
4.1 INTACT STRENGTH 55
4.2 FABRIC ANALYSIS 56
4.3 ROCK MASS ANALYSIS 58
4.4 STRESS 60
4.5 NUMERICAL MODELING - CONSTITUTIVE MODELS AND MA TERJAL PROPERTIES 60
4.6 EMPIRICAL STOPE DESIGN - MUSSEL WHITE'S STABILITY DA TABASE 61
4.7 STATISTICAL ANALYSIS OF THE MUSSELWHITE DATABASE 63

5 DESIGN GUIDELINES FOR CEMENTED ROCKFILL SILL MATS 66

5.1 INTRODUCTION 66
5.2 LOAD OF BACKFILL 66
5.2.1 Vertical Stress Comparison using different Coefficients of Lateral Earth Pressure 70
5.3 VERTICAL STRESS USING NUMERICAL MODELING 71
5.3.1 Introduction 71
5.3.2 Stress Distribution as a Function of Stope Dip Angle 74
5.3.3 Vertical Stress as a Function of Stope Span 75
5.3.4 Vertical Stress as a Function of Rockfdl 's Density 76
5.3.5 Vertical Stress as a Function of Rockfdl's Friction Angle 76
5.3.6 Comparison of Analytical and Numerical Methods to Determine Vertical Stress 77
5.3.7 Proposed Analytical Equation for Inclined Stopes 79
5.4 STABILITY OF CEMENTED ROCKFILL SILL MATS 80
5.4.1 Introduction 80
5.4.2 Sill Mat Strength - Friction and Cohesion (<j> ) c 81
5.4.3 Backfill Load. 82
5.4.4 Proposed Rotational Analytical Formulation 82
5.5 FAILURE MODES-NUMERICAL MODELING SOLUTION 85
5.5.1 Model Construction 85
5.5.2 Constitutive Equations - Strain Softening 86
5.5.3 Model Execution 87
5.5.4 Caving Failure 88
5.5.5 Flexural Failure 88
5.5.6 Sliding Failure 89
5.5.7 Rotational - Crushing Failure 90
5.5.8 Rotational — Breaking Failure 90
5.5.9 Sill Mat Design Curves 91
6 EFFECT OF DELAYED BACKFILL ON OPEN STOPING 97
6.1 INTRODUCTION 97
6.2 BACKFILL AS LOCAL SUPPORT 97
Table of Contents vi

6.3 USE OF THE STABILITY GRAPH METHOD 99

6.4 QUANTIFYING THE EFFECT OF BACKFILL ON LONG HOLE OPEN STOPING 100
6.5 UP DA TING MUSSEL WHITE 'S STABILITY GRAPH 100
6.6 NEURAL NETWORK TRAINING AND RESULTS 104
6.6.1 Relative Importance of Inputs 104
6.6.2 Neural Network Predictions 107
6.7 EFFECT OF BACKFILL IN STABILITY OF OPEN STOPES - NUMERICAL MODELING ANALYSIS 110
6.7.1 Radius Factor Behavior with Stope Strike Length 112
6.7.2 Horizontal Displacement Behavior considered with Stope Strike Length /15
6.7.3 Design Curves - Avoca Mining Method 116
7 CONCLUSIONS AND RECOMMENDATIONS 120
8 FUTURE WORK 125
REFERENCES 128
APPENDIX A 134
PROPOSED ANALYTICAL EQUATION TO DETERMINE VERTICAL LOAD OF BACKFILL.... 134
APPENDIX B 138
PROPOSED ANALYTICAL EQUATION TO DETERMINE ROTATIONAL SILL MAT FAILURE
MODES 138
APPENDIX C +
139
INTERFACE ELEMENTS 139
List of Tables vii

LIST OF TABLES

Table 2-1: Support strength properties 11

Table 2-2: Development support 12
Table 2-3: Mobile Equipment List 16
Table 4-1: Mean uniaxial compressive strength, standard deviation and ISRM hardness
55
Table 4-2: Mean Young's Modulus (E) and Mean Poisson's Ratio (v) 56
Table 4-3: RMR and Q' of a typical foorwall stope 59
Table 4-4: Typical RMR ranges for the different Musselwhite rock types 59
Table 4-5: Musselwhite's far field stress state 60
Table 4-6: Mohr-Coulomb stress-strain and strength parameters for backfill and host
rock 61
Table 4-7: Updated Musselwhite stability database 63
Table 5-1: Example of Musselwhite's geometry, rockfill loading, and sill mat strength
properties 86
Table 6-1: Average and standard deviation for the Musselwhite's stability database. 102
 Table of Figures viii

T A B L E OF FIGURES

Figure 2-1: Location of Musselwhite mine 5

Figure 2-2: Long section of Musselwhite orebody 5
Figure 2-3: Geological cross-section through the T-Antiform Northern Iron Formation6
Figure 2-4: Geological cross-section showing the Wa, T, C and S zones 7
Figure 2-5: Longitudinal retreat open stoping -- Mining adjacent to backfill 9
Figure 2-6: Transverse open stoping -- Mining adjacent to backfill 10
Figure 2-7: Cemented rockfill sill mats constructed: 275, 300 and 375 m Levels 14
Figure 3-1: Purposes of backfilling 18
Figure 3-2: Mohr-Coulomb failure envelope for a cohesive fill material 24
Figure 3-3: Mohr-Coulomb failure envelope for a cohesionless fill material 25
Figure 3-4: Angle of repose < >
| for cohesionless fill material 26
Figure 3-5: Sill mat pillar cross-section 28
Figure 3-6: Natural angle of repose of the material (left) and the oblique, normal and
tangential components of the force (right) 29
Figure 3-7: Vertical stress at increasing depth for a confined material 31
Figure 3-8: Pressure on a horizontal slice of thickness 5z 31
Figure 3-9: Relationship of earth pressures to wall movements 35
Figure 3-10: Comparison of at-rest earth and active earth pressure coefficients 35
Figure 3-11: Differential slice in a silo 40
Figure 3-12: Mathews Stability Graph (Mathews et al., 1981) 46
Figure 3-13: Modified Stability Graph (Porvin and Milne, 1992) 47
Figure 3-14: Modified Stability Graph with support (Nickson, 1992) 47
Figure 3-15: ELOS dilution design method (Clark, 1998) 48
Figure 3-16: Empirical estimation (a) of wall slough (ELOS) expressed in terms of
Radius Factor (b) 49
Figure 4-1: Equal area stereonet representing the three major joint sets 57
Figure 4-2: Wedge formed on the back of a drift running north-south 58
Figure 4-3: Geological section obtained using Vulcan database 60
Figure 4-4: Statistical Analysis on Musselwhite stability database 64
Figure 5-1: Coefficient of lateral earth pressure A" for cohesionless material 67
Figure 5-2: Comparison of analytically and numerically determined coefficients of
lateral earth pressure 68
Figure 5-3: Coefficient of lateral earth pressure obtained using numerical modeling and
best fit curve derived 69
Figure 5-4: Coefficient of lateral earth pressure obtained using numerical modeling and
best fit curve derived for the rockfill friction angle range 69
Figure 5-5: Horizontal vs. vertical stress at different rockfill friction angles 70
Figure 5-6: Vertical stress comparison for different K values 71
Figure 5-7: 25 meters (5 layers) and 35 meters (7 layers) of unconsolidated backfill
deposited in the stope 73
Figure 5-8: Vertical stress distribution along span at different stope dips 74
Figure 5-9: Stress contour for a stope dipping at 70 degrees 75
Figure 5-10: Vertical stress at various stope spans for different rockfill heights 75
Figure 5-11: Vertical stress vs. rockfill density at varying rockfill heights 76
 Table of Figures ix

Figure 5-12: Vertical stress vs. rockfill height at varying friction angles 77
Figure 5-13: Vertical stress over rockfill height comparing analytical and numerical
results 78
Figure 5-14: Vertical stress versus stope span using different methods 78
Figure 5-15: Vertical stress for a 10 meter stope span using different methods 79
Figure 5-16: Analytical vs. numerical modeling comparison for different stope dip
angles SO
Figure 5-17: Geometry, strength and stress component and failure modes in a sill mat
pillar g
Figure 5-18: Rotational failure considering shear strength in the hangingwall of the sill
mat g
Figure 5-19: Rotational failure considering shear strength in the hangingwall of the sill

Figure 5-20: Factor of safety for sill mat failure modes 84

Figure 5-21: Strain softening model - cohesion example 87
Figure 5-22: Caving failure mode - planar and semi circular crack 88
Figure 5-23: Sill mat flexural failure mode - grid elements and plasticity state 89
Figure 5-24: Sill mat sliding failure mode - grid elements and plasticity state 89
Figure 5-25: Sill mat rotational crushing failure mode - grid elements and plasticity
state
Figure 5-26: Sill mat rotational breaking failure mode - grid elements and plasticity
state 91
Figure 5-27: Sill mat stability for 90° - No strength on HW (x = 0%)
t 92
Figure 5-28: Sill mat stability for 90° - 50% sill mat strength on HW (x, = 50%) 92
Figure 5-29: Sill mat stability for 90° - 100% sill mat strength on HW (T, = 100%) 92
Figure 5-30: Sill mat stability for 85° - No strength on HW (x = 0%)
t 93
Figure 5-31: Sill mat stability for 85° - 50% sill mat strength on HW (x, = 50%) 93
Figure 5-32: Sill mat stability for 85° - 100% sill mat strength on HW (x, = 100%) 93
Figure 5-33: Sill mat stability for 80° - No strength on HW (x = 0%)
t 94
Figure 5-34: Sill mat stability for 80° - 50% sill mat strength on HW (x = 50%)
t 94
Figure 5-35: Sill mat stability for 80° -100% sill mat strength on HW (x = 100%)
t 94
Figure 5-36: Sill mat stability for 75° - No strength on HW (x, = 0%) 95
Figure 5-37: Sill mat stability for 75° - 50% sill mat strength on HW (x = 50%)
t 95
Figure 5-38: Sill mat stability for 75° -100% sill mat strength on HW (x = 100%)
t 95
Figure 6-1: Horizontal stress versus rockfill height at different stope spans 98
Figure 6-2: Stope adjacent to rock (isolated) or backfill abutment(s) - Modified from
Atlas Copco Drawing 99
Figure 6-3: Stability Graph update - longitudinal stope data (AVOCA) 101
Figure 6-4: Stability Graph update - transverse stope data 102
Figure 6-5: Stability Graph update - transverse and longitudinal stope data 103
Figure 6-6: Relative importance of inputs on stope stability for stopes dipping more
than 90 degrees (footwalls) V
Figure 6-7: Relative importance of inputs on stope stability for stope dips under 90
degrees (hangingwalls) 1<
Figure 6-8: Relative importance of inputs on stope stability considering rock type, ore
dip and undercut Ii
Table of Figures
x

Figure 6-9: Effect of undercutting the hangingwall: various types of failure 107
Figure 6-10: Predicted ELOS values of transverse vs. longitudinal stopes 108
Figure 6-11: Wall instability due to structure on hangingwall 109
Figure 6-12: Sequence without backfill 110
Figure 6-13: Sequence with backfill 110
Figure 6-14: Mining of the top stope only Ill
Figure 6-15: Horizontal displacement profile Ill
Figure 6-16: Schematic of a stope of increasingly larger strike length 112
Figure 6-17: Radius factor measurement at distance from front face 113
Figure 6-18: Radius factor generation 114
Figure 6-19: Radius factor versus stope strike length at different distance from front
face 114
Figure 6-20: Horizontal displacement measurement at various distances from front face
115
Figure 6-21: Horizontal displacement versus stope strike length at different distances
from front face 116
Figure 6-22: Radius factor increment - stope of 30 meter height and 400 meter strike
length 117
Figure 6-23: Radius factor increments for a stope of 30 meter height and 400 meter
strike length 118
Figure 6-24: Radius factor increments for a stope of 40 meter height and 400 meter
strike length 119
Acknowledgments xi

ACKNOWLEDGMENTS

I would like to extend my deepest gratitude to all those who have helped to make this

thesis possible.
For his assistance and on-going support, and for giving me the opportunity to create this
thesis, I thank first and foremost Dr. Rimas Pakalnis. His input and influence represent the
sine qua non of my work and will never be forgotten.
Enrique Rubio enthusiastically encouraged me to pursue this degree and I could never
fully express my debt of gratitude to him. Thank you beyond words.
Many kind individuals graciously leant their support and the benefit of their experience,
including Robert McDonald, Rod Gray, Tim Sanford, Don Peterson, Joe Hunter, Ken
Strobbe, Jean-Marc Dallaire, Cameron Chapman, Sam Mah, and other personnel of Placer
Dome's Musselwhite mining operation that I may be remiss in not naming specifically.
Forgive me.
For their instrumental recommendations and counsel, I extend my sincere thanks to
committee members, Dr. Malcolm Scoble, Dr. Mario Morin, and David Sprott.
I am enormously grateful for the financial support provided to me by the Department of
Mining and Mineral Processing Engineering at U B C , as well as by N S E R C and Placer
Dome's Musselwhite Mine. Without this support, no thesis, indeed no career, would be
possible.
I would also like to extend my sincerest appreciation to my parents for their continuous
support and encouragement during my studies.
I am infinitely thankful to my patient, forgiving, and un-selfishly supportive wife
Veronica, and to my inspiration to do all that I am capable of doing, my precious daughter
Victoria.
For pointing out errors and omissions, and tactfully criticizing the presentation of my
ideas, research, and theoretical conclusions on paper, I would also like to thank my editor,
Trenton McColl.
Chapter 1 - Introduction 1

1 INTRODUCTION

1.1 Preface

Research into design methodologies for open stope mining is essential to overall mine
efficiency and productivity. New, more accurate, design guidelines will prove more cost
effective than those in existence, thus effecting a more competitive mining operation.
Additionally, these new guidelines will reduce mine waste while improving safety
conditions.
To arrive at new design guidelines, the first component in stope design to be considered
is analysis of cemented rockfill, for the loading case (i.e., loading of uncemented rockfill
material), and for failure mode study (i.e., the limits of the design).
The second component is analysis of backfill adjacent to open workings (i.e., stopes)
and its influence on design.
A method based upon past practice and observed measurements (combined with
modeling analysis) will be employed to arrive at a new methodology that accounts for an
increase in overall stability.

1.2 Open Stopes

Open stoping accounts for over 60% (De Souza, et al., 1998) of the total underground
tonnage in Canadian mining operations.
Open stoping in Canada is one of the most cost effective and productive underground
mining methods. This thesis focuses on the geomechanics of mining by longhole open
stoping, adjacent to backfill. The objective is to improve recovery and reduce dilution.
Dilution and recovery are crucial factors, and may determine the viability of a given mining
operation. A mine may fall below profitability where dilution is higher than expected or
recovery is lower than predicted, because of excess ore left behind in the form of sills and/or
rib pillars. These factors are investigated in detail.

A l l field research was completed at Placer Dome's Musselwhite mine. Musselwhite is a

4000 tpd, underground gold mine, located 500 km north of Thunder Bay, Ontario.
The underground methods in use at Musselwhite are longitudinal retreat and/or
transverse open stope mining, and voids are subsequently backfilled using development
Chapter I - Introduction 2

waste rock and open pit waste rock. The waste is augmented by adding cement, and used for
either backfilling primary stopes or in sill mat construction. Chapter 4 details these activities.
In the first case (Chapter 5), the research focuses on sill mat failure modes by analyzing
current conventional methods (re: Musselwhite). The second case (Chapter 6) concentrates
on rock behavior when mining adjacent to backfilled stopes (primary, and indirectly,
secondary stopes) and where longitudinal retreat with delayed backfill is employed. The
effect of backfill on longitudinal retreat mining (Avoca) is examined in light of its influence
on stability design.
Empirical tools, numerical modeling, and equilibrium analyses are supplemented by
field data, illustrating how backfill affects the stability of mining adjacent to, and below, a
filled stope.

1.3 Background

Mining open stopes in the proximity of backfill has become an increasingly popular
mining method over the past 10 years. This is principally due to recent technical innovations
such as the introduction of pastefill and cemented rockfill. In the past, cemented hydraulic
stopes had less capacity for quality control compared with that of recent innovations in stope
design. Various methods of sizing stopes have evolved since 1982, when Mathews
introduced the "Stability Graph". However, these methods focus on isolated stopes, which
have solid pillar abutments along the strike/dip.
Substitution of backfill for a rock pillar abutment intuitively implies compromised
overall stability of the open stope. This thesis addresses this concern and presents an updated
amendment to the "Stability Graph Method", where mining occurs adjacent to a backfilled
stope. Open stoping under consolidated backfill poses additional risk in terms of back
collapse, and in turn dilution, and subsequent delays to the "mining cycle", which includes
drilling, blasting, mucking, and so on.

The design methodologies of sill mats are reviewed in the context of open stope "non-
entry" mining at the Musselwhite mine. Deficiencies are addressed and a unique method of
design is introduced via numerical and analytical simulations. Musselwhite and other
underground operations are discussed in detail.
Chapter 1 - Introduction 3

1.4 Thesis Overview

Chapter 2 details the mining method currently in use at the Musselwhite mine. Chapter 3
reviews existing literature detailing the types and uses of backfill, and backfill's intrinsic
properties. This chapter also addresses existing analytical equations, and introduces two new
analytical equations, to estimate rockfill loads and sill mat stability. A brief description of the
software used in this study is given. Chapter 4 comprises all of the information gathered from
the Musselwhite mine, including stress, structures, fabric analysis, cavity monitoring system
surveys, geological sections, and others. Chapter 5 introduces relevant cemented rockfill
guidelines by analyzing the stability of sill mats and the variables controlling stability, such
as the vertical load exerted by unconsolidated backfill material. Chapter 6 further expands
our understanding of the stability of open stopes in those cases where mining occurs adjacent
to backfill material, by analyzing the effects of backfill on exposed wall dimensions. Chapter
7 provides a summary of the theoretical, empirical, and practical considerations of this thesis
in concise, conclusive form. Chapter 8 proposes on-site instrumentation to evaluate the
actual, practical effects of backfill material, and suggests possibilities for future work.

1.5 Contributions Made by Thesis

The principal contributions to the field of study are: a more efficient and cost effective
design of sill mats within a mine situation; and a new method of evaluating the effect of
backfill on open stope design. These are unique to the present state of knowledge for open
stope operations.
Replacing an ore sill pillar with a cemented rockfill sill mat pillar will yield a higher
overall recovery, i.e., improve the economics of the mining operation.
Where mining encounters a highly stressed back (e.g., the production drift ceiling)
and/or weak rock mass, safety becomes a critical issue. In this case, the replacement of an
existing ore pillar with a consolidated backfill pillar can improve the overall safety of the
mine.
The methodology proposed in this thesis for the design of open stopes can be used to
effectively reduce dilution as it exists in current mine designs, by accounting for the effect of
backfill on wall stability. A technique based upon past practice and observed measurements,
Chapter 1 - Introduction 4

coupled with numerical modeling, will be employed in the thesis to arrive at a new
understanding of how backfill affects overall wall stability.
The models developed in this thesis attempt to reproduce as accurately as possible the
behavior of the rock mass at the Musselwhite mine, i.e., stress regime, rock properties (e.g.,
strength), and so on. Therefore, these models relate to the Musselwhite mine operation itself,
and should be considered applicable primarily to this specific case.
Although the models are intended to reflect accurately the reality of a particular mining
situation, these models nevertheless require formal validation, i.e., calibration, by the
acquisition of actual mine data, accomplished via proper instrumentation, conducted at the
mine.
Additionally, this real world data application will enhance the accuracy of these
numerical models, as it is carried out in future research, thus progressively increasing the
usefulness of the design curves developed in this thesis.
Chapter 2 - The Musselwhite Mine 5

2 T H E MUSSELWHITE MINE

2.1 Introduction

Placer Dome Inc.'s Musselwhite Mine is located 430 km northwest of Thunder Bay,
Ontario (figure 2-1).

Figure 2-1: Location of Musselwhite mine

The first underground production stope blast was made in March of 1997. The mill
tonnage today is 4200 tonnes daily, of which 4000 tonnes of muck are extracted from
underground, and the remainder comes from open pit. Figure 2-2 portrays a long section of
the mine's orebody plunging at 12 to 15 degrees.

Figure 2-2: Long section of Musselwhite orebody

 Chapter 2 - The Musselwhite Mine.

2.2 Geology

2.2.1 Regional Geology

The Musselwhite property is located in the central region of the Weagamow, North
Caribou Lake metavolcanic, metasedimentary greenstone belt. The property is located in an
area where the package has been isoclinally folded into a series of northwesterly trending
antiforms and synforms (Musselwhite Project Feasibility, Geology Report, 1995). These
structures plunge 12 to 15 degrees to the northwest. Important lithological units, referenced
in this thesis, are:

- 4ea: Iron Formation, chert-grunerite-garnet-amphibole;

- 4f: Iron Formation, garnet-biotite schist;
- 4b: Iron Formation, chert-magnetite;
- A Vol: Volcanic, intermediate-mafic; and
- BVol: Volcanic, intermediate-felsic.
Figure 2-3 portrays a geological cross-section of the T-Antiform in the Musselwhite gold
deposit along grid line 10050 North. The cross-section depicts the distribution of gold
bearing zones within the mixed basaltic and northern Iron Formation sequences.

Figure 2-3: Geological cross-section through the T-Antiform Northern Iron Formation
 Chanter 2 - The Musselwhite Mine
1

2.2.2 Mine Geology

This study deals principally with the gold mineralized zone in the T-Antiform. The gold
mineralization is found predominantly within the sediment rich iron foundation called the
"4ea". More specifically, the gold is located in four subsidiary fold closures within the 4ea
lithology, named the Wa, T, C, and S zones (figure 2-4). The Wa, T, and C zones are
approximately located on the westerly limb of the antiform, and the S zone is located on the
faulted, eastern limb.

Figure 2-4: Geological cross-section showing the Wa, T, C and S zones

In figure 2-4, the Wa, T, and C zones can be generally described as sub-parallel, steeply
dipping, planar bodies of varying thickness. The Wa zone is the lowest and most westward
zone on the fold limb, the T zone is in the middle, and C zone is at the top of the antiform.

These three ore zones are typically found adjacent to each other, but frequently have
several meters of less significant mineralized 4ea between them. The zones vary in thickness
from 0.5 meters to 25 meters, and typically represent a sectional height of 200 meters from
the bottom of the Wa zone to the top of the C zone. The three zones lay in close proximity to
each other.
Chapter 2 - The Musselwhite Mine 8

For example, the Wa and T zone widths vary from 0 meters to 17 meters of waste
materials (and typically, widths fall between 0 meters and 6 meters). The T and C zone
widths vary from 0 meters to 13 meters of waste materials (and typically, widths fall between
0 meters and 9 meters). The S zone is a steeply dipping body positioned on the eastern limb
of the anticline to the south, and running up to straddle both sides of the fold to the north
(depicted in figure 2-5). The S zone is comprised of a short (45 meters), thick (10 meters)
body in the south, and a thinner (6 meters), tall (150 meters), steeply dipping body to the
north.
Chapter 2 - The Musselwhite Mine

2.3 Underground Mining Methods

2.3.1 Avoca

Two underground longhole mining methods are currently employed at the Musselwhite
mine: Avoca, or longitudinal retreat open stoping, with delayed backfill; and transverse open
stoping (Musselwhite Project Feasibility, Mining Report, 1996). Longhole open stoping is
the principal method employed at Musselwhite, because of its safety, productivity, and
economic characteristics.
Where the ore is between 4 meters and 12 meters in width, Avoca, or longitudinal retreat
is used. This method employs conventional longhole benching where the stopes are
backfilled with waste rock composed of development waste and crushed material from open
pit stripping.
Stopes are mined upwards, where the ore is mucked on top of the previously backfilled
stopes below. The broken ore is usually remotely mucked on the lower sill. Rockfill material
is dumped, using either scoops or trucks, from the top sill, filling the stope to just short of
where the hanging wall exposure would exceed its predicted maximum stable span. The
Avoca method demands periodic re-deposit of backfill to prevent the stope walls from
collapsing into the open stope. The mine operation comprises sequentially: blasting,
mucking, and backfilling. Slot raises are drilled and blasted at the end of each stope, to create
the free face necessary for subsequent production blastholes to be fired.

