Numerical Modelling of Surface Subsidence Associated With Block Cave Mining Using A Finite Element / Discrete Element Approach

NUMERICAL MODELLING OF SURFACE SUBSIDENCE
ASSOCIATED WITH BLOCK CAVE MINING USING A

FINITE ELEMENT / DISCRETE ELEMENT APPROACH
by
Alexander Vyazmensky
Master of Applied Science, University of British Columbia, 2005
THESIS SUBMITTED IN PARTIAL FULFILLMENT OF

THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
In the
Earth Sciences
© Alexander Vyazmensky 2008
SIMON FRASER UNIVERSITY
Fall 2008
All rights reserved. This work may not be

reproduced in whole or in part, by photocopy
or other means, without permission of the author.
http://sites.google.com/site/alexvyazmensky/
ABSTRACT
Recent years have seen a major increase of interest in the block cave mining
method which is characterized by extraction of a massive volume of rock usually
accompanied by the formation of a significant surface depression above and in the
vicinity of the mining operation. The ability to predict surface subsidence is
important for mine planning, operational hazard assessment and evaluation of
environmental and socio-economic impacts. Owing to problems of scale and lack
of access, the fundamental understanding of the complex rock mass response
leading to subsidence is limited as are current subsidence prediction capabilities.
Through the use of an integrated FEM/DEM-DFN modelling technique this thesis

presents a new approach to simulation of block caving induced surface
subsidence allowing physically realistic simulation of subsidence development
from caving initiation to final subsidence deformation. As part of the current
research, a fundamental issue in modelling, the selection of representative
equivalent continuum rock mass modelling parameters, is investigated and a
procedure for calibration of modelling parameters devised. Utilizing a series of
conceptual numerical experiments our fundamental understanding of the
mechanisms and the role of the factors controlling block caving subsidence
development is investigated. Valuable insights gained from this work are
summarized in a preliminary subsidence classification and an influence
assessment matrix of the governing factors. These are intended as an aid to
engineering judgment for decision makers at the pre-feasibility and mine design
stages.
This study also addresses one of the most challenging problems in mining rock
engineering - the interaction between block cave mining and a large overlying
open pit, focusing on caving induced step-path failure initialization. Using a novel
approach to modelling data analysis a clear link between caving propagation, step-
ii
path failure development within the slope, and the resultant surface subsidence is
established. In addition, FEM/DEM-DFN modelling is applied to the preliminary
analysis of the block caving triggered slope failure at Palabora open pit.
This research represents a valuable contribution to block caving geomechanics

and is a major step forward in the understanding of complex block caving
subsidence phenomena, paving the way to more reliable assessment of caving
induced subsidence deformations.
Keywords: rock mechanics; numerical modelling; block cave mining; block

caving subsidence; open pit - caving mining interaction; FEM/DEM-DFN
iii
DEDICATION
I dedicate this work to my parents, Mikhail and Sofia, who offered unconditional
love and support and have always been there for me. Thank you so much.
iv
ACKNOWLEDGEMENTS
It is my pleasure to acknowledge the roles of several individuals who were

instrumental for completion of my Ph.D. research.
First of all, I would like to express my gratitude to Dr. Doug Stead, who encouraged
me to pursue this project and taught me the art of rock engineering. I truly enjoyed
working in a research environment that stimulates original thinking and initiative,
which he created. Dr. Stead‟s skilful guidance, innovative ideas and stoic patience
are greatly appreciated.
I would like to acknowledge the valuable input of Dr. Davide Elmo, who contributed
to many discussions that helped to shape this project, assisted with FracMan and
was always willing to help resolving the most difficult modelling issues.
I would also like to acknowledge helpful suggestions from my committee members:

Dr. Erik Eberhardt, Dr. Scott Dunbar, Dr. Malcolm Scoble and Dr. Diana Allen.
This work would not materialize without the financial support of Rio Tinto Pty and
technical support by Rockfield Technology Ltd and Golder Associates Ltd. I would
like to recognize the important roles of Allan Moss, Andre van As, Dr. Jon Rance,
Dr. Melanie Armstrong and Dr. Steve Rogers.
These acknowledgements would not be complete without mentioning my research

lab colleagues: David van Zeyl, Matthieu Sturzenegger and Andrew Beveridge. It
was a great pleasure working with them and I appreciate their ideas, help and good
humour. Special thanks goes to Roderick Tollenaar for sharing valuable data.
My deepest appreciation belongs to my family for their patience and understanding.
With regards to numerous questions about my future academic endeavours from

family and friends I shall answer in the words of Sir Winston Churchill: “Now, this is
not the end. It is not even the beginning of the end. But it is, perhaps, the end of
the beginning”.
v
TABLE OF CONTENTS
Approval ...................................................................Error! Bookmark not defined.

Abstract ............................................................................................................... ii
Dedication .......................................................................................................... iv
Acknowledgements ............................................................................................ v
Table of Contents .............................................................................................. vi
List of Figures ..................................................................................................... x
List of Tables ................................................................................................. xviii
Chapter 1: Introduction ...................................................................................... 1
1.1 Block Caving Mining and Associated Surface Subsidence ................ 1
1.2 Research Objectives .......................................................................... 3
1.3 Thesis Organization ........................................................................... 4
Chapter 2: Literature Review ............................................................................. 6
2.1 Introduction ........................................................................................ 6
2.2 Stages of Subsidence Development and Associated Rock
Mass Failure Mechanisms .................................................................. 7
2.3 Subsidence Characterization ............................................................. 9
2.4 Factors Governing Subsidence Development .................................. 12
2.5 A Block Caving Subsidence Analysis Toolbox ................................. 15
2.5.1 Empirical Methods ........................................................................ 15
2.5.2 Limit Equilibrium Techniques ........................................................ 17
2.5.3 Numerical Approaches.................................................................. 21
2.6 Summary .......................................................................................... 29
Chapter 3: Research Methodology and Theory ............................................. 31
3.1 Introduction ...................................................................................... 31
3.2 Modelling Approach Selection .......................................................... 31
3.2.1 Analysis of Available Modelling Approaches ................................. 31
3.2.2 Fracture Mechanics Based Hybrid Continuum/Discontinuum
Approach ....................................................................................... 33
3.3 Fundamentals of Fracture Mechanics .............................................. 34
3.4 Implementation of the Hybrid Continuum-Discontinuum with
Fracture Approach in the ELFEN Code .............................................. 36
3.4.1 Solution Procedure for Continuum to Discontinuum
Transition ...................................................................................... 37
3.4.2 Constitutive Models for Rock Material........................................... 38
vi
3.4.3 Continuum Degradation through Fracturing .................................. 41

3.4.4 Contacts Interaction Handling ....................................................... 43
3.4.5 Applications of ELFEN in Rock Engineering ................................. 44
3.5 Rock Mass Representation Options in Context of Fracture
Mechanics Based FEM/DEM Modelling ........................................... 45
3.6 Integrated Use of Discrete Fracture Network Models and
ELFEN Code .................................................................................... 47
3.6.1 DFN Approach to Discontinuity Representation ............................ 47
3.6.2 Integration of DFN Models into ELFEN ......................................... 49
3.7 General Considerations in Modelling Methodology .......................... 51
3.8 Summary .......................................................................................... 51
Chapter 4: Use of Equivalent Continuum Rock Mass Properties in the
Discrete Fracture Network FEM/DEM Modelling of Block Caving
Induced Subsidence ......................................................................................... 52
4.1 Introduction ...................................................................................... 52
4.2 Rock Mass Equivalent Continuum Mechanical Properties ............... 52
4.3 Derivation of Rock Mass Mechanical Properties ................................. 54
4.3.1 Analytical Methods ........................................................................ 54
4.3.2 In-situ Testing ............................................................................... 55
4.3.3 Numerical Modelling ..................................................................... 55
4.3.4 Empirically Based Rock Mass Classification Systems. ................. 56
4.4 Comparative Study of Rock Mass Strength and Deformability
Parameters Derived through RMR76, GSI and Q Rock Mass
Classification Systems ...................................................................... 60
4.4.1 Properties Derivation .................................................................... 60
4.4.2 Rock Mass Deformation Modulus ................................................. 64
4.4.3 Rock Mass Friction Angle ............................................................. 65
4.4.4 Rock Mass Cohesion .................................................................... 66
4.4.5 Rock Mass Tensile Strength ......................................................... 66
4.4.6 Summary ...................................................................................... 67
4.5 Evaluation of Applicability of Rock Mass Classifications Derived
Equivalent Continuum Properties for Modelling of Block Caving
Induced Surface Subsidence .............................................................. 68
4.5.1 Modelling Methodology ................................................................. 68
4.5.2 Constraining Criteria ..................................................................... 74
4.5.3 Modelling Results.......................................................................... 76
4.5.4 Conclusions .................................................................................. 89
4.6 Summary .......................................................................................... 89
Chapter 5: Conceptual Modelling Study of the Factors Controlling
Block Caving Subsidence Development ......................................................... 91
5.1 Introduction ...................................................................................... 91
5.2 Model Setup and Analysis Strategy .................................................. 91
5.3 Influence of Jointing ......................................................................... 93
5.3.1 Effect of Joint Orientation .............................................................. 93
5.3.2 Effect of Joint Persistence .......................................................... 108
vii
5.3.3 Effect of Joint Shear and Normal Stiffness ................................. 111

5.4 Influence of Faults .......................................................................... 114
5.4.1 Effect of Fault Location ............................................................... 114
5.4.2 Effect of Fault Inclination ............................................................. 127
5.5 Influence of Rock Mass Strength and Deformability
Characteristics................................................................................ 137
5.6 Influence of Stress Environment..................................................... 142
5.7 Influence of Varying Lithological Domains ...................................... 147
5.7.1 Effect of Varying Strength between Ore and Host Rock ............. 147
5.7.2 Effect of Varying Joint Orientation within the Ore and Host
Rock............................................................................................ 154
5.8 Influence of Block Depth and Excavation Volume .......................... 160
5.9 Results Synthesis ........................................................................... 169
5.10 Summary ........................................................................................ 178
Chapter 6: Block Caving Induced Instability in Large Open Pit Slopes .... 179
6.1 Introduction .................................................................................... 179
6.2 Transition from Open Pit to Block Cave Mining .............................. 179
6.3 Characteristic Slope Failure Mechanisms in Large Open Pits ........ 181
6.4 Conceptual Study of Block Caving Induced Step-path Driven
Failure in Large Open Pit Slope ..................................................... 187
6.4.1 Modelling Methodology ............................................................... 187
6.4.2 Modelling Results........................................................................ 190
6.4.3 Conclusions ................................................................................ 200
6.5 Preliminary Modelling of Block Caving Induced Failure of the
North Wall, Palabora mine ............................................................. 201
6.5.1 Problem Description.................................................................... 201
6.5.2 General Approach in the Current Modelling Analysis.................. 207
6.5.3 Model Setup ................................................................................ 208
6.5.4 Modelling Results and Discussion .............................................. 211
6.5.5 Conclusions ................................................................................ 215
6.6 Summary ........................................................................................ 215
Chapter 7: Conclusions and Recommendations for Further Research ..... 217
7.1 Conclusions .................................................................................... 217
7.1.1 Current State of Knowledge of Block Caving Induced
Subsidence ................................................................................. 217
7.1.2 FEM/DEM-DFN Approach to Subsidence Analysis..................... 218
7.1.3 Modelling Input Parameters ........................................................ 219
7.1.4 Factors Governing Block Caving Subsidence Development ....... 220
7.1.5 Large Open Pit - Block Caving Interaction .................................. 221
7.2 Key Scientific Contributions ............................................................ 222
7.3 Recommendations for Further Research ....................................... 223
7.3.1 Equivalent Continuum Rock Mass Properties ............................. 223
7.3.2 Modelling of 3D Aspects of Block Caving Subsidence
Development ............................................................................... 223
7.3.3 ELFEN Code Enhancements ...................................................... 223
viii
7.3.4 In-situ Subsidence Characterization ........................................... 224

Reference List ................................................................................................. 225
Appendix A. Preliminary Statistical Summary of Conceptual Study
Modelling Results ............................................................................................ 237
Appendix B. DFN Parameters for Palabora Model ....................................... 239
Appendix C. Development of North Wall Failure at Palabora mine
(photos) ........................................................................................................... 240
ix
LIST OF FIGURES
Fig. 1.1 Typical block caving mine layout. .......................................................... 2

Fig. 1.2 Caving initiation and ore extraction........................................................ 2
Fig. 1.3 Block caving mining and associated surface subsidence. ..................... 2
Fig. 2.1 Development of surface subsidence induced by block caving ............... 8
Fig. 2.2 Block caving subsidence characterization terminology ........................ 10
Fig. 2.3 Example of surface subsidence zonation at Northparkes mine ........... 10
Fig. 2.4 Relative frequency of break angles for different rock mass
qualities (Laubscher‟s RMR), in the subsidence craters of
underground mines by caving methods ............................................... 12
Fig. 2.5 Beneficial and detrimental effects of major weakness planes ............. 15
Fig. 2.6 Empirical chart relating MRMR and cave (break) angle ...................... 16
Fig. 2.7 Progressive hanging wall failure sequence with increasing depth
of mining .............................................................................................. 17
Fig. 2.8 Idealised model for limit equilibrium analysis of progressive
hangingwall caving .............................................................................. 18
Fig. 2.9 Toppling of steeply dipping hangingwall strata .................................... 20
Fig. 2.10 Example of problem discretization using FEM ..................................... 22
Fig. 2.11 Various approaches to model rock engineering problems: (a)
continuum, (b) discrete and (c) hybrid ................................................. 23
Fig. 2.12 Design chart for estimation the angle of break in a transition from
open pit to underground mining by block/panel caving ........................ 27
Fig. 3.1 Fracture mechanics basic crack tip loading modes (based on
Whittaker et al., 1992). ........................................................................ 34
Fig. 3.2 Rankine Rotating Crack model: (a) yield surface and (b)
softening curve .................................................................................... 39
Fig. 3.3 Yield surface for the conventional Mohr Coulomb model .................... 39
Fig. 3.4 Yield surfaces for (a) Mohr Coulomb and (b) Mohr Coulomb with
Rankine tensile cut-off criteria ............................................................. 40
Fig. 3.5 Crack insertion in ELFEN .................................................................... 43
x
Fig. 3.6 Comparison of contacts interaction handling in DEM and