Drilling Equipment Truck backfills after

most ore is mucked
— \ Retreating.

Backfill

L H D Equipment
Floor can be of any type: Ore, backfill or sill (mat) pillar

Figure 2-5: Longitudinal retreat open stoping ~ Mining adjacent to backfill

Chapter 2 - The Musselwhite Mine 10

2.3.2 Transverse Retreat Open Stoping

Where the ore width exceeds 12 meters, it is typically mined transversely within 50
meter or 25 meter high stope blocks. The transverse stopes are backfilled with cemented and
uncemented rockfill. Cemented rockfill is used in primary stoping blocks and uncemented
rockfill is used in secondary stopes. Primary stope rockfill is crushed and screened to less
than 5 inches, yielding a workable size that achieves the strength needed when mixed with
cement binders. The material is dropped down a raise fill to the underground batching plant.
Next, waste material is discharged through a chute into the truck box, and cemented slurry is
then added via spray bars. Slot raises are drilled and blasted at each stope, creating new faces
for each primary and secondary stope.

; Open Stope
Secondary Primarj Primar)
U R F Backfill CRF Backfill ! Blasted Ore C R F Backfill

L H P Equipment

Figure 2-6: Transverse open stoping ~ Mining adjacent to backfill

2.3.3 Drilling and Blasting Methods

The limits of the orebody define whether parallel or fanned 10 centimeter (4 inch)
blastholes are drilled. These blastholes can be drilled as either down holes or up holes,
depending on the location of the stope and its sequence.
Historically, ground control problems associated with drilling and blasting have been:
inaccurate drilling; faulty timing of the holes; and/or discovery of unfamiliar geological
structures. These are all important factors in wall stability.
Chapter 2 - The Musselwhite Mine 11

2.3.4 Ore Handling

Ore is handled underground, by remote-controlled LHDs with 40-tonne-capacity

articulated low profile trucks. They are used to haul the ore to the surface or to feed the bin of
an underground crusher, 460 meters beneath the surface. Remote-controlled LHDs enhance
both safety and productivity measures. Remotely, the operator controls the fully-functional
L H D , ensuring maximum safety.

2.3.5 Ground Support

Consistent with Placer Dome's support guidelines, pattern bolting sufficient to prevent
small scale local failure, is used in all cases (Geotechnical Assessment of the Musselwhite
Project, Golder Associates, 1996). The following table summarizes the type and pattern of
development support employed:

Table 2-1: Support strength properties

R O C K B O L T PROPERTIES
Bolt Strength Yield Strength Breaking Strength
(tonnes) (tonnes)
5/8" mechanical 6.1 10.2 (Grade 690MPa)
Split Set (SS-33) 8.5 10.6
Standard Swellex N/A 11
Yielding Swellex N/A 9.5
Super Swellex N/A 22
20mm rebar (#6) 12.4 18.5
25mm rebar (#8) 20.5 30.8
#6 Dywidag 11.9 18
#7 Dywidag 16.3 24.5
#8 Dywidag 21.5 32.3
#9 Dywidag 27.2 40.9
#10 Dywidag 34.6 52
1/2" Cable bolt 15.9 18.8
5/8" Cable bolt 21.6 25.5
#6 refers to 6/8", #7 refers to 7/8", #8 refers to 8/8" diam etc.

BOND STRENGTH
Split Set 0 . 7 5 - 1.5 tons/ft 0 . 7 - 1 . 4 tonnes/ft
Swellex 3 - 5 tons/ft 2 . 7 - 4 . 6 tonnes/ft
Cable Bolt (5/8") 29 tons ~ 3ft 10.2 (Grade 690MPa)
Rebar 18 tonne 20 tons - 12" 18 tonnes ~ 12" (granite)
Chapter 2 - The Musselwhite Mine 12

S C R E E N - B A G S T R E N G T H 4ft X 4ft P A T T E R N
4' X 4" Welded wire mesh (4 gauge) bag strength = 3.6 tonne
4' X 4" Welded wire mesh (6 gauge) bag strength = 3.3 tonne
4' X 4" Welded wire mesh (9 gauge) bag strength =1.9 tonne
4' X 2" Welded wire mesh (12 gauge) bag strength = 1.4 tonne
2" chainlink - 11 gauge bare metal bag strength = 2.9 tonne
2" chainlink - 11 gauge galvanized bag strength = 1.7 tonne
2" chainlink - 9 gauge bare metal bag strength = 3.7 tonne
2" chainlink - 9 gauge galvanized bag strength = 3.2 tonne
4 gauge = 0.23" diam, 6 gauge = 0.20", 9 gauge = 0.16"
11 gauge = 0.125", 12 gauge = 0.11" diam

2
shotcrete shear strength = 2MPa = 200 tonnes/m

2.3.5.1 Development Support

Placer Dome's support guidelines indicate that pattern bolting sufficient to prevent small
scale local failure be used in all cases. The type of support that should be employed
depending on the span of the opening, is summarized in the following table:

Table 2-2: Development support

pattern
#6 rebar 1.4 x 1.4 m

pattern
#8 rebar 1.8 x 1.8 m

Fully Grouted

Face plate

pattern
Cable Bolts 3.6 x 3.6 m

Span

<5.5m X X X
5.5m - 7.2m X X X
7.2m < X X X X

2.3.5.2 Stope support - Stope backs and walls

It is standard practice in Canada to use support in all working areas. The accepted
convention is that the length of the support should not be less than one-third of the minimum
span of the supported excavation. Assuming that rock bolts with a maximum length of 2.4
Chapter 2 — The Musselwhite Mine 13

meters are used, cables should be employed in all excavations where the minimum span
exceeds 7.2 meters.
Cables effectively support hanging walls and foot walls, and prevent cave-in, which
would affect production and increase dilution if not addressed sufficiently.

2.3.6 Backfill

The Musselwhite orebody is of variable width and dip. Accordingly, there are a number
of different stope layouts. The two main layouts are longitudinal retreat, where the fill
exposure is of limited span and up to 30 meters high, and transverse stoping where the fill
spans up to 30 meters with heights of up to 25 or 50 meters.
Two backfill stations are located at the 150 mL and the 200 mL. Given that the bulk of
the mine production today comes from the 400 mL and below, backfilling has become a time
consuming and costly task.
Open pit, screened, cemented rockfill is used for backfilling transverse primary stopes.
As waste is mined, selected rock is stockpiled for crushing. Crushed rock is dumped into a
fill raise and then transferred to trucks via a chute at the bottom of the raise. Cement is
transferred via a borehole to an underground cement slurry station, or batch plant, located
adjacent to the fill raise, where the cement is mixed with rockfill. Spray bars are used to coat
the rockfill with a pre-measured amount of cement slurry, in accordance with relevant
strength requirements. Trucks haul the fill to the stope and dump the fill at the top of the
stope.
The system is comprised of several components:
- crusher;
- on-site cement storage (cement and fly ash silos);
- boreholes for cement transfer underground;
- raises for transfer of rockfill underground; and
- underground fill stations (including the slurry plant).

Generally, the upper bound for particle size delivered to the top of the fill raise is set
within a range of 150 mm to 200 mm. Waste rock consists typically of meta-volcanic rock
with estimated uniaxial strengths in excess of 150 MPa, which is ideal for CRF composition.
Chapter 2 - The Musselwhite Mine 14

2.3.6.1 Mining Underneath Backfill - CRF Sill Mats

A sill mat pillar is a cemented, enriched, bottom floor of a backfilled stope. It is used in
place of an ore sill pillar, maintaining stability when mining the stope below. Musselwhite's
sill mats are constructed from several batches of screened rockfill material, recovered from
run of mine development waste and/or open pit crushed waste rock (crushed to < 12.5
centimeters). Binder is mixed at the surface plant. The resultant slurry is transported below
ground via boreholes. Each truck-load is coated with the binder mixture using spray bars,
transported and end-dumped onto the floor in a retreating fashion. A scoop (LHD) is used to
bring the sill mat to proper elevation, as required.

To date, 3 sill mats have been constructed: the 275 mL, the 300 mL, and the 375 mL,
portrayed in figure 2-7.

h- — > rJ CRFS ill Mats

\
225 L e v e l

250 L e v e l 100m
5 Level
*
— /
\ 330 L e v e l

375 L e v e l

— 4.00 L ( » V P I

425 L e v e l

Figure 2-7: Cemented rockfill sill mats constructed: 275, 300 and 375 m Levels

2.3.6.2 Mining Adjacent to Backfill

Stopes are mined upwards, using sub-level open stoping methods. The ore is mucked on
the floor of backfill material and deposited on the level below. The broken ore is normally
remotely mucked on the bottom sill. Rockfill material is dumped, using either scoops or
Chapter 2 - The Musselwhite Mine 15

trucks, from the top sill to the point just before the hanging wall exposure would exceed its
predicted maximum stable span.
The longitudinal retreat and transverse open stoping methods both require periodic
placement of backfill to support the stope walls.
This mine operation consists sequentially of: blasting; mucking; and backfilling. Slot
raises are drilled and blasted at the end of each stope to create the free face needed for
subsequent production blastholes. Remote scoops are utilized to muck ore from the blasted
stopes. Use of remote-control mucking improves both safety and productivity.
Rockfill used for filling mined stopes comes from underground waste development, or
from open pit rock material. Open pit rockfill material is crushed and screened to a maximum
size of 10 cm, providing crushed rock dimensions consistent with strength guidelines for
mixing with cement (< 5% for primary stopes; < 8-10% for sill mat pillars). The material is
dropped down a raise fill to the underground batching plant, where it is discharged through a
chute and onto a truck or scoop. Cemented slurry is then applied with spray bars to meet
shear and tensile strength requirements.
Maintaining a high quality backfill product is crucial to the safety and efficiency of the
mining operation, while maximizing ore recovery, and leaving behind minimal ore pillars.
Chapter 2 - The Musselwhite Mine 16

2.4 Equipment

Mobile equipment in use is summarized by table 2-3.

Table 2-3: Mobile Equipment List

Type of Equipment Make/Model Number of Units

Drilling Equipment
Production drills Tamrock Solo 1000 Sixty 2
Development drills Tamrock Minimatic H205D 3
Bolters Tamrock Robolt H320-30C 3
Miscellaneous drills MacLean Blockholer 1
Boart BCI-2 1
Production Scoops
3 yd" Tamrock EJC-130 1
8yd J
Tamrock Toro T500D 3
9yd J
Tamrock Toro T650D 5
11 yd J
Tamrock Toro 0011 1
Production Trucks
30 tonne Tamrock EJC 430 1
40 tonne Tamrock Toro 40D 9
Ancillary Equipment
Explosives loader Tamrock ALB45 2
Scissor lift Teledyne SL6-812 3
Boom truck Teledyne 1
Grader Caterpillar M-120 1
Personnel vehicle Toyota Landcruiser 17
Bulk explosives truck ICI U-101-1 1

This chapter reviewed Musselwhite's mining methods, types of support, handling of ore

and waste materials, and types of equipment in current use. The next chapter will discuss

design methodologies in current use.

Chapter 3 - Review of Design Methodologies 11

3 REVIEW OFDESIGN METHODOLOGIES

This review of pertinent, existing literature is sub-divided into the following topics:
- uses, types, and properties of backfill;
- vertical load exerted by rockfill;
- analytical sill mat design;
- empirical stability graph methods; and
- software utilized in the analysis.

3.1 Backfdl

Backfill is an increasingly important component of underground mining operations

around the world. Following is a brief overview of current methods of backfill technology,
focusing on rockfill. Rockfill is the principal backfill material used at the Musselwhite
operation.
Disposal of backfill underground not only reduces the environmental impact and
footprint, but also provides the basis of an engineering material that can be used to improve
both the ground conditions, and the economics of mining. Carefully engineered and
efficiently operated backfill systems can significantly enhance the productivity of a given
mining operation. On the other hand, badly engineered backfill system can become a serious
impediment to mining operations and, in the worst case scenario, compromise safety.
Whether composed of tailings, alluvial sand, or coarse rockfill, backfill can be treated as
a special form of soil. Therefore, several of soil's mechanical properties and relationships can
be applied to backfill. Note that mechanical and cure properties of fill deposited underground
may vary considerably, depending upon the fill material's intrinsic properties, the method of
preparation, and the location and condition of the mine environment.

3.1.1 Purpose of Backfdl

Backfill has multiple purposes, as indicated in figure 3-1. The primary purposes are to
improve hangingwall stability, and to permit increased ore extraction. Other important
application purposes include dilution control and regional support. Although much less
Chapter 3 - Review of Design Methodologies 18

common, some interesting applications might include ventilation improvement, fire control,
and environment protection (De Souza, et al., 1998).

Purpose of backfilling

Figure 3-1: Purposes of backfilling

For wall stability using backfill, the objective is not to transmit rock stresses, but to
reduce the uncontrollable convergence of rock mass so that the rock itself will retain its load
carrying capacity, and in turn, decrease cracking propagation of crown pillars and abutments
(Barrett et al., 1978). This also helps to prevent cave-ins and roof falls, minimize surface
subsidence, and enhance pillar recovery (Coates, 1981). The ultimate result is less
deterioration of ground conditions in the mine, and thus operation and safety are improved
overall.
Adding small percentages of ordinary Portland cement provides cohesive strength, and
self-support when exposed in vertical faces to adjacent pillar mining, as well as to the mining
conducted below. The self-supporting nature of backfill permits higher recovery of pillars.
This, in turn, allows greater exploitation of the mining reserves, and improves the economics
of the mining operation. Increased ore recovery results in a longer mine life.
Chapter 3 - Review of Design Methodologies 11

In some mining methods, backfill forms a working platform for mine personnel and
equipment. Backfill must be capable of supporting this traffic. Cement is usually not required
in such an application.
Depositing backfill underground directly reduces the quantity of waste to be disposed of
on the surface. This results in direct operating and capital cost benefits, and reduces future
rehabilitation costs.
Nantel (1998) initiated a trend now common in Canada, where environmental
authorizations for future underground mines require the return of the maximum available
quantity of mine wastes to the underground operation. This trend has reached its logical limit,
as seen in the recent Australian government recommendation to approve a project for the
proposed Jabiluka uranium mine, which is encircled by the world famous Kakadu National
Park (inscribed on the World Heritage List). For this project, a maximum percentage of
milling wastes are required to be re-deposited underground.
Research by Y u (unpublished) and Stone (1993) provides a good background reference
source regarding the placement of cemented rockfill, typical strength characteristics and its
applications.

3.1.2 Types of Backfill

Hassani and Archibald (1998) stated that backfill refers to any waste materials re-
deposited into voids mined either for disposal or some other engineering task. These
materials include waste development rock, deslimed and whole mill tailings, quarried and
crushed aggregate and alluvial or aeolian sands. Other exotic backfill components may
include ice and salt. The waste materials are often combined with cement or other pozzolanic
(volcanic ash) binders to improve their strength properties.
The three most common backfill types are:

- Hydraulic backfill Deslimed mill tailings slurry, with densities up to

70%cw (i.e., concentration of solids by weight) — the coarser fractions are
deposited underground as hydraulic backfill and the slimes rejected to the
surface dam;
Chapter 3 - Review of Design Methodologies 20

- Rock/ill Surface or underground waste rock, crushed to a

maximum size -- can be deposited as is or combined with cemented hydraulic
backfill slurry, or cement water slurry; and

- Paste backfill Total mill tailings filtered or thickened to approx.

80%cw, to which cement and water is added, achieving required rheological
and strength characteristics. (Note: pastefill is a type of backfill in current use,
although this thesis does not deal specifically with pastefill.)

3.1.2.1 Rockfdl

Rockfill can be grouped into unconsolidated and consolidated types of material. The
latter comprises binding mixtures that achieve the required strength for a given purpose.
Consolidated, or cemented, rockfill (CRF) includes classified or unclassified aggregate,
mixed with a variety of types and quantities of binder materials. Typically, unconsolidated,
or uncemented, rockfill (URF) is pre-mixed with cement binder, usually in slurry form, prior
to its entry into the stope, forming a reasonably homogenous material at that time.
Consolidated rockfill is commonly employed in large exposures where undermining can
occur, future exposure of fill walls is expected, and significant resistance to wall movement
is desired. The rockfill stiffness so achieved produces significant advantages to operations
seeking ground control at sites subject to high ground stresses.
URF also facilitates an inexpensive method of disposing of development waste or
surface open pit waste.

3.1.2.2 Uncemented Rockfdl - URF

URF is waste material deposited without additives. Although U R F has a limited ground
support capability because of its minimal resistance to closure, URF will prevent wall slough.
The free standing height capacity of such fill is negligible. URF is used when the filled stope
won't be exposed to future pillar recovery operations. In this case, stopes surrounding the
filled stope will have been mined and filled before pillar recovery. URF is generally used for
void filling and to provide some measure of passive wall support to resist localized ground
Chapter 3 - Review of Design Methodologies 21

movement. It is a relatively low cost fill medium, which is both expedient and relatively easy
to place in situ.

3.1.2.3 Cemented Rockfill with Portland cement Slurry - CRF

A variation on cemented rock backfill widely used in Canada makes use of crushed
waste rock and Portland cement slurry aggregated at rates of approximately 5% cement by
weight. Consequently, the overall sizing curve is low in fines content (i.e., size distribution of
rockfill). This reduces the workability of the deposited backfill and demands higher cement
aggregation to provide equivalent strength performance. The technique is currently in use at
the Musselwhite mine, where a surface batching plant produces slurry, which is then
delivered via boreholes to an underground mixing plant.
Note that CRT yields a higher strength fill, demanding lower quantities of cementing
agents in comparison to cemented hydraulic fill materials (including pastefill). Aggregated
with equivalent binder components, CRF exhibits uniaxial compressive strengths that can be
two to three times higher than consolidated hydraulic fill (including pastefill). CRF rockfills
also exhibit higher moduli of elasticity, cohesion, and friction angle characteristics than
hydraulic fill mixtures (including pastefill) composed of similar cement contents.

3.1.3 Backfill Properties

Rockfill particles are in constant contact with each other, i.e., particles do not collide but
rather roll, rub, and scrape against one another. This type of stress is called frictional stress,
as opposed to collisional or kinetic stress, which occurs in more diluted solutions.
When granular material does not move or flow to any extent, the only force acting on the
grains is gravity. Basic engineering theory states that for any fluid, the total stress at the
bottom of a granular pile of height h would be:

P = p-g-h 3-1

Where:

g is the acceleration of gravity; and

P is the bulk density of the rockfill material.

Chapter 3 - Review of Design Methodologies 22

Theoretically, the total stress at the bottom of the waste fill is a normal stress along the
vertical direction (h), and therefore is an isostatic pressure, solely resulting from the weight
of the rockfill pile. In practice this is not true. In fact, beyond a sufficient height of the
rockfill column, the pressure reaches a maximum value and will not increase further,
regardless of the height of the fill. This occurs because rockfill supports frictional shear
stress, even in a static situation. Additionally, i f side walls exist within the stope, they can
support the extra weight of the rockfill column. And so, the total stress at the bottom of the
rockfill pile is a combination of normal stress and shear stress.
In order to predict the ground support capacity of a backfill system, it is essential to first
define, and to understand, the significant properties of the fill material. Since most fill
material can be defined as either granular or fine-grained soils, evaluation of any fill property
involves the principles of soil mechanics. However, soil mechanics alone cannot describe all
of the properties of cemented fill. Aggregation of cement and other pozzolans to a fill,
transforms the fill from natural material, e.g., soil, to an engineered material, making the
process of characterizing fill properties more complex. Extensive evaluation of the effects of
cement aggregation is therefore necessary.
Existing detailed studies and laboratory work examine cemented fills and the properties
significant in their design. Laboratory testing indicates that the relevant properties of
cemented fill (Knissel and Helms, 1983) are:
- strength (i.e., uniaxial compressive strength);
- deformation behavior;
- cohesion and angle of internal friction;
- density and porosity; and
- consistency of the mixture.

As with uncemented fill, strength is the principle concern in cemented fills. Certain
properties of the fill material including friction angle, density, and porosity, have the same
effect on cemented fill strength as on uncemented fill strength.
Cement aggregation improves the strength properties of backfill material. The most
evident increase in strength of a cemented fill over an uncemented fill is in shear strength.
Chapter 3 - Review of Design Methodologies 23

Cement bonds formed between fill particles produce a cohesive component in the fill's
shear strength, absent in an uncemented fill.
Chapter 3 - Review of Design Methodologies 24

The relationship between shear stress and normal stress is commonly referred to as the
Mohr-Coulomb relationship for frictional shear stress.
In its basic form, it is:

r = c + <r„-tan^ 3-2

Where:
- x is the fictional shear stress (MPa, psi, etc.);

- rj n is the normal frictional stress (MPa, psi, etc.); and

- c is the cohesion of the material (MPa, psi, etc.).

Cohesion, c, is a known material property (describing the cohesive state of grains)

normally negligible in uncemented fill material, and (j) is the angle of repose (i.e., the angle of
internal friction of the material).
The following figure describes equation 3-2:

Frictional Stress: Mohr-Coulomb Plane

ress

V) T = c + On tan((|)) —
Sheai

\ ()) = Friction Angle

c = cohesive shear

0 Normal Stress an

Figure 3-2: Mohr-Coulomb failure envelope for a cohesive fill material

At yield, the higher the frictional stress, the higher the shear stress. The more
interconnected the particles, i.e., the more cohesive, and/or the higher the angle of friction,
the higher the shear stress, as indicated in figure 3-2.
Chapter 3 - Review of Design Methodologies 25

The Mohr-Coulomb law described by equation 3-2 and portrayed in figure 3-2, is a
yielding law asserting that a material will yield by shearing on a surface element i f x attains a
critical value defined by equation 3-2. This linear relationship is sometimes called the "yield
line". Below the yield line, the material response will be rigid or elastic and will not typically
undergo strain, and i f it does it is merely elastic strain. Elastic strain is negligible in
uncemented rockfill material.
Figure 3-3 portrays the Mohr-Coulomb envelope for a cohesionless fill material. If the
shear stress is increased for a given normal stress so that the stress state of the material
remains at yield, then plastic strain, i.e., yielding, will result. A state of stress cannot exist
above the Mohr-Coulomb yield line. When yield stress is reached, particles will simply slide
over one another.

Frictional S t r e s s : M o h r - C o u l o m b P l a n e

0 Normal Stress On

Figure 3-3: Mohr-Coulomb failure envelope for a cohesionless fill material

In this case, shear stress of the rockfill material is due solely to the frictional component
and is mobilized by the normal stress.
Chapter 3 - Review of Design Methodologies 26

The angle of repose (or angle of internal friction) is evident in the following figure:

Figure 3-4: Angle of repose <

>
| for cohesionless fill material

The angle of repose, portrayed in figure 3-4, is low when grains are smooth, coarse, or
rounded, and high for sticky, sharp, irregular, or very fine particles. Typically, it is between
15 and 50 degrees. Musselwhite rockfill friction angles range from 35 to 40 degrees.

3.1.4 Parameters Affecting Cemented Rockfill Strength

The following parameters are considered essential to final rockfill strength (Kuganathan,
et al., 2001):
- maximum size of the aggregate;
- grading of the aggregate;
- binder content of the mix; and
- water content of the mix.

Suitable grading of the aggregate is important, not only to reduce porosity, i.e., the void
ratio of the fill, but also to increase the number of rock to rock contacts per unit area. The
strength of the fill depends on the number of cement bonds that develop at these contact
points. The strength and quality of CRF depends on how well the rockfill particles are coated
with the cement slurry before they are deposited in the stope. In a well controlled operation,
Chapter 3 - Review of Design Methodologies 27

the resultant fill will exclude uncemented rockfill pockets, and a nearly homogeneous CRF
fill mass is thus achievable.
In the majority of underground mining operations reviewed, sized rockfill aggregate is
mixed with cement slurry, typically at a cement content of 5 to 6 percent, by weight(cw), and
a pulp density of 50 to 60%(cw). This type of fill may exert active pressure on contact with
stope walls, providing not only ground support but also improvement in the inherent wall
rock strength. In the case of consolidated rockfill, there normally occurs no drainage
problem, and a high fill quality can often be achieved i f the materials are mixed properly.
With these rockfill materials however, segregation control can be difficult, and quality
control may be variable. In this regard, Stone (1993) presents a methodology for the
optimization of mix designs for cemented rockfill, and suggests laboratory evaluations for
verification of the design parameters. Furthermore, a program of quality control testing and
performance monitoring is outlined to improve the reliability of the design.