FEM/DEM methods ............................................................................. 44
Fig. 3.7 DFN model development workflow ...................................................... 49
Fig. 3.8 Example of a FracMan model (adapted after Elmo, 2006, with
permission) .......................................................................................... 50
Fig. 4.1 RMR, Q and rock mass deformation modulus correlations ................. 58
Fig. 4.2 GSI and rock mass deformation modulus correlation proposed
by Hoek & Diederichs (2006) ............................................................... 58
Fig. 4.3 Comparison of rock mass strength and deformability
characteristics derived using RMR76, Q and GSI RMC systems ......... 63
Fig. 4.4 Percentile change in properties magnitudes with increasing rock
mass rating .......................................................................................... 64
Fig. 4.5 ELFEN model setup and response evaluation procedure ................... 70
Fig. 4.6 Physical model of progressive caving development (adapted
after Kvapil, 2004 with permission). ..................................................... 71
Fig. 4.7 Derivation of joints shear stiffness using Barton‟s (1982)
correlation ............................................................................................ 73
Fig. 4.8 Constraints for ELFEN modelling of block caving induced
subsidence .......................................................................................... 76
Fig. 4.9 Caving response with RMR equivalent continuum properties ............. 80
Fig. 4.10 Caving response with Q equivalent continuum properties (90%
tensile c.o.) .......................................................................................... 81
Fig. 4.11 Caving response with Q equivalent continuum properties (70%
tensile c.o.) .......................................................................................... 82
Fig. 4.12 Caveability assessment using Laubscher‟s chart for studied
equivalent continuum properties .......................................................... 83
Fig. 4.13 Simulation of cave development progression for RMR equivalent
continuum properties. .......................................................................... 84
Fig. 4.14 Simulation of cave development progression for Q (90% tensile
cut-off) equivalent continuum properties. ............................................. 85
Fig. 4.15 Simulation of cave development progression for Q (70% tensile
cut-off) equivalent continuum properties. ............................................. 86
Fig. 4.16 Simulation of subsidence development with RMR equivalent
Fig. 4.17 Simulation of subsidence development with Q equivalent
Fig. 5.1 Assumed fracture orientations for BC (a), J1 (b), J2 (c), J3 (d)
and J4 (e) models ................................................................................ 94
xi
Fig. 5.2 Subsidence crater formation for Base Case (a) and J1(b) models ...... 96
Fig. 5.3 Subsidence crater formation for J2 (a) and J3 (b) models ................... 97
Fig. 5.4 Subsidence crater formation for J4 model ........................................... 98
Fig. 5.5 Variation of vertical stress (Pa) contours with caving at 5% ore
extraction for Base Case and J2 models ............................................. 98
Fig. 5.6 Subsidence at 100% ore extraction for BC (a), J1 (b), J2 (c), J3
(d) & J4 (e) models ............................................................................ 101
Fig. 5.7 Surface profiles at the end of ore extraction for BC, J1, J2, J3
and J4 models ................................................................................... 102
Fig. 5.8 Subsidence characterization for Base Case, J1, J2, J3 and J4
models ............................................................................................... 104
Fig. 5.9 Evolution of zone of major (≥10cm) vertical (YY) and horizontal
(XX) surface deformations with continuous ore extraction for
Base Case (a), J1 (b), J2 (c), J3 (d) and J4 (e) models ..................... 105
Fig. 5.10 Rate of growth of 10cm surface displacement zone west of the
block centre vertical axis with continuous ore extraction for Base
Case, J1, J2, J3 and J4 models......................................................... 106
Fig. 5.11 Total vertical (a) and horizontal (b) surface displacements at the
end of ore extraction at different distances from block centre for
Base Case, J1, J2, J3 and J4 models ............................................... 106
Fig. 5.12 Subsidence at 100% ore extraction for J5 (a) and J6 (b) models ...... 109
Fig. 5.13 Subsidence characterization for J2, J5 and J6 models ...................... 110
J2, J5 and J6 models......................................................................... 110
Fig. 5.15 Subsidence at 100% ore extraction for J7 and J8 model ................... 112
Fig. 5.16 Subsidence characterization for Base Case, J7, J2 and J8
models ............................................................................................... 112
Base Case, J1, J2, J3 and J4 models ............................................... 113
Fig. 5.18 Assumed fracture orientations and fault positions for F1 to F9
models ............................................................................................... 115
Fig. 5.19 Subsidence crater formation for F1 model ......................................... 116
Fig. 5.20 Subsidence crater formation for F2 model......................................... 117
xii

Fig. 5.24 Subsidence at 100% ore extraction for F1, F2, F3, F4 and F5
model ................................................................................................. 121
Fig. 5.25 Subsidence characterization for Base case, F1, F2 and F3 .............. 124
Fig. 5.26 Subsidence characterization for J2, F4 and F5 ................................. 124
Base Case, F1, F2 and F3 models .................................................... 125
J2, F4 and F5 models ........................................................................ 125
Fig. 5.29 Differential XY displacements for surface points on the fault
hanging and footwalls: (a) F1, F2 and F3; (b) F4 and F5 models ...... 126
Fig. 5.34 Subsidence at 100% ore extraction for F6, F7, F8 and F9 models .... 132
Fig. 5.35 Subsidence characterization for BC, F2, F6 and F7 models ............. 134
Fig. 5.36 Subsidence characterization for J2, F4, F8 and F9 models............... 135
Base Case, F2, F6 and F7 models .................................................... 135
J2, F4, F8 and F9 models .................................................................. 136
Fig. 5.39 Differential XY displacements for surface points on the fault
hanging and foot walls for F2, F6 and F7 models .............................. 136
Fig. 5.40 Stages of subsidence crater formation at different percentages of
ore block extraction for model RM2 ................................................... 139
Fig. 5.41 Subsidence at 100% ore extraction for models RM1, RM2 and
RM3 ................................................................................................... 140
Fig. 5.42 Subsidence characterization for models J2, RM1, RM2 and RM3..... 141
models J2, RM1, RM2 and RM3........................................................ 141
xiii
Fig. 5.44 Subsidence at 100% ore extraction for models S1, S2, S3 and
S4 ...................................................................................................... 144
Fig. 5.45 Subsidence characterization for models Base Case, S1 and S2 ....... 145
Fig. 5.46 Subsidence characterization for models J2, S3 and S4 .................... 145
Fig. 5.47 Total horizontal surface displacements at the end of ore
extraction at different distances from block centre for models
Base Case, S1 and S2 ...................................................................... 146
end of ore extraction at different distances from the block centre
for models J2, S3 and S4 .................................................................. 146
Fig. 5.49 Assumed geometries for models GD1 to GD4................................... 147
Fig. 5.50 Subsidence crater formation for models (a) GD1 with a caprock
and (b) GD2 without a caprock .......................................................... 149
Fig. 5.51 Subsidence crater formation for models (a) GD3 with a caprock
and (b) GD4 without a caprock .......................................................... 150
Fig. 5.52 Subsidence at 100% ore extraction for models (a) GD1, (b) GD2,
(c) GD3 and (d) GD4 ......................................................................... 151
Fig. 5.53 Subsidence characterization for models Base Case, GD1 and
GD2 ................................................................................................... 153
Fig. 5.54 Subsidence characterization for models J2, GD3 and GD4 .............. 153
models J2, GD3 and GD4 ................................................................. 154
Fig. 5.56 Assumed modelling geometries for models (a) GD5 and (b) GD6..... 155
Fig. 5.57 Subsidence crater formation for models GD5 and GD6 .................... 157
Fig. 5.58 Subsidence at 100% ore extraction for models GD5 and GD6 .......... 158
Fig. 5.59 Subsidence characterization for models Base Case, J2, GD5
and GD6 ............................................................................................ 158
models Base Case, J2, GD5 and GD6 .............................................. 159
Fig. 5.61 Subsidence crater formation for models BD1 and BD2 ..................... 161
Fig. 5.62 Subsidence at 100% and 150% for BD1 model ................................. 162
Fig. 5.63 Subsidence at 150% ore extraction for Base Case model ................. 163
Fig. 5.64 Subsidence at 100% (a) and 150% (b) for BD2 model ...................... 163
Fig. 5.65 Subsidence at 150% ore extraction for J2 model .............................. 164
xiv
Fig. 5.66 Comparison of surface profiles and angles limiting major surface
deformations for models with varying block depth and joint
inclination at 100% ore extraction ...................................................... 165
Fig. 5.67 Surface profiles at the end of ore extraction for BC (100%), BC
(150%), BD1 (100%) and BD1 (150%) models.................................. 165
Fig. 5.68 Surface profiles at the end of ore extraction for J2 (100%), J2
(150%), BD2 (100%) and BD2 (150%) models.................................. 165
Fig. 5.69 Subsidence characterization for models BC (100%), BC (150%),
BD1 (100%) and BD1 (150%)............................................................ 167
Fig. 5.70 Subsidence characterization models J2 (100%), J2 (150%), BD2
(100%) and BD2 (150%) ................................................................... 167
BC (100%), BC (150%), BD1 (100%) and BD1 (150%) models ........ 168
J2 (100%), J2 (150%), BD2 (100%) and BD2 (150%) models .......... 168
Fig. 5.73 Comparative summary of modelling results - total extent of major
(≥10cm) surface displacements ......................................................... 171
Fig. 5.75 Comparative summary of modelling results - minimum angles
delineating the extent of major (≥10cm) surface displacements,
asymmetry index ............................................................................... 173
Fig. 5.76 Comparative summary of modelling results – volume of the rock
mass mobilized by the caving ............................................................ 174
Fig. 6.1 Slope failure modes (based on Sjöberg, 1999).................................. 183
Fig. 6.2 Comparison of limit equilibrium analysis of planar and step-path
failure (modified after Eberhardt et al., 2004a, with permission) ........ 184
Fig. 6.3 The pit-slope proximity problem: preliminary brittle fracture
analysis (adapted after Stead et al., 2007, with permission) ............. 186
Fig. 6.4 Typical model geometry for simulation of block caving induced
step-path failure (cases with two rock bridges shown). ..................... 188
Fig. 6.5 Block caving induced step-path failure in large open pit slope
(model M1) ........................................................................................ 191
Fig. 6.6 Maximum principal stress contours (Pa) - tensile stress
concentrations (red) in rock bridges prior to failure ............................ 192
Fig. 6.7 Typical rock bridge failure development ............................................ 192
Fig. 6.8 Stress/displacement analysis of caving - open pit slope
interaction (rock bridges failure, model M1) ....................................... 193
xv

interaction (rock bridge failure, model M2) ........................................ 194
interaction (rock bridge failure, model M3) ........................................ 194
Fig. 6.11 Block caving induced rock step-path failure in large open pit
slope (model M4) ............................................................................... 196
Fig. 6.12 Block caving induced step-path failure in large open pit slope
(model M5) ........................................................................................ 197
Fig. 6.15 Development of step-path failure in the open pit slope during
caving mining with relation to crown pillar geometry and stress
level for simulations with different % of rock bridges in step-path
failure surface .................................................................................... 199
Fig. 6.16 Variation of vertical stress in the crown pillar (50m below pit
bottom) for models M1-M5................................................................. 200
Fig. 6.17 3D view of Palabora pit and cave mine (adapted after Brummer
et al., 2005, with permission) ............................................................. 202
Fig. 6.18 General geology and pit slope geometry at Palabora mine
(adapted after Moss et al., 2005, with permission) ............................ 204
Fig. 6.19 Major geological structures at Palabora mine (based on data
provided by Palabora Mining Company Limited) ............................... 204
Fig. 6.20 Plan view of North Wall failure at Palabora mine ............................... 206
Fig. 6.21 Preliminary DFN model of Palabora mine North Wall section ........... 209
Fig. 6.22 ELFEN model of Palabora mine NW-SE section (section A-A in
Fig. 6.20). .......................................................................................... 210
Fig. 6.23 Pit slope deformation at cave breakthrough for model P1 ................. 213
Fig. 6.24 Pit slope deformation at 40% ore extraction for model P1 ................. 213
Fig. 6.25 Pit slope deformation at cave breakthrough for model P2 ................. 214
Fig. 6.26 Pit slope deformation at 40% ore extraction for model P2 ................. 214
Fig. A.1 Probability (relative frequency) of the caving induced total extent
of major (≥10cm) vertical surface displacements based on 37
model runs. ........................................................................................ 237
Fig. A.2 Probability of the caving induced total extent of major (≥10cm)
horizontal surface displacements. ..................................................... 237
xvi
Fig. A.3 Probability of rock mass volume mobilized by caving, in percent

of extracted ore volume. .................................................................... 238
Fig. A.4 Probability of minimum angles delineating the extent of caving
induced major surface displacements. .............................................. 238
Fig. A.5 Probability of subsidence asymmetry index. ...................................... 238
Fig. C.1 Bench failure at the bottom of the pit - July 2004............................... 240
Fig. C.2 Developing failure of the North Wall - October 7th, 2004. .................. 241
Fig. C.3 Failed North Wall – late May, 2005.................................................... 241
xvii
LIST OF TABLES
Table 2.1 Degree of influence of various factors affecting surface

subsidence development suggested by Flores & Karzulovic
(2002). ................................................................................................. 13
Table 2.2 Evolution of limit equilibrium methods of subsidence analysis ........... 19
Table 2.3 Numerical methods and proprietary codes widely applied in
rock engineering .................................................................................. 24
Table 2.4 Numerical studies of caving induced surface subsidence. ................. 25
Table 3.1 Hybrid continuum-discontinuum with fracture modelling
approach ............................................................................................. 33
Table 3.2 Fracture toughness and fracture energy values for various rock
types .................................................................................................... 36
Table 4.1 Rock mass parameters used in rock mass rating by RMR76, Q
and GSI ............................................................................................... 56
Table 4.2 Empirical relationships for derivation of estimates of basic rock
mass strength and deformability properties based on the RMR, GSI
and Q rock mass classification systems ................................................ 57
Table 4.3 Rock mass properties derived using RMR76 ....................................... 61
Table 4.4 Rock mass properties derived using Q ............................................... 62
Table 4.5 Rock mass properties derived using GSI ............................................ 62
Table 4.6 List of principal modelling scenarios and corresponding input
parameters........................................................................................... 74
Table 5.1 Modelling input parameters ................................................................ 92
Table 5.2 Modelling scenarios for analysis of the effect of joint orientation ......... 93
Table 5.3 Modelling scenarios for analysis of the effect of fault location ........... 114
Table 5.4 Modelling scenarios for analysis of the effect of fault inclination ........ 127
Table 5.5 Modelling scenarios for analysis of the effect of stress
environment ....................................................................................... 142
Table 5.6 Modelling scenarios for analysis of the effect of varying strength
between ore and host rock ................................................................. 148
Table 5.7 Modelling scenarios for analysis of the effect of varying joint
orientation within the ore and host rock .............................................. 155
xviii
Table 5.8. Preliminary classification of caving induced surface subsidence

for cases with no major geological discontinuities ................................ 175
Table 5.9. Preliminary classification of the influence of major geological
discontinuities on caving induced surface subsidence.......................... 176
Table 5.10. Initial influence assessment matrix of factors contributing to
block caving induced surface subsidence ............................................ 177
Table 6.1 Modelling input parameters used in conceptual modelling ................ 189
Table 6.2 Modelling scenarios ......................................................................... 189
Table 6.3 Description of rock units present in the North Wall............................ 203
Table 6.4 Modelling input parameters for preliminary Palabora failure
simulation........................................................................................... 211
Table B.1 Description of rock units present in the North Wall ........................... 239
xix
CHAPTER 1: INTRODUCTION
1.1 Block Caving Mining and Associated Surface Subsidence

Block caving is one of the most cost effective underground mining techniques; it
largely relies on managing the forces of nature to extract the ore. Block caving is
typically employed to mine massive low grade copper, gold and molybdenum
mineralization and diamond-bearing kimberlite pipes. High efficiency and low
production costs coupled with ever growing demand on natural resources are
making the block caving method increasingly important for the mining industry. It
is anticipated that in the next ten years the number of major block caving
operations will double. This will involve development of new block caving mines,
as well as the transition of existing large open pit mines reaching economic limits
to underground mining, utilizing the block caving method.
A typical block caving mine layout, shown in Fig. 1.1, consists of two mining levels
(production level and undercut level) placed within the ore column. In block caving,
ore is mined sequentially in large sections (blocks), with areas of several thousands
of square metres. Caving is initiated by blasting an extensive horizontal panel
(undercut) under the mined block. Stress redistribution and gravity combine to trigger
progressive fracturing and caving of the ore into the undercut. As caving of the ore is
initiated, the undercut is connected with the production level by blasting bell-shaped
ore passages, called drawbells, each consisting of at least two drawpoints (see Fig.
1.2). Broken ore falls through the drawpoints into the production level where it is
picked up and transported to the crusher and subsequently brought to the surface.
As broken ore is removed from the drawpoints, the ore above continues to break and
cave in by gravity, as illustrated in Fig. 1.3. Caving progressively extends upwards
as the ore is extracted, causing significant surface depression, or subsidence, above
the undercut and in the adjacent areas. In open pit/caving environments this may
trigger instability in open pit slopes.
1
surface subsidence
Fig. 1.1 Typical block caving mine layout.
caving ore
Fig. 1.3 Block caving mining and associated surface

subsidence.
Note: Figs. 1.1 - 1.3 are block caving animation screenshots

Fig. 1.2 Caving initiation and ore extraction. modified after Sandvik Corp. with permission.
2
According to Brady & Brown (2004) some instances of block caving mining
induced subsidence have had dramatic consequences, including loss of life, loss
of parts of producing mines and loss of major surface installations. In addition,
as indicated by Blodgett & Kuipers (2002), changes to surface landforms brought
about by block caving subsidence can be quite dramatic and may lead to a
pronounced environmental impact.
1.2 Research Objectives