Segregation of consolidated rockfill is often unavoidable but can be minimized when fill
operations are well planned and closely monitored.
Chapter 3 - Review of Design Methodologies 28

3.2 Vertical Load on Cemented Rockfdl Sill Mat Pillars

3.2.1 Sill Mat Pillar Definition

Sill pillars are ore blocks left between working levels in an underground mine to support
the overlying mine backfill, during removal of the underlying ore. Steeply dipping ore bodies
should be mined bottom up but for economic reasons and/or because of stability constraints
(i.e., underhand cut and fill), upper ore levels are usually mined first. Sill pillars are
commonly used in steeply dipping ore zones of limited width.
A typical situation is portrayed in figure 3-5:

Backfill

Sill Mat Pillar

Stope

Figure 3-5: Sill mat pillar cross-section

Accordingly, mineable values in the ore generally make it economically feasible to mine
the sill pillar. It is necessary to create an artificial sill, which will support the overlying
unconsolidated fill, after the stope below is mined. Such sill mats can be constructed from
either cemented rockfill or hydraulic and paste backfill material.
In analyzing sill mat stability, the strength of backfill necessary to maintain stability is a
function of the stresses generated within and around the sill mat and the fill mass.
These stresses are caused by:
- self-weight of the sill mat;
- vertical load of unconsolidated backfill;
- blast damage/abrasion; and
- ground movements.
Chapter 3 - Review of Design Methodologies 29

The most significant design loading used in this thesis is the weight of the fill material.
The stresses generated by this weight are a function of fill density, fill friction angle, height,
and span.
A number of methods have been proposed for analyzing the stability of cemented sill
mats. These generally use free standing wall or two-dimensional stope stability concepts.
Mitchell's (1981) approach is reviewed in this chapter.

3.2.2 Background - Loads Acting by Unconsolidated Material

When backfill material is dumped onto a horizontal surface, it heaps into a volume
conical in shape, forming a specific angle <>
f with the horizontal plane. This angle is
characteristic of each type of material and is called the natural angle of repose of the
material. This is the angle of internal friction. If the material is dumped into a confined space
such as that of a stope, it will exert pressure on the walls and on the floor of the stope. The
resultant thrust, due to the friction of the material on the walls, is oblique in relation to the
surface of the walls. This thrust has two components, one, N , normal to the wall, and the
other, T, parallel to the wall (Reimbert, 1976).

The normal pressure is also called lateral thrust (figure 3-6):

Figure 3-6: Natural angle of repose of the material (left) and the oblique, normal and
tangential components of the force (right)
Chapter 3 - Review of Design Methodologies 30

The material's angle of friction on the walls is termed <)>,' and the corresponding
coefficient of friction is tan <)>.' As a function of the oblique thrust Q defined above, the two
components N and T are:

N = Q-cosf 3-3
T = Q-sm</>' 3-4

Therefore:

T = N-tan.fi 3-5

Where T is the load balanced by the friction corresponding to the thrust N .

At a given depth of the stope, the load, i.e., total vertical pressure, is the difference
between the total weight of the backfill material and the total load offset by the friction of the
material on the walls.
In situ measurements tell us that pressure increases with depth. However, because of the
friction of the material on the walls, the pressure on the stope floor is merely a fraction of the
weight of the confined material. Furthermore, at great depths the pressure reaches a constant
maximum value (Reimbert, 1976). The curve representing pressure (plotting depth on the x-
axis, and pressure on the y-axis) is asymptotic parallel to the x-axis, in correspondence with
the maximum vertical pressure (portrayed in figure 3-7).
Chapter 3 - Review of Design Methodologies 11

Vertical Stress at Deoth

for Confined Material

Vertical Stress rjyy

/ Maximum Vertical Pressure

Depth z

Figure 3-7: Vertical stress at increasing depth for a confined material

3.2.3 Maximum Horizontal Pressure on the Walls of the Stope

A behavior similar to that portrayed by figure 3-7, can be inferred from the curve
representing the lateral thrust on the walls. This curve is asymptotic parallel to the x-axis
where the y-axis plots maximum thrust.
The value of the maximum horizontal thrust can be determined by the following method
(figure 3-8):

* •
L

Figure 3-8: Pressure on a horizontal slice of thickness 8z

Chapter 3 - Review ofDesien Methodologies 21

The weight of a horizontal slice of thickness dz in the interior of the stope is:

S-dz-y 3-6

Where:
- S = internal area of the cross-section of the stope;
- y = unit weight of the backfill material; and
- dz = thickness of the horizontal slice.
The weight exerts a lateral pressure on the walls over the entire perimeter:

cr {z)-P-dz
x 3. 7

Where:

- CT (z) =
x lateral thrust at a depth z; and

- P = internal perimeter of the stope.

Pressure increases with depth up to the maximum (7 _M ,

x ax at which point the friction on

the walls balances the actual weight of the slice considered. Therefore:

°_ x Max -P-dz-tan(p'=S-dz- r 3-8

From which the value of the maximum thrust can be determined:

cr = r
' S
3-9
X
- Max
P-tanp'

Note that, for stopes of sufficient strike length (i.e., in theory, infinite), the hydraulic
radius S/P tends toward one half the span of the stope (see equation 3-9). Equation 3-10
indicates the maximum lateral thrust within a stope of sufficiently long strike:

a = y L
3-10
xx Max ,

2 • tan cp
Where:
Chapter 3 - Review of Design Methodologies 33

- L = stope span.

Up to this point in the analysis, only the maximum lateral thrust has been derived
analytically. Determining the stability of CRF sill mats deals with the vertical load exerted on
the top of the sill. The coefficient of lateral earth pressure implies a relationship between the
vertical and horizontal pressures. This relationship is considered next.

3.2.4 Coefficient of Lateral Earth Pressure

Estimating lateral earth pressures is not a trivial problem. Nor is it a new problem.
Couplet and Coulomb developed their theories in 1726 and 1776 (respectively), and Rankine
developed his theory in 1857. Flexural movement of the structure must occur for the
predictions of these theories to be valid, and whether it is translation/rotation into, or out of,
the soil mass in question, active (or passive) lateral earth pressure will develop.
The coefficient of lateral earth pressure implies a constant relationship between the
average horizontal, OH, and vertical, oy, stresses, independent of the geometry of the fill
volume. While there is no physical reason for such a relationship to hold, in practice, routine
industrial calculations indicate that the results implied by this relationship are reasonably
accurate empirically.

Equation 3-11 portrays this relationship (the coefficient of lateral earth pressure):

K =— 3
" n

3.2.4.1 Categories of Lateral Earth Pressure

There are three categories of lateral earth pressure and each depends upon the movement
experienced by the vertical wall upon which the pressure is acting.
These categories are:
- active earth pressure;
- passive earth pressure; and
- at-rest earth pressure.
Chapter 3 - Review of Design Methodologies 34

Active pressure develops when the wall is free to move outward. E.g., a retaining wall
where the soil mass may stretch sufficiently to mobilize its shear strength. On the other hand,
if the wall moves into the soil, then the soil mass is compressed sufficiently to mobilize its
shear strength where passive pressure develops.
This problem might occur along the section of the wall that is below grade, on the
opposite wall from that of the higher section.
In order to develop full active pressure or full passive pressure, the wall must move. If
the wall does not move sufficiently, full pressure will not develop. If full active pressure does
not develop behind a wall, then the pressure will be higher than the expected active pressure.
Significant movement is necessary to mobilize the full passive pressure. This is illustrated in
figure 3-9.
Most earth retaining problems involve the movement of the structure away from the soil
mass, known as the "active condition". Terzaghi (1920) demonstrated the significance of
lateral earth pressures developed on structures that experience no movement, and termed
them "at-rest earth pressures". Note the at-rest condition, portrayed by figure 3-9(b), where
the wall rotation is equal to zero (i.e., the condition for zero lateral strain).
Figure 3-9 shows that:
- as the wall moves away from the soil backfill (see figure 3-9(c)), the active
condition develops and the lateral pressure against the wall decreases with
wall movement, until the minimum active earth pressure force is reached;
- as the wall moves toward (i.e., into) the soil backfill (see figure 3-9(a)), the
passive condition develops and lateral pressure against the wall increases with
wall movement until maximum passive earth pressure is reached - the
intensity of the active/passive horizontal pressure, which is a function of the
applicable earth pressure coefficient, depends on wall movement, as it affects
the degree of shear strength mobilized in the surrounding soil; and

- at-rest pressure develops when the wall experiences no lateral movement,

which typically occurs when the wall is restrained from movement, as in the
case of a stope (see figure 3-9(b)).
Chapter 3 - Review of Design Methodologies 35

Figure 3-9: Relationship of earth pressures to wall movements

Coefficient of Lateral Earth Pressure K

- Ko at rest

- Ka active

20 25 35 40
Friction Angle (deg)

Figure 3-10: Comparison of at-rest earth and active earth pressure coefficients

3.2.4.2 Calculating Lateral Earth Pressure Coefficients

Lateral earth pressure is related to vertical earth pressure by coefficients, termed as

follows:
- active earth pressure coefficient (Ka);
- passive earth pressure coefficient (Kp); and
- at-rest earth pressure coefficient (Ko).

Lateral earth pressure is equal to vertical earth pressure multiplied by the appropriate
earth pressure coefficient.
 36
Chapter 3 - Review of Design Methodologies

Since soil backfill is typically granular material, such as crushed rock, sand, silty sand,
sand with gravel, etc., the backfill material that exerts pressure on the wall can be treated as
coarse-grained non-cohesive material.

3.2.4.2.1 Active and Passive Earth Pressure Coefficients

When discussing active and passive lateral earth pressure, there are two relatively
straightforward classical theories (among others) in widespread use: Rankine earth pressure,
and Coulomb earth pressure.
The Rankine theory assumes that:
there is no adhesion or friction between the wall and soil;
lateral pressure is limited to vertical walls;
failure (in the backfill) occurs as a sliding wedge along an assumed failure
plane defined by (j);
lateral pressure varies linearly with depth, and the resultant pressure is located
one-third of the height above the base of the wall; and
the resultant force is parallel to the backfill surface.

The Coulomb theory is similar to the Rankine theory, except that:

there is friction between the wall and soil and accounts for this by using a soil-
wall friction angle of 8 (8 ranges from §12 to 2cp/3 and a 5 equal to 2<J>/3 is
commonly used);
lateral pressure is not limited to vertical walls; and
the resultant force is not necessarily parallel to the backfill surface because of
the soil-wall friction value 5.

The Rankine active and passive earth pressure coefficients for a horizontal backfill
surface are calculated as follows:

1 - sin(fz))
(Active) —» Ka = = tan ( 4 5 - ^ / 2 )
2
3-12
1 + sin(c^)
Chapter 3 - Review of Design Methodologies 21

1 + sin(^)
(Passive) -> Kp = tan (45 + ^/2)
2

3-13
1 - sin(^)

The Coulomb active and passive earth pressure coefficients are more complex
expressions that depend on the angle of the back wall, the soil-wall friction value, and the
angle of the backfill. The Coulomb active and passive earth pressure coefficient for the
specific case of a vertical back wall angle and horizontal backfill surface (not shown) yields
results equivalent to the Rankine method (equations 3-12 and 3-13).

3.2.4.2.2 At-Rest Coefficient

Generally, at-rest earth pressure is the horizontal component of the in situ stress state, or
the horizontal pressure acting on an earth retaining structure. At-rest lateral earth pressures
can be shown to be some multiple of the vertical stress at any point, illustrated in the
following ratio (Terzaghi, 1920):

3-14

Where:
is the horizontal pressure;

is the vertical pressure; and

Ko is the coefficient of earth pressure at-rest.

A more specific definition is offered by Bishop (1958): "The coefficient of earth

pressure at-rest, is the ratio of the lateral to the vertical effective stresses in a soil
consolidated under the condition of no lateral deformation, the stresses being principal
stresses with no shear stress applied to the planes on which these stresses act" or:

3-15

Where:
- a'h is the horizontal principal effective stress; and
Chapter 3 - Review of Design Methodologies 38

- a'v is the vertical principal effective stress.

Existing published relationships depend upon the soil's engineering values to calculate
the at-rest earth pressure coefficient. One common earth pressure coefficient for the at-rest
condition used with granular soil is called the "Neutral Earth Pressure Method", derived by
Tschebotarioff(1973):

K = — 3-16
0
l-v

Where:

- v is the Poison's ratio.

The assumption is that the force of gravity affects the elastic mass of material and
therefore lateral movement is prevented.
A n alternate solution for Ko can be found via Jaky's (1944) equation:

K =l-sm(0)
o 3-17

Where:
- <j) is the soil friction angle value.

3.2.5 Maximum Vertical Load Exerted by Backfill on the Floor of the Stope

After determining the maximum horizontal pressure or lateral thrust a _Max,

x shown in
equation 3-9, on the wall of the stope, the next step is to find the maximum vertical pressure
exerted by the backfill. From the (previously defined) coefficient of lateral earth pressure K,
corresponding with the horizontal to vertical stress ratio, maximum vertical stress can be
found using equation 3-18:

^ y _Max
rl 3-i8
P-K-\M$)

Strike length of sufficient magnitude allows for the hydraulic radius to be considered as
one-half of the stope span. The maximum vertical stress will be:
Chapter 3 - Review of Design Methodologies 39

3-19
y _Max
2-#-tan(^)

3.2.5.1 Terzaghi's Formulation

Another approach to estimating the vertical load exerted by backfill material is found in
Terzaghi et. al. (1948). Terzaghi derives a solution to the first order linear differential
equation:

5^ 2-c 2• K• a • tan0 3-20

dy
r—

Where:
is cemented rockfill cohesion.

For cohesionless soils, the solution to the first order differential equation yields the
equation for maximum vertical stress:

y_Max (*) = 3-21

2-.K-tan(0)

3.2.5.2 Blight's Formulation for Inclined Stopes

Blight (1984) proposed an approximate solution for the stresses in an inclined planar
stope. This solution applied to the design of inclined rectangular stopes, produces the
equation:

y _Max •sin(/7)
2-K- tan(^) 3-22

Where:
P is the inclination of stope.
Chapter 3 - Review of Design Methodologies 40

3.2.6 Vertical Load of Unconsolidated Material as a Function of the Material's Height

Employing the maximum vertical load rather than the actual vertical load, dependent
upon the height of the backfill, overestimates the strength required to withstand the load on a
sill mat. This in turn leads to an increase in cost. For that reason the preferable application of
vertical loads is as a function of the backfill height.
There are numerous formulations of silo theory, two of which are discussed here. Silo
theory only applies to vertical silos and thus is not valid for inclined stopes. Silo theory,
however, can enhance our understanding of how vertical stresses act within open stopes.
Janssen's method, developed in 1895, and Reimbert's method, developed in 1953 (and
subsequently further developed in 1976), are discussed next.

3.2.6.1 Th e Janssen Meth od - Silo Th eory

Janssen's method is based on the equilibrium of a thin horizontal slice, expressed as a

differential equation (3-23).

Figure 3-11: Differential slice in a silo

da K • tan(^) • tr 3-23
dy • + HR—--y
: = o
Chapter 3 - Review of Design Methodologies 41

This yields a solution of the form:

y • HR K-ian(<f))-z
CT(Z) = 1-exp 3-24
K-tan(<j>) HR

or,

CT (Z) = CT y _Max 1 -exp 3-25

HR

Where:
Y unit weight of the uncemented rockfill material;
HR hydraulic radius of the stope;
K coefficient of lateral earth pressure;
b Mid-span of the stope (shown in figure 3-11):
rockfill friction angle; and
z rockfill height.

The overall implication of this equation is that below a certain depth in the granular
material, all weight supported by the stope walls, via friction, generates either an active or a
passive state of stress (as illustrated above).

3.2.6.2 The Reimbert Method - Silo Theory

Reimbert's method is a modification of the Janssen method based on empirical

observations. The Reimbert method assumes that the vertical stress, which is the difference
between the hydrostatic vertical stress and the load balanced by the frictional force
(asymptotic with depth), can be represented by an experimental curve. For a silo undefined
between two parallel vertical walls the vertical stress at depth (z) is expressed as:

()
-i"
z +1 >
1 2

[Aij

Where:
Chapter 3 - Review of Design Methodologies 42

Ai is the characteristic abscissa corresponding to the silo.

The characteristic abscissa is defined as:

2-L
Ai = 3-27
n • tan(^) • K

Combining equations 3-26 and 3-27, yields:

-i-i
( n • tan(^) • K- z
+1 3-28
2TZ
V J

3.3 Stability of Cemented Rockfill Sill Mats -Analytical Solution

One of the key components of sill mat stability, analyzed previously, is the vertical load
exerted by the backfill material. The vertical load is a crucial input parameter of the sill mat
stability analytical solution. The specific modes relevant here are: block caving, flexural
failure, block sliding (shear failure parallel to the hanging wall within the fill), and rotation
failure (tensile/shear failure of the roof as it rotates away from the flat dipping hanging wall).
Schematics of these failure modes can be found in Chapter 5.

3.3.1 Caving Failure

Assuming that caving would extend to a stable arch height L/2 (for a semi-circular arc),
all unreinforced sills would form a depth, d > L/2, and caving would occur when (Mitchell,
1991):

L-y>S-a,/Tc 3-29

Where:

tensile strength of the cemented sill;

unit weight of the material; and

stope span.
Chapter 3 - Review of Design Methodologies 43

The driving forces represented by the weight w of the arc of radius L/2 are defined as:

w=
8 •r 3-30

Sill mat tensile strength a over the entire hanging wall to footwall span L, represents
h

the resisting force. The factor of safety (F.S.) is derived by the forces resisting movement
over the driving forces:

8-CT,
F.S.= 3-31
y-n-L

3.3.2 Flexural Failure

A wide, thin sill mat is susceptible to flexural failure due to the relatively low tensile
strength of cemented backfill. Using standard flexural formulae for a fixed-end uniformly-
loaded beam, failure is predicted when (Mitchell, 1991):

2-(cr +cr )
3-32
t c

>
w

or:

2-(<X,+CT ) C

> 3-33
(7 +d-y
v

Where:
tensile strength of the cemented sill;

a c
horizontal confining stress;
w uniform loading, which includes the self-weight of the sill mat;
L stope span; and
d sill depth.
Chapter 3 - Review of Design Methodologies 44

3.3.3 Sliding Failure

From equilibrium, block sliding of the sill as a result of side shear failure occurs when
(Mitchell, 1991):

f - ^ , d \ 3-34
(a +d-y)>2-
v

sin (/3)
2

Where:

shear strength of the cemented sill; and

P dip of the stope.

3.3.4 Rotational Failure

Rotational failure is most likely to occur when the shearing resistance at the hanging
wall contact is too low, as a result of poor quality hanging wall rock and/or low dip angles.
Separation subsequently occurs. For low dip angles, an approximate prediction can be made
by assuming the sill mat tensile failure results in a complete separation in hanging wall
contact (i.e., a gap forms).
Rotational failure develops when (Mitchell, 1991):

d a. 3-35
(cr v + d • y) >
2 • L • (L - d • cot(/3)) • sin (J3) 2

Technically, this formula has an embedded factor of safety of 2.0. The formula should
therefore be of the form:

d2
-a. 3-36
(a +d-y)>
v

L-(L-^-cot(/3))-sin (/3) 2

3.3.5 Strength Properties

The tensile strength of the fill is crucial in its capacity to resist damage due to blast
vibration, and to resist failure where undercutting is effected. Tensile strength of the fill is
Chapter 3 - Review of Design Methodologies 45

limited to, and is primarily a function of, binder content. Resistance to blast vibration damage
is a function of dynamic tensile strength, which can be approximated as 5% to 15% of
unconfined compressive strength. Arioglu (1983) concluded that tensile strength of cemented
aggregate fill occurs at approximately 15% of UCS. Smith (1982) concluded that tensile
strength of cemented tailings fill occurs at approximately 12% of UCS.
Y u (unpublished) states that the actual strength of CRF placed in a mine will be
approximately 2/3 of the laboratory value that is obtained from standard 6 inch diameter
concrete test cylinders, but will be about 90% of the value obtained from 12-inch diameter
cylinders.
For stability analysis of CRF sill mats, this study uses a tensile strength of 10% of UCS.
A l l other strength properties were obtained using Mohr-Coulomb stress strength criteria.
The strength properties used to calculate the factors of safety are obtained by the
analytical equations:

Cohesion, c U£S_ ^ + ^ 3 3 ?

2 2

Tensile Strength, <r t

10

Shear Strength, r t C + CT / ; -tan(^) 3

" 3 9

Normal Stress, a„ - i • K • y • d • s i n ( J3)

a
2 3 4
°

3.4 Empirical Stability Design Meth ods

This thesis also addresses the stability of open stopes adjacent to backfill. Empirical
stability design methods are the basis for the development of a solution for stopes with
backfilled abutments.
Chapter 3 - Review of Design Methodologies 46

3.4.1 The Stability Graph Method^

The Stability Graph Method for open stope design was initially proposed by Mathews, et
al. (1981), depicted in figure 3-12, and subsequently modified by Potvin (1988), using 242
case histories (176 unsupported, 66 supported) and by redefining some of the adjustment
factors. The result was the modified stability number N ' . The influence of cable bolt support
was re-examined by Potvin & Milne (1992), depicted in figure 3-13, and Nickson (1992),
figure 3-14, to arrive at the Modified Stability Graph. Stability was qualitatively assessed as
being: stable, potentially unstable, or caved.
Research at the University of British Columbia (1988) quantified the degree of slough by
the introduction of the term "Equivalent Linear Overbreak/Slough," ELOS, (Clark, 1998) as
depicted in figure 3-15. Empirical estimation of wall slough employs:
- Stability Number N (or N ' , i.e., modified); and
- Adjusted Rock Mass Rating with respect to the Hydraulic Radius (Clark,
1998).

1000

0.1-I 1 1 1 1 1

0 5 10 15 20 25
Shape Factor, S (m)

Figure 3-12: Mathews Stability Graph (Mathews et al., 1981)

Chapter 3 - Review of Design Methodologies 42

1000

0 5 10 15 20 25
Hydraulic Radius, HR (m)

Figure 3-13: Modified Stability Graph (Potvin and Milne, 1992)

H Y D R A U L I C RADIUS, HR (m)

Figure 3-14: Modified Stability Graph with support (Nickson, 1992)

Chapter 3 - Review of Design Methodologies 48

Numbers showing measured ELOS values

0.1 L
- 1
' ' 1

0 5 10 15 20

H Y D R A U L I C R A D I U S (m)

Figure 3-15: E L O S dilution design method (Clark, 1998)

3.4.2 The Stability Graph Method - Radius Factor f

The Stability Graph, later augmented by Mah (1997), was first introduced in terms of
Radius factor, by Milne (1997). Radius factor accounts for complex mining geometries
whereas Hydraulic Radius is restricted to rectangular surfaces (Area/Perimeter). The radius
factor has the same form as hydraulic radius; however, instead of being based on the average
of four measurements of the supporting abutments, several measurements are taken at small
angular increments, as shown in figure 3-16(b). This allows greater flexibility in assessing
the geometry of the support potential at a specific point on the surface (Milne, 1997),
involving complex shapes, providing "weightings" (i.e., RF values).

The Stability Number N ' employed in figure 3-16(a) is similar to that of figure 3-15.
The only significant difference is in the calculation of stope surface geometry, which
employs the radius factor parameter.
t from "Pakalnis et al.", (1998).
Chapter 3 - Review of Design Methodologies 4£

MODIFIED STABILITY GRAPH (ELOS) - 9 6 0 D S

liilllio! I! n i t i>

Figure 3-16: Empirical estimation (a) of wall slough (ELOS) expressed in terms of
Radius Factor (b)

3.5 Software and Device Utilized

3.5.1 Flac 2D
- Two Dimensional Explicit Finite Difference Method* 1

Flac 20
(version 4.0) is the numerical modeling tool used in this study to examine the
load exerted by backfill. Flac 2D
(Fast Lagrangian Analysis of Continua) is a two-dimensional
explicit finite difference program used in engineering mechanics computations.
The program can simulate the behavior of structures built of soil, rock, or other
materials, which may undergo plastic flow when their yield limits are reached. Materials are
represented by elements, or zones, which form a grid that can be adjusted to fit the shape of
the object to be modeled.
Each element behaves according to a prescribed linear or nonlinear stress/strain law in
response to the applied forces or boundary restraints.
The material can yield and flow and, i f set to do so, the grid can deform and move the
material that is represented. Plastic collapse and flow are modeled very accurately using the
explicit Lagrangian calculation scheme and the mixed-discretion zoning techniques in
Flac .2D
Flac 2D
is also the numerical modeling tool used to derive the design charts for
stability of the C R F sill mat. Flac 2D
is an effective tool for modeling non-linear gravity
Chapter 3 - Review of Design Methodologies 50

driven materials like uncemented and cemented rockfill. Within appropriate quality control
guidelines, the material (of required strength for a self-weight supported structure like the
CRF sill mat) should behave as an inelastic material, once cured. Even so, the two
2D

dimensional version of Flac will overestimate results given that the third dimension is
considered to be infinite, and therefore Flac is considered to provide a conservative
estimate.
At Musselwhite, the stope strike length, where sill mats were constructed, is typically
three or more times greater than the stope height, and the use of plain-strain numerical
modeling is effective and practical, for understanding its behavior. A n essential aspect of
modeling backfill using Flac lies in the fact that backfill is deposited as a material with no
initial stress state. The only stress comes from its own weight, which allows the distribution
of vertical loads to be a function of Hangingwall (HW) and Footwall (FW) host-rock/backfill
interactions, and backfill properties.
Another important feature of Flac , for the purpose of this analysis, is its capacity to
2D

display actual deformations of grid zones caused by material failure under different stress
states and material properties. The explicit, time-marching solution permits analysis of
progressive failure and collapse, significant phenomena in studies of mine design.
It is also possible to perform parametric studies, conveniently, by restarting the analysis
at a previous state. Results are given in Chapter 5.
t l technical source: Itasca Consulting Group Ltd. (2002)

3.5.2 Map3D - Three Dimensional Boundary Element Method^ 2

Map3D is a comprehensive, fully three-dimensional rock stability analysis package.

Map3D can construct models as well as analyze and display displacements, strains, stresses,
and strength factors. Map3D is effective for modeling rock engineering design problems
involving both large tabular orebodies and irregular massive excavations.
The program features:
- full three-dimensional stress analysis;
- tabular stress analysis with yielding pillars;
- elastic, non-linear, creep and thermal/fluid flow options;
- fault slip and fracture analysis capability;
Chapter 3 - Review of Design Methodologies 51

- simulation of stiff dykes, weak schist zones and backfill;

- external loading effects;
- structural support placement; and
- seismic database visualization.

Models may be made of rock slopes, open pits, tunnels, fractures, and underground
excavations, with yielding (i.e., non-linear) zones of different moduli (e.g., stiff dykes or soft
ore zones) and loads resulting from steady-state thermal/fluid flow. Excavations can be
intersected by multiple discrete faults (i.e., non-planar and gouge filled) that slip and open.
Map3D can simulate ground support elements such as arches, steel sets, props, thick liners,
chalks, strong backfill, etc.
Map3D is based on three-dimensional boundary element formulation. Both displacement
discontinuity and force discontinuity (fictitious force) formulations are available in the
software package. Special proprietary elements are used for simulation of thermal/fluid flow
and non-linear effects. Because automated lumping is built into the program, Map3D can
accommodate more than 300,000 elements (and over 1,000,000 degrees of freedom) using
merely desktop computers. Larger models can be created as required..
t2 technical source: Mine Modeling Pty Ltd. (2004)
Map3D-SV (standard version), release 5.0, is used in this study to calculate the
displacements on the stope wall for different strike lengths. The results and analysis are given
in Chapter 6.

3.5.3 NeuroSh ell Predictor - Neural Networks fS

Neural network technology mimics the brain's own problem solving system. Just as
humans apply knowledge gained from past experience to new problems or situations, a
neural network takes previously solved examples to build a system of "neurons" that makes
new decisions, classifications, and forecasts.
A Neural network looks for patterns within training sets of data, then learns these
patterns, and develops the ability to correctly classify new patterns, or to make forecasts and
predictions. Neural networks excel at problem diagnosis, decision making, prediction,
Chapter 3 - Review of Design Methodologies 52

classification, and other problems where pattern recognition is vital but precise
computational answers are not required.
The neural network begins by evaluating linear relationships between the inputs and the
output. Weights are assigned to the links between the input and output neurons. After those
relationships are identified, neurons are added to the "hidden layer", so that non-linear
relationships can then be searched for. Input values in the first layer are multiplied by the
weights and passed to the second (hidden) layer. Neurons in the hidden layer "fire," i.e,
produce outputs that are based on the sum of the weighted values passed to them. The hidden
layer passes values to the output layer in the same fashion, and the output layer produces the
desired results, i.e., its predictions.
The network "learns" by re-adjusting the weights between layers. The answers the
network produces are continually compared with the correct answers, and each time the
connecting weights are adjusted slightly in the direction of the correct answers. Additional
hidden layer neurons are added as needed to isolate structures in the data set.
Eventually, i f the problem can be learned, a stable set of weights evolves and will
produce useable answers for all sample decisions, or predictions. The real power of neural
networks becomes evident when the trained network is able to produce useable results for
data that the network has never "seen" before.
t3 technical source: Ward System Group Inc. (1997)
NeuroShell Predictor version 5.1 is used in this study to compare ELOS (Equivalent
Linear Overbreak Slough) data for transverse primaries, or isolated stopes, against
longitudinal stopes, where backfill is adjacent to the exposed wall. Results are given in
Chapter 6.

3.5.4 Unwedge - Underground Wedge Stability Analysis* 4

Unwedge (version 2.2) is a three-dimensional stability analysis and visualization

program for underground excavations in rock, containing intersecting structural
discontinuities. Safety factors are calculated for potentially unstable wedges so that support
requirements can be modeled using various types of pattern and spot bolting, and shotcrete.
Unwedge can perform safety factor analysis, in accordance with reinforcements employed,
and interpret the results.
Chapter 3 - Review of Design Methodologies 53

The graphical data interpreter provides a rich set of tools, including 3D animation,
providing for convenient display of wedges surrounding the excavation.
Unwedge incorporates enhanced support models for bolts, shotcrete, and support
pressures, and has the ability to optimize tunnel orientation. There is also an option to
examine different combinations of three joint sets based on a list of multiple joint sets.
Unwedge uses a new analysis engine based on Goodman and Shi's block theory, including
the ability to evaluate induced stress around the excavation and its effects on stability.
Unwedge includes new strength models such as Barton-Bandis and Power Curve, and the
ability to improve the scaling and sizing of wedges.
t4 technical source: Rocscience Inc.(2004)
Unwedge is used in this study to represent structural blocks, and support pattern
performance in a typical production tunneling scenario.

3.5.5 Dips - Graphical and Statistical Analysis of Orientation Data 14

Dips is designed for the interactive analysis of orientation-based geological data and
allows the user to analyze and visualize structural data, following the same techniques used
in manual stereonets. Additionally, it has several computational features, such as statistical
contouring of orientation clustering, mean orientation and confidence calculation, cluster
variability, and qualitative and quantitative feature attribute analysis.
Dips is designed for analysis of parameters related to the engineering analysis of rock
structures, however, the free format of the Dips data file permits analysis of any orientation-
based data (Rocscience Inc., 2004).
+4 technical source: Rocscience Inc.(2004)
Dips version 5.0 is used in this study in the modeling of Musselwhite's fabric analysis
(see 4.2).

3.5.6 Methods Ground Control Assessment - Cavity Monitoring System Surveys fS

Optech's Cavity Monitoring System (CMS) is an advanced, reflectorless, laser-based

auto scanning system that provides efficient and accurate measurement of dangerous and
inaccessible cavities. CMS collects thousands of data points per minute. The data is used to
Chapter 3 - Review of Design Methodologies 54

determine stope volume and stope dilution, sloughing/backfill volumes and mine
measurements, and to create detailed profiles of pillars, orepasses, raises, drifts, and
drawpoints. Mine surveyors obtain detailed 3D survey data images of underground
production stopes. Mine planners use the data to determine the exact area mined, and
quantify volumes of mined/unmined ore. Geologists use the data to match month end
reconciliation. Rock mechanics then use the data to predict rock failure (e.g., hangingwall
and footwall).
After a set of blast rings is designed, blasted, excavated and surveyed, the results are
compared to the original blast design. The results are then analyzed for tonnage of lost ore
(underbreak), tonnage of waste rock or unplanned dilution (overbreak), and accuracy of
grades sent to the mill (reconciliation). Ultimately, the results are analyzed for the success, or
lack of success, of drilling and blasting, and to predict dilution in future stopes (Optech Inc.).
t5 technical source: Optech Incorporated.(2004)
Chapter 4 - Database 55

4 DATABASE

4.1 Intact Strength

A total of 32 additional uniaxial compressive strength tests were carried out, augmenting
the existing 12 results on both iron formation and intermediate-mafic/felsic volcanic rocks.
Four more samples of the iron formation were tested with extensometers, allowing for elastic
modulus determination. The tests were performed as recommended in the ISRM Suggested
Methods (Brown, E.T. (Editor), 1981). The test result averages are presented in Table 4-1
and Table 4-2.

Table 4-1: Mean uniaxial compressive strength, standard deviation and ISRM hardness

Rock Type
Number of UCS Standard Deviation ISRM Hardness
Iron Formation
Samples (MPa) (MPa)
4ea 7 194 46 Very Strong Rock - R5

4f 12 92 31 Strong Rock - R4

4e 3 116 15 Very Strong Rock - R5

4b 5 156 46 Very Strong Rock - R5

4e4f 3 86 11 Strong Rock - R4

Intermediate-Mafic
Volcanic
AVol 9 181 43 Very Strong Rock - R5

Intermediate-Felsic
Volcanic
BVol 5 117 22 Very Strong Rock - R5
Chapter 4 - Database 56

Table 4-2: Mean Young's Modulus (E) and Mean Poisson's Ratio (v)

Rock Type Young Modulus Standard Poisson's ratio Standard

E Deviation V Deviation
(GPa) (GPa) () ()
4ea 84 13 0.21 0.03
4f 72 7 0.21 0.02
AVol 85 N/A 0.23 N/A
BVol 79 N/A 0.24 N/A

The tests were performed using an electro-hydraulic servo-controlled stiff testing

machine, used for comparing the various physico-mechanical properties of rocks under
different conditions of loading to ISRM standards.

4.2 Fabric Analysis

Generally, three joint sets are prominent in all rock types: parallel to the ore,
perpendicular to the ore, and flat. Some random, typically vertical, faulting also exists as
does the 4f (garnet-biotite-schist), which is also vertical and parallel to the ore zone. A
structurally intense domain seems to be characteristic of the BVol (footwall drift) in which
the three prominent joint sets exist, except with tighter spacing, resulting in blocky ground.
The A V o l has characteristically tight foliation (north-south, vertically dipping), which ranges
in spacing from below one centimeter to five centimeters, or more. Flat and crosscutting
joints in the A V o l are spaced from one to two meters or more as opposed to the BVol.
Ground conditions at the back of the A V o l are generally very good. Because there is tight
foliation, the walls are prevented from buckling.

The ore zone consists of three predominant joint sets in dip/dip direction, global
orientation format:
83° / 90° (north-south, vertical joints);
- 75° / 177° (east-west, vertical joints); and
2 5 ° / 2 1 0 ° (flat jointing).
Chapter 4 - Database 57

Figure 4-1 portrays the equal area projection stereonet, obtained by employing
Rocscience's Dips program on the three major joint sets.

Figure 4-1: Equal area stereonet representing the three major joint sets

In the ore zone, the three joint sets typically have an average spacing of 1 - 1.5 meters.
The joint sets have slightly rough surfaces and little or no separation with hard joint wall
rock. In the B V o l , the joints are typically spaced more tightly and have thin chloritic-coated
surfaces.
There is the potential for wedge or block failures in these areas, both when drifting and
when longholing. Wedge failures may occur at the hangingwall or the back of the open stope.
Wedges in the drifts are normally supportable with standard 1.8 meter (#6) support.
Figure 4-2 portrays a 10-tonne wedge, supported by standard 1.4 meter by 1.4 meter
pattern, fully-grouted resin rebar. The factor of safety for the standard bolting pattern
employed at Musselwhite produces a high degree of confidence. Stability analysis was
effected using Rocscience's Unwedge program on three major joint sets (previously defined)
with 1 meter spacing each, considering no cohesion and a 35 degree friction angle for the
joint sets' infilling material. The tunnel trends to the north.
Chapter 4 - Database 58

UNMEDCE - STABILITY ANALYSIS

P e i c s p o o t rue uii<t|

/
£01

lit U.I'M

IJ'.K
1-fK.L MP A IfOMHi
TO V.Q13H
IIM I.HIUrl Kik'i\
TO HOTftT J"

> S«Uct button to l n a t « l 1 '*uj»i»or't (CESC1 to oxit)

Figure 4-2: Wedge formed on the back of a drift running north-south

Failures in the open stope due to random faulting and jointing produce the potential for
unplanned dilution. Some failures of up to 4.5 meters deep have been observed occasionally,
due to the poor hangingwall rock mass quality 4f, which runs parallel to the hangingwall.

4.3 Rock Mass Analysis

Bieniawski's Geomechanics Classification (1976), or the Rock Mass Rating (RMR), of

the stopes varies between 40% - 70% for the hanging wall, 60% - 80% for the footwall, 65%
- 70% for the back of the stope, and 70% - 75% for the ore itself. Table 4-3 illustrates an
example of the R M R where R4 rock strength represents uniaxial compressive strength
between 100 and 250 MPa, and the Rock Quality Designation (RQD) provides a quantitative
estimate of rock mass quality from drill core logs. RQD is defined as the percentage of intact
core pieces longer than 100 mm (4") within the total length of the core. Barton's Tunneling
Quality Index Q, was also measured, or derived, for each stope, to estimate its stability and
determine the approximate degree of hangingwall and footwall slough.
Chapter 4 - Database 59

Table 4-3: RMR and Q' of a typical footwall stope

Parameter Description Rating (%)

Strength R4 12
% RQD 60-70% 12-14
Joint Spacing 0.25-0.5m 11-17
Joint Condition Slightly rough/smooth 15-17
Ground Water Dry 10
Total R M R 60-70%
Total Q' 5.9-18.0

The following table summarizes the average range of Rock Mass Rating (Bieniawski,
1976) for the different rock types found at Musselwhite throughout the life of the mine.

Table 4-4: Typical RMR ranges for the different Musselwhite rock types

Rock Type BVol AVol 4f 4b

120 12 12 12
Strength
(240 MPa) (180 MPa) (100 MPa) (150 MPa)

18 18 17-10 17-10
RQD
(75-90) (75-90) (50-90) (50-90)

18-13 18-13 11-8 11-8

Spacing
(0.3-lm) (0.3-lm) (50-300mm) (50-300mm)

17-12 17-12 15-7 15-7

Condition
(tight-sep<lmm) (tight-sep<lmm) (tight-slick) (tight-slick)

10 10 10 10
Ground Water
(dry) (dry) (dry) (dry)

75-65 75-65 65-47 65-47

RMR
(typical: 70) (typical:70) (typical:60) (typical:55)
Chapter 4 - Database 60

Some of the R M R points were derived by interpreting geological sections using the
Vulcan geological database. Thus, the contact rock type and thickness was associated with
the R M R range in each case (see Table 4-4).

Figure 4-3: Geological section obtained using Vulcan database

4.4 Stress

The following table summarizes the stress state of Musselwhite orebodies at a depth of
500 meters (Arjang et. al., 1997):

Table 4-5: Musselwhite's far field stress state

Type In-situ Stresses e.g., at 500m

CTh (E-W)
Max 2.5 x a MPa/m depth
v
38MPa

0-h (N-S)
Min 1.5 xaV MPa/m depth 23MPa

rjv(Vertical) 0.03 MPa/m depth 15MPa

4.5 Numerical Modeling - Constitutive Models and Material Properties

Two different constitutive models were used to characterize both material types, host
rock and backfill. Host rock material was assigned an elastic and isotropic model. This model
provides the simplest representation of material behavior. The model is valid for
Chapter 4 - Database 61

homogeneous, isotropic, continuous materials that exhibit linear stress-strain behavior.

Backfill material was represented by the Mohr-Coulomb model, which is the conventional
model used to represent shear failure in soils and rocks.

Table 4-6: Mohr-Coulomb stress-strain and strength parameters for backfill and host
rock

Parameter Rockfill Host Rock

Unit weight (kJN/m3), y 19.67 26.48
Cohesion (MPa), c 0.0 -
Friction angle (degree), § 37.0 -
Tensile strength (MPa), at 0.0 -
Young modulus (MPa), E 300.0 74,000
Poisson's ratio ( ), v 0.25 0.22
Bulk modulus (MPa), K 200 44,047
Shear modulus (MPa), S 120 30,327

4.6 Empirical Stope Design - Musselwhite's Stability Database

The stability of underground excavations at Musselwhite is determined by three

components:
- characterization of the surrounding rock mass;
- geometry and relative orientation of the opening; and
- induced stress-state about the opening.

The first component is the identification and quantification of the rock mass properties,
e.g., rock type, support, R M R , and/or Barton's Tunneling Quality Index Q. The second
component is the size of the planned opening and the dip of the stope, i.e., strike length and
slanted height of the opening. From this, the hydraulic radius (or radius factor) of the
exposed wall is determined. The third aspect considered is the degree of relaxation of the
stope wall and the major principal stress (versus minor principal stress) about the opening.
The modified stability number N ' is adjusted for the new stress conditions.
By identifying and quantifying rock mass properties and the planned geometry it is
possible to derive empirically the stability of the stope, using the Modified Stability Graph
Chapter 4 - Database 62

(Potvin, 1998) and/or determining the amount of wall slough using the Equivalent Linear
Overbreak Slough (ELOS) Graph (Clark, 1998).
Table 4-7 portrays Musselwhite's mine stability database, comprised of the 69 stopes
investigated. For each stope, the following data are indicated: length, height, dip, R M R , and
ELOS, for both hangingwall and footwall. The ELOS value was calculated by comparing the
drill layout and/or geological block model with the CMS survey profile for each stope. The
following table was derived specifically for this study:
Chapter 4 - Database 63

Table 4-7: Updated Musselwhite stability database

L HHW Dip™ Positive D i p ^ RMR™ QHW N'HW HW E L 0 S HFW DiPm Positive Dippw H.R.pw RMR™ QFW N'™ FW E L 0 S