The ability to predict surface subsidence associated with block caving mining is
important for mine planning, operational hazard assessment and evaluation of
environmental and socio-economic impacts. Owing to problems of scale and
lack of access, our fundamental understanding of the complex rock mass
response leading to subsidence development is limited at best as are our current
subsidence prediction capabilities.
In light of increasing use of the block caving mining method and the importance of
knowledge of potential surface subsidence there is a genuine need for an inclusive
study on surface subsidence associated with block caving. Current knowledge of
subsidence phenomena can be improved by employing numerical modelling
techniques. These provide a convenient framework for enhancing our understanding
of the basic factors governing the mechanisms of subsidence, an essential
prerequisite if advances in the prediction of subsidence are to be made.
This thesis has three principal research objectives:
1. Introduce a new methodology for the numerical analysis of surface

subsidence associated with block caving mining;
2. Improve understanding of subsidence phenomena and rock mass

behaviour in block caving settings through thorough analysis of factors
that may govern subsidence development;
3. Investigate mechanisms of block caving induced instability in large open pit

mine slopes.
3
1.3 Thesis Organization
The dissertation consists of seven chapters. This introductory chapter, which
poses the research problem and thesis objectives is followed by a
comprehensive literature review, presented in Chapter 2. This review
summarizes the current state of knowledge of subsidence phenomena, critically
evaluates the available methods of subsidence analysis and outlines previous
work undertaken in this area.
Chapter 3 discusses the available modelling approaches for simulation of block

caving surface subsidence phenomena and introduces the finite/discrete element
method (FEM/DEM) with fracture propagation code used in the current analysis.
An overview of the fundamentals of fracture mechanics is provided as well as an
outline of the implementation of this approach in the proprietary code ELFEN.
The advantages of this type of geomechanical modelling are highlighted and its
integration with Discrete Fracture Networks (DFN) is discussed and the proposed
modelling methodology is outlined.
Chapter 4 addresses the question of selection of equivalent rock mass properties

for the adopted FEM/DEM-DFN modelling technique. It discusses available
approaches to the derivation of jointed rock mass mechanical properties and
examines the properties output from rock mass classification systems. An analysis
of rock mass representation options within FEM/DEM modelling is given and a
novel approach to the derivation and calibration of rock mass properties
specifically tailored for block caving subsidence analysis using integrated DFN-
FEM/DEM modelling is presented.
Chapter 5 gains fundamental understanding of the block caving subsidence

phenomenon through a thorough conceptual FEM/DEM-DFN modelling analysis
of the factors controlling surface subsidence development, including: geological
structure (jointing and faults), rock mass strength, in-situ stress level, mining depth,
volume of extracted material and varying lithological domains. The discussion and
conclusions on the role of individual factors are provided. Synthesis of the
modelling results is presented in terms of preliminary classification of the caving
4
induced surface subsidence, classification of the role of major geological
discontinuities and an influence assessment matrix providing qualitative
assessment of the relative importance of individual factors affecting block
subsidence.
Chapter 6 addresses the problem of slope instability/subsidence in a large

overlying open pit associated with block caving mining, focusing on the analysis of
caving induced step-path failure in the pit slope. It uses a comprehensive “total
interaction” analysis approach, allowing to relate cave propagation, development of
the failure within the slope and the resultant surface subsidence. The FEM/DEM-
DFN modelling methodology is applied to the analysis of the complex mechanism
of caving induced open pit slope failure at Palabora mine.
Finally, Chapter 7, contains research conclusions, outlines key scientific

contributions of this work and discusses further research avenues.
5
CHAPTER 2: LITERATURE REVIEW
2.1 Introduction
Brady & Brown (2004) defined subsidence as the lowering of the ground surface
following the underground extraction of ore or another reserve. They suggested
that there are two types of subsidence deformation: continuous and discontinuous.
Continuous subsidence is usually associated with longwall mining, employed for
the extraction of thin, horizontal or flat-dipping orebodies overlain by weak
sedimentary strata, and is characterized by a formation of relatively smooth
surface depression. Block caving mining is typically utilized in hard rock
environments and involves extraction of massive volumes of ore and is
characterized by discontinuous subsidence. In this case, large scale surface
displacements manifest in formation of a crater and stepped terraces or major
discontinuities on the surface profile.
Underground mining induced surface subsidence has been discussed extensively

in the literature, however this has primarily focused on the analysis of continuum
subsidence associated with longwall coal mining. There is a wealth of literature on
this topic including textbooks (Kratzsch, 1983; Whittaker & Reddish, 1989; Peng,
1992) and a subsidence engineering handbook (National Coal Board, 1975). In
marked contrast the literature on block caving induced discontinuum surface
subsidence is more limited and our state of understanding of the phenomena
certainly does not allow confident prediction in the form of handbooks.
The objective of this chapter is to summarize the current state of knowledge on

caving induced discontinuum surface subsidence. The following themes are
discussed:
stages in surface subsidence development and associated rock mass

failure mechanisms;
6
subsidence characterization;
factors influencing surface subsidence development; and
existing methods of caving induced subsidence analysis. (Block Caving
subsidence toolbox).
2.2 Stages of Subsidence Development and Associated Rock

Mass Failure Mechanisms
Block caving derived surface movements are caused by the progressive failure of
the caved rock into the void left by extracted ore. A conceptual model of block
caving induced subsidence development is shown in Fig. 2.1 based on key stages
of subsidence after cave initiation as proposed by Abel & Lee (1980).
1. Collapse of rock progresses upward from the extraction level as ore is

withdrawn from below. The resulting column of caved and broken rock is
restricted above the area of extraction.
2. The ground surface does not measurably begin to subside until caving has
so thinned the overlying cap rock that it cannot transfer the load from this
rock onto the adjacent solid cave walls. The overlying rock will thus begin
to deflect downward toward the caved rock below. Lateral movement of
adjacent rock into the ore column is resisted by the active (and possibly
passive) pressure of the caved muck pile.
3. Further extraction of caved ore from below results in increased

subsidence of the ground surface above and adjacent to the area of
extraction. The overlying intact rock is progressively thinned by further
propagation of the cave back.
4. Continued extraction of ore will result in breaching of the surface. The

initial breach is typically in the form of a circular pit, commonly referred to
as a chimney cave breakthrough. The breakthrough is roughly centered
over the mining area although offsets may occur if geologic weaknesses
are present.
7
5. If ore extraction continues, the surface breach will grow laterally near the
surface. The rock adjacent to the subsided chimney either slides along
geologic weaknesses, such as joints or faults, or topples into the open crater.
a. Caving initiation b. Cave advance towards surface
c. Formation of surface depression d. Cave breakthrough and crater

development
Fig. 2.1 Development of surface subsidence induced by block caving

(modified after van As, 2003, with permission)
Woodruff (1966), Hoek (1974), Brown & Ferguson (1979) and Singh et al. (1993)
stated that during caving the main rock failure mechanisms are shear, tension or
a combination of the two. Based on numerical modelling studies Singh et al.
(1993) suggested that in the absence of major geologic structures, the dominant
mode of rock failure in progressive caving is tension. The gravity load results in
tensile stresses in the rock mass close to the surface, when the strain limit of the
rock mass is exceeded, tensile fractures develop. If major geological structures
are present, and appropriately oriented, the dominant failure mode is likely to be
shear failure along these surfaces.
8
2.3 Subsidence Characterization
Following surveys of the recent literature on block caving subsidence van As
(2003) and Flores & Karzulovic (2004) noted a scatter of opinions and to some
degree even confusion in the use of caving subsidence terminology (principally
due to a lack of standardization across the mining industry). These authors
unfortunately proposed two somewhat different subsidence characterization
terminology standards and a consensus has yet to be attained. In the current
study, a more descriptive block caving subsidence terminology proposed by van
As (2003), shown in Fig. 2.2, is adopted. An example of using this terminology for
characterization of subsidence at Northparkes Mines (Australia) is illustrated in Fig.
2.3.
The terminology proposed by van As (2003) is largely based on the work of Lupo
(1998), who conceptualized surface deformations observed and/or measured at
several caving mines into three distinctive zones:
Caved rock zone
Large-scale cracking zone (fractured zone in van As (2003) terminology)
Continuous subsidence zone
The caved rock zone corresponds to a zone of active cave movement. It is

typically situated above the active caving footprint and usually is manifested as a
crater. The material in this zone comprises caved ore and waste rock which has
broken up into irregular blocks and fines. As indicated by Laubscher (1981) this
caved material provides lateral restraint to the crater wall reducing the extent of
subsidence deformations. With continuing ore extraction this restraint is
gradually removed and the extent of surface deformations will grow. Total
surface subsidence in the caved rock zone over the life span of the mine may
reach several tens of metres.
The large scale cracking or fractured zone is characterized by an irregular broken

surface with scarps (produced as a result of rotational overturning), large open
tension cracks, benches and large blocks undergoing shear-rotational and toppling
failure into the crater.
9
Fig. 2.2 Block caving subsidence characterization terminology
(modified after van As, 2003, with permission)
continuous subsidence zone Stable Zone

tension cracks
Continuous Subsidence Zone
Stable Zone
Stable Zone
fractured zone
Unstable Zone
Unstable Zone
Breakthrough Zone
caved rock or crater zone
Fig. 2.3 Example of surface subsidence zonation at Northparkes mine

(modified after van As, 2003 with permission)
10
The continuous subsidence zone is characterised by a gentle surface depression
which may also involve minor cracking. The surface deformations within this
zone are relatively small in comparison to other zones, although, they should not
be neglected in mine planning. Lupo (1998) reported subsidence of up to
200mm in a continuous subsidence zone which caused significant damage in
nearby structures.
As illustrated in Fig. 2.2 the boundaries between different deformation zones are
defined by three limiting angle drawn from the edge of the undercut: angles of
break, angle of fracture initiation and angle of subsidence.
The angle of break measures the extent of the caved rock zone or the boundary
of the caved material. According to van As (2003) mining experience suggests
the following common trends regarding this angle:
it increases with depth; and
it tends to be sub-vertical in strong rocks with no significant inclined
discontinuities, and inclined where mining depths are shallow and/or
overburden rocks are weak.
Flores & Karzulovic (2002) based on the subsidence data from ten caving
operations, related rock mass quality and break angle as shown in Fig. 2.4. As
apparent from this figure:
the angle of break tends to increase with rock mass quality;
most angle of break data are in a range of 50° to 90°; and
for a rock mass with a Modified Rock Mass Rating (MRMR) > 70 the angle
of break tends to be larger than 60°.
The fracture initiation angle measures the extent of major surface cracking,
where the ground is already failed or potentially unstable. In his summary of
block caving subsidence observations, based on limited sources, van As (2003)
reported fracture initiation angles in the range of 25° to 80°.
The angle of subsidence defines the limits of surface deformations. It should be

noted that this information is rarely collected. Based on the data from studied caving
11
mines Lupo (1998) suggested that the surface deformation limits may extend from
50m to more than 250 metres away from the extent of large scale surface cracking
zone.
0.70
0.65
0.60
0.55
0.50
0.45
0.40
0.35
0.30
0.25
0.20
0.15
70 < RMR
0.10 60 < RMR < 70
50 < RMR < 60
0.05 40 < RMR < 50
0.00
0 10 20 30 40 50 60 70 80 90
RMR
Fig. 2.4 Relative frequency of break angles for different rock mass qualities (Laubscher‟s
RMR), in the subsidence craters of underground mines by caving methods
(adapted after Flores & Karzulovic, 2002, with permission)
2.4 Factors Governing Subsidence Development

Laubscher (1994) suggested that major geological structures, rock mass
strength, induced stresses and depth of mining are the most important
parameters influencing block caving subsidence development. Flores &
Karzulovic (2002), based on data from ten caving operations ,classified causes of
subsidence by the degree of their influence as summarized in Table 2.1.
12
Table 2.1 Degree of influence of various factors affecting surface subsidence
development suggested by Flores & Karzulovic (2002).
Degree of influence on surface subsidence

High Moderate
Geological Structures Water conditions
Rock mass quality Footprint geometry
Block height Caving initiation
Draw rate Mining sequence
Draw management Undercutting management
According to Brown (2003) there are a number of features of the orebody and the
local geology and topography which can influence subsidence development
including the:
dip of the orebody;
plan shape of the orebody;
depth of mining and the associated in situ stress field;
strengths of the caving rock mass and of the rocks and soils closer to the
surface;
nature of the slope of the ground surface;
presence of major geological features such as faults and dikes
intersecting the orebody and/or cap rock;
prior surface mining;
accumulation of caved or failed rock, or the placement of fill, in a pre-
existing or a newly formed crater; and
presence of nearby underground excavations.
It should be emphasized that the published literature sources, surveyed as part

of this research, focus on the importance of geologic structure on subsidence
development and provide minimal or no elaboration on the effect of other factors.
Crane (1929) carried out measurements of caving at a number of iron ore mines
in Michigan. He suggested that rock breaks according to a systematic
arrangement of planes of weakness (joints) with slight irregularities due to
breaking between joints. In the absence of faults and dykes, joint dip governs
13
the angle of break. The angle of break for a mine should be equal to the dip of
the most prominent joint, where prominence is a measure of number of
observations, joint persistence and character of the joint surface. Wilson (1958)
studied geological factors influencing block caving at San Manuel mine, USA
where it was noted that that there is close agreement between trends of
subsidence deformations and local fracture systems. Parker (1978) indicated
that in weak rocks there may be no significant geologic structure, hence the
reported cave angles are usually consistent and can be predicted with
reasonable confidence. In stronger rocks, however, the angle of break is usually
controlled by geological structures. A well defined fault plane, which is parallel to
a mining face and steep to moderately inclined, will result in a cave which
propagates to surface fairly rapidly and is defined on the surface by the trace of
the fault plane. If the predominant joints and faults are roughly perpendicular to
the mining front, caving may be inhibited and negative break angles may occur.
As noted by Stacey & Swart (2001) and van As (2003) in most cases when a
mining face encounters a significant discontinuity, such as a fault, with moderate
to steep dip, movement will occur on the fault regardless of the cave angle
through the intact rock. A stepped topographical crack will appear where the
fault daylights at surface. If mining is only on the hangingwall side of the fault
there will only be surface movements on the one side. If the fault dip is steeper
than the cave angle the extent of surface subsidence will be reduced.
Conversely, if the fault dip is less than the cave angle the extent of surface
subsidence will be increased, as illustrated in Fig. 2.5. The intersection of major
geological structures in a subsidence crater wall may define a wedge, within which
the rock mass has greater freedom to deform. The initiation of localised failure and
its progression will be facilitated within this zone of greater freedom of movement,
allowing the ultimate development of crater wall failure.
According to van As (2003) many observations of the influence of discontinuities

have been made; however, only a modest amount of research work has been
carried out to qualify and quantify the influence on surface subsidence
development. Based on the analysis of subsidence in coal mines Hellewell
14
(1988) asserted that quantifying the effects of discontinuities is complicated by
the fact that many features have not reacted adversely when subjected to
subsidence and the results of scientific investigations are in some instances
contradictory. Similar reasons explain the apparent lack of quantitative data from
block caving operations.
UNIFORM ROCK MASS subsidence

crater
disturbed zone
subsidence
EFFECT OF STEEPLY crater
DIPPING FAULT
fault
subsidence
EFFECT OF SHALLOW crater
DIPPING FAULT disturbed zone
fault
Fig. 2.5 Beneficial and detrimental effects of major weakness planes