S100T535 20.5 44.5 74.4 74.4 7.0 60.0 5.9 9.4 0.8 47.3 -65.0 115.0 7.2 70.0 18.0 35.9 0.5
S100t515 20.0 44.5 81.6 81.6 6.9 70.0 18.0 32.0 0.6 45.2 -77.1 102.9 6.9 70.0 18.0 35.9 0.4
S100T495 20.0 42.0 81.0 81.0 6.8 70.0 18.0 31.7 0.5 42.5 -77.5 102.5 6.8 60.0 5.9 11.8 0.6
S100T475 20.0 42.0 81.0 81.0 6.8 70.0 18.0 31.7 0.5 42.1 -80.0 100.0 6.8 70.0 18.0 35.9 0.3
S100T455 20.0 42.0 81.0 81.0 6.8 70.0 18.0 31.7 0.4 42.6 -76.6 103.4 6.8 70.0 18.0 35.9 0.3
S100T435 20.0 42.0 85.7 85.7 6.8 70.0 18.0 33.9 0.6 42.1 -84.6 95.4 6.8 70.0 18.0 35.9 0.3
S125T570 42.5 36.5 79.1 79.1 9.8 55.0 3.4 5.8 1.4 36.4 -80.0 100.0 9.8 70.0 18.0 35.9 0.4
S150T625 25.0 54.0 80.6 80.6 8.5 55.0 3.4 6.0 1.5 54.1 -80.0 100.0 8.5 70.0 18.0 35.9 0.5
S150T600 25.0 52.0 80.2 80.2 8.4 45.0 1.1 1.9 2.6 53.9 -72.0 108.0 8.5 70.0 18.0 35.9 0.4
S150L510SZ 40.0 32.2 -88.0 92.0 8.9 70.0 18.0 35.9 0.4 33.1 76.8 76.8 9.0 55.0 3.4 5.6 1.4
S175L610 47.5 29.0 75.0 75.0 9.0 70.0 18.0 29.0 0.4 28.4 80.7 80.7 8.9 45.0 1.1 2.0 2.3
S175L505SZ 22.5 23.4 -79.4 100.6 5.7 70.0 18.0 35.9 0.4 24.3 71.4 71.4 5.8 60.0 5.9 9.0 0.7
S175L940 12.5 30.5 68.3 68.3 4.4 70.0 18.0 26.0 0.3 28.7 -81.1 98.9 4.4 70.0 18.0 35.9 0.3
S175940CZ 12.5 30.2 67.8 67.8 4.4 70.0 18.0 25.8 0.3 28.3 -81.2 98.8 4.3 70.0 18.0 35.9 0.3
S200T785 20.0 32.8 90.0 90.0 6.2 50.0 1.9 3.9 1.7 32.8 90.0 90.0 6.2 55.0 3.4 6.8 0.8
S200L455 15.0 32.2 -83.0 97.0 5.1 70.0 18.0 35.9 0.4 32.3 82.0 82.0 5.1 60.0 5.9 10.6 0.5
S225T888 20.0 57.4 77.0 77.0 7.4 70.0 18.0 29.9 0.4 57.2 -77.8 102.2 7.4 70.0 18.0 35.9 0.3
S225T842 22.5 54.4 80.0 80.0 8.0 45.0 1.1 1.9 1.6 54.4 -80.0 100.0 8.0 70.0 18.0 35.9 0.2
M250L250 17.0 22.0 64.9 64.9 4.8 45.0 1.1 1.5 2.0 20.0 -83.9 96.1 4.6 70.0 18.0 35.9 0.3
M250T110 35.0 27.0 82.0 82.0 7.6 60.0 5.9 10.6 0.9 30.8 -60.3 119.7 8.2 60.0 5.9 11.8 0.5
M250T090 20.0 49.5 78.5 78.5 7.1 70.0 18.0 30.6 0.4 50.3 -74.6 105.4 7.2 70.0 18.0 35.9 0.3
S250T980 22.5 49.3 86.5 86.5 7.7 60.0 5.9 11.3 0.8 49.3 -86.5 93.5 7.7 70.0 18.0 35.9 0.5
S250T957 22.5 50.5 83.8 83.8 7.8 55.0 3.4 6.2 1.0 50.4 -85.4 94.6 7.8 60.0 5.9 11.8 0.7
S250T935 17.5 52.0 83.6 83.6 6.5 60.0 5.9 10.8 0.7 52.0 -84.0 96.0 6.5 70.0 18.0 35.9 0.4
S250L820 32.5 29.0 78.4 78.4 7.7 45.0 1.1 1.9 2.0 28.6 -82.9 97.1 7.6 60.0 5.9 11.8 0.6
S250L820 72.5 28.5 85.0 85.0 10.2 45.0 1.1 2.1 1.8 28.5 -84.0 96.0 10.2 55.0 3.4 6.8 0.9
S250L735 35.5 46.0 -86.8 93.2 10.0 55.0 3.4 6.8 0.7 46.0 86 8 66.8 10.0 70.0 18.0 34.4 0.4
S250L735 37.5 44.0 90.0 90.0 10.1 55.0 3.4 6.8 0.9 44.3 83.4 83.4 10.2 70.0 18.0 32.8 0.4
S250L550 80.0 39.7 -87.6 92.4 13.3 70.0 18.0 35.9 0.5 39.9 83.6 83.6 13.3 60.0 5.9 10.8 0.4
M275L440CZ 50.0 27.5 73.7 73.7 8.9 55.0 3.4 5.4 1.0 27.2 -76.0 104.0 8.8 55.0 3.4 6.8 0.8
M275L440 17.5 23.1 74.6 74.6 5.0 55.0 3.4 5.4 0.8 22.6 -79.5 100.5 4.9 60.0 5.9 11.8 0.6
M275L350 20.0 23.5 72.4 72.4 5.4 55.0 3.4 5.2 1.3 22.7 -80.1 99.9 5.3 60.0 5.9 11.8 0.5
275 POCZ 35.0 24.0 73.3 73.3 7.1 45.0 1.1 1.8 3.0 23.0 90.0 90.0 6.9 70.0 18.0 35.9 0.4
M275L260 35.0 27.0 80.4 80.4 7.6 70.0 18.0 31.5 0.4 26.6 -87.7 92.3 7.6 70.0 18.0 35.9 0.4
M275L048 55.0 52.0 78.0 78.0 13.4 70.0 18.0 30.3 0.5 50.9 -89.8 90.2 13.2 70.0 18.0 35.9 0.4
M275T003 37.5 55.0 76.1 76.1 11.1 70.0 18.0 29.5 0.4 53.4 90.0 90.0 11.0 60.0 5.9 11.8 0.7
M275L920 53.0 30.0 84.0 84.0 9.6 70.0 18.0 33.1 0.6 29.9 -86.0 94.0 9.6 60.0 5.9 11.8 0.5
M300L450 35.0 27.9 69.6 69.6 7.8 70.0 18.0 26.5 0.6 26.8 -77.5 102.5 7.6 70.0 18.0 35.9 0.3
M300L350CZ 105.0 31.6 68.5 68.5 12.1 45.0 1.1 1.6 2.4 30.7 -73.3 106.7 11.9 55.0 3.4 6.8 0.7
M300T162 45.0 54.6 80.4 80.4 12.3 70.0 18.0 31.5 0.4 55.2 -77.3 102.7 12.4 70.0 18.0 35.9 0.2
M300L152 35.0 55.7 78.1 78.1 10.7 70.0 18.0 30.4 0.4 54.7 -84.8 95.2 10.7 70.0 18.0 35.9 0.3
M300L092 45.0 55.7 79.1 79.1 12.4 70.0 18.0 30.8 0.5 54.7 -88.1 91.9 12.3 70.0 18.0 35.9 0.3
M330L250 25.0 38.1 84.6 84.6 7.5 70.0 18.0 33.4 0.4 37.9 90.0 90.0 7.5 70.0 18.0 35.9 0.3
M330L550 30.0 38.6 74.4 74.4 8.4 70.0 18.0 28.7 0.3 38.3 -75.9 104.1 8.4 70.0 18.0 35.9 0.2
M330L135 30.0 36.4 90.0 90.0 82 70.0 18.0 35.9 0.3 36.4 88.8 88.8 8.2 70.0 18.0 35.4 0.4
M330L035WA 50.0 37.0 90.0 90.0 10.6 70.0 18.0 35.9 0.4 37.1 -85.0 95.0 10.7 60.0 5.9 11.8 1.0
M330L070 86.0 38.0 -84.3 95.7 13.2 70.0 18.0 35.9 0.4 38.0 84.0 84.0 13.2 70.0 18.0 33.1 0.6
M375L235 44.0 36.0 90.0 90.0 9.9 70.0 18.0 35.9 0.2 36.0 90.0 90.0 9.9 70.0 18.0 35.9 0.2
M375L235 40.0 31.7 -88.3 91.7 8.8 55.0 3.4 6.8 1.6 31.7 86.5 86.5 8.8 70.0 18.0 34.3 0.3
M330L550 35.0 38.6 76.5 76.5 9.2 70.0 18.0 29.7 0.4 38.5 -77.0 103.0 9.2 70.0 18.0 35.9 0.3
M375L740 72.5 38.5 77.4 77.4 12.6 60.0 5.9 9.9 1.4 38.3 -78.5 101.5 12.5 60.0 5.9 11.8 0.5
M375L550 50.0 36.7 82.0 82.0 10.6 45.0 1.1 2.0 3.7 36.6 -83.3 96.7 10.6 70.0 18.0 35.9 0.3
M375L550 86.0 37.5 75.1 75.1 13.1 70.0 18.0 29.0 0.4 37.7 -74.0 106.0 13.1 70.0 18.0 35.9 0.3
M375L550 47.5 36.4 81.6 81.6 10.3 60.0 5.9 10.5 0.9 36.5 -80.8 99.2 10.3 70.0 18.0 35.9 0.3
M375L420CZ 48.0 37.4 77.9 77.9 10.5 45.0 1.1 1.9 5.5 37.1 -80.0 100.0 10.5 70.0 18.0 35.9 0.0
M375L235WA 37.5 35.8 -87.6 92.4 9.2 70.0 18.0 35.9 0.4 36.1 82.0 82.0 9.2 70.0 18.0 32.2 0.6
M375L175WA2 22.5 35.1 -89.1 90.9 6.9 70.0 18.0 35.9 0.4 35.2 85.2 85.2 6.9 70.0 18.0 33.7 0.6
M400L340WA 35.0 29.1 -88.7 91.3 7.9 60.0 5.9 11.8 0.7 29.2 84.1 84.1 8.0 70.0 18.0 33.2 0.3
M400L240WA2 57.5 29.2 -79.8 100.2 9.7 60.0 5.9 11.8 1.0 29.2 80.2 80.2 9.7 55.0 3.4 5.9 1.4
M400L340WA 37.5 29.1 89.1 89.1 8.2 70.0 18.0 35.5 0.5 29.2 84.5 84.5 8.2 70.0 18.0 33.4 0.2
J475L285SZN P4 37.5 35.3 81.0 81.0 9.1 70.0 18.0 31.7 0.3 35.6 -78.3 101.7 9.1 70.0 18.0 35.9 0.3
J475L285SZN S3 37.5 37.0 67.9 67.9 9.3 45.0 1.1 1.8 2.4 35.3 -76.5 103.5 9.1 60.0 5.9 11.8 0.8
J475L140SZN P3 35.5 35.4 78.9 78.9 8.9 60.0 5.9 10.1 1.3 37.0 -70.0 110.0 9.1 70.0 18.0 35.9 0.6
475PO 140 SZ 33.5 33.7 73.7 73.7 8.4 45.0 1.1 1.8 1.9 33.2 -76.7 103.3 8.3 70.0 18.0 35.9 0.4
J475L040 30.0 30.8 87.6 87.6 7.6 70.0 18.0 34.8 0.0 30.8 -88.8 91.2 7.6 70.0 18.0 35.9 0.2
J500L455SZN P3 15.0 35.8 77.7 77.7 5.3 70.0 18.0 30.2 0.4 35.4 -81.2 98.8 5.3 70.0 18.0 35.9 0.3
J500P3 32.5 35.8 74.9 74.9 8.5 55.0 3.4 5.5 1.5 35.4 -77.3 102.7 8.5 70.0 18.0 35.9 0.3
J500L370S1 32.5 36.2 71.6 71.6 8.6 60.0 5.9 9.0 1.1 35.8 -73.5 106.5 8.5 70.0 18.0 35.9 0.0
J500L255SZN P1 41.0 36.3 74.0 74.0 9.6 45.0 1.1 1.8 2.6 35.4 -80.0 100.0 9.5 70.0 18.0 35.9 0.3

4.7 Statistical Analysis of the Musselwhite Database

The following figures portray frequency of occurrence histograms from the Musselwhite
mine database. These histograms include a normal distribution curve for each parameter. The
histograms were sorted into hangingwall and footwall data, except for stope length, which
was considered as being unique for each stope.
Chapter 4 - Database 64

Figure 4-4: Statistical Analysis on Musselwhite stability database

Chapter 4 - Database 65

From the previous statistical analysis, it is obtained the ranges of stope configurations
(height, span and dip) which will be used in the following chapters as input parameters for
numerical modeling and neural network analysis.
Chapter 5 — Design Guidelines for Cemented Rockfill Sill Mats 66

5 DESIGN GUIDELINES FOR CEMENTED ROCKFILL SILL MATS

5.1 Introduction

Theories of cemented rockfill sill mat design demand the optimization of fill
requirements, achieving the strength needed, while meeting safety guidelines and minimizing
mining costs.
Cement is the largest material cost component of backfill. Therefore, strategies aimed at
reducing cement content are of the highest benefit to mining operations. This chapter defines
and compares the relevant analytical relationships and numerical modeling results of backfill
behavior on sill mat stability. The analysis is done via observations of practitioners and
researchers working in the mining field.
Note that more than 1000 numerical models have been executed, as a part of this thesis,
that directly relate to these analytical relationships, including models to determine vertical
stress and models to determine the stability of cemented rockfill mats.

5.2 Load of Backfdl

One of the key factors in assessing cemented rockfill stability is quantifying the vertical
stress acting on the top of the sill mat. In situ measurements and analytical predictions
derived by various researchers imply that arching of unconsolidated rockfill material
decreases the vertical load on the sill by transferring part of the total load to the hangingwall
and footwall. Methods to better estimate the true load lead to more reliable and cost effective
solutions in the design of sill mats. Overestimation of the vertical load produces unnecessary
expense by increasing the quantity of cement used and/or the height of the sill mat required
to withstand the supposed load. Underestimating can cause a premature failure of the sill mat
while mining is being conducted below it, resulting in ore lost, ore dilution, cycle
interruption, damage to equipment, and a compromise in personal safety.

Theoretical considerations derived from Janssen (1895), Terzaghi (1948), Reimbert

(1976), and Blight (1984) are compared with numerical modeling results to determine the
theoretical true load exerted by the fill. To analyze backfill load, first it is necessary to
determine accurately the coefficient of lateral earth pressure, K, an input variable of the
analytical equation. Failing to estimate accurately the value of K produces unreliable vertical
Chapter 5 - Design Guidelines for Cemented Rockfill Sill Mats

load calculations. The derived curve for K can be used when numerical modeling software is
not available and/or to compare vertical loads corresponding to different K values.

5.2.1 Coefficient of Lateral Earth Pressure K

This thesis determines the coefficient of lateral earth pressure K using numerical
modeling and proposes that an effective computational method is capable of representing the
behavior of cohesionless soils. Moreover, the software applied here can be used to
characterize the deformations of backfill using only the effect of gravity and its interaction
with the stope walls. The computational tool employed is the finite difference program
known as Flac . 2D

Figure 5-1 illustrates a Flac 2D

graph of stress measurements at a specific point within a
backfilled stope. The stope is backfilled in layers of 5 meters each, to a maximum height of
35 meters. The y-axis corresponds to the horizontal stress, a^, and the x-axis corresponds to
the vertical stress, Oyy. The resultant slope represents the coefficient of lateral earth pressure,
K.

FLAC (Version 4.00)

LEGEND

16-May-M 13:47
step 25092

HISTORY PLOT
Y-axis :
Ave.SXX ( 31. 1)
X-axis.
Ave. SYY (31. 1)

Figure 5-1: Coefficient of lateral earth pressure K for cohesionless material

Figure 5-2 displays the graph of the coefficient of lateral earth pressure at different
friction angles, comparing the results of analytical equations of the coefficients for the active
Chapter 5 - Design Guidelines for Cemented Rockfill Sill Mats 68

(Ka), and at-rest (Ko) cases, taking into consideration the settling effect (Ks), with the results
obtained from numerical modeling.

Coefficient of Lateral Earth Pressure K

Figure 5-2: Comparison of analytically and numerically determined coefficients of

lateral earth pressure

Figure 5-2 portrays the values obtained using numerical modeling. The values are
consistent with the active case, up to the point where the friction angle reaches 30 degrees. At
friction angles greater than 30 degrees, the K value becomes constant, and the curve tends
toward the at-rest condition. This behavior might occur because at lower friction angles
greater deformation of the material takes place. The rockfill material behaves as i f the wall is
in motion, re-forming itself by the action of gravity, and the material is supported by the
stope walls, generating the active state of stress. On the other hand, at friction angles greater
than 30 degrees the deformation tends to decrease and the interaction between rockfill and
the stope walls decreases accordingly, tending toward the at-rest state of stress.
Figure 5-3 and figure 5-4 portray the second order polynomial curve that best fits the
data obtained using numerical modeling (K Derived). The coefficient of lateral earth pressure
is plotted against the sine of the friction angle and the best fit curve determined.
Chapter 5 - Design Guidelines for Cemented Rockfill Sill Mats 61

Coefficient of Lateral Earth Pressure K

Sin(4>)

Figure 5-3: Coefficient of lateral earth pressure obtained using numerical modeling and
best fit curve derived

0.40 0.45 0.50 S i H (<()) ° 5 5

° 6 0
° 6 5

Figure 5-4: Coefficient of lateral earth pressure obtained using numerical modeling and
best fit curve derived for the rockfill friction angle range

Equation 5-1 represents the best fit curve:

K = 1 .4 • Sin (<p) - 2 • Sin {</>)+ \

2
5 A

Note that equation 5-1 is only applicable for cohesionless backfill material with friction
angles ranging from 0 degrees to 40 degrees.
Chapter 5 - Design Guidelines for Cemented Rockfill Sill Mats 10

Figure 5-5 portrays the variation of the slope (K) of the horizontal versus vertical stress
for different rockfill friction angles.

Horizontal Stress vs. Vertical Stress

at different Friction Angles

0.800 , — -

Vertical Stress ayy (MPa)

Figure 5-5: Horizontal vs. vertical stress at different rockfill friction angles

5.2.1 Vertical Stress Comparison using different Coefficients of Lateral Earth Pressure

Variations in vertical stress obtained via different coefficients of lateral earth pressure K
are portrayed by figure 5-6. It shows the plot of maximum vertical stress for the active case,
the at-rest case, and the derived case using numerical modeling while considering the
following backfill properties and stope configuration:
y = 19.67 kN/m 3

4 = 37.0°
L = 10.0 m
(3 = 90°
Chapter 5 - Design Guidelines for Cemented Rockfill Sill Mats

Vertical Stress C o m p a r i s o n
Using different K values

^ 600
CO
0.
s
— 500

<» 400
200 kPa difference
CB
k_
4-»
CO
—
ra
300
HHlifl jijijiiiiii
u
t

200
E lililpiiplipJIl ••Sp
3

E 100
"><
ra lliliPill

K active K at-rest KFIac

Figure 5-6: Vertical stress comparison for different K values

As portrayed in figure 5-6, there is a considerable difference between the maximum

vertical pressure using the active versus the at-rest coefficients of lateral earth pressure. The
variation of approximately 200 kPa can be attributed to the difference between the designs of
a stable versus an unstable sill mat. Results obtained using numerical modeling produce an
intermediate value, which falls between the active and the at-rest cases. The most important
outcome here is that the solution obtained using numerical modeling accounts for the
deformation of the material within the stope. The behavior of this cohesionless material is
affected by the magnitude of the friction angle at the contact between the stope walls and the
rockfill material.

5.3 Vertical Stress Using Numerical Modeling

5.3.1 Introduction

This section addresses the total load on the floor of the stope given that it is this load that
determines the stability of CRF sill mats.
Rockfill material deposited into stopes constitutes a granular, or pseudo-solid, mass, for
which equilibrium laws still apply. Given a plastic, homogeneous, and isotropic material, a
solution can be derived from numerical modeling results, and is here so derived for this case.
Chapter 5 - Design Guidelines for Cemented Rockfill Sill Mats 11

The problem defined by the equilibrium of confined rockfill material is complex, and
does not well fit the theory of elasticity. It is extremely difficult to determine accurately the
loads applied to the stope walls and floor via existing theories. A numerical modeling
solution capable of accounting for the friction that exists between the fill and the stope walls,
and so making possible comparison of the resultant curves with existing analytical equations,
is enormously valuable.
The process of modeling backfill as a function of the fill height proceeds as follows: a
five meter Hangingwall (HW) to Footwall (FW) layer of backfill is deposited on the floor of
the stope, the model is executed, and the vertical and horizontal loads are then measured.
Incrementally, five meter layers of backfill are added on top of previous layers. The model is
executed and loads are measured for each increment to the point where the stope has been
filled with seven layers of backfill material.
This analysis does not incorporate wall convergence, under the assumption that by the
time waste rock is backfilled, wall displacement has already occurred.
Chapter 5 - Design Guidelines for Cemented Rockfill Sill Mats 11

Figure 5-7 (a) and figure 5-7 (b) portray two models backfilled at two different stages:
25 meters and 35 meters of backfill. The vertical stress contour represents the stress exerted
by the cohesionless backfill material with a 37 degree friction angle. From this, the arching of
the fill due to the material's frictional component and its interaction with the stope walls can
be seen.

(a) (b)
FLAC (Version 4.00)
r-LACfVerafon4.nl;
35m layer rockfill —

;.i v., i.n ism

T 3Gm layer rockfill-
-I nSE+OKx* 3 179E+-01
4384E-01 <>< 4 402E*01 :ttl step 25092
-1.179E+01 <x< 3 179E*01
* 3U4E-G1 «y< fl«02EHI1 :5mlay8rrockti»-

-3.SCE-0!
-uoe-in •: 3.L u- :

v i. I iscir-c 1
37 Ode g Friction Angle
| •i Density 2.0ttn
3

-i.sie.oi • -i -.a. v.'

• ;!::• ri 15m layer roc Will -
i.'U:-:
•5 ' j ' ^ - ' ;
li::i]i:t!).:
10m layer rocWill -

5m layer rocktiB -

Figure 5-7: 25 meters (5 layers) and 35 meters (7 layers) of unconsolidated backfill

deposited in the stope
Chapter 5 - Design Guidelines for Cemented Rockfill Sill Mats 1A.

5.3.2 Stress Distribution as a Function of Stope Dip Angle

Various stope geometries are analyzed to determine the maximum stress and the stress
distribution along the stope span. Variation in stope dip angle is modeled to analyze the
effects on stress and on stress distribution. The stress distribution along the stope span for a
range of stope dip angles is portrayed by figure 5-9. Note that the vertical stress decreases as
the stope dip decreases, and the maximum stress tends to locate near the stope's footwall.
The importance of this graph is in that it produces a reference point as to where the material's
center of mass should be located.

HW Rockfill FW

10m
Vertical Stress Distribution along Span
at Different Stope Dips
HW

4.0 5.0 6.0

Distance Along Span (m)

Figure 5-8: Vertical stress distribution along span at different stope dips

Figure 5-9 portrays the stress contour of a stope dipping at 70 degrees.

Chapter 5 - Design Guidelines for Cemented Rockfill Sill Mats 21

F i g u r e 5-9: Stress c o n t o u r for a stope d i p p i n g at 70 degrees

5.3.3 Vertical Stress as a Function of Stope Span

Figure 5-10 portrays the vertical stress of various rockfill heights for different stope

spans. This graph, and those to follow, deal with vertical stopes. The graphs dealing with

inclined stopes appear later. Note that vertical stress is recorded at mid-span, on the floor o f

the stope.

Vertical Stress vs. Rockfill Heiaht

for different Scans
O GOO

Musselwhite vertical stress range

on top of CRF Sill Mat
^ 16m __*i2m
Vertical Stress rjyy (MPa)
<

i
o

^ g > £ ^ — • 8m
o

j * ^ - 5
* " ^ — — - • * — " * • " " '
o

J^Z" *•*"*'"*'
§

^ —• « • 4m
a

i
p

i
§
o _
p

0 5.0 100 15.0 20.0 25.0 30.0 35.0 40.0 45.0 5C0

Rockfill Vertical Height h (m)

F i g u r e 5-10: V e r t i c a l stress at various stope spans for different r o c k f i l l heights

Chapter 5 - Design Guidelines for Cemented Rockfill Sill Mats 76

The numerical results predict a decrease in vertical stress as rockfill height increases, as
a result of arching of the material as it interacts with the stope walls. Similar asymptotic
behavior can be seen in the analytical solutions derived by Janssen and Reimbert. For two
stope spans, 10 and 12 meters, the fill heights extend beyond 35 meters, as these are possible
scenarios for Musselwhite's constructed sill mats.

5.3.4 Vertical Stress as a Function of Rockfill's Density

Several models are executed evaluating the behavior of unconsolidated rockfill for
different material properties. The results are compared with the analytical formulations given
in Chapter 3. The following sections discuss vertical stress variation for each parameter
assessed. The analytical and numerical results both imply linear behavior at varying rockfill
densities, as indicated in figure 5-11.

Vertical Stress vs. Rockfill Density

at different Rockfill Heiahts
0.700

0600 35m
Musselwhite's range of density . 30m
«" for rockfill material
Q- 0^00
25m
r. ; 4^1
S
20m
h D4O0
rtical Stress <

8i

• 1 0 m
3

>
• 5m
0.100

00 0.5 1.0
—III! 1.5 2.0
Density 5 (t/m )
3
2.5 30 35

Figure 5-11: Vertical stress vs. rockfill density at varying rockfill heights

5.3.5 Vertical Stress as a Function of Rockfill's Friction Angle

Figure 5-12 portrays the variation of vertical stress at different friction angles, for a
given height. As expected, for friction angles approaching 0 degrees, the vertical stress is
equivalent to the hydrostatic pressure of a material of 2t/m density. The higher the friction
angle of the material, the lower is the load exerted on the floor of the stope.
Chapter 5 - Design Guidelines for Cemented Rockfill Sill Mats 21

Vertical Stress vs. Rockfill Height

at different Friction Angles

Rockfill Height h (m)

Figure 5-12: Vertical stress vs. rockfill height at varying friction angles

5.3.6 Comparison of Analytical and Numerical Methods to Determine Vertical Stress

The following figures compare the vertical stress exerted on the floor of a backfilled
stope using analytical and numerical results. Figure 5-13 displays the curves of vertical stress
at different rockfill heights, comparing Janssen's and Reimbert's analytical equations with
Flac2D
results. Figure 5-14 displays the analytical formulations derived by Mitchell, Blight,
Janssen, and Reimbert, and compares them to numerical results at different stope spans.
Figure 5-15 compares vertical stresses for a 10 meter stope span, for each method. Figure 5-
15 illustrates that there are substantial differences in vertical stress calculations among the
different methods analyzed.
Note the 0.12 M P a difference between Flac 2D
and the most commonly used method,
Mitchell (1991). This difference implies over-estimation of the strength required, when using
Mitchell's method. This thesis proposes that numerical modeling is the most reliable method,
as it accounts for deformation of cohesionless material, as well as interaction between the
material and the stope walls. Reimbert's solution yields very similar results as those of
numerical modeling, but it is limited in that it was developed for use with vertical silos.
Chapter 5 - Design Guidelines for Cemented Rockfill Sill Mats JZ8_

Vertical S t r e s s C o m p a r i s o n

0.450
Assuming:
0.400 - Density of 2.0 t/m3
-10m stope span
0.350 - 90deg stope dip
- 37deg friction angle
0.300 - v = 0.29

15 20 25
Rockfill vertical height (m)

Figure 5-13: Vertical stress over rockfill height comparing analytical and numerical
results

Vertical Stress C o m p a r i s o n

0.9000

Assuming: Mitchelj^-*
0.8000
-Density of 2.0 t/m3
- 90deg stope dip
0.7000
- 37deg friction angle
0.6000 - v = 0.29

0.5000

0.4000

0.3000

0.2000

0.1000

10.0 12.0 14.0

Stope Span (m)

Figure 5-14: Vertical stress versus stope span using different methods
Chapter 5 - Desien Guidelines for Cemented Rockfill Sill Mats 21

Vertical S t r e s s C o m p a r i s o n

Assuming:
- Density of 2.0 t/m3
0.50 - 10m stope span
- 90deg stope dip
0.12 MPa Difference

S. 0.40

35m

o
2 35m 35m

Blight (1984) Mitchell (1991) Mitchell (1991) Janssen (1895) Reimbert (1953) Flac

Figure 5-15: Vertical stress for a 10 meter stope span using different methods

5.3.7 Proposed Analytical Equation for Inclined Stopes

The following equation here derived applies to an inclined differential slice, thereby
accounting for inclined stopes. The derivation itself can be found in Appendix A . Note that
this expression can be considered an approximation only.

y-L f
2-£-tanQ)-z^
sin (/?)
2
1-exp 5-2
2-£-tan(^) Z-sin (/?) 2

In comparison to Blight's analytical formulation for inclined stopes, equation 5-2 differs
by the factor: sin ((3). The following figure compares equation 5-2 with the maximum vertical
stress that occurs employing numerical modeling for different stope dip angles.
Chapter 5 - Design Guidelines for Cemented Rockfill Sill Mats 80

Vertical Stress Comparison

Analytical vs. Numerical Modeling

0.45

0.40

0.
0.35
S
>. 0.30
>>
b 0.25
in
in
a 0.20
i—
to
cal

0.15
t0)
0.10
> -Analytical Equation
0.05
- Numerical Modeling
0.00
45 50 60 65 70 75

S t o p e Dip A n g l e (deg)

Figure 5-16: Analytical vs. numerical modeling comparison for different stope dip
angles

Results from the proposed analytical equation and from numerical modeling are
consistent with each other.

5.4 Stability of Cemented Rockfdl Sill Mats

5.4.1 Introduction

The stability of rockfill sill mats depends on two principal factors:

- sill mat strength; and
- the weight of backfill.

The next section considers these two factors in detail.

Figure 5-17 portrays stope geometry, strength, stress components, and failure modes in a
sill mat pillar.
Chapter 5 - Design Guidelines for Cemented Rockfill Sill Mats 81

Figure 5-17: Geometry, strength and stress component and failure modes in a sill mat
pillar

5.4.2 Sill Mat Strength - Friction and Cohesion (fy

The strength of rockfill material can be substantially improved by the addition of a

binding material. The most evident improvement (between an uncemented fill and a fill
treated with cement) is in shear and tensile strength. Cement bonds that form between fill
particles introduce a cohesive component to the fill's shear and tensile strength, and it is this
cohesive component that is absent in an uncemented fill.

The following two soil components directly determine backfill strength:

- frictional forces, proportional to the internal angle of friction (§) resulting

from interlocking solid particles; and
- Portland cement (or similar soil binder) which significantly increases
actual fill cohesion by binding the solid particles to each other.

The effects of grain interlocking, and hence the magnitude of the internal friction angle,
depend upon grain shape, overall particle size, and packing density (Thomas, et al., 1979).
The elastic modulus of backfill increases with the addition of a binding agent. Backfill
typically has a low modulus of elasticity of approximately 1000 M P a to 100 MPa, and is
normally one to two orders of magnitude less stiff than the surrounding rock.
 Chapter 5 - Design Guidelines for Cemented Rockfill Sill Mats 82

5.4.3 Backfill Load

The load of backfill exerted on the top of the CRF sill mat was analyzed in detail in the
previous section. This load is dependent upon the backfill properties: density and friction
angle; and the geometry of the stope which includes the height of backfill material, span of
the stope, and stope dip.

5.4.4 Proposed Rotational Analytical Formulation

The original rotational analytical formulation was derived for pastefill. At the
Musselwhite mine (rockfill), the shearing resistance of a given hangingwall contact surface
may not be particularly low, as would be expected for pastefill. Also, the three sill mats in
existence at Musselwhite do not have low dip angles.
For these reasons, the following analytical formulation incorporates shearing resistance
at the hangingwall contact surface. This case does not yield an especially conservative result
(in shearing resistance). Safety considerations demand conservative estimation so that, in this
particular case, only some extent of the shearing resistance should be taken into account,
whereas full shearing resistance is indicated in figure 5-18. Additionally, consistent with this
proposition, the research developed by Dirige, De Souza, and Chew (2001) implied that
centrifuge modeling studies of pastefill sill mats indicate that, in stopes with smooth rock
wall conditions, sill mat failure is caused by the fill self-weight, i.e., failure has little
dependency upon fill binder content. Furthermore, Dirige, De Souza, and Chew concluded
that, in stopes with rough rock wall conditions, the wall roughness itself contributes
significantly to the stability of the sill mat during undercut mining. The present study implies
that backfill sill mats can be designed with a high degree of stability using standard
centrifuge physical modeling in combination with numerical modeling. This approach
suggests that, for the sill mat rotational failure mode, shear strength should be considered a
component of the analytical solution derived by Mitchell (1991).

A multiplier factor, a, ranging from 0 to 1, is added to the equation, corresponding to the

estimated contact length at the hangingwall between the rock wall and the cemented rockfill
material. As an alternative, the contact quality itself may be estimated to arrive at a. After
visual inspection, i f it is determined that there is poor quality contact between the rock wall
 Chanter 5 - Design Guidelines for Cemented Rnrkfill Sill Mats 83

and the CRF, then an a equal to zero should be used, yielding a result similar to that given by
equation 3-36. Alternatively, estimation of the a factor could be made by measuring the
horizontal load exerted by the CRF material on the hangingwall/sill mat contact (see Chapter
8: Future Work), which would be an indirect measurement of the quality of the contact.

Figure 5-18: Rotational failure considering shear strength in the hangingwall of the sill
mat

The following equation can still be considered conservative given that sill mat failure
(when a circular shear failure occurs) produces additional shearing resistance as the pillar
slides against the rock wall (r,), as portrayed in figure 5-19.

(cr + a • Y) > —
L-(L-d-cot(j3))-sm (f3)
2

Also, tensile failure could occur at an angle differing from /3 and a location differing
from point "O", as predicted by numerical modeling (discussed later). If tensile failure does
occur at point " O " (figure 5-18), then only the failing mass (portrayed by geometric figure u)
should be considered and not the total mass of the sill mat, as implied by Mitchell's
simplification. Doing so thus accounts for less weight than the value denoted by w. Also, the
vertical stress u should be measured near the footwall, not near the center, as in Mitchell
v

(1991).
Chapter 5 - Design Guidelines for Cemented Rockfill Sill Mats 84

Figure 5-19: Rotational failure considering shear strength in the hangingwall of the sill
mat

The following figure (5-20) displays factors of safety for the different failure modes
incorporating Musselwhite's sill mat properties, vertical stress, and stope geometry. The
minimum safety factor required is 1.2 in the design of non-entry methods, and 1.5 for entry
methods (Pakalnis, course notes, 2004).

Sill Mat Failure Modes

11.0

10.0 •

9.0 •

8.0 •
re 7.0-I
F.S. > 1.2 for design (non-entry methods)
a= 1
6.0

5.0

4.0 -

3.0 -

2.0 -

1.0 -

0.0 -
Caving Flexural Sliding Rotational Including
Shear Strength

Failure M o d e

Figure 5-20: Factor of safety for sill mat failure modes

 Chapter 5 - Design Guidelines for Cemented Rockfill Sill Mats 85

5.5 Failure Modes - Numerical Modeling Solution

A series of models is run to obtain design curves representing sill mat failure modes.
These parametric studies account for variation in sill mat strength via variations in cohesion,
and via stope geometry (characterized by span, and dip angle) variations. In all cases, in the
interest of time efficiency of the numerical modeling analysis, a constant vertical stress of
0.41 MPa (in the case of a vertical stope of 10 meter span and 35 meter rockfill height) acting
along the entire stope span is applied to the sill mat pillar. As indicated previously, the
distribution of the vertical stress along the stope span is not constant, and decreases with
stope dip. However, this assumption (using a constant of 0.41 MPa) yields an extra factor of
safety in the sill mat design.

5.5.1 Model Construction

The following sections deal with a non-standard grid of CRF sill mat pillar stability.
There is no conventional standard for a grid portraying measurements of stability in CRF sill
mat pillars, due to the fact that the pillar dimensions (height and span) vary according to
strength properties assigned to the grid. These pillar dimensions result from a trial and error
execution of the model with given strength properties.
A n important consideration of the models is the interface elements designed to simulate
distinct planes along which slip and/or separation can occur. More about interface elements
can be found in Appendix C.

Once the grid is constructed, material properties must be assigned to the individual grid
elements. Elastic host rock properties are assigned to the outer grid elements. In situ host
rock properties are obtained form intact rock cores (i.e., in laboratory testing) and results are
subsequently scaled to the in situ reality. For details of host rock properties, refer to Chapter
4: Database.

For the elastic model, the relevant properties are: density; bulk modulus; and shear
modulus.

Even though backfill properties could be obtained from existing literature, it is the
implication(s) of varying these parameters that is the principal objective here. These
implications are discussed in detail in the next section.
Chapter 5 - Design Guidelines for Cemented Rockfill Sill Mats 86

The following table (5-1) portrays an example of Musselwhite's sill mat geometry,
rockfill loading, and strength properties.

Table 5-1: Example of Musselwhite's geometry, rockfill loading, and sill mat strength
properties

Sill Mat Geometry

Span, L 10.0 m (variable)
Height of Sill, d 7.0 m (variable)
Stope Dip, (3 85° (variable)
Rockfill Loading
Vertical Stress, a v 0.41 MPa (constant)
Sill Mat Strength Properties
Unconfined Compressive Strength Test, U C S 8.0 MPa (variable)
Cohesion, c 2.0 MPa (variable)
Tensile Strength, rj, 0.8 MPa (variable)
Shear Strength, T, 2.1 MPa (variable)
Friction angle, <j) 38° (constant)
Density, 8 2.0 t/m (constant)
3

5.5.2 Constitutive Equations - Strain Softening

Swan and Brummer (2001) analyzed a strain softening model where a cohesive sill will
lose integrity at a plastic strain of 1.5%. The CRF sill mat properties assigned by this thesis
are of a Mohr-Coulomb type of material with strain-softening behavior that loses its
cohesion, and tensile strength, from the given value to zero (MPa, psi, etc.), at a plastic strain
of 1.5%. Correspondingly, the friction angle falls from the original value of 37 degrees (at a
plastic strain of 0%) to 15 degrees (at a plastic strain of 1.5%). At a plastic strain of 1.5%,
cohesion will be zero but there is still some strength because of the friction angle. For the
Mohr-Coulomb plasticity model, the relevant properties are density, bulk modulus, shear
modulus, friction angle, cohesion, dilation angle, and tensile strength. Figure 5-21 illustrates
an example of a strain-softening model for the cohesion strength parameter.
Chapter 5 - Design Guidelines for Cemented Rockfill Sill Mats 87

Strain-Softening M o d e l

0.0% 0.2% 0.4% 0.6% 0.6% 10

.% 12
.% 14
.% 16
.%
Strain ()

Figure 5-21: Strain softening model - cohesion example

5.5.3 Model Execution

For a given value of cohesion, span, and dip of the stope, the maximum height of the
CRF sill pillar that results in failure can be determined by trial and error. Setting the actual
stable height 25 centimeters above the height determined, results in a factor of safety of at
least 1.0. In this case, 25 centimeters is arbitrarily determined, yet this choice represents an
efficient compromise between the accuracy of a low value (time consuming) and the time
saved in using a high value (which yields inaccuracy). Note that, for flexural, sliding, and
rotational failure modes, the same model is applied where a specific failure mode results,
according to the strength parameters: cohesion, tensile strength, and friction angle. For these
three failure modes, two plots are presented: first, a plot of the grid elements and the applied
vertical load, and second, a plot that illustrates the plasticity state of the grid elements.

Since the parameters are purposely set low enough to result in failure, the equilibrium
solution is clearly not the objective here. When failure does occur, the model does not
converge to the equilibrium state. Rather, the model is stepped through the simulation
process, plotting the resulting collapse as it occurs. The minimum safety factor is thus
achieved (>1.0).
 Chapter 5 - Design Guidelines for Cemented Rockfill Sill Mats 88

5.5.4 Caving Failure

Even though caving failure is the least probable failure mode for a sill mat of given
strength, this mode is analyzed here theoretically, as i f the strength path is dependent on the
shape of the fracture caused by a caving failure. No difference between planar and semi-
circular fractures can be found. Each caved pillar represents the same overall weight, but the
difference is in the length of contact over the distance of the fracture.
Figure 5-22(a) portrays a planar fracture and figure 5-22(b) portrays a semi-circular
fracture. From this analysis, no conclusions can be drawn regarding path dependence. It is
apparent that both failures yield the same results.

(a) (b)
X,
P Y Y Y Y Y Y Y V Y y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y
<
Y

X
X <
X <
X <
X, K
X. <
i.

\ A. A , a A A , A, A,
A /J, A A
A (A ' A A A
fa A"
Aj A A,/
\ ^ A A A A A "' A
A A A,1 A A < /
tJ

\ iA A A A l\ /I, ,f i, A
^
\ A A, ^ A
^ *V A *-\ *V
A ix A ,f i, ,ri, i' 'i
i A A
1 1 1

1 1 1 1 1 1 1
A A A A, A
' i 1 ,f i. A
M
S A
S A ij, , A Xi, A; A
^ fa ^/
/.i, A
\ •v r
i\ "' A "> A A A
\ ^,A A
1 1 1
'\ \
1 1
^
A A fa
1

A A
A A
fafaAA '''
f

Aj u, fa r" fa fa fa
A fa fafa
1
\ r-\ XI, Xi, Xi, XI,
XJ, Xl, XL 1 1
\/\ A A A A A A lA A A x^ A A'
A A A A fa A A A
A A fa A A A fafaA '
A 1

s/\ A A A A A
A A A A A A A A A A A A A A fa ^faA A A A fat
\]A A A A A A A A A A A A A A A A A A A A A A fat
A A A A A fa A
10m
A
A

Figure 5-22: Caving failure mode - planar and semi circular crack

5.5.5 Flexural Failure

Figure 5-23 (a) and figure 5-23(b) portray a thin sill pillar failing as a result of flexing. It
is evident that the top grid elements fail in compression, and that tension cracks form on the
floor of the pillar, subsequently causing its collapse. Plasticity indicators (figure 5-23(b)) are
included. These identify the type of failure, e.g., shear, i.e., compressive failure, (represented
by black in figures 5-23(b), 5-24(b), and 5-25(b)) or tensile failure (represented by white in
figures 5-23(b), 5-24(b), and 5-25(b)), and whether the stress state in the zone is currently at
the yield surface ("at yield"), or has previously reached the yield surface but is currently
 Chanter 5 - Design Guidelines for Cemented Rockfill Sill Mats RQ

below the yield surface ("at yield in past"). (Note: shaded area represents elements that do
not fail.)

Figure 5-23: Sill mat flexural failure mode - grid elements and plasticity state

5.5.6 Sliding Failure

The following figures illustrate shearing failure of hangingwall and footwall sill-host
rock contact. These figures represent a typical shearing failure case (out of many
possibilities).

Elastic
At Yield in Shear or Vol,
Elastic. Yield in Past
At Yield in Tension

0.100 0.300 o.wo

Figure 5-24: Sill mat sliding failure mode - grid elements and plasticity state
Chapter 5 - Design Guidelines for Cemented Rockfill Sill Mats 90

5.5.7 Rotational - Crushing Failure

Rotational-crushing failure mode is not conceived as presented here, in the analytical

solution previously discussed. However, rotational-crushing failure modeling illustrates that
failure occurs when the sill in contact with the hangingwall crushes (i.e., fails in
compression) as the sill mat rotates. Tensile failure of the footwall/sill contact results in the
rotation of the sill with respect to the bottom footwall/sill contact. These failures must occur
simultaneously in both the hangingwall and the footwall.

Figure 5-25: Sill mat rotational crushing failure mode - grid elements and plasticity
state

5.5.8 Rotational - Breaking Failure

A similar result occurs in the rotational-breaking failure mode, where the centrifuge
model (Mitchell, 1991) predicts that a tensile failure propagates at the same angle as the dip
of the stope, rotating about point "O". Numerical modeling predicts that this failure occurs at
approximately mid-span, with no particular angle of failure. The numerical solution implies
that failure occurs where the sill, in contact with the hangingwall, crushes (i.e., fails in
compression), as the sill mat rotates and breaks.
 Chanter 5 - Deswn Guidelines for Cemented Rockfill Sill Mats
91

Y'.V Y 3

1 Elastic
I At Yield in Shear or Vol.
I Elastic, Yield in Past
]Al Yield in Tension

O-10O 0.300 0.300 0.700

f~1C*1)

Figure 5-26: Sill mat rotational breaking failure mode - grid elements and plasticity
state

5.5.9 Sill Mat Design Curves

Figures 5-27 to 5-38 represent a series of design curves that predict, in each case, the
point of failure and the failure mode, for a given strength and specific geometric parameters.
In each graph, the y-axis portrays (in meters) the minimum height, or depth, of a CRF sill
mat necessary to be in a stable state. The graphs are of three types: no strength on
hangingwall contact, 50% strength, and 100% strength on hangingwall contact, as
corresponding to the percentage of the CRF cohesion value. E.g., for an 85 degree stope, with
10 meter span, when hangingwall/sill contact equals 0.75 MPa, i.e., 50% of full cohesion
(1.50 MPa), rotational breaking occurs below 2.5 meters (see figure 5-31). Each individual
graph represents a specific stope dip angle, ranging from 90 degrees (vertical) to 75 degrees.
The individual curves within each graph represent specific stope spans, ranging from 6
meters to 12 meters. The type of failure occurring (below the minimum sill mat vertical
height necessary to be stable) is indicated for each curve by the following notations: " S " for
sliding, " R c " for rotational-crushing, "Rb" for rotational-breaking, and " F " for flexural,
failure modes. A l l cases assume uniform vertical stress acting on the top of the CRF sill mat.
A stable CRF sill mat is represented as being anywhere above the individual design curve in
question (e.g., 6, 8, 10, or 12 meter span), implying a safety factor of 1.0 or greater.
Chapter 5 - Design Guidelines for Cemented Rockfill Sill Mats 21

Figure 5-27: Sill mat stability for 90° - No strength on HW (x = 0%)

t

SIM Mat Failure Graph for 90deq Stope Dip

at different Stope Spans
7.0 .,

C o h e s i o n (MPa)

Figure 5-28: Sill mat stability for 90° - 50% sill mat strength on HW (x, = 50%)

SHI Mat Failure Graph for 90deq Stope Dip

at different Stope Spans
5.0 -|

C o h e s i o n (MPa)

Figure 5-29: Sill mat stability for 90° - 100% sill mat strength on HW (x = 100%)
t
Chapter 5 - Design Guidelines for Cemented Rockfill Sill Mats 93

Figure 5-30: Sill mat stability for 85° - No strength on HW (x, = 0%)

Figure 5-31: Sill mat stability for 85° - 50% sill mat strength on HW (T, = 50%)

Figure 5-32: Sill mat stability for 85° - 100% sill mat strength on HW (x, = 100%)
Chapter 5 - Design Guidelines for Cemented Rockfill Sill Mats 21

SHI Mat Failure Graph for 80deq Stope Dip

at different Stope Spans

E F.S. = 1
I

T,=0%

HW contact
cohesion value
1.50 1.75 2.00
Cohesion (MPa)

Figure 5-33: Sill mat stability for 80° - No strength on HW (x = 0%) t

SIM Mat Failure Graph for 80deq Stope Dip

at different Stope Spans
10.0

9.0 F.S. = 1
8.0

7.0

6.0 STABLE.
5.0 T,=50%
4.0

3.0

2.0
FAIL
' HW contact
" cohesion value
1.50 1.75 2.00
Cohesion (MPa)

Figure 5-34: Sill mat stability for 80° - 50% sill mat strength on HW (x, = 50%)

SMI Mat Failure Graph for 80deq Stope Dip

at different Stope Spans

F.S. = 1

STABLE

12m
" 10m
Rb
8m
_6m
FAIL
_ HW contact
cohesion value
1.50 1.75 2.00
Cohesion (MPa)

Figure 5-35: Sill mat stability for 80° - 100% sill mat strength on HW (x = 100%) t
Chapter 5 - Design Guidelines for Cemented Rockfill Sill Mats 95

Figure 5-36: Sill mat stability for 75° - No strength on HW (x = 0%)

t

Figure 5-37: Sill mat stability for 75° - 50% sill mat strength on HW (r = 50%)
t

Figure 5-38: Sill mat stability for 75° - 100% sill mat strength on HW (x = 100%)
t
Chapter 5 - Design Guidelines for Cemented Rockfill Sill Mats 96

This series of design curves enables the operator to select the sill dimensions, given the
cohesive strength of the fill material, and the parameters considered of greatest significance
to the design such as:
- span;

- wall dip; and

- H W shear strength.

The Flac 20
approach used in this study does not incorporate a factor of safety in its
method of design. However, by applying a factor of safety to the cohesion parameter, the
total factor of safety in the overall design will be improved.
Chapter 6 - Effect of Backfill on Open Stoping 97

6 E F F E C T OF DELAYED BACKFILL ON OPEN STOPING

6.1 Introduction

A significant question at the Musselwhite mine and for open stope operators in general,
is what is the effect of mining adjacent to previously backfilled stopes? Design methods such
as The Stability Graph have been derived from a database of isolated stopes. These are stopes
that have solid abutments adjacent to the stope being mined.
The Stability Graph Method makes the assumption that the adjacent backfilled stope
does not reduce overall stability.
This thesis, alternatively, investigates the effect of backfill on adjacent stopes in terms of
resultant wall slough.

6.2 Backfill as Local Support

Backfill should generate induced stresses on the order of 30 MPa, to provide for
significant local support (Giirtunca et al., 2001). However, the research by Gtirtunca implies
that backfill affects displacement and stresses within the rock mass at a much lower backfill
stress level. This research indicates, in terms of backfill providing local support, that backfill
starts to induce changes within the rock mass when backfill stresses approach 1 M P a and
furthermore, that rock mass induced effects reach their maximum when backfill stresses
approach 2 - 3 MPa. These effects are measured via reduced wall closure, reduced bed
separation, reduced dilation, and the effect of induced wall stresses.

Reduced closure and dilation within the rock mass limits stope wall
disturbance/conditioning, and in turn lowers wall slough. This level of
disturbance/conditioning is a function of distance to the backfill front. Backfill strain at a
specific distance near the open stope/fill face should yield (a critical) 2 - 3 M P a of induced
backfill stress, in theory. If this critical 2 - 3 MPa backfill stress occurs, but a distance much
greater than that of the exposed strike length of the stope, then clearly these conditions do not
adequately influence the exposed area, and do not render stability within the rock mass. If the
selected mine layout does not include within its parameters sufficient closure, then the use of
backfill in such a low closure environment is not fully effective and consequently provides
negligible support to the adjacent stope.
Chapter 6 - Effect of Backfill on Open Stoping 98

Consequently, it is evident that, when mining adjacent to backfill (e.g., Avoca and
secondary transverse open stoping), the radius factor (or hydraulic radius) of the exposed
area is only one of several geometric factors determining stability in open stopes. A
combination of the radius factor (or hydraulic radius) of the exposed zone and the backfilled
zone should be used to account for the stability of open stopes.
For mining via the Avoca method utilizing unconsolidated rockfill, or in the case of
transverse open stope mining with cemented rockfill primaries, the main consideration is the
effect of mining adjacent to backfill. Taking into account that, at the actual depth of mining
(200 - 600 meters below the surface), wall closure is negligible, and that rockfill will
generate minimum horizontal stresses, less than the minimum threshold of 1 M P a is
generated, i.e., not enough to provide local support. As a result, it is necessary to adjust the
stability graph curves (or modify the radius factor) in accordance with this minimum support
condition. Radius factor (or hydraulic radius) must be increased in order to account for the
backfilled zone and its effects on the overall exposed surface.

Figure 6-1 displays the horizontal stresses exerted by the backfill in a 35 meter high
stope (unconsolidated rockfill) at different stope spans. It is evident that the horizontal stress
values are very low and are far below that needed to provide local support ( 2 - 3 M P a of
horizontal stress).

Horizontal Stress vs. Rockfill Height

at different Spans
0.18 i

0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 45.0 50.0
Rockfill Vertical Height h(m)

Figure 6-1: Horizontal stress versus rockfill height at different stope spans
 Chapter 6 - Effect of Backfill on Open Stnpinv 99

6.3 Use of the Stability Graph Method

A l l of the known Empirical Design Stability Methods (re: Chapter 3) were developed for
isolated stopes and not for stopes mined adjacent to backfill (Figure 6-2). This leads to
fundamental errors i f these existing methods are not calibrated to fit the given (non-isolated)
specific mining conditions (e.g., mining adjacent to backfill). The calibration of the stability
graph, the modified stability graph, or the hangingwall stability rating graph (re: Chapter 3),
when mining adjacent to a backfilled stope, is accomplished by modifying the radius factor
(or hydraulic radius) thereby increasing its value, and then applying the method to the
circumstance of mining adjacent to a backfilled stope. Interpolation (rather than
extrapolation) is appropriate to empirical methods. New data points thus enhance the existing
database. Numerically modeling the significance of mining adjacent to backfill will improve
our understanding of stope stability.