(adapted after Stacey, 2007, with permission)
2.5 A Block Caving Subsidence Analysis Toolbox
2.5.1 Empirical Methods
Empirical methods are traditionally used in rock engineering and are based on a
synthesis of past observations, usually in similar settings, to describe the tentative
response trends associated with the studied phenomenon. Empirically based
block caving subsidence estimates include “rules of thumb” and experience based
design charts linking angle of break, rock mass rating and other parameters.
One example of an empirical approach is given in “The Hard Rock Miner‟s

Handbook - Rules of Thumb” (McIntosh Engineering, 2003) where it is stated that
15
the design of a block cave mine should assume a potential subsidence zone of 45
degrees from the bottom of the lowest mining level. Although it is unlikely that actual
subsidence will extend to this limit, there is a high probability that tension cracking
will result in damage to underground structures (such as a shaft) developed within
this zone, no permanent surface structures should be placed within this limit
(Hartman, 1992). Clearly such a “rule of thumb” may be rather conservative, non-
specific and provide only a crude estimate of potential subsidence.
The most commonly used empirical method in cave mining for estimating
subsidence damage limits is the Laubscher method (Laubscher, 2000). Laubscher
proposed a design chart (see Fig. 2.6) that relates the predicted cave angle (angle
of break) to the MRMR (Laubscher‟s Mining Rock Mass Rating), density of the
caved rock, height of the caved rock and mine geometry (minimum and maximum
span of a footprint).
Fig. 2.6 Empirical chart relating MRMR and cave (break) angle
(after Laubscher, 2000)
Flores & Karzulovic (2004) noted that this method does not take into account the
effect of geological features like faults which may influence the dip of the cave
16
angle, nor does it consider the apparent difficulty in determining density of the
caved rock. Overall, the Laubscher chart is a useful tool for preliminary estimates
of the angle of break, although it is too general to be relied solely upon in design.
The estimated break angles should be adjusted to account for the effect of
geological structures. This however is not a trivial exercise and requires sound
engineering judgement.
2.5.2 Limit Equilibrium Techniques
The concept of limit equilibrium has been widely applied in the analysis of soil
and rock engineering problems and is in essence based on a series of analytical
solutions aimed at the evaluation of rigid body stability above a defined failure
surface.
The initial limit equilibrium model for the analysis of tensile fracturing limits
associated with the progressive sub-level caving of an inclined orebody was
developed by Hoek (1974) for the analysis of subsidence at the Grängesberg mine
in Sweden. Hoek (1974) proposed a conceptual mechanism of hanging wall
failure during progressing downhole mining as illustrated in Fig. 2.7.
(a) (b) (c)

ground surface
orebody
(d) (e)
Fig. 2.7 Progressive hanging wall failure sequence with increasing depth of mining
(a) mining from outcrop; (b) failure of overhanging wedge; (c) formation of steep face;
(d) development of tension crack and failure surface; (e) development of second tension
crack and failure surface. (adapted after Hoek, 1974, with permission)
17
It was assumed that at each stage of vertical retreat a tension crack and a failure
surface form in the hanging wall at a critical location determined by the strength
of the rock mass and the imposed stresses. A limit equilibrium solution was
developed to determine the stability of the block of rock isolated by the new
tension crack and failure surface formed at each stage of mining. Using Hoek‟s
analysis it is possible to predict the angle of inclination of the new failure surface,
new tension crack depth and angle of break as the depth of mining increases,
Fig. 2.8. The analysis assumes a flat ground surface and full drainage
throughout the caving mass. Hoek‟s method is applicable to progressive
hangingwall failure only, and requires input of initial subsidence conditions.
Z 2 sin( p2 )
tan b tan p2 Tension
Tension crack
crack from
from previous
previous failure
failure
H 2 sin( New tension crack
0)
cos p2 Z 2 cos
sin 0
H1 Previous mining level depth

H2 Current mining level depth
Hs Depth to the caved material surface
Caved
Hc Caved material height
material
z1 Previous tension crack depth
z2 New tension crack depth
ψ0 Dip of the orebody
ψb Break angle
ψp1 Inclination of the previous failure plane
ψp2 Inclination of the new failure plane
φw Friction angle between caved material and
rock wall
θ Inclination of the line of action of T
Wc Weight of the caved material
W Weight of the potentially unstable block
T Lateral force due to Wc on the potentially Working face Footwall
unstable block
Tc Lateral force due to Wc on the footwall
γc Unit weight of the caved material Orebody
γ Unit weight of rock mass
c Cohesion and rock mass
φ Angle of friction of rock mass
Fig. 2.8 Idealised model for limit equilibrium analysis of progressive hangingwall caving
proposed by Hoek (1974)
Several authors have modified Hoek‟s method to incorporate various additional

parameters and mining geometries. Table 2.2 summarizes the available limit
equilibrium approaches to the analysis of caving induced subsidence developed
after Hoek (1974).
18
Table 2.2 Evolution of limit equilibrium methods of subsidence analysis
(modified after Flores & Karzulovic, 2004 with permission)
Author(s) Brief Method Description and Applications
Extended Hoek‟s limit equilibrium model to account for a sloping

surface and groundwater pressures in the tension crack and on
Brown &
the shear plane. This model was used to evaluate the
Ferguson (1979)
progressive failure of the hanging wall at Gath‟s mine in
Rhodesia.
Used Hoek‟s limit equilibrium model to include the progressive

failure occurring in both hanging wall and footwall in a very
Kvapil et al. (1989) steeply dipping orebody. This model was applied at El Teniente
mine, in Chile, to evaluate the subsidence generated by
underground block and panel caving operations.
Used Brown and Ferguson‟s limit equilibrium model to predict

discontinuous subsidence associated with block caving at Rio
Karzulovic (1990)
Blanco mine in Chile. This model was developed to evaluate
subsidence in a vertical orebody.
Proposed a simplified geometrical model for the calculation of

geometrical factors affecting the stability of hanging walls in an
inclined orebody using a sublevel caving method. This limit
Herdocia (1991)
equilibrium model was used to evaluate the hanging wall
stability at Grängesberg, Kiruna and Malmberget mines, in
Sweden.
This model considers the failure of the hanging wall using the
limit equilibrium equations derived by Hoek (1974) but
considering an active earth pressure coefficient, and the limit
equilibrium equations derived by Hoek (1970) for excavated
Lupo (1996) slopes in open pit mines to analyse the footwall. The use of an
active earth pressure coefficient is intended to include the effect
of the movement of broken rock during draw. This method was
applied to the analysis of sublevel caving at the Kiruna mine, in
Sweden.
Developed a model for a case of transition from open pit to

underground mining by caving methods. Extended Karzulovic
(1990) limit equilibrium solution to account for the geometry with
an existing open pit. Considered a sub-vertical orebody, sloping
Flores & Karzulovic
ground, influence of surface traction of caved rock and a pore
(2004)
water pressure. Based on a limit equilibrium model and
supplemental FLAC modelling a series of design charts relating
angle of break, zone of influence and rock mass quality and
crater depth were derived.
19
Heslop & Laubscher (1981) indicated that a governing factor in hanging wall
failure is rock structure. The presence of faults may provide preferential shear
failure planes. Persistent discontinuities having similar dip to an orebody may
create a tendency for toppling failure, as illustrated in Fig. 2.9. Woodruff (1966)
postulated that the tension cracks surrounding a caved or subsidence area do
not necessarily represent planes of movement extending from ground surface to
undercut level. It appears that the mechanism of failure behind Hoek‟s (1974)
limit equilibrium approach may not be applicable in all cases. It should be noted
that the predictive capabilities of limit equilibrium techniques are restricted to the
estimation of the angle of break. Overall, even the advanced limit equilibrium
approaches should be treated as very approximate as they do not account for
complex rock structure, in-situ stresses and neglect stress-strain relationships.
Fig. 2.9 Toppling of steeply dipping hangingwall strata

(after Brady & Brown, 2004).
van As (2003) suggested that given the significant cost implications of locating
major excavations and infrastructure beyond the influence of the cave subsidence
limits it is well worth the effort of using numerical modelling to ensure that the
empirical or analytical methods are not overly conservative in their predictions.
Similarly Brown (2004) recommended using a combination of empirical, analytical
and numerical methods for subsidence predictions. It was suggested to derive a
preliminary estimate of the angle of break using Laubscher‟s chart and then calibrate
it against observed angles of break in similar mining settings. The estimated angle
of break should then be checked against limit equilibrium approaches. The
20
estimated value of the angle of break can be adjusted to take into account local
geological features and the amount of broken material in the crater. Finally,
numerical methods should be used to confirm the estimate of the angle of break and
to estimate the stresses and displacements induced in the rock mass around the
caved zone.
2.5.3 Numerical Approaches
Numerical methods offer a powerful, sophisticated and flexible framework for the
analysis of engineering systems allowing consideration of complex geometries,
factors and mechanisms unattainable through use of empirical and analytical
techniques. Rapid advances in computing technology over the last two decades
have made numerical methods often routine in rock engineering analysis. Here
Section 2.5.3.1 provides a concise introduction to numerical techniques and
Section 2.5.3.2 discusses block cave subsidence related modelling studies.
2.5.3.1 Numerical Modelling in Rock Engineering
According to Hoek et al. (1991) all numerical methods adopt the approach of
dividing the problem into smaller physical and mathematical components and then
summing their influence to approximate the behaviour of the combined system.
The most common way to compute the series of equations formed during this
process is to assemble them into matrices, which can be solved by a variety of
techniques. This usually requires storage of large systems of matrix equations,
and the technique is known as the “implicit” solution method. The implicit method
is most efficient for solving problems with comparatively simple constitutive laws.
For more complex problems where reformulation of multiple matrices is required,
the implicit technique is less efficient. An alternative method, known as the
“explicit” solution scheme, is based on the assumption that a disturbance at a point
in space is initially felt only by points in its immediate surrounding. With
continuous computational steps the disturbances spreads through the whole
system until equilibrium is reached. This method usually involves solution of the
full dynamic equations of motion, even for problems that are essentially static or
21
quasi-static. The explicit methods do not require the storage of large systems of
equations.
Hoek et al. (1991) classified available numerical methods into two categories,
boundary and domain methods:
In boundary methods only the problem boundaries are defined and

discretized. The boundary element method (BEM), including the
displacement discontinuity method (DDM), is based on this proposition.
In domain methods, the whole modelling domain is divided into a finite

number of sub-regions or elements, as illustrated in Fig. 2.10. Finite
element (FEM), finite difference (FDM) and discrete element (DEM)
methods belong to this category.
Fig. 2.10 Example of problem discretization using FEM

(modified after Brady & Brown, 2005, with permission).
Detailed descriptions of these numerical methods may be found in a large

number of textbooks (e.g., Becker, 1992; Zienkiewicz et al., 2005; Davis, 1986;
Jing & Stephansson 2007).
Generally, there are three possible approaches to model rock engineering

problems: Continuum, Discrete, and Hybrid, as illustrated in Fig. 2.11. In the
rock mechanics context the use of BEM, FEM, and FDM methods is generally
based on the assumption that the rock mass behaves as a continuum medium,
22
i.e. it cannot be torn open or broken into pieces. Use of the DEM implies the rock
mass behaves as a discontinuum medium, consisting of a finite number of
interacting bodies (e.g., blocks, particles and etc.). Hybrid approaches attempt to
capitalize on the advantages of the continuum and discontinuum methods by
combining them into a single model, allowing, for instance representation of the
near field excavation area with discrete elements and the use of a continuum to
represent the far-field material (see Fig. 2.11(c)) therefore improving
computational efficiency. It should be noted that hybrid models have not to date
found widespread use in rock engineering with most analyses carried out using
either continuum or discontinuum approaches.
(a) FEM / FDM / BEM (b) DEM
(c) combination of FEM / FDM / BEM and DEM
Fig. 2.11 Various approaches to model rock engineering problems: (a) continuum, (b)
discrete and (c) hybrid
The numerical methods and their implementation in the most commonly

employed in rock engineering proprietary numerical codes are summarized in
Table 2.3.
23
Table 2.3 Numerical methods and proprietary codes widely applied in rock engineering
Numeri- Modelling Numerical code Rock Mass Rock Mass

cal Approach Represen- Failure
Method tation Realization
FLAC 2D/3D
FDM, (Itasca, 2007); Flexural
deformations,
FEM Phase2 plastic yield
Continuum (RocScience Ltd., 2007)
Map3D Continuum
(Mine Modelling Pty Ltd, 2007) medium
NFOLD Elastic
BEM
(Golder Associates, 1989) deformations
MULSIM
(Zipf, 1992)
Assembly
Blocks
UDEC/3DEC of
movements
Discontinuum
deformable
(Itasca, 2007) and/or blocks
or rigid
deformations
DEM blocks
Assembly Bond
PFC2D/3D of rigid breakage,
(Itasca, 2007) bonded particle
particles movements
Further information on the numerical modelling techniques applied in rock

engineering can be found in Cividini (1993), a recent review by Jing (2003) and the
respective manuals for the proprietary codes listed in Table 2.3. It should be
emphasized that the choice of the suitable modelling technique should always be
problem specific, and in some cases, may require the use of a combination of
different methods.
2.5.3.2 Numerical Modelling of Caving Induced Subsidence
A comprehensive literature survey reveals that there are relatively few published
accounts describing the modelling of surface subsidence associated with caving
mining. The most important of these are summarized in Table 2.4.
Singh et al. (1993) used the continuum finite difference code FLAC to simulate
progressive development of fractures in the hanging wall and footwall with
24
increase in mining depth in sublevel caving. The analysis was carried out at
Rajpura Dariba (India) and Kiruna (Sweden) mines. The modelling results were
found to be in reasonable agreement with field measurements.
Table 2.4 Numerical studies of caving induced surface subsidence.