Figure 6-2: Stope adjacent to rock (isolated) or backfill abutment(s) - Modified from
Atlas Copco Drawing
Chapter 6 - Effect of Backfill on Open Stoping 100

6.4 Quantifying the Effect of Backfill on Longhole Open Stoping

Two mining methods are employed at Musselwhite: Avoca (or longitudinal retreat), and
transverse open stoping. Effects of mining adjacent to backfill are assessed for isolated
stopes, and for stopes mined adjacent to backfill, via the following procedure:
1. Compilation of the Musselwhite database in terms of hydraulic radius (converted to
radius factor) and the modified stability number N ' ;
2. Analysis of other parameters that affect the stability of open stopes such as
undercutting of stope walls;
3. Conducting structural mapping within the different rock units;
4. Determining Equivalent Linear Overbreak Slough (ELOS) for the compiled database
using C M S survey profiles;
5. Establishing, using neural networks, the relative importance of individual parameters
and their significance on wall stability (ELOS) for isolated stopes versus stopes
mined adjacent to backfill;
6. Determining via numerical modeling (finite difference method), the resultant decrease
on wall closure for stopes with and without backfill;
7. Determining wall closure using three dimensional numerical modeling (3D boundary
element method) in order to assess the effect of wall closure as further mining of the
stope progresses; and
8. Determining the change in radius factor throughout the stope wall as mining
progresses, i.e., away from the original radius factor reference point.

6.5 Updating Musselwhite's Stability Graph

Musselwhite's stability database is plotted on the Stability Graph, in terms of radius

factor, with the equivalent linear overbreak curves superimposed onto it. Figure 6-3 portrays
this updated Stability Graph for Avoca (or longitudinal retreat) stope data, and figure 6-4
portrays the updated Stability Graph for transverse stope data. 90 and 48 points are plotted
for longitudinal and transverse stopes respectively, for a total of 138 points, all of which are
plotted in figure 6-5.
Chapter 6 - Effect of Backfill on Open Stoping 101

Empirical Estimation of Overbreak/Slough

Musselwhite Mine (Longitudinal Stopes Data) 90 cases

0.1 i 1 1 1 1 1
0 5 10 15 20
Radius Factor (m)

Figure 6-3: Stability Graph update - longitudinal stope data (AVOCA)

Chapter 6 - Effect of Backfill on Open Stoping 102

Empirical Estimation of Overbreak/Slouah

Musselwhite Mine (Transverse Stopes Data) 48 cases

Radius Factor (m)

Figure 6-4: Stability Graph update - transverse stope data

The data points o f the updated database are summarized in the following table:

Table 6-1: Average and standard deviation for the Musselwhite's stability database

Longitudinal stopes Transverse stopes

Average Std Deviation Average Std Deviation
Length FW 42.0 20.8 26.5 9.0
Length HW 41.4 20.0 26.5 9.0
Height FW 33.9 8.0 44.6 8.3
Height HW 34.3 7.6 44.2 8.5
Dip FW 92.4 8.7 101.4 7.0
Dip HW 82.9 10.6 80.4 3.8
Hydraulic Radius FW 8.9 2.5 8.0 1.6
Hydraulic Radius HW 8.9 2.4 8.0 1.6
Radius Factor FW 9.8 2.7 8.8 1.8
Radius Factor HW 9.8 2.6 8.8 1.8
ELOS FW 0.5 0.4 0.4 0.2
ELOS HW 1.0 1.0 1.0 0.7
Chapter 6 - Effect of Backfill on Open Stoping 103

Empirical Estimation of Overbreak/Slouqh

Musselwhite Mine (All Stopes Data) 138 c a s e s

100

•
: i><7 % \ /-\0
• • •• »•
A

* • A AAA« J

a)
10
' . * A A 4 s"*^ •

-Q
E ^
z
>» * ^^^^^^
-«—'

!5
TO 0>
ft * I*
*
"
o *
0) »
TJ
(D
ft
i£
f;
X> 1

O
_

Leaena
# # # ELOS <= 0.5m
• • • 0.5m < ELOS <=1.0m -
• • • 1.0m < ELOS <=1.5m -
-fr -sir -sir
1.5m < ELOS <= 2.0m
CI CI CI 2.C)m < ELOS
I I I I I I I I I I I ! i i i i

0.1
10 15 20
Radius Factor (m)

F i g u r e 6-5: Stability G r a p h update - transverse a n d longitudinal stope data

Chapter 6 — Effect of Backfill on Open Stoping 104

6.6 Neural Network Training and Results

Neural networks are used to determine the relative importance of the various inputs
related to the stability of open stopes, and to determine the effect of backfill on open stope
stability where mining adjacent to backfill occurs. The significance of individual inputs is
assessed in terms of its effect on wall slough (ELOS).

6.6.1 Relative Importance of Inputs

Figure 6-6 implies that the dip of the orebody highly affects stope stability when dip
angle is greater than 90 degrees (gemometric footwall). From a geometric perspective, this is
the point at which a hangingwall becomes a footwall.

•
Relative Importance of Inputs

•• ||
Dip: Stope dip angle

H.R.: Hydraulic radius

•• |
Q: Barton's rock mass

- •
quality number

Dip H.R. Q

Figure 6-6: Relative importance of inputs on stope stability for stopes dipping more
than 90 degrees (footwalls)

Conversely, at orebody dip angle of less than 90 degrees, the quality of the rock mass is
the principle determining stability parameter. Figure 6-7 portrays the relative importance of
the parameters that determine stope stability.
Chapter 6 - Effect of Backfill on Open Stoping 105

Relative Importance of Inputs

Dip: Stope dip angle

• H.R.: Hydraulic radius

. . • Q: Barton's rock mass

quality number
._ — •

•
•

Dip
•
H.R.
•
Q
•

Figure 6-7: Relative importance of inputs on stope stability for stope dips under 90
degrees (hangingwalls)

The parameters known to determine stope stability are (in no particular order): drilling
and blasting, wall undercut, orebody dip, hydraulic radius of the stope walls, quality of the
rock mass, stress state, geological heterogeneity, and level of support, among other less
significant factors.
Current tools in use for the design of open stopes oversimplify all these parameters as
being represented by quality of rock mass, N ' , and the distance to the abutments (hydraulic
radius). The assumption is that the quality of the rock mass subtends all parameters excluding
hydraulic radius.
Chapter 6 - Effect of Backfill on Open Stoping 106

Figure 6-8 portrays the database of measurements gathered by Musselwhite's

engineering staff and incorporates other factors that influence stope stability. One significant
factor is the thickness of the 4f rock, whose R M R value can be as low as 45%. Another
variable determining stope stability, when combined with rock quality, is the degree of
undercutting of the stope walls.
The quality of 4f rock usually determines the stability of the stope. Exposing the 4f rock
may produce caving, especially when this rock type has been undercut. This is due to the
foliated nature of 4f rock and its structure, where, typically, the joint sets with low strength
sericitic-alteration-infilling, run parallel to the orebody.

| Relative Importance of Inputs

4f: Average 4f thickness

Dip: Stope dip angle

U/C: Undercut of HW/FW

4f Dip U/C

Figure 6-8: Relative importance of inputs on stope stability considering rock type, ore
dip and undercut

Figure 6-9 illustrates how numerical modeling can be used to determine stability in the
hangingwall where the (poor-quality) rock mass is undercut. The modeling results imply that
when this poor-quality rock type is not undercut, the stope walls are stable. Naturally, this
behavior may vary significantly in a real environment, due to the numerous variables
involved in a real-world situation. Some elements of these real-world variables are difficult,
if not impossible, to represent by numerical modeling.
 Failures

I
ftjijjjj
rrjff I

Figure 6-9: Effect of undercutting the hangingwall: various types of failure

6.6.2 Neural Network Predictions

The "neural net" analysis is trained on two databases, independent of each other:
- Isolated stopes (transverse primaries stopes); and
- Stopes adjacent to backfill (Avoca).

This analysis is employed to determine i f any relationship exists between ELOS values
when mining adjacent to backfilled stopes versus mining adjacent to solid abutments. ELOS
predictions are made for various hydraulic radius and stability values. The trained results
generated for longitudinal stopes are expected to be higher in terms of E L O S than results
generated for transverse primary stopes, but no such difference can be found where backfill
is used in actual practice, even though radius factor measurements should be somewhat
higher (re: Section 6.7).

Figure 6-10 portrays the predicted ELOS values for transverse primaries versus
longitudinal stopes.
Chapter 6 - Effect of Backfill on Open Stoping 108

P r e d i c t e d E L O S (not to scale)
T r a n s v e r s e S t o p e s vs. L o n g i t u d i n a l S t o p e s

Stopes Adjacent to Backfill

0.5 1.0 1.5

E L O S of L o n g i t u d i n a l S t o p e s (m)

Figure 6-10: Predicted ELOS values of transverse vs. longitudinal stopes

Figure 6-10 illustrates the random scatter generated from neural net analysis, as opposed
to the expected distribution of predicted data which should have been located primarily
below the one-to-one regression. It is not clear that longitudinal stopes (adjacent to backfill)
would yield higher ELOS values than would transverse primary stopes (i.e., isolated), as
expected. This result can be justified by the many other factors that affect stability such as
drilling and blasting, geological heterogeneity, undercutting of the stope wall, and others. A n
example of unexpected wall slough is portrayed by figure 6-11, where a C M S survey of a
section of the 500SZN-P1 zone shows that dilution occurs due to a geological structure.
Chapter 6 - Effect of Backfill on Open Stoping 109

500SZN-P1 Looking North

Geological structure

Unexpected wall slough

Figure 6-11: Wall instability due to structure on hangingwall

Chapter 6 - Effect of Backfill on Open Stoping 110

6.7 Effect of Backfill in Stability of Open Stopes - Numerical Modeling Analysis

2D

The following sequences are tested using Flac to quantify the effect of mining
adjacent to backfill. The first sequence (figure 6-12), corresponds to the excavation of a 40
meter high bottom stope and subsequent excavation of the top stope (of the same height)
without using backfill for the bottom stope. The next sequence (figure 6-13) is the same as
the first but in this case, after the bottom stope has been mined, it is then backfilled. The third
sequence (figure 6-14) considers only mining of the top stope.

Figure 6-12: Sequence without backfill

Figure 6-13: Sequence with backfill

Chapter 6 - Effect of Backfill on Open Stoping 111

Figure 6-14: Mining of the top stope only

Figure 6-15 implies that backfill has a minimal effect when compared with solid rock
abutments. These results are based solely on an elastic, numerical model. In practice, this is
no the case, as backfill provides confinement, in terms of stability. The assumption that it is
indeed a minor effect on elastic wall closure explains why there is an increase in radius factor
(or hydraulic radius) when mining adjacent to backfill. It is important to note that this is a
parametric study, measuring the effect of backfill in an elastic, homogeneous, isotropic
model.

Wall Closure Profile

0 1 2 3 4 5 6 7 8 9
Wall C l o s u r e Profile (cm)

Figure 6-15: Horizontal displacement profile

Chapter 6 - Effect of Backfill on Open Stoping 112

6.7.1 Radius Factor Behavior with Stope Strike Length

Radius factor is calculated for various points from the front face, while increasing the
stope strike length until the radius factor reaches a constant value, beyond a specific strike
length. Beyond this strike length, radius factor variation is minimal and begins to behave
asymptotically. The assumption being made here is that, for a given exposed strike length,
the measuring point is one-half of the maximum unsupported span, and beyond that
maximum unsupported span the stope is considered to be backfill. Figure 6-16 portrays a
measurement point at 5 meters from the front face, where it can be assumed that the
maximum unsupported strike length is 10 meters, beyond which the stope is considered to be
backfilled. At 5 meters from the front face, the radius factor is calculated for a stope of 10,
20, 30, 40, 50, 100, 200, and 400 meter strike lengths.

10m 0

«—> Long Section

10m Strike Length - Displacement and
Radius Factor measured at 5m from face
BackfiTj begins here 30m stope height

20m Strike Length - Displacement and

Radius Factor measured at 5m from face
Backfill 30m stope height

30m Strike Length - Displacement and

Radius Factor measured at 5m from face
30m stope height

,'>ys/\ 'J
r,y,f - j ., yJ-,*w.»-„.- . ..
40m Strike Length - Displacement and
Radius Factor measured at 5m from face
* J/ / ?/ / * J* 30m stope height

f / fJ / ~ J' / /
e J«- 1

50m Strike Length - Displacement and

•v.','' > Radius Factor measured at 5m from face
tr *V ' j. " 30m stope height

oo
Theoretical affective Un-affective backfill
backfill (varies case
by case)

Figure 6-16: Schematic of a stope of increasingly larger strike length

Chapter 6 - Effect of Backfill on Open Stoping 113

Figure 6-17 illustrates a radius factor measurement at 5 meters from the front face, for a
stope of 40 meter height and 100 meter strike length. The same procedure is used to measure
the radius factor at 6, 7, 8, 9, 10, 15, and 20 meters from the front face, representing 12, 14,
16, 18, 20, 30, and 40 meters, respectively, of unsupported strike length. Beyond these
distances (12, 14, 16, etc.) the stope is considered to be backfill (figure 6-17 portrays the 5
meter case only).

40m

i — i

5m vr10m
100m

Figure 6-17: Radius factor measurement at distance from front face

Figure 6-18 illustrates the generation of radius factors by an AutoLISP program (Milne
and Lunder, 1994) over an entire area of a stope of 40 meter height and 100 meter length.
Chapter 6 - Effect of Backfill on Open Stoping 114

5 0,7 0,8 0,9 0,9 0.9 0,8 0,6 0,9 O.S 0,9 0,9 0,S 0,8 0,9 0.B 0,8 0,8 O.B 0,7 0,
C.7 4.3 5.S 5.6 5.7 5.8 5.9 5,9 6.0 6.0 6.0 6.0 6,0 5,9 5.9 5.S 5.7 5.6 5.2 4.3 0.'
C.B 5.6 7.7 8.6 9.1 9.4 9.6 9.7 9.8 9.6 9.B 9.8 9.8 9.7 9.6 9.4 9.1 8.6 7.7 5.6 0.3
C,B 6,2 9,0 10,5 11,3 11,8 12,112,312,412,512,512,512,412.312,1 11,8 11,3 10,5 9,0 6,2 0,3
C.8 6.4 9.6 11.5 12.513,2 13.613.B14.0 14,0 14,1 14,0 14,013.813.613.212.5 11,5 9,6 6,4 0,3 40m
C.8 6.5 9.S ll.B 12.913.614.014.314.514.614.614.614.514.314.013.612.9 ll.B 9.8 6,5 0.3
C.8 6,4 9.6 11.5 12.513.213,613.814.0 14.0 14.1 14.0 14.013.813,613,212,5 11.5 9.6 6.4 0.3
C.8 6,2 9.0 10.5 11.3 11.8 12.112.312.412.512.512.512.412.312.1 11.8 11.3 10.5 9.0 6.2 0.3
C.8 5.6 7.7 8.6 9.1 9.4 9.6 9.7 9.8 9.a 9.8 9.a 9.B 9.7 9.6 9.4 9.1 8.6 7.7 5.6 O.J
C.7 4.3 5.2 5.6 5.7 5.8 5.9 5.9 6.0 6,0 6.0 6.0 6.0 5.9 5.9 5.8 5.7 5.6 5.2 4.3 0.'
0:5 0.7 0.8 0.8 0.3 0,8 0.8 O.S 0.8 0.9 0.8 0.8 0.8 0.8 0.8 0.8 0.8 O.S 0.8 0.7 O.g
10m
5m
100m

Figure 6-18: Radius factor generation

Figure 6-19 portrays the relative increment of radius factor at increasingly larger stope
strike length measured at a distance from the front face. Radius factor behaves asymptotically
with the stope strike length.

Radius Factor versus Stope Strike Length (AutoLisp)

— 20m from face

—
-«-15m from face
-»- 10m from face
-*- 9m from face
j ^ — . . .
—• * — 8m from face
2 8.0 a.^-— ----7m from face
•» 6m from face
-*- 5m from face

0 50 100 150 200 250 300 350 400 450

S t o p e Strike Length (m)

Figure 6-19: Radius factor versus stope strike length at different distance from front
face
Chapter 6 - Effect of Backfill on Open Stoping 115

6.7.2 Horizontal Displacement Behavior considered with Stope Strike Length

The procedure is repeated but using the Map3D boundary element program in order to
measure horizontal wall closure at 5, 6, 7, 8, 9, 10, 15 and 20 meters from the front face.
These measurements represent 10, 12, 14, 16, 18, 20, 30 and 40 meters respectively of
maximum unsupported strike length. As before, the assumption is that, beyond these
distances, in each case, the stope can be considered as a backfilled stope. There is a
decreasingly minor (asymptotic) effect on wall closure. Figure 6-20 portrays horizontal wall
closure measurement at 5 meters from the front face for a stope of 40 meter height and 100
meter strike length.

Grid line - wall closure measuring point

(5m c u r v e f o r 100m strike length - p r o g r e s s i v e l y m i n e d stope)

|Map30"SV50 IC:VDocu»onli and Setling*\Ciitlian C*ceie»\M> Docunmnli\The»i«\Map3DMhiee.001112005:01:04]

Figure 6-20: Horizontal displacement measurement at various distances from front face

Figure 6-21 illustrates the relative incremental decrease in wall closure at progressively
larger stope strike lengths. From this analysis it is evident that radius factor is correlated with
horizontal wall closure. Wall closure behaves asymptotically with stope strike length. The
increase in wall displacement is proportional to the degree of failure, which itself is limited in
degree (i.e., displacement) by the distance to the abutments.
Chapter 6 - Effect of Backfill on Open Stoping 116

Horizontal Wall Closure versus Stope Strike Length (Map3D)

— 20m from face

1
15m from face
-•-10m from face
-*- 9m from face
- - 8m from face
- - 7m from face
• 6m from face
-»- 5m from face

100 150 200 250 300 350

S t o p e Strike L e n g t h (m)

Figure 6-21: Horizontal displacement versus stope strike length at different distances
from front face

No significant effect of backfill on wall displacement was found from an elastic,

homogeneous, isotropic, numerical modeling perspective. This parametric assessment
analyzes the effect of backfill only, and does not incorporate other factors such as blasting
damage, rock heterogeneity, etc. The model is an approximation only.
However, this model predicts, that greater displacement leads to a proportional increase
in the degree of failure. Backfill alone does not affect the degree of failure in a meaningful
way. The following conclusion is reached: radius factor for a stope mined adjacent to backfill
should consider the exposed wall and at the same time consider the backfilled wall (in terms
of its effect on stability).

6.7.3 Design Curves -Avoca Mining Method

Figures 6-22 and 6-24 illustrate the increase of radius factor from its initial value
(represented by the isolated stope curve) to its maximum value (represented by the 380 meter
backfill curve). Figures 6-22 and 6-24 portray radius factors for stopes of 30 meter and 40
Chapter 6 - Effect of Backfill on Open Stoping 117

meter heights, respectively. The dashed line ( — ) represents the maximum radius factor
obtained at the middle of the stope, i.e., 200 meters from each end.

Radius Factor
Stope of 30m Height - 400m Strike Length »

Distance to Front Face (m)

Figure 6-22: Radius factor increment - stope of 30 meter height and 400 meter strike
length
Chapter 6 - Effect of Backfill on Open Stoping ill

Example calculation: radius factor measurement is obtained for point R of the isolated
stope A B C D (figure 6-23) and is also measured for the exposed stope plus the backfilled
stope A E F D (figure 6-23) which can then in its entirety be assumed to be open. In the case of
a 30 meter stope height, the radius factor increases from an initial value of 6.5 meters for a 20
meter strike length to a constant value of 8.4 meters at a 400 meter strike length, as illustrated
by figure 6-22 and depicted three dimensionally in figure 6-23.

20m

Figure 6-23: Radius factor increments for a stope of 30 meter height and 400 meter
strike length

For a stope of 30 meter height, where radius factor is measured at a mid-span of 10

meters (20 meter strike length), radius factor increases from an initial value of 6.5 meters (in
the isolated stope) and reaches the constant value of approximately 8.4 meters at 400 meter
strike length (mid-span remains 10 meters) for 380 meters of backfill.
Chapter 6 - Effect of Backfill on Oven Stopinz 111

Radius Factors
Stope of 40m Height - 400m Strike Length
in

4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
Distance to Front Face (m)

Figure 6-24: Radius factor increments for a stope of 40 meter height and 400 meter
strike length

This approach shows that it is numerically possible to assess the increase in radius factor
due to mining adjacent to backfill. After applying neural network analysis, the mine database
(see Chapter 4, Table 4-7) does not conclusively reflect a quantitative procedure, as a
consequence of the many parameters that affect dilution.
Chapter 7 - Conclusions and Recommendations 120

7 CONCLUSIONS AND RECOMMENDATIONS

Backfill plays an essential role in the efficiency of mining operations at the Musselwhite
ore deposit. Monitoring the true performance of backfill, both from safety and operational
perspectives, should be a routine part of operations to ensure that the quality of the mine's
final product results as planned.

Cement is the principle cost component of a backfill system. Strategies that optimize the
use (i.e., amount) of cement, while maintaining a high degree of quality in backfill material,
are of substantial benefit to the overall mining operation.

In order to determine analytically the vertical load exerted by unconsolidated backfill,

the coefficient of lateral earth pressure, K, must be obtained. A n inaccurate estimation of K
will in turn produce a sill mat design differing from the optimized solution. Numerical
modeling was used to determine directly the coefficient K by measuring the horizontal and
vertical stresses of a given grid element, and subsequently calculating the ratio between these
two stresses. The coefficient of lateral earth pressure K obtained via numerical modeling falls
between the active value and the at rest value. Additionally, a curve was fitted to the points
obtained from this numerical modeling and an equation was subsequently determined. This
equation can be used to obtain the K value for unconsolidated rockfill material confined
within stope walls.
The equation is:

K = 1.4 • Sin (</>)- 2 • Sin(0) +1

2
with 0° < <b <= 40°

Also, an analytical equation to determine the vertical load of unconsolidated backfill as a

function of backfill height was derived in this thesis. This equation accounts for the decrease
in vertical load as a result of stope inclination. Although this equation yields an
approximation, the numerical modeling conducted was shown to be highly correlated with
analytical results.
The equation is:
Chapter 7 - Conclusions and Recommendations 121

r-L f 2-K-tan(<f>)-z
•sin (/?)
2
1 -exp
L-sm\B)

The existing analytical equation for the rotational failure mode does not consider shear
strength on hangingwall/sill mat contact, due to the low strength of pastefill and/or the gap
that may occur at the contact surface (as a result of low dip angle), given the method of
pastefill delivery itself (i.e., via piping), and so it becomes difficult to maintain firm contact.
Methods of constructing sill mat pillars employing rockfill most likely can overcome this
problem, in nearly all cases, and so a new analytical equation that considers shear strength of
the wall contact was developed. As a result of this new analytical equation, there is a
significant increase in the factor of safety of a given sill mat.
The equation is:

, , C T -d +a-2-T
2
• d-L-sm (B) 2

(cj+d-y)> —— ? -—r — —
L-(L-J-cot(^))-sin (^) 2

Although analytical equations are highly useful in the design of sill mat pillars, this
thesis proposed that numerical modeling is a more accurate method of designing sill mat
pillars given numerical modeling's capacity of incorporating several additional factors that
influence stability. These factors include: material deformation, and the reduction in strength
properties throughout the modeling process (as a function of the strain on grid elements).
Among the strategies assessed, the finite difference program Flac 2D
was considered to be the
most suitable program of all those available, in addressing and understanding the mechanics
of the various failure modes of cemented rockfill sill mats, and the parameters involved in
rockfill sill mat design.

Design curves for the stability of cemented rockfill sill mats were developed to aid in the
design of new pillars or to reassess the performance of existing pillars at the Musselwhite
mine. It is imperative that, when mining under backfill, no personnel are exposed to the
danger, and remote mining methods are employed (with a minimum factor of safety of 1.2
for non-entry methods). This is the lowest useable factor of safety to ensure that the sill mat
remains integral during its exposure. The above statement notwithstanding, via numerical
Chapter 7 - Conclusions and Recommendations 122

modeling it is possible only to determine sill mat pillar stability with a factor of safety of 1.0,
and by incorporating a suitable factor of safety on the input strength parameters (cohesion)
the operator is able to produce a design within existing construction guidelines. It is also
essential that stability calculations dictating variation in fill recipes be confirmed by full-
scale field trials and verified by observations during actual mining. If at all possible,
numerical modeling is the preferred method of analysis (over analysis using analytical
equations).

The graphs shown in detail in Chapter 5 portray the stability curves developed via
numerical modeling. The x-axis represents the strength of the sill mat pillar (in terms of
cohesion) and the y-axis represents the minimum required vertical height of the pillar to be
stable. Each graph includes four stability curves for spans ranging from 6 to 12 meters. The
graphs are divided into 3 categories: zero strength in the hangingwall/sill mat contact, 50%
strength in the contact, and 100% strength assumed in the contact (cohesion value). Graphs
were developed for 90 degrees (vertical), and for 85, 80 and 75 degree stopes.

The rotational failure mode obtained via numerical modeling differs from that obtained
in previous existing research employing centrifuge models. Two rotational failure modes
were generated via numerical modeling, rotational-crushing and rotational-breaking. Neither
implied tensile failure at the footwall propagating at an angle of the same magnitude as the
stope dip angle, but in a direction defined by the supplementary angle, as predicted by
Mitchell's (1991) rotational failure mode equation. Rotational-crushing failure mode predicts
tensile failure at the footwall contact and shear (or volumetric) failure at the hangingwall
contact, whereas the rotational-breaking failure mode predicts a tensile failure propagating
vertically at mid-span, and a shear (or volumetric) failure at the hangingwall contact.

Musselwhite's Stability Graph was updated using the cavity monitoring system survey
database. A total of 138 cases were analyzed. These were divided into isolated stopes (48
cases) and stopes mined adjacent to backfill (90 cases). The points were plotted into the
empirical estimation of wall slough (ELOS) expressed in terms of radius factor (portrayed in
Chapter 7 - Conclusions and Recommendations 123

detail in Chapter 6), since the adjustment factor was obtained only for the radius factor case
(which would not be possible in the case of hydraulic radius).

Neural network analysis was implemented to determine i f isolated stopes would account
for less wall slough compared with stopes of the same exposed wall size, but where the
mining occurs adjacent to backfill. As a consequence of the many variables involved
(principally, drilling, blasting and geological heterogeneity), no conclusive results were
obtained from this analysis in order that the assumption could be made that stopes mined
adjacent to backfill will result in a greater degree of wall slough than isolated stopes.

The existing empirical design-stability methods were mostly developed for isolated
stopes. Although it is possible to calibrate the empirical design stability curves to site specific
conditions (e.g., mining adjacent backfilled stopes), the method proposed in this thesis uses
the same design curves developed for isolated stopes, but adjusts the magnitude of the radius
factor of the excavation to account for the portion of the stope containing backfill. The
figures shown in detail in Chapter 6, portray the incremental increase in radius factor that
must be progressively applied as mining advances, where mining adjacent to backfilled
stopes occurs. The x-axis represents the distance to the front face. This thesis made the
assumption that, for a given exposed strike length, the measuring point is one-half of the
maximum unsupported span, and beyond that maximum unsupported span the stope is
considered to be backfill. The y-axis displays the radius factor values. Two curves are
plotted, the lower representing radius factor for isolated stopes, and the upper representing
maximum radius factor obtained for a 380 meter backfilled stope (strike length) that can be
considered infinite. Two graphs were developed, one for a stope of 30 meter height and the
other for a stope of 40 meter height.

The increase in radius factor will vary depending upon exposed strike length. There is a
negligible incremental difference in the radius factor value for a backfilled stope of 100
meter strike length compared to a backfilled stope of "infinite" strike length. A stope of 400
meter strike length is considered to be the same mathematically as a stope of infinite strike
length.
Chapter 7 - Conclusions and Recommendations 124

A hypothetical effect should be that secondary stopes, where both abutments are CRF
pillars, will have greater instability as consequence of higher radius factor, and in turn will
suffer greater dilution (when compared to stopes with a single backfill abutment).

Radius factor calculations obtained in this thesis using the AutoLISP procedure (Milne
and Lunder, 1994), always yield the maximum value for each location measured (since the
stope is considered entirely without backfill). For this reason, the radius factor value obtained
in each case is the maximum value attainable. Consider a reduced percentage of the radius
factor maximum value as being effective in reducing wall closure, due to backfill's minimal
effects on wall closure as predicted by the numerical modeling results in this thesis.

The increased use of pastefill/CRF in mining operations demands a greater

understanding of the geomechanical effects of backfill and their implications on design. This
area of application will gain increasing acceptance in the future, as mining deposits are found
deeper underground and ground conditions prove to be more hazardous.
Chapter 8 - Future Work 125

8 FUTURE WORK

Implementing a full instrumentation campaign for determining in situ properties (i.e.,

strength, stress and strain) of cemented and uncemented rockfill is highly recommended, as
initially proposed in the research by Scoble, et al. (1987). A better understanding of these
properties will contribute to the optimization of CRF sill mat designs. Additionally, optimal
design of stopes, in terms of excavation dimensions, leads to a reduction in wall instability,
where mining adjacent to backfilled stopes occurs. A n optimized design directly reduces
mining costs.

Unconsolidated rockfill vertical loads along the span of a backfilled stope could be
determined using an array of earth pressure cells installed on top of the sill mat pillar. In
practice, this is a very difficult task due to the physical hazards of entering an open stope.
Remoted equipment would thus be employed to install this equipment, minimizing the risk to
mine personnel.

CRF sill mat strength properties could be determined via an in situ pressuremeter
apparatus. Pressuremeters have cylindrical rubber membranes that are inflated with
pressurized fluid (e.g., water, gas, or oil) to measure volume change. Entire stress-strain-
strength curves could then be derived, as well as in situ total horizontal stress, shear modulus,
shear strength, and limit pressure. Types of pressuremeter testing include: the pre-bored
(Menard), the push-in device, the self-boring, and the full-displacement, types (e.g., cone
pressuremeter or pressiocone). Pre-bored types are preferred considering the extreme
difficulty of penetrating cemented rockfill.

The strength on the hangingwall/sill mat contact could be indirectly inferred using earth
pressure cells. The higher the load exerted by C R F backfill on the contact, the higher the
inferred strength value. Horizontal stress, as well as the stress parallel to the hangingwall
contact, should increase during mining of adjacent stopes (wall convergence). These load
measurements would help in determining the crushing mode instability of sill mats (not
analyzed in this research). Another possibility for determining the strength on
hangingwall/sill mat contact is to obtain core samples through the contact and, using a shear
Chapter 8 - Future Work 126

box apparatus, test the shear strength of the sample obtained. Samples should be cored at
various depths of the contact and also at various points along the strike length of the sill mat
pillar, in order to obtain a representative measurement of the strength of the hangingwall/sill
mat pillar contact.

The effects on sill mat stability incorporating the actual vertical load distribution along
the stope span, as well as, greater vertical height of rockfill, could be effected by numerical
modeling analysis. In this thesis, the analyses were completed using a constant vertical load
along the span.

A three dimensional numerical modeling program could be used to model backfill

behavior. Three dimensional numerical modeling would help in that it can model the depth
dimension, which cannot be modeled using the two dimensional version. The two
dimensional version cannot accurately model the overall three dimensional geometry
associated with A V O C A mining.

Also, numerical modeling to determine stope stability, where mining adjacent to

backfill, could be re-run in three dimensions. The results may help determine how backfill
affects displacement around an opening adjacent to the backfill itself. Instrumentation of the
stope wall by employing extensometers could be used to determine the real displacements of
both isolated stopes and of stopes mined adjacent to backfill.

Stopes with backfill below (longitudinal mining method with backfill on the floor of the
stope) could be analyzed. The analysis performed in this thesis deals only with radius factor
adjustments for stopes backfilled along the strike. Backfill on the floor of the stope would
have a direct effect on the magnitude of the radius factor.

To determine solely the influence of backfill on wall stability, it would be necessary to

isolate all external factors that account for wall instability. Those factors that can be isolated
are mainly drilling and blasting. In order to isolate these factors, it would be necessary to
survey all blastholes (or at least those blastholes drilled adjacent to the hangingwall and
Chapter 8 - Future Work 127

footwall of the orebody). A direct measurement of blasting performance could be obtained

via seismic wave method. Results from this analysis could be combined with drill deviation
to account for unplanned wall slough as a result of dynamic effects. Doing so, would make it
possible to determine the actual, real-world effect of backfill on wall stability.
References 128

REFERENCES

Arioglu, E., 1983. "Engineering Properties of Cemented Aggregate Fill for Uludag
Tungsten Mine of Turkey". Mining with Backfill, Proc. Of the Int. Symp., Lulea, Sweden,
June 7-9, 1983, publ. by A . A . Balkema, Rotterdam, Neth., pp. 3-9.

Arjang B., Herget G., 1997. "In Situ Ground Stresses in the Canadian Hardrock Mines:
An Update". Int. J. Rock Mech. Min. Sci. Vol. 34, No. 3/4, p. 652, paper No. 15, 1997.

Clark, L . M . , 1998. "Minimizing Dilution in Open Stope Mining with a Focus on Stope
Design and Narrow Vein Longhole Blasting". MASc. Thesis, Faculty of Graduate Studies,
Department of Mining and Mineral Process Engineering.

Coates, D.F., 1981. "Caving, Subsidence, and Ground Control", Rock Mechanics
Principles, Minister of Supply and Services Canada, Chapter 5.

Barrett, JR, Coulthard, M A and Dight, P.M., 1978. "Determination of Fill Stability,
Mining with Backfill". 12th Canadian Rock Mechanics Symposium, C I M Special Volume 19,
Sudbury, Ontario. May 23-25

Beer, G. and Meek, J.L., 1982. "Design Curves for Roof and Hangingwalls in Bedded
Rock Based on Voussoir Beam and Plate Solutions". Trans Instn Min. Metall.

Bishop, A . W . , 1958. "Test Requirements for Measuring the Coefficient of Earth

Pressure at Rest". Proceedings, Brussels Conference on Earth Pressure Problems, Vol 1, pp.
2-14.

Blight, G.E., 1984, "Soil Mechanics Principles in Underground Mining",

Geotechniques, Vol. 110, No. 5, pp. 567-581.
References 129

Brown, E.T. (ed.) 1982: "Rock characterization, testing and monitoring". International
Society for Rock Mechanics suggested methods. Pergamon Press, Oxford., p. 211 (Contains
15 suggested methods.)

De Souza E., Archibald J.F., Dirige A . P . 1998. "Economics and Perspectives of

Underground Backfdl Practices in Canadian Mines". Mine Backfill 1998, The Canadian
Institute of Mining, Metallurgy and Petroleum, C I M , Canada.

Dirige A.P., De Souza E. and Chew J., 2001. "Centrifuge Physical Modeling of Paste
Fill Sillmats". Proceedings of the 7 International Symposium of Mining with Backfill. USA.
th

Graham S. and Brummer R.K.., 2001. "Backfdl Design for Deep, Underhand Drift-and-
Fill Mining". Proceedings of the 7 International Symposium of Mining with Backfill. USA.
th

Grice T., 1998. "Underground Mining with Backfdl". The 2nd Annual Summit - Mine
Tailings Disposal Systems, Brisbane, Australia.

Gurtunca R.G., Toper A.Z., Squelch A.P. and Dlokweni T., 2001. "Rock Engineering
Criteria for Backfill in Ultra-Deep Level Mining Environments". Proceedings of the 7 th

International Symposium of Mining with Backfill. USA.

Hassani F., Archibald J., 1998. "Mine Backfdl". C D - R O M format. The Canadian
Institute of Mining, Metallurgy and Petroleum Bulletin CIM, Canada.

Jaky, J., 1944. "The Coefficient of Earth Pressure at Rest", Journal for Society of
Hungarian Architects and Engineers, October, pp. 355-358.

Janssen, H . A . , 1895. "Versuche uber getreidedruck in silozellen." Zeitschrift des

Vereines Deutscher Ingenieure, Berlin, Germany, 39(35), pp. 1045-1049.

Knissel, W. and Helms, W., 1983. "Strength of Cemented Rockfill from Washery Refuse
Results from Laboratory Investigations". Proceedings of International Symposium on Mining
with Backfill, A A Balkema, Rotterdam, pp. 31-37.
References 130

Kuganathan K . and Sheppard I., 2001. "A Non-Segregating "Rock Paste Fill" (RPF)
Produced by Co-Disposal of Cemented De-Slimed Tailings Slurry and Graded Rockfill".
Proceedings of the 7 International Symposium of Mining with Backfill. U S A .
th

Kuganathan K . , 2001. "A Method to Design Optimum Pastefdl Mixes through Flow
Cone and Mini Slump Cone Testing". Proceedings of the 7 th
International Symposium of
Mining with Backfill. USA.

Lewis T.D., 1998. "Musselwhite: An Exploration Success". The Canadian Institute of

Mining, Metallurgy and Petroleum Bulletin CIM, Canada. Issue Number: 1017.

Ley G . M . M . , Steed C M . , Bronkhorst D., Gustas R., 1998. "Mining Under Backfill". The
Canadian Institute of Mining, Metallurgy and Petroleum Bulletin C I M , Canada. Issue
Number: 1020.

Mah S., 1997. "Quantification and Prediction of Wall Slough in Open Stope Mining
Methods". M A S c Thesis, University of British Columbia, p. 290.

Marcinyshyn, K., Pakalnis, R., Dunbar, S. and Vongpaisal, S., 1996. "Sill Mat Design".

Mathews, K.E., Hoek, E., Wylie, D. and Stewart., 1981. Prediction of stable excavation
spans for mining at depths below 1000m in hard rock. C A N M E T , DSS Serial No. 0SQ80-
00081.

Milne, D. 1997. "Underground Design and Deformation Based on Surface Geometry".

PhD Thesis, University of British Columbia, p. 220.

Mitchell R.J., 1991. "5/7/ Mat Evaluation Using Centrifuge Models". Mining Science
and Technology, Elsevier Science Publisher B.V., 13: pp. 301-313.

Nantel J. H., 1998. "Recent Developments and Trends in Backfill Practices in Canada".
Australian Institute of Mining and Metallurgy Symposium on Mining with Backfill,
Brisbane, Australia, April, 1998.
References 131

Newman P.L., Robert J. and Ross K., 2001. "The Optimization of High Density Backfill
at the Stratoni Operations", Greece. Proceedings of the 7 th
International Symposium of
Mining with Backfill. USA.

Nichols, H.R., Johnson, C.F., and Duvall, W.I., 1971, "Blasting Vibrations and Their
Effect on Structures", Bulletin 656, Bureau of Mines, Washington, D.C.

Nickson, S. 1992. "Cable support guidelines for underground hard rock mine
operations". M A S c Thesis, University of British Columbia, p. 223.

Pakalnis, R., and Vongpaisal, S., 1998. "Empirical design methods - UBC
Geomechanics". Proceedings of the 100th AGM-CIM, Montreal, 3 - 7 May 1998., pp. 1-12

Potvin, Y . , 1988, "Empirical Open Stope Design in Canada". PhD Thesis, Faculty of
Graduate Studies, Department of Mining and Mineral Process Engineering.

Quesnel, W., 1988. "Optimum Grading for Hemlo CRF (Consolidated Rock Fill)",
Internal Report, Lac Minerals.

Quesnel, W. J. F.; de Ruiter, H . & Pervik, A . , 1989. "The Assessment of Cemented

Rockfill for Regional and Local Support in a Rockburst Environment". Proceedings, 4th
International Symposium on Mining with Backfill, Montreal, Quebec.

Reimbert, M . & A., 1976. "SILOS Theory and Practice", Trans Tech Publications, Vol.
1, No 3, p. 250.
Scoble, M . , Piciacchia, L . and Robert, J.M., 1987. "In Situ Testing in Underground
Backfilled Stopes". C I M Bulletin, 80(903), July, pp. 33-38.

Smith, J.D. & Mitchell, R.J., 1982. "Design and Control of Large Hydraulic Backfill
Pours", C I M Bulletin, Vol. 75, No. 838, Feb. 1982, pp. 102-111

Stone, D.M.R., 1993. "The Optimization of Mix Designs for Cemented Rockfill",
Minefill 93, Johannesburg, S A I M M , 1993, pp. 249-253
References 132

Terzaghi, K., 1920. "Old Earth Pressure Theories and New Test Results", Engineering
News Record, Vol. 85, p. 632.

Terzaghi, K., Peck, R.B., 1948. "Soil Mechanics in Engineering Practice". Wiley and
Sons, New York, p. 549.

Thomas, E.G., Nantel, J.H. and Notley, K.R., 1979. "Fill Technology in Underground
Metalliferous Mines". International Academic Services Limited, Kingston, Ontario, Canada,
p. 293

Thomas, E.G., and Vance W.E., 1969. "Characteristics of hydraulic fill - Portland
cement mixtures". Proc. Rock Mech. Symp., Sydney University.

Tschebotarioff, G.P., 1973. "Foundations: retaining and earth structures: the art of
design and construction and its scientific basis in soil mechanics ". New York: McGraw-Hill
[1973] First ed. published in 1951 under title: Soil mechanics, foundations, and earth
structures.

Yu, T.R., (unpublished). "ConsolidatedRockfill".

References 133

Sources

Rocscience Inc., 2004. "Dips and Unwedge - User's Guide".

Itasca Consulting Group, 2002. "Flac2D

Fast Lagrangian Analysis of Continua - User's
Guide ".

Mine Modelling Pty Ltd., 2004. "Map3D - User's Guide ".

MTS Systems Corporation, 2004. "Stiff Testing Machine - UCS tests".

Optech Incorporated, 2004. "Cavity Monitoring System - CMS surveys ".

Ward Systems Group Inc., 1997. "NeuroShell Predictor - User's Guide".

Appendix A - Analytical Equations to Determine Vertical Load of Backfill 134

APPENDIX A

Proposed Analytical Equation to Determine Vertical Load of Backfill

This method is based on the Janssen's method of silo theory, but considers inclined
walls.

Differential inclined slice in a silo

This method considers the same characteristics of Janssen's differential slice but
incorporates the following assumptions: the horizontal pressure is equal in magnitude and in
opposite direction on both silo walls; and the tangential stress is equal in magnitude for both
sides of the silo walls. This will give the following "approximation" for the force equilibrium
expression:

2-b-a -2-b-(a da )- ' ' '

y y+ y
2 dy
o
a
+ 2-b- dy
r =0
sin(/?)

Simplifying the above equation and dividing by 2b*dy gives the differential form:

r =o
dy b • sin(/?)
Appendix A - Analytical Equations to Determine Vertical Load of Backfill 135

A further generalization of the above equation is obtained by defining the hydraulic

radius of the silo for any cross section of any shape as HR = S / P, where S is the area and P
is the perimeter. With this notation, the equilibrium equation becomes:

der.
-r = o
dy HR • sin(/?)

The friction law would be:

a, =cr -tan(^)
n

However, in the inclined case a n relates to a x with the angle of the wall (3, and then the
friction law can be written as:

a, =
sin (/?)
2

The two unknowns, the average vertical stress a y and the average horizontal stress a x

are necessary to solve equation 3. To obtain a solution, it is necessary to employ the

coefficient of lateral earth pressure K previously defined.

K-tm(0)-a v

sin (/?)
2

Combining equations 3 and 6, yields:

da K-tm(<f>)-cT
+ — = 0
dy HR- sin (/?) 2

The integrating factor would be:

K •tan(^) K •tan(^)•y
«O0 = exp J- dy = exp,
HR- sin (j3)
2
HR- sin (/?)
2
Appendix A - Analytical Equations to Determine Vertical Load of Backfill 136

The solution of the first order differential equation would be:

f
^ - t a n f ^ V ^ dy
r
v l HR- sin
eXP 2
(/?)
r
K • tan(<f>) • y^
exp
HR-sin 2
(8)

Solving the integral yields to the following equation:

r -HR- sin (/?)

2
K • tanQ) • y f
K-\an((j>)-y^
exp + C exp 10
K • tan(» HR- sin (/?)
2
HR- sin (j3)
2

Using the boundary conditions, the following solution is obtained:

^ y-HR N
' A: • tan(^) • z ^
sin (/?)
2
1-exp 11
/C-tan(^) HR-sin (J3)
2

or,

CT
vO) = °-v Ato - s i n ( ^ ) ' 1-exp
2 12
HR-sin 2
(B)

Where:

- P = stope's dip

Now, when infinite stope strike length is considered:

L-SL L 13
HR lim SL—><x>2(1 + SL) 2

Where:
L = stope's span; and
SL = stope's strike length.
Appendix A - Analytical Equations to Determine Vertical Load of Backfill 137

Then equation 11 can be simplified to:

Y-L f
2 tariff)-z^
o\0) = sin (/?) 1 - e x p
2
14
I-sin (/?)
2
Appendix B - Analytical Equations to Determine Sill Mat Failure Modes 138

APPENDIX B

Proposed A n a l y t i c a l E q u a t i o n to Determine R o t a t i o n a l S i l l M a t F a i l u r e M o d e s

Driving momentum with respect to point O is:

L
CT-L- + d-L-y
2 2 • tan(/?) 2 2 • tan(/?)

Whereas resisting momentum with respect to O is:

d f
d ^
cr. +a• T, L • sin(/?)
sin(A) 2-sin(/?) sin(/?)

Then, rotational failure would develop when:

f
d 2
^
G + a • r, • d • L
' 2 v sin (/?)2
y
(cr +d-y)>
v
f
i . l . L-
2 tan(/?)

Simplifying equation 3 yields:

f
d 2
^
+a • 2• T •d•L
sin (/5)y
2

(cr +d-y)>
V

L•

or:

a, • d + a- 2-T • d•L-sin (/J)

2 2

K + d • r) >
(

L-(L-d-cot(j3))-sin (j3) 2
Appendix C - Interface Elements 139

APPENDIX C f

Interface Elements

A n interface or slip-plane model is available to represent distinct interfaces between two

or more portions of the grid. The interfaces are planes upon which slip and/or separation are
allowed, thereby simulating the presence of faults, joints or frictional boundaries.
A n interface is represented as a normal and shear stiffness between two planes which
may contact one another (figure A).

Figure A: An interface represented by sides A and B, connected by shear (k ) and s

normal (k„) stiffness springs

Where:
- S = slider;
- T = tensile strength;
- k = normal stiffness; and
n

- k = shear stiffness.
s

Three options are available for specifying the conditions of the interface, as follows:

Glued Interfaces - i f interfaces are declared glued, no slip or opening is allowed, but
elastic displacement still occurs, according to the given stiffnesses.
Coulomb Shear-Strength - the Coulomb shear-strength criterion limits the shear force
by the following relation:
Appendix C - Interface Elements 140

Fs max = c- L + Fn- tan(^) 1

Where:

- c = cohesion (in stress units) along the interface,

- L = effective contact length, and

- § = friction angle of interface surfaces. In addition, the interface may dilate at the
onset of slip (non elastic sliding). Dilation is governed in the Coulomb model by a
specified dilation angle, v|/.

Tension Bond - If a (positive) tensile bond strength is specified for an interface, each
segment of the interface acts as if it is glued (elastic response only) while the magnitude of
the tensile normal stress is below the bond strength. If the magnitude of the tensile normal
stress of a segment exceeds the bond strength, the bond breaks for that segment, and the
segment behaves thereafter as unbonded (separation and slip allowed, in the normal way).

The apparent stiffness (expressed in stress-per-distance units) of a zone in the normal

direction is:

4 ^
f

K + --G
3
max
Az mm
V J

Where:

- K = bulk modulus,

- G = shear modulus, and

- Az m i n = is the smallest width of an adjoining zone in the normal direction.

Bulk modulus, K, and shear modulus, G, are related to Young's modulus E, and
Poisson's ratio v, by:
Appendix C - Interface Elements 141

K=-
3-(l-2-v)

G=
2-(\ + v)

+ Itasca Consulting Group, 2002. "Flac2D

Fast Lagrangian Analysis of Continua -
User's Guide ".