Author(s) Code Type of analysis
FLAC Site specific: Rajpura Dariba and Kiruna
Singh et al. (1993)
mines
Karzulovic et al. (1999) FLAC Site specific: El Teniente mine
reported by van As (2003) FLAC 3D Site specific: Northparkes mine
Cavieres et al. (2003) 3DEC Site specific: El Teniente mine
FLAC & Conceptual
Flores & Karzulovic (2004)
FLAC 3D
Gilbride et al. (2005) PFC 3D Site specific: Questa mine
Brummer et al. (2006) 3DEC Site specific: Palabora mine
Itasca Ltd. FLAC 3D Site specific: San Manuel mine
reported by Elmo et al.
(2007a)
Phase2, Site specific: Kiruna mine
Villegas & Nordlund (2008a,b)
PFC 2D
Karzulovic et al. (1999) performed a study using Karzulovic‟s (1990) limit

equilibrium model to predict the evolution of the subsidence crater at El Teniente
mine (Chile), and conducted numerical modelling using FLAC to assess the
extent of the zone of influence. The angle of break calculated using the limit
equilibrium approach was adjusted to take into account major geological
structures. The numerical models were calibrated against field observations to
define the limits of the influence zone.
van As (2003) reported that subsidence modelling of the Northparkes (Australia),

Lift 1 block cave was undertaken using FLAC 3D in an attempt to define the
extent of the cave deformation and subsidence limits (for both underground and
surface), and to quantify the increase in the abutment stresses subsequent to
cave break-through. The results of the modelling proved far more reliable and
“closer to reality” than those derived using empirical methods. van As (2003) also
25
postulated that caution must be exercised when using numerical models to ensure
that a reliable geological model forms the framework of the numerical model.
Cavieres et al. (2003) employed the 3DEC code to carry out analysis of the
evolution of fracturing limits in Braden Pipe associated with caving mining at El
Teniente mine. The mine scale model was built and calibrated against observed
subsidence damage, and then subsequently used to make forward predictions. It
was concluded that the main mechanism governing the growth of the fracturing
limit is a loss of confinement of the Braden Pipe‟s wall resulting from mining at its
perimeter. The authors found the contours of total strain to be good indicators of
subsidence limits. Based on the modelling results a mine subsidence monitoring
program, focussing on strain measurements was recommended.
As a part of Stage II of the International Caving Study, Flores & Karzulovic (2004)
conducted conceptual FLAC/FLAC3D modelling of the surface subsidence during
block caving, for the case with an existing open pit, varying rock mass strength,
open pit depth and undercut level depth. Based on their modelling results
combined with limit equilibrium analysis, a series of design charts were
developed correlating angle of break and zone of influence of caving with
undercut level depth and crater depth for rock with varying rock mass quality.
Fig. 2.12 shows an example of such a chart. It should be noted that in their analysis
Flores & Karzulovic (2004) did not account for the presence of geological structures.
Moreover, the validity of these charts is yet to be confirmed by mining experience.
Gilbride et al. (2005) conducted a three-dimensional study of surface subsidence

at the Questa mine (USA) using PFC3D. The rock mass was simulated as an
assembly of bonded spheres with diameters of 13 to 20 metres. The authors
stated that the true advantage of a spherical particle code (PFC3D) for modelling
surface subsidence in block caving settings lies in it‟s ability to simulate large-
displacement mass flow simultaneous with elastic and small-strain, inelastic
deformation. Simulated rock mass behaviour was found to be reasonably
realistic and subsidence trends were found to be in good agreement with surface
deformation measurements at the site. The authors noted that the assumption of
26
the large diameter spheres prevented some potentially influential smaller-scale
deformation mechanisms from developing in the cave and near the surface. It
was emphasized that it remains a challenge to achieve both computational
efficiency and reasonable simulated behaviour with PFC3D for mine scale
subsidence problems. In this context a spherical particle size that is “too small”
has yet to be attained.
1700
1600 HT
1500
b
1400
1300
1200
UCL Depth, HT (meters)
1100
1000
900
800
700
600
500
400
300
200
Rock Mass Geotechnical Quality
Poor to Fair
100 Fair to Good
Good to Very Good
0
45 50 55 60 65 70 75 80 85 90
Angle of Break, b (degrees)
Fig. 2.12 Design chart for estimation the angle of break in a transition from open pit to
underground mining by block/panel caving
(adapted after Flores & Karzulovic, 2004 with permission)
Brummer et al. (2006) carried out 3-dimensional 3DEC analysis of open pit wall
failure mechanisms associated with block caving at the Palabora mine (South
Africa). This study was instigated by a massive failure at the North wall of the
27
Palabora mine during block caving operations. Mine scale 3DEC modelling
incorporating major geological structures (faults and joint sets) showed that
observed movements were most likely caused by the fact that pervasive joint
sets may form wedges that daylight into the cave region below the pit. The draw
of the ore into the cave zone undermines the pit wall, and appears to have a
direct control on the movements.
As reported by Elmo et al. (2007a), Itasca Ltd. carried out back analyses of
subsidence due to caving at San Manuel mine (USA) using FLAC3D. Being a
continuum code, FLAC3D, cannot explicitly model discontinuous behaviour. It
was necessary to incorporate the effects of jointing for the model to be
representative. Based on Clark (2006) the Equivalent Rockmass Model (ERM)
was proposed. To account for joint fabric, randomly oriented ubiquitous joint
planes were distributed through every zone in the model according to the
mapping data. This allows for the larger scale behaviour to be affected by small
scale effects. Initial simulations based on FLAC3D mine scale model and the
ERM concept have yielded promising results with the subsidence profile and
breakthrough matching observations reasonably well.
Villegas & Nordlund (2008a) carried out numerical analysis of hanging wall failure
at Kiruna mine using Phase2. The caving process was simulated by adding
voids moving upwards from the extraction level and changing the properties of
the material when the void was filled. Based on the estimated failure location on
the ground surface, the break angle and the limit angle were calculated for
different mining levels. The results indicated that the break angle and the limit
angle are almost constant for deeper mining levels. However, the limit angle
differs between sections with different rock mass strength. Moreover, the break
angle could be altered by large geological structures. It should be emphasized
that the Phase2 model exhibited excessive tensile deformations at the surface,
particularly in far-field. Villegas & Nordlund (2008b) also analyzed subsidence
deformations at Kiruna mine using PFC2D. The modelling indicated that
although tension cracks develop at the surface the primary failure mechanism in
the hangingwall is shear. In addition, it was shown that the caved rock and the
28
backfill in the pit provide support to the footwall and hangingwall therefore
reducing the magnitude and extent of surface subsidence.
Overall the following can be noted with respect to the previous subsidence
modelling attempts:
the majority of subsidence modelling studies have attempted to simulate
mine scale geometries in 3D and consequently were based on fairly
coarse models;
all the modelling studies, with the exception of that by Flores & Karzulovic
(2004), focused on back analysis or predictive modelling of particular mine
sites.
2.6 Summary
A comprehensive literature survey has shown that our knowledge of rock mass
behaviour leading to surface subsidence in block caving settings is rather
tentative and primarily founded on empirical, mostly qualitative, observations.
This is perhaps not surprising given the scale of the block caving problem, the
complexity of rock mass response and the multitude of factors affecting
subsidence. The literature focuses on the importance of the effect of geological
discontinuities in subsidence development, stopping short of elaborating on the
effect of other factors.
Available methods of subsidence analysis include empirical, analytical and numerical

approaches. The empirical methods are not particularly reliable. The analytical
approaches are restrictive, being based on Hoek‟s (1974) assumed failure
mechanism, and are able to provide only estimates for the angle of break. The
numerical approaches being inherently more flexible and sophisticated offer an
opportunity to improve our understanding of block caving subsidence phenomena and
increased accuracy of subsidence predictions. However, previous modelling studies
were largely oriented towards providing subsidence predictions for a particular site.
The modelling study by Flores & Karzulovic (2004) was the first attempt after
Laubscher (2000) to provide non-site specific guidance for subsidence analysis. It
29
should be noted that their modelling was limited to the case with an open pit and no
consideration was given to the effects of significant factors including rock structure. It
appears that to date no comprehensive attempt has been made to evaluate the
general principles characterizing surface subsidence development in block caving
settings and evaluate the predominant factors governing subsidence phenomena.
30
CHAPTER 3: RESEARCH METHODOLOGY AND THEORY
3.1 Introduction
This chapter discusses the available modelling approaches for simulation of
block caving surface subsidence phenomena and introduces the finite/discrete
element method with fracture propagation code used in the current analysis. An
overview of the fundamentals of fracture mechanics is provided as well as an
outline of the implementation of this approach in the proprietary code ELFEN.
The advantages of this type of geomechanical modelling are highlighted and its
integration with Discrete Fracture Networks (DFN) is discussed. Finally the
proposed modelling methodology is outlined.
3.2 Modelling Approach Selection
3.2.1 Analysis of Available Modelling Approaches
Block caving subsidence is a product of a complex rock mass response to

caving. This response comprises massive failure of the rock mass driven by
brittle fracture, both in tension and compression, along existing discontinuities
and through intact rock bridges. Moreover, block caving subsidence
development almost invariably involves complex kinematic mechanisms.
As discussed in Section 2.5.3.1 traditional rock mechanics modelling techniques

include a variety of continuum and discontinuum methods. In the FDM and FEM
based continuum approaches the effect of brittle fracturing can be accounted for
implicitly by use of plasticity models or a damage mechanics failure criterion
(Hajiabdolmajid et al., 2001 and Fang & Harrison, 2002). The BEM based
continuum approach allows, for pre-existing flaws, explicit simulation of rock
fracture initiation and growth under mode I (tensile), II (shear) and mixed fracture
modes. Among various techniques to simulate fracture propagation, the
31
displacement discontinuity method (Crouch & Starfield, 1983) is the most efficient.
Several researchers, including Scavia (1990, 1995), Scavia & Castelli (1996) and
Muller & Martel (2000) have applied BEM based fracture codes to analyse failure
initiation in slopes. Although as noted by Jing (2003), to date, due to the
computational difficulties involved in the analysis of BEM fracture growth, the
majority of studies were often performed considering a small number of isolated,
non-intersecting fractures, frequently laboratory scale 2-D models. These models
involved local failure mechanisms, such as borehole breakout and step-path
fracture propagation (see for instance works by Backers et al., 2006, Yan, 2008).
According to Choi (1992), despite many adaptations, continuum models are not
suitable for modelling a rock mass where large scale sliding, separation, and
rotation will occur along discontinuities. It should be recognized that even
advanced continuum techniques are unable to capture the true failure kinematics.
In DEM based discontinuum approaches the effect of brittle fracturing can be

simulated indirectly through particle assembly analysis (Potyondy & Cundall, 2004)
and a sliding joint model (Pierce et al., 2007) using PFC 2D/3D. One drawback of
this approach is that particle bond breakage or a sliding joint model are not based on
fundamental fracture mechanics but governed by simple bonding models. An
alternative DEM approach is based on the UDEC Voronoi tessellation algorithm
which divides rock mass into polygonal blocks, allowing simulation of crack
propagation and fracturing occurring when strength between blocks is exceeded.
Fairhurst (2006) and Christianson et al. (2006) applied this approach for analysis of
lab and slope scale problems. Similar to the PFC based approaches these modelling
techniques did not consider fracture mechanics principles. Voronoi analysis trials
were carried out in the current research as part of the methodology selection process
and demonstrated the very low efficiency of the Voronoi mesh generator, i.e. a
geometry of medium complexity could take up to several hours to mesh. Kemeny
(2005) using both fracture mechanics and the UDEC discontinuum approach
developed a methodology for time-dependant analysis of rock bridge failure.
Overall, both continuum and discontinuum approaches to rock mass modelling

provide a convenient framework for the analysis of many engineering problems
32
however they are not always applicable to highly complex engineering problems
such as block caving subsidence due to the above limitations. It appears that
realistic simulation of the subsidence phenomena necessitates consideration of
fracture mechanics principles for brittle fracturing simulation and a blend of
continuum and discontinuum approaches to capture the complex failure
mechanisms.
3.2.2 Fracture Mechanics Based Hybrid Continuum/Discontinuum Approach
In the current study a state-of-the-art hybrid continuum-discontinuum approach

based on the finite/discrete element method (Munjiza et al., 1995) and incorporating
fracture mechanics principles is adopted. In the combined finite-discrete element
method the finite element-based analysis of continua is merged with discrete
element-based transient dynamics, contact detection and contact interaction
solutions (Munjiza, 2004). Use of fracture mechanics principles in combination with
the finite-discrete element method allows the caving process to be simulated in a
physically realistic manner. Rock mass failure is realized through a brittle fracture
driven continuum to discontinuum transition with the development of new fractures
and discrete blocks, and full consideration of the failure kinematics. Table 3.1
summarizes the hybrid continuum-discontinuum with fracture approach.
Table 3.1 Hybrid continuum-discontinuum with fracture modelling approach

Modelling Numerical Rock Mass Rock Mass Failure Realization
Approach Method Representation
Fract. Mech. Degradation of a continuum into
based Hybrid discrete deformable blocks
FEM/DEM Continuous medium
Continuum- through fracturing and
Discontinuum fragmentation
Intensive research carried over the last 15 years has facilitated major advances in
FEM/DEM theory and led to the development of several codes including Y (Munjiza
et al., 1999), VGW (Munjiza & Latham, 2004) and ELFEN (Rockfield Software Ltd.,
2006). So far ELFEN is the only fully featured commercially available code and
therefore was adopted for the current study. ELFEN is a multipurpose FEM/DEM
software package that utilizes a variety of constitutive criteria and is capable of both
33
implicit and explicit analyses in 2-D and 3-D space. It has the capability to simulate
continuum materials, jointed media and particle flow behaviour. The current study
uses only the hybrid FEM/DEM features of the code.
3.3 Fundamentals of Fracture Mechanics

Before introducing the ELFEN code it is important to outline the relevant basic
elements of fracture mechanics. This discipline deals with crack propagation and
provides a quantitative description of the transformation of an intact structural
component into a fractured medium by crack growth.
In the fracture mechanics of a solid three basic crack loading modes (or their
combinations) are assumed (see Fig. 3.1):
Mode I: opening or tensile mode (the crack faces are pulled apart);
Mode II: sliding or in-plane shear (the crack surfaces slide over each other);
Mode III: tearing or anti-plane shear (the crack surfaces move parallel to
the leading edge of the crack and relative to each other).
Mode I Mode II Mode III
Fig. 3.1 Fracture mechanics basic crack tip loading modes (based on Whittaker et al., 1992).
There is some debate in the literature as to which mode of fracturing is predominant

in specific rock mechanics problems. Whittaker et al. (1992) concluded that the
majority of fracture problems in rock mechanics involve a combination of modes I-II.
Liu (2003) indicated that in mixed I-II modes and even in pure mode II, tensile
failure is still the overwhelming failure mode. Scholz (2002) states that it is not
possible for a shear crack in an isotropic elastic medium to grow in its own plane.
Instead, the propagation of a shear crack due to 1 (major principal stress) occurs
34
by the generation of mode I cracks parallel to 1. In other words, what appears to
be shear fracturing on macroscopic level is realized by formation of an echelon
tensile crack system on a microscopic level, with increasing loading these tensile
cracks link up and form a shear plane.
The fracture insertion algorithm in the current ELFEN code is based on mode I or
tensile fracturing (development of a mode II fracture capability is ongoing).
Considering the above discussion and the notion by Singh et al. (1993) that in
caving mining a tensile mode of fracturing is predominant; the assumption of purely
tensile failure for the block caving subsidence analysis appears to be reasonable.
Griffith (1921) established a relationship between fracture stress and the crack size
that led him to state that a crack in a material will propagate if the total energy of the
system is lowered with crack propagation. That is, if the change in elastic strain
energy due to crack extension is larger than the energy required to create new crack
surfaces then crack propagation will occur. Griffith`s theory was developed for
perfectly brittle materials. Irwin (1957) extended the theory for quasi-brittle
materials. He postulated that the energy due to plastic deformation must be added
to the surface energy associated with the creation of new crack surfaces. He
recognized that for quasi-brittle materials, the surface energy term is often negligible
compared to the energy associated with plastic deformation. Furthermore, he
defined a quantity G, the strain energy release rate or "crack driving force," which is
the total energy absorbed during cracking per unit increase in crack length and per
unit thickness. Irwin further showed that the energy approach is equivalent to the
stress intensity approach and that crack propagation occurs when a critical strain
energy release rate G (or in terms of a critical stress intensity, KC) is achieved.
Whittaker et al. (1992) postulated that a fundamental feature of rock fracture

mechanics lies in its ability to establish a relationship between rock fracture strength
and the geometry of cracks or between the cracks and the fracture toughness or
critical stress intensity, KC. Proposed by Irwin (1957), this is the most fundamental
parameter of fracture mechanics and is used to describe the resistance of a crack to
propagation. The fracture toughness represents the extreme value of the stress
35
intensity factor which is a measure of stress intensity in an elastic field. The fracture
toughness can be considered as the limiting value of stress intensity just as the yield
stress might be considered the limiting value of applied stress.
According to the fracture modes described earlier, fracture toughness may be either
mode I (KIC), mode II (KIIC) or mode III (KIIIC), or any mixed mode fracture toughness.
Intrinsic interrelationships exist between the fracture parameters of different fracture
modes. Fracture toughness for a particular rock type can be established through
laboratory testing adopting ISRM suggested methods (ISRM, 1988).
For the Mode I fracture initiation the critical strain release (fracture) energy and the
critical stress intensity factor (fracture toughness) are related according to the
following relationship:
GIC = KIC 2/E (3.1)
where E is Young‟s modulus of the solid.
In ELFEN, fracture propagation is controlled by this criteria and the fracture

energy G (GIC) is one of the input parameters.
Table 3.2 shows fracture mechanics parameters for selected hard to soft rocks.
Further examples can be found in Atkinson (1987) and Whittaker et al. (1992).
Table 3.2 Fracture toughness and fracture energy values for various rock types
(based on Scholz, 2002)
Rock type Fracture toughness Fracture energy

(MPa√m) (Jm-2)
Black gabbro 2.88 82
Westerly granite 1.74 56
Solnhofen limestone 1.01 19.7
3.4 Implementation of the Hybrid Continuum-Discontinuum with

Fracture Approach in the ELFEN Code
A comprehensive discussion of the finite/discrete element method and its
implementation in the ELFEN code can be found in Munjiza et al. (1995), Yu
36
(1999), Klerck (2000), Munjiza (2004) and Owen et al. (2004a,b). Here only salient
features of the code relevant to the current study are introduced.
3.4.1 Solution Procedure for Continuum to Discontinuum Transition
According to Owen & Feng (2001), problems involving continuum to discontinuous

transition are often characterized by the following:
they are highly dynamic with rapidly changing domain configurations;
sufficient mesh discretization is required;
multi-physics phenomena are involved;
the domination of contact/impact behaviour gives rise to a very strong
non-linear system response.
These factors dictate that there is no alternative to employing time integration

schemes of an explicit nature to numerically simulate such problems.
The solution procedure at each computational time step involves the following steps:
1. Finite element and fracture handling:

computation of internal forces of the mesh;
evaluation of material failure criterion;
creation of new cracks if any;
global adaptive re-meshing if necessary.
2. Contact/interaction detection:
spatial search: detection of potential contact/interaction pairs among
discrete objects;
interaction resolution: determination of actual interaction pairs through
local resolution of the kinematic relationship between (potential) interaction
pairs;
interaction forces: computation of interaction forces between actual
interaction pairs by using appropriate interaction laws.
3. Global solution:
computation of velocities and displacements for all nodes.
37
4. Configuration update:
update of coordinates of all finite element nodes and positions of all
discrete objects.
3.4.2 Constitutive Models for Rock Material
The simulation of fracturing, damage and associated softening in ELFEN is

achieved by employing a fracture energy approach controlled by a specified
constitutive fracture criterion. There are two main constitutive fracture models
implemented in the code:
Rankine rotating crack model
Mohr-Coulomb model with a Rankine cut-off
In both constitutive models, fracturing is associated with extensional strain and

thus occurs parallel to the localized loading direction.
According to Klerck (2000) for the Rankine rotating crack model failure occurs
according to the following principles:
a failure threshold is a function of the material tensile strength;
prior to reaching the failure threshold the material is considered to be
homogeneous (i.e. without any faults or defects), and to be linearly elastic;
after violation of the failure threshold or yielding condition, material
damage is initiated (the yield surface is shown in Fig. 3.2(a)).
the failure process is completed once the material strength is totally lost.
At complete failure it is assumed that a physical crack is created through
the material point;
stress softening behaviour is a function of the fracture energy release rate Gf
(see Fig. 3.2(b))
The Mohr Coulomb with Rankine cut-off criteria is a more sophisticated model
that is able to consider material failure both in tension and compression. As
indicated by Klerck et al. (2004) this model is based on the assumption that
quasi-brittle fracture is extensional in nature, i.e. any phenomenological yield
surface is divided into regions in which extensional failure can be modelled
38
directly, as in the case of tensile stress fields and indirectly, as in the case of
compressive stress fields (Klerck et al., 2004).
For degradation and subsequent discrete fracturing of the material in
compression the model employs the Mohr-Coulomb failure criterion in the form of
a softening, isotropic, non-associated elasto-plasticity model (Fig. 3.3).
(a) (b)
Fig. 3.2 Rankine Rotating Crack model: (a) yield surface and (b) softening curve
(adapted from ELFEN User manual; Rockfield, 2005, with permission).
Fig. 3.3 Yield surface for the conventional Mohr Coulomb model
(adapted from ELFEN User manual, Rockfield, 2005, with permission).
39
The Mohr-Coulomb yield surface in tension cannot reasonably represent the
physically observed plastic flow directions normal to the mutually orthogonal
principal tensile planes. To improve the tensile description, a three-dimensional
rotating crack model is coupled with the Mohr-Coulomb yield surface. According
to Klerck (2000), only the mutually orthogonal tensile planes of the isotropic
Rankine yield surface are able to recover the correct plastic flow directions in
tension. Thus as an approximation to the anisotropic softening response of
physical quasi-brittle materials, the isotropic non-hardening Rankine tensile cut-
off emerges as the most feasible tensile cut-off formulation. Fig. 3.4 shows the
conventional Mohr Coulomb tensile yield surface in a 3D space (Fig. 3.4a),
together with the Mohr Coulomb tensile yield surface with Rankine tensile corner
(Fig. 3.4b).
Fig. 3.4 Yield surfaces for (a) Mohr Coulomb and (b) Mohr Coulomb with Rankine
tensile cut-off criteria
(adapted from ELFEN User manual with permission, Rockfield, 2005).
According to Klerck (2000) fracturing due to dilation is accommodated by

introducing an explicit coupling between the inelastic strain accrued by the Mohr-
Coulomb yield surface and the anisotropic degradation of the mutually orthogonal
tensile yield surfaces of the rotating crack model. An explicit coupling between
compressive stress induced extensional strain and tensile strength degradation
permits the realisation of discrete fracturing in purely compressive stress fields.
The Rankine tensile corner introduces additional yield criteria defined by:
40
i – ft = 0; i = 1,2,3 (3.2)
where i refers to each principal stress and ft is tensile strength.

Although at present no explicit softening law is included for the tensile strength,
indirect softening does result from the degradation of cohesion according to the
following criteria:
t c 1 sin cos (3.3)
This ensures that a compressive normal stress always exists on the failure shear
plane.
According to Klerck (2000), the isotropic non-hardening Rankine tensile cut-off for
the isotropic Mohr-Coulomb yield surface constitutes a means of incorporating the
return-mappings needed for the implementation of the compressive fracture model.
Klerck et al. (2004) demonstrated that the combined Mohr-Coulomb with Rankine
tensile cut-off model is able to effectively simulate realistic fracture propagation in
rock under both tensile and compressive stress fields. This model is employed in
the current study.
The following input parameters are required for Mohr-Coulomb model with a
Rankine cut-off:
Young‟s modulus, E
Poisson‟s ratio,
Density, ρ
Cohesion, c
Friction angle,
Dilatancy angle,
Tensile strength, t
Fracture energy, Gf
3.4.3 Continuum Degradation through Fracturing
One of the critical issues of continuum to discontinuum transition is how to

convert the continuous finite element mesh to one with discontinuous cracks.
41
The fracture algorithm employed in ELFEN inserts physical fractures or cracks
into a finite element mesh such that the initial continuum is gradually degraded
into discrete bodies. At some point in the analysis when the strength limit has
been reached, the adopted constitutive model predicts the formation of a failure
band within a single element. Anisotropic damage evolution is initiated by
degrading the elastic modulus, E, in the direction of the major principal stress
invariant. As described by Owen et al. (2004a) a special damage indicator Fk
f
defined as a ratio of the inelastic fracturing strain and the critical fracturing
f
strain c is employed for damage evaluation:
f f
Fk c k (3.4)
where k is a Gauss point of an element in the model and the critical fracturing
strain is given by
f
c 2Gf / hcft (3.5)
where Gf is fracture energy, hc is crack band width, and ft is tensile strength
The load carrying capacity across a failure band decreases to zero as damage
increases until eventually the energy needed to form a discrete fracture is released
and the damage indicator reaches unity. At this point the topology of the mesh is
updated, initially leading to fracture propagation within a continuum and eventually
resulting in the formation of discrete elements as the rock fragments are formed.
As described by Owen et al. (2004a), a discrete crack is introduced when the

tensile strength in a principal stress direction reaches zero and is orientated
orthogonal to this direction. The crack is then inserted along the failure plane.
The failure plane is defined in terms of a weighted average of the maximum
failure strain directions of all elements connected to the node. There are two
options for crack insertion: intra-element and inter-element, as shown in Fig. 3.5.
If a crack is inserted exactly through the failure plane (Fig. 3.5b), some ill-shaped
elements may be generated and local re-meshing is then needed to eliminate
them. Alternatively, the crack is allowed to propagate through the most closely
42
aligned element boundary (Fig. 3.5c). In this way, new elements are not created
and the updating procedure is simplified. However, this approach necessitates a
very fine mesh discretization around the potential fracture area.
n
Failure plane
Failed nodal
point
Fig. 3.5 Crack insertion in ELFEN

(a) weighted average nodal failure direction; (b) intra-element fracture description; (c) the
inter-element fracture description. (adapted after Yu, 1999 with permission).
It should be noted that preliminary ELFEN modelling trials carried out as part of the
current study showed that use of the intra-element option was prohibitive due to
numerical instability leading to termination of computations. Therefore, all modelling
analyses undertaken during this study are based on inter-element fracture insertion
and use the finest mesh discretization allowed by the available computing
capabilities.
3.4.4 Contacts Interaction Handling
Crack insertion is followed by application of contact conditions on the crack

surfaces. In ELFEN the interaction of contact surfaces is governed by the
penalty method. Surface penetration that violates the impenetrability constraint
invokes normal penalty (contact) forces Pn that prompt surface separation.
Similarly, tangential penalty forces Pt are invoked by the relative tangential
displacement between contacting surface entities. Normal and tangential
penalties have a similar meaning to the normal (kn) and tangential (ks) stiffness
employed in DEM modelling, as shown in Fig. 3.6. Using the procedure
proposed by Hudson & Harrison (1997), Elmo (2006) back-calculated normal
stiffness values for a series of pillar models containing a single set of horizontally
43
spaced discontinuities and found a ratio of kn / Pn less than or equal to 1.11. It
was suggested that these parameters can be effectively considered as being
equivalent in magnitude.
σn Normal
displacement
s
Aperture ks kn Shear
displacement
Fig. 3.6 Comparison of contacts interaction handling in DEM and FEM/DEM methods
(a) meaning of normal and shear stiffnesses for a block containing a single discontinuity
(based on Bandis, 1993); (b) penalty contacting couple in ELFEN as an equivalent
spring system (based on Klerck, 2000) – adapted after Elmo (2006), with permission
ELFEN uses a Mohr-Coulomb joint slip model to simulate shear failure along joint
contacts. The contact parameters required by ELFEN are:
Normal penalty coefficient, Pn
Tangential penalty, Pt
Cohesion, c
Friction angle, υ
3.4.5 Applications of ELFEN in Rock Engineering
The FEM/DEM plus fracture methodology used in the ELFEN code has been
extensively tested and validated against controlled laboratory tests (Yu, 1999;
Klerck, 2000, Klerck et al., 2004 and Stefanizzi, 2007). In addition, recent work
by Yan (2008) has illustrated that ELFEN simulations of laboratory scale step-
path failure under axial compression are in good agreement with actual physical
tests and correlate well with modelling results obtained by other codes (Phase2,
UDEC and FRACOD).
Numerous researchers including Coggan et al. (2003), Klerck et al. (2004), Cai
& Kaiser (2004), Stead et al. (2004), Eberhardt et al. (2004), Elmo et al. (2005),
Coggan & Stead (2005), Stead & Coggan (2006), Pine et al. (2007), Stefanizzi et
44
al. (2007), Karami & Stead (2008), and Yan (2008) have demonstrated the
capabilities of the ELFEN code for the analysis of a number of rock mechanics
problems of various complexity involving brittle failure. Initial applications of the
code for the analysis of block caving by Rockfield Software Ltd (2003), Esci &
Dutko (2003), Pine et al. (2006), Vyazmensky et al. (2007) and Elmo et al.
(2007b), have provided encouraging results.
3.5 Rock Mass Representation Options in Context of Fracture

Mechanics Based FEM/DEM Modelling
In the context of finite/discrete element modelling there are three possible
approaches to the representation of the rock mass:
Equivalent continuum;
Discrete fracture network; and
Mixed discrete fracture network/equivalent continuum approach.
In the Equivalent continuum approach, employing continuum modelling techniques,

the jointed rock mass system is represented as an equivalent continuum by means
of reducing intact rock properties to account for the presence of discontinuities.
The degree to which the intact rock properties are scaled can be deduced from one
of a number of rock mass classification systems. It should be noted that the
mechanical behaviour of a jointed rock mass is strongly influenced by the presence
of discontinuities which provide kinematic control and in many cases govern the
operative failure mechanisms. In the Equivalent continuum approach, the
kinematic controls resulting from jointing to some degree can be accommodated by
utilizing a constitutive model that allows directional strength properties.
However, it is suggested that the true kinematic controls of discontinuities cannot

be captured without explicit inclusion of pre-inserted fractures into the models. In
this sense the Discrete Fracture Network approach, where the rock mass is
represented as an assembly of pre-inserted discontinuities and intact rock regions,
is closer to reality. The intact rock properties can be established based on
laboratory tests and the discontinuity characteristics can be determined from field
45
mapping/borehole logging data or stochastic modelling. Fracture representation
must be adequate to capture the realistic behaviour of the specific problem under
investigation. Clearly it is not feasible to consider every single discontinuity within
the rock mass; however the resolution of fractures should be sufficient to capture
salient features of the simulated behaviour.
In some circumstances, such as the simulation of large scale problems, to achieve

reasonable computational efficiency, discontinuities must be placed fairly sparsely.
In this context representation of the rock between fractures as an intact material
may produce an overly stiff response. A mixed approach, where the rock mass is
represented as an assembly of numerically practical spaced discontinuities and
intervening regions with reduced intact properties, is a necessary compromise
between the first two approaches. This allows consideration of the kinematic
effects of discontinuities and adequate computational efficiency. In this approach
the appropriate combination of discrete fracture network/reduced intact rock
properties should be chosen so that the salient features of the simulated response
are captured.
As part of this research Vyazmensky et al. (2007) carried out ELFEN modelling
trials to evaluate the general applicability of the above methods of rock mass
representation to the modelling of block caving induced subsidence. The modelling
based on an equivalent continuum approach demonstrated that use of rock mass
properties derived through available rock mass classification systems produce
results which corresponded reasonably well with some trends observed at actual
block caving mines. Modelling conducted using the Discrete Fracture Network
approach showed that use of intact rock properties in combination with limited
discontinuity input leads to an overly stiff response so that rock mass caving cannot
be adequately captured, thus prohibiting subsidence analysis. An increase in the
number of input discontinuities to allow reasonable simulation of caving behaviour
would require a very fine mesh discretization and lead to excessively long
simulation run-times. Initial modelling, employing the mixed approach using
equivalent continuum properties in combination with pre-inserted discontinuities,
allowed realistic simulation of block caving induced surface subsidence. It was also
46
found that a combination of RMR based properties with an assumed fracture
network resulted in an overly soft response. This indicates that the selection of
equivalent continuum properties for a combined system of pre-inserted fractures
and an equivalent continuum rock mass requires further analysis. Pre-inserted
discontinuities played an important role in subsidence development. In the
equivalent continuum approach tensile cracks must be formed for toppling to occur
whereas in the mixed approach the kinematic conditions for toppling were already
in place, creating preferential directions of failure.
As discussed in Section 2.4, mining experience suggests that block caving

subsidence development is strongly affected by the presence of discontinuities,
and initial FEM/DEM with fracture modelling also supports this observation. It
appears that explicit consideration of discontinuities in the block caving
subsidence models is essential and therefore the mixed discrete fracture
network/equivalent continuum approach to rock mass representation was
adopted for the block caving subsidence analysis.
3.6 Integrated Use of Discrete Fracture Network Models and

ELFEN Code
Adequate representation of discontinuities in rock engineering models is essential
for capturing realistic failure mechanisms. In the current study an approach is
taken that integrates the use of the finite element/discrete element code ELFEN
with discrete fracture network (DFN) models capable of providing geologically
sound representation of natural discontinuities. The fundamentals of the DFN
approach can be found in Long et al. (1982) and Dershowitz & Einstein (1987).
3.6.1 DFN Approach to Discontinuity Representation
Accurate characterization of the discontinuities in a jointed rock mass is not a

trivial task due to their inherent three dimensional nature and the frequent
limitations in exposure to spatially isolated surface outcrops, boreholes and
stopes. A number of techniques have been proposed to develop 3D fracture
networks from collected discontinuity data using stochastic modelling. Studies
47
show that, among the different approaches developed to characterize fracture
networks, the discrete fracture network (DFN) model is the most appropriate to
simulate geologically realistic networks (Dershowitz et al., 1996). Initially applied
to the problems of fluid flow transport (Dershowitz, 1992) and hydrocarbon
reservoir modelling (Dershowitz et al., 1998), in recent years the DFN approach
has found wider use in rock engineering analysis (see Staub et al., 2002; Elmo,
2006; Rogers et al., 2006; Rogers et al., 2007 and Elmo et al., 2007b).
DFN models seek to describe the heterogeneous nature of fractured rock

masses by explicitly representing key elements of the fracture system as discrete
objects in space with appropriately defined geometries and properties (Rogers et
al., 2006). By building geologically realistic models that combine the larger
observed deterministic structures with smaller stochastically inferred fractures,
DFN models capture both the geometry and connectivity of the fracture network,
the geometry of the associated intact rock blocks and also the nature of intact
rock bridges connecting fractures (Rogers et al., 2007).
Building DFN models is an iterative procedure that involves multiple random

sampling aiming to build upon limited measured data and eventually replacing it
with a more detailed pseudo-replicate model. Rogers et al. (2007) described the
process of building a DFN model which involves four main stages: data analysis,
conceptual model development, model building and validation. Fig. 3.7 shows
the work flow in this process. Of these stages, validation is particularly important
to ensure a good fit between simulated and collected data. Examples of DFN
model development can be found in Pine et al. (2006) and Elmo (2006). A DFN
model based on this rigorous process is not just a statistical description but an
ordered geologically realistic representation of the distribution, interaction and
geometry of rock mass fractures (Rogers et al., 2007).
48
Main Data Sources
Geophysics Borehole Data
Well Testing Mapping / Imaging
Validation Loop
Geological Model Data Analysis & DFN Inputs
Conceptual Fracture Model
DFN Model Model Validation
Fig. 3.7 DFN model development workflow

(modified with permission after Rogers et al., 2007)
3.6.2 Integration of DFN Models into ELFEN
In this thesis the Discrete Fracture Network (DFN) code FracMan (Golder, 2005;
Dershowitz et al., 1998) is employed. FracMan is a convenient tool to generate 3D
stochastical models of fracture networks based on collected discontinuity data; it
allows export of 2D and 3D fracture data into the ELFEN code. Integrated use of
ELFEN and FracMan was first proposed and applied by Elmo et al. (2006) and Pine
et al. (2007). An example of 3D FracMan model and derived 2D trace planes for
ELFEN modelling is given in Fig. 3.8. Elmo (2006) and Rance et al. (2007)
described the methodology for transfer of the FracMan DFN data into the ELFEN
finite element/discrete element model including the following steps:
1. The fracture geometry data are exported from the FracMan system in
Autocad dxf files defining fracture planes within a rock mass on a full three
dimensional basis. These planes may intersect at arbitrary angles and do
not normally traverse the entire region. This information is imported using
a specific modelling interface module in ELFEN in which the joints are
represented as lines for 2D and planar surfaces for 3D situations.
49
2. Imported fracture surfaces are then assigned appropriate interface
properties such as friction coefficient, cohesion, normal and tangential
penalty values.
3. Fracture entities are first constructed independently as a network, accounting
for intersections of lines or surfaces, including the partial intersection of
surfaces in 3D and the intersection with material region boundaries.
4. Once the network has been constructed it is embedded within the
computational solid ELFEN model of the rock mass by inserting fractures
with both sides of the fracture represented as free surfaces. The crack
aperture is normally (as in the current study) assumed closed, but may be
set to a prescribed value.
5. This solid model is then discretized into a finite element mesh employing
triangular (in 2D) and tetrahedral elements (for 3D problems).
Imported discontinuity data may require minor editing to allow discretization.

This may include merging very closely spaced fractures and/or snapping together
fractures terminating in close proximity to each other. Care should always be
taken to ensure that the editing does not introduce significant geomechanical
changes to the rock mass.
(a) (b)
Set 1a Set 2a Set 3a

Set 1b Set 2b Set 3b
Bedding
Fig. 3.8 Example of a FracMan model (adapted after Elmo, 2006, with permission)
(a) FracMan 3D model of a Middleton mine pillar developed based on scanline mapping
data; (b) sampling planes used to define the 2D fracture traces models for ELFEN
50
3.7 General Considerations in Modelling Methodology
One of the main objectives of this thesis is to improve our general understanding
of the block caving induced subsidence phenomenon. It should be noted that the
ELFEN code is capable of modelling fracturing in both 2D and 3D space.
Although eventually full 3D mine scale analysis of block caving subsidence is
undoubtedly desirable, available modelling tools have yet to reach the
computational efficiency to allow a detailed and realistic 3D analysis. It is
suggested that parametric 2D analyses are an essential prerequisite to complex
three dimensional models if we are to further understand the factors controlling
caving and induced subsidence. In the current 2D modeling study emphasis is
given to the representation of the maximum level of detail possible with the
currently available computing power. The future development of 64 bit parallel
FEM/DEM with fracture codes will without doubt allow improved modelling in both
two and three dimensions. It is intended that the work presented in this thesis
will both demonstrate the immense potential of the FEM/DEM plus fracture
approach and provide a foundation for future more computationally intensive
research. Notwithstanding it should be stated that the models presented as part
of this thesis were in general run on high end PC‟s with the maximum available
RAM to 32 bit computers and involved runtimes of up to 25 days.
3.8 Summary
This chapter outlined the fundamental principles and advantages of the hybrid
FEM/DEM fracture mechanics based approach and its implementation in the
numerical code ELFEN. Clearly, application of this approach in combination with
DFN based fracture systems provides the opportunity to undertake physically
realistic modelling of a highly complex process such as block caving induced
surface subsidence development.
51
CHAPTER 4: USE OF EQUIVALENT CONTINUUM ROCK
MASS PROPERTIES IN THE DISCRETE FRACTURE
NETWORK FEM/DEM MODELLING OF BLOCK CAVING
INDUCED SUBSIDENCE
4.1 Introduction
One of the key aspects in the numerical modelling of rock engineering problems
is establishing representative rock mass material strength and deformability
characteristics. This chapter
discusses available approaches to the derivation of jointed rock mass
mechanical properties;
examines the properties output from rock mass classification systems;
analyzes rock mass representation options within FEM/DEM plus
fracture modelling; and
presents a novel approach to the derivation and calibration of rock mass
properties specifically tailored for block caving subsidence analysis using
integrated DFN-FEM/DEM modelling.
4.2 Rock Mass Equivalent Continuum Mechanical Properties

On a scale of typical rock engineering analysis, such as the design of a civil/mine
opening or slope stability assessment, the rock mass material represents a
system comprised of intact rock regions separated by natural, pre-existing
discontinuities. The mechanical response of a jointed rock mass system is highly
complex and the material properties of the rock mass differ quantitatively by a
substantial but often unknown amount from laboratory tested intact rock
specimens. Usually in modern rock mechanics the rock mass is treated as an
equivalent continuum with scaled down strength and deformability characteristics
in order to account for the presence of discontinuities. The main equivalent
52
continuum rock mass mechanical properties required by most engineering
models include rock mass deformability and strength. Their description is given
below.
Rock mass deformability represents the combined deformability of the intact rock
material and the deformability of the discontinuity interfaces. The deformability of
the intact rock component is governed by its Young‟s modulus and Poisson‟s
ratio, whereas the normal and shear stiffness of a discontinuity interface governs
its deformation characteristics. The higher the density of the discontinuities,
generally, the more deformable the rock mass.
Rock mass strength parameters include uniaxial compressive strength, cohesion

and frictional strength, and uniaxial tensile strength. For the current ELFEN
analysis, uniaxial compressive strength is not used as a direct input parameter.
The rock mass cohesion is governed by the strength of the intact rock bridges
and the cohesive strength along the discontinuities interfaces. Fewer and less
persistent joints would result in higher cohesive strength of the rock mass. Good
joint surface contacts also contribute to higher cohesion. Similarly, the rock
mass friction angle is governed by two contributing components, the intact rock
internal friction angle and the friction along the discontinuity surfaces. In most
cases it is the frictional strength available of the joint surface which is the critical
factor. Rock mass friction generally depends on the orientation and persistence
of discontinuities, joint surface conditions (especially roughness) and presence of
infill.
Lastly, the intact rock can have a substantial tensile strength. On the rock mass
scale, typically, only a small component of the intact tensile strength can be
considered. Generally the tensile strength of the rock mass would depend on
persistence and density of discontinuities as well as amount of rock bridges and
quality of the jointing infill. The range in rock mass tensile strength could be from
nearly zero for heavily fractured rock masses up to nearly the intact rock tensile
strength for massive rock masses with very few discontinuities.
53
4.3 Derivation of Rock Mass Mechanical Properties
According to Jing (1998) the continuum approach trades material complexity for
geometrical simplicity, requiring proper homogenization techniques to identify the
material parameters associated with specified constitutive equations for the
equivalent continuum; the homogenization process is usually very complex and
valid only over a certain representative elementary volume. Cai et al. (2004)
indicated that there are so many parameters that affect the deformability and
strength of an arbitrary rock mass, it is generally impossible to develop a
universal law that can be used in any practical way to predict the mechanical
behaviour of the rock mass. As noted by Brady & Brown (2004) the
determination of the global mechanical properties of a large mass of
discontinuous in-situ rock remains one of the most difficult problems in the field of
rock mechanics. Clearly, establishing representative rock mass properties for
numerical modelling input is not a trivial task. Despite the apparent difficulty,
addressing this issue is unavoidable in any practical rock engineering analysis.
The principal approaches for determination of the basic mechanical characteristics

of a jointed rock mass include:
analytical solutions;
large scale in-situ testing;
numerically simulated large scale testing; and
empirical relationships based on rock mass classification systems.
4.3.1 Analytical Methods
Several closed form analytical solutions exist for description of the behaviour of a
jointed rock mass, among them Jaeger‟s (1960) and Wei‟s (1988) theories which
address the mechanical response of an axially loaded rock sample containing
several variously oriented discontinuities. Jaeger‟s plane of weakness theory
can be applied in several parts to determine the strength of the jointed rock
sample. Wei‟s theory gives a complete solution for derivation of the jointed rock
sample deformability, allowing incorporation of the effect of varying persistence.
54
It should be noted that these solutions have to date found limited practical
application as the required input data is seldom readily available.
4.3.2 In-situ Testing
It is possible to determine the rock mass strength and deformability characteristics of

a rock mass through “large scale” tests carried out in-situ. Traditional methods to
determine these parameters include compression tests for deformation modulus
(e.g. ASTM D4395 - 04) and block shear tests (e.g. Szymakowski, 2007) for shear
strength parameters. Generally the use of in-situ testing is complicated by the
difficulty in selecting and accessing the representative rock mass segment. These
methods are associated with high costs and therefore are applied only in rare cases.
4.3.3 Numerical Modelling
In recent years, numerical modelling methods are being increasingly employed for
deriving rock mass properties through simulated rock mass tests. The Particle
Flow Code, PFC, (Itasca, 2007), based on a “Synthetic Rock Mass” approach
(Pierce et al., 2007; Mas Ivars et al., 2007; Cundall, 2008) is a particularly
interesting development. It involves construction of a synthetic “sample” of the
rock mass in two or three dimensions by bonding together thousands of circular or
spherical particles. DFN based pre-inserted joints are introduced by debonding
particles along specific joint surfaces and employing a sliding joint model within the
PFC code. This approach overcomes the limitations of other numerical modelling
approaches to rock mass simulation in which the results are governed by a user-
specified constitutive model. Through strain-path driven loading of the “Synthetic
Rock Mass” samples the estimates of rock mass strength and brittleness can be
derived and used as a direct input to standard continuum models. Reyes-Montes
et al. (2007) performed a validation study of the Synthetic Rock Mass approach
and found a good correlation between the fracturing orientation and modes
predicted by the PFC models and those observed from in-situ recorded
microseismicity at Northparkes block cave mine. Overall, it should be emphasized
55
that this approach is still being refined and further work is ongoing to illustrate its
applicability in the analysis of a range of practical rock engineering problems.
4.3.4 Empirically Based Rock Mass Classification Systems.
Empirical rock mass classification systems are perhaps the most widely used tool
for the assessment of rock mass engineering behaviour. According to Brady &
Brown (2004), classification systems seek to assign numerical values to those
properties or features of the rock mass considered likely to influence its behaviour,
and to combine these individual values into one overall classification rating for the
rock mass. There are more than 20 different rock mass classification systems (see
review by Edelbro et al., 2006). The most commonly used are the Rock Mass
Rating RMR76 (Bieniawski, 1976), the Tunnelling Quality Index Q (Barton et al.,
1974) and the Geological Strength Index GSI (Hoek, 1994). Table 4.1 lists
parameters utilized in the rating calculation for these systems.
Table 4.1 Rock mass parameters used in rock mass rating by RMR76, Q and GSI
RMC system Classification parameters Rating values
UCS, RQD, Joint spacing, Joint conditions,
RMR76 Ground water conditions, (adjustments for joint 8 … 100
orientation)
Joint set number (Jn), RQD, Joint roughness (Jr),
Q Joint alteration (Ja), Joint water reduction factor 0.001 … 1000
(Jw), Stress reduction factor (SRF)
GSI Rock mass blockiness, joint surface conditions 5 … 100
The RMR and Q systems were initially developed to provide assessment of rock
mass quality for tunnelling, their application later expanded to a wider field of rock
engineering problems. Accumulated experience of applying these rating systems
has led to the development of empirical correlations between rock mass rating and
basic mechanical properties. The GSI rock mass classification system was
developed specifically to provide input data for engineering analysis. It relies on
fewer parameters and is built around the Hoek-Brown failure criterion (Hoek et
al., 2002). Empirical correlations exist between the GSI rating, rock mass
deformability and Hoek-Brown failure criterion parameters. A summary of RMR,
56
Q and GSI based empirical relationships for estimation of mechanical properties
of the rock mass is given in Table 4.2. To appreciate the applicability and
limitations of these relationships it is important to understand their origin.
Table 4.2 Empirical relationships for derivation of estimates of basic rock mass strength and
deformability properties based on the RMR, GSI and Q rock mass classification systems
RMC Estimates of Rock Mass Strength and Deformability Reference
System Characteristics
Serafim &
Em 10 ( RMR 10) / 40 (GPa) Pereira
RMR 5 RMR / 2 (1983)
c 5RMR (kPa) Bieniawski
(1989)
1
Em 10 Qc 3 (GPa)
1 Jr Jw
" " tan
Ja 1
Q RQD 1 c Barton
" c" (MPa) (2002)
Jn SRF 100
where:
Qc Q( c / 100) - normalized Q;
c – uniaxial compressive strength (MPa)
1 D/2
Em Ei 0.02 (( 60 15 D GSI ) / 11)
(MPa)
1 e
' a 1
' 6amb s mb 3n
a sin
21 a 2 a 6amb s mb ' a 1 Hoek et al.
3n
(2002)
' ' a 1
' ci 1 2a s 1 a mb 3n s mb 3n
c (MPa)
GSI ' a 1 Hoek &
1 a 2 a 1 6amb s mb 3n 1 a 2 a
Diederichs
(2006)
where:
Ei – intact rock Young‟s modulus;
D - disturbance factor;
a, s, mb – material constant;
' '
3n 3 max / c, 3 max - upper limit of confining stress
The RMR and Q correlations for the Young‟s modulus originated from the dataset
of in-situ tests compiled by Bieniawski (1978) and Serafim & Pereira (1983), both
57
correlations were derived through exponential fit to the data, Fig. 4.1. Recent
GSI correlations are founded on a different dataset compiled from in-situ testing
data from China and Taiwan. The Hoek & Diederichs (2006) GSI correlation of
rock mass Young‟s modulus to GSI rating is based on a sigmoidal fit to the
experimental data in order to constrain the increase of the modulus as the rock
becomes more massive, Fig. 4.2.
Fig. 4.1 RMR, Q and rock mass deformation modulus correlations
Fig. 4.2 GSI and rock mass deformation modulus correlation proposed by Hoek &
Diederichs (2006)
Bieniawski (1979) suggested approximate ranges of cohesion and friction angles

for particular RMR ranges. In a later version of the RMR classification system
58
(Bieniawski, 1989), used these ranges to propose simple relationships correlating
RMR, rock mass cohesion and friction.
The GSI does not provide a direct estimate of rock mass cohesion and friction, as
there is no direct correlation between the linear Mohr-Coulomb criterion and the
non-linear Hoek-Brown criterion. The estimates of these parameters are
computed through a procedure that involves simulating a set of triaxial strength
tests using Hoek-Brown criterion then fitting the Mohr-Coulomb linear failure
envelope to the resulting Mohr‟s circles through linear interpolation, thus allowing
computation of c and υ. Care should be exercised when selecting the „best‟ linear
line in fitting the Mohr circles. Generally it depends on the level of confining stress
for a particular engineering application. For a slope problem, the confining stress
may vary from zero to some level of stress (usually related to the height of the
slope). Curvature of the non-linear Hoek-Brown strength envelope is greatest at
low confinement stresses, and for this region the fitting procedure is particularly
sensitive to the assumed stress. For a tunnel problem, if the depth and stress
range are known, the linear envelope should be fitted best for the Mohr circles in
the stress region of interest (Hoek & Brown, 1997; Hoek et al., 2002).
Barton (2002) gives a detailed rationale for the proposed Q correlations for
estimation of jointed rock mass cohesion and friction. Jointed rock mass cohesion
in Q correlation is derived from the “massiveness” of the rock mass, expressed
through the RQD/Jn ratio. Massive, highly stressed rock masses with high
cohesive strength suffer the greatest reduction in block-size and cohesive strength,
as a result of stress-induced fracturing around deep excavations, therefore it is
important to adjust the cohesion estimates for the effect of stress. Generalization
of the cohesion estimate is achieved through normalization with c/100. The
jointed rock mass friction in the Q correlation is governed by the conditions at the
discontinuity contacts. Barton (2002) argued that the Jr/Ja ratio closely resembles
the dilatant or contractile coefficient of friction for joints and filled discontinuities.
Back-calculation of case records showed that tan-1(Jr/Ja) provided realistic
estimates of the friction angle on the discontinuity surfaces. To account for the
59
potential effect of ground water the Jw reduction factor was included in the
correlation. It was noted that the Jr and Ja ratings should be derived for the
discontinuities most affecting the result of the particular loading direction.
There is an extensive discussion in the literature about the use and misuse of rock
mass classification systems (see Milne et al., 1998; Stille & Palmstrom, 2002;
Marinos et al., 2005). The validity of mechanical properties derived using different
systems is also a subject of considerable debate. Marinos et al. (2005) argued
that the ground water and structural orientation parameters in RMR and the
groundwater and stress parameters in Q are dealt with explicitly in effective stress
numerical analyses and the incorporation of these parameters into the rock mass
property estimate results is inappropriate. Barton (2007) asserted that: “GSI-
based Hoek-Brown formulations for “simple” geotechnical input data for the rock
mass, such as deformation modulus, cohesion and friction angle, appear to have
reached „black-box‟ levels of complexity, which seems to be detrimental to the idea
of rock engineering, if engineering judgement is still to be exercised in this
rewarding field of engineering”.
Notwithstanding the uncertainties and limitations, rock mass classification systems

remain a primary source for estimation of equivalent continuum rock mass strength
and deformability characteristics in practical rock engineering analyses. The
following section compares the properties derived from rock mass properties
correlations given in Table 4.2.
4.4 Comparative Study of Rock Mass Strength and Deformability

Parameters Derived through RMR76, GSI and Q Rock Mass
Classification Systems
4.4.1 Properties Derivation
It is a common practice in rock engineering to use the rating from one

classification system to estimate the value of another. A number of cross-
correlations between RMR, GSI and Q systems ratings were suggested over the
years (see review by Milne et al., 1998; Ramamurthy, 2004). According to Milne
60
et al. (1998), the equation proposed by Bieniawski (1976), linking RMR76 and Q,
is the most popular correlation:
RMR76 9 ln Q 44 (4.1)
Hoek et al. (1995) suggested that for dry RMR76 > 18:
GSI RMR76 (4.2)
Given the intrinsic interrelationship between the RMR, GSI and Q ratings, it is
possible to compare the rock mass properties estimated by these classification
systems for the same rating value. Based on this proposition, a fictitious dry, hard
rock mass with fair joint surface conditions was considered. For this rock mass,
based on the RMR, GSI and Q systems, four cases with varying rock mass quality
ratings equivalent to RMR76 80, 70, 60 and 50 were derived and corresponding
rock mass properties were estimated, as shown in Tables 4.3 to 4.5. Here it should
be noted that extreme care was taken to maintain consistency in the rock mass
conditions between the fictitious cases across studied classifications systems. The
comparison discussed here is based on an assumption that comparative RMR76, Q
and GSI ratings should yield similar mechanical properties.
Table 4.3 Rock mass properties derived using RMR76

Rating RMR76 Rock Mass
Criteria RMR 80 RMR 70 RMR 60 RMR 50
UCS, MPa 100 - 250 (12)* 100 - 250 (12) 50 - 100 (7) 50 - 100 (7)
RQD 75%-90% (17) 50%-75% (13) 25%-50% (8) <25% (3)
Spacing of
0.6 – 2m (15) 0.2 -0.6m (10) 0.2 -0.6m (10) <0.06m (5)
discont.
Length 1-3m (4) Length 1-3m (4) Length 1-3m (4) Length 1-3m (4)
Separ. 0.1-1mm Separ. 0.1-1mm Separ. 0.1-1mm Separ. 0.1-1mm
Joint (4) (4) (4) (4)
conditions Smooth (1) Smooth (1) Smooth (1) Smooth (1)
No infill (6) No infill (6) No infill (6) No infill (6)
Unweath. (6) Unweath. (6) Unweath. (6) Unweath. (6)
Water cond Dry (15) Dry (15) Dry (15) Dry (15)
E=56GPa E=32GPa E=18GPa E=10GPa
Estimated
c=0.4MPa c=0.35MPa c=0.3MPa c=0.25MPa
rock mass
=45 =40 =35 =30
properties
σt =0.33MPa** σt =0.33MPa σt =0.31MPa σt =0.29MPa
2c cos
* - RMR rating value; ** - derived using Mohr-Coulomb criterion: t
1 sin
61
Table 4.4 Rock mass properties derived using Q
Q Rock Mass
Rating
Q 47.5 Q 17.5 Q 6.2 Q 0.67
Criteria
(RMR79) (RMR 70) (RMR 60) (RMR 50)
RQD=95% RQD=70% RQD=37% RQD=12%
Block size 1 joint set, Jn=2 2 joint sets, Jn=4 2 joint sets + 3 joint sets, Jn=9
random, Jn =6
Smooth, planar Smooth, planar Smooth, planar Smooth, planar
Inter block
joints, Jr =1; joints, Jr =1; joints, Jr =1; joints, Jr =1;
shear
Unaltered joint Unaltered joint Unaltered joint Unaltered joint
strength
walls, Ja =1 walls, Ja =1 walls, Ja =1 walls, Ja =1
Dry: Jw=1; Dry: Jw=1; Dry: Jw=1; Dry: Jw=1;
Active
Medium stress: Medium stress: Medium stress: Medium stress:
Stress
SRF=1 SRF=1 SRF=1 SRF=1
for σc =110MPa for σc=100MPa for σc=90MPa for σc=80MPa
c=52MPa c=15MPa c=4.7MPa c=1.6MPa
=45 =45 =45 =45
Estimated
σt=43MPa* σt=13MPa σt=3.9MPa σt=1.3MPa
rock mass
(no cut off) (no cut off) (no cut off) (no cut off)
properties
σt=22MPa σt =6.3MPa σt =1.8MPa σt =0. 7MPa
(50% cut off) (50% cut off) (50% cut off) (50% cut off)
σt=4.3MPa σt =1.3MPa σt =0.4MPa σt =0.1MPa
(90% cut off) (90% cut off) (90% cut off) (90% cut off)
* - derived using Mohr-Coulomb criterion
Table 4.5 Rock mass properties derived using GSI

GSI Rock Mass
Rating
GSI 80 GSI 70 GSI 60 GSI 50
Criteria
(RMR 80) (RMR 70) (RMR 60) (RMR 50)
UCS, MPa 110 100 90 80
mi 15 15 15 15
γ, MN/m3 0.026 0.026 0.026 0.026
Ei, GPa 60 60 60 60
D 0 0 0 0

Estimated
c=4.9MPa c=2.9MPa c=2 MPa c=1.4MPa
rock mass
=57 =57 =55 =53
properties
σt =1.6MPa* σt =0.8MPa σt =0.4MPa σt =0.2MPa
* - derived using Mohr-Coulomb criterion
The strength and deformability property estimates for the RMR and Q systems
were computed directly based on the equations given in Table 4.2, and the GSI
derived properties were established using the RocLab v1.031 program
(Rocscience Inc., 2007), which is based on the GSI equations given in Table 4.2.
62
It should be noted that the GSI allows property estimates to be tailored to slope
stability analyses, as well as for tunnelling. The current study deals with
underground mining simulation and therefore assumed a tunnel located at 200m
(i.e. corresponding to a moderate stress level in the Q system correlation).
Fig. 4.3 compares the estimates of rock mass Young‟s modulus, friction angle,
cohesion and tensile strength derived using RMR76, Q and GSI systems, expressed
through equivalent RMR76 rating. Fig. 4.4 shows the relative change in the
magnitude of the above mentioned parameters with increasing rock mass rating.
(a) Deformation modulus (b) Friction angle

60 60
56 55 57
RMR 57
53
GSI - tunnel (-200m) 53
50 50
45 45 45
Q
Deformation modulus, GPa
44 45
Friction angle, degrees

40 40 45
40
26 37
31 35
30 30
30
32
20 20
18
18 RMR
11
10 10 GSI - tunnel (-200m)
10
Q
0 0
50 60 70 80 50 60 70 80
RMR RMR
(c) Cohesion (d) Tensile Strength

52 / 80 43 / 80 22 / 80
20 15
RMR RMR
18 GSI - tunnel (-200m)
GSI - tunnel (-200m)
Q 13
16 Q
15 Q - 50% cut-of f
14
Tensile strength, MPa
10 Q - 90% cut-of f
Cohesion, MPa
12
10
6.3
8
5
6 4 4.3
4.7 4.9
4 2.9
2 1.3 1.8 1.3
2 1.6 1.4 0.8 1.6
0.35 0.4 0.7 0.4
0.3 0.2 0.33
0 0.25 0 0.31 0.33
0.1
50 60 70 80 50 60 70 80
RMR RMR
Fig. 4.3 Comparison of rock mass strength and deformability characteristics derived
using RMR76, Q and GSI RMC systems
(a) deformation modulus; (b) friction angle; (c) cohesion; (d) tensile strength
63
(a) RMR (b) GSI
1000 1000
Magn. increase in %of equiv. RMR 50 value

Magnitude increase in %of RMR 50 value
Def orm. modulus Def orm. modulus
Cohesion Cohesion 841
800 Friction 800 Friction
Tensile Strength Tensile Strength
600 600
462
400 400 371
216 250
200 200 187
139
78 112 107
40 33 60 50 70 43
20 17 8 13 15 7.4 8.2
4.4
0 0
60 70 80 60 70 80
RMR Equivalent RMR
(c) Q
3150 3238
1000
Magn. increase in % of equiv. RMR 50 value
Def orm. modulus

Cohesion 843 861
800 Friction
Tensile Strength
600
400
234
194 200
200
133
59
0 0 0
0
60 70 80
Equivalent RMR
Fig. 4.4 Percentile change in properties magnitudes with increasing rock mass rating
(a) RMR based properties; (b) GSI based properties; (c) Q based properties
4.4.2 Rock Mass Deformation Modulus
According to Fig. 4.3(a), the estimates of rock mass deformation modulus

derived using the three different rock mass classification systems exhibit
generally similar trends. The GSI correlation yields deformation modulus values
in a range of 18 to 53 GPa which are consistently higher than those by other
systems, with the exception of an equivalent RMR of 80 where the RMR
predicted modulus is about 3 GPa higher. The Q correlation yields values which
are about 30-40% lower than those predicted by GSI (12 - 37 GPa). The RMR
64
correlation curve follows the Q curve for RMR 50 to 60 and then steeply rises.
Fig. 4.4 indicates that with the increase of rock mass rating from RMR 50 to RMR
80, the RMR based deformation modulus estimate increases by more than five
times (from 10 to 56 GPa), whereas GSI and Q based estimates only doubled. It
is rather difficult to make a definitive judgement on which system yields more
appropriate modulus values, as all systems are based on interpretation of actual
measured data. The recently proposed GSI correlation is based on a large
number of case histories and has been rigorously verified, as reported by the
authors. However, GSI estimates appear to be less conservative. More practical
applications are needed to gain confidence in GSI based modulus correlation.
Overall, the Q correlation appears to yield adequately conservative values, with
respect to deformation modulus, throughout the studied rating range.
4.4.3 Rock Mass Friction Angle
As follows from Fig. 4.3(b) the estimates of rock mass friction angle derived using
RMR and GSI exhibit an increase in rock mass friction angle with increase in
rock mass rating. Friction angle estimates in the Q correlation are solely
dependant on joint surface conditions. For the purpose of the current analysis,
identical joint surface conditions were assumed, with Q correlations yielding a
constant friction angle of 45 degrees. It should be noted that sole dependence of
friction estimates on joint surface conditions in some cases may lead to very low
estimates of friction angle. As reported by Barton (2002), estimates of the friction
angle of discontinuity interfaces with clay infill can be as low as a few degrees. It
appears unreasonable to assume that an equivalent continuum rock mass would
have such a low friction angle. If the discontinuities with very low frictional
characteristics are present they must be incorporated into the equivalent
continuum analysis explicitly, or alternatively, a discontinuum analysis should be
considered. GSI correlations yielded high friction angle estimates of more than
50 degrees, although only a slight change in the magnitude (less than 10%) was
observed with increasing rock mass rating. Justifying such high friction angle
values is rather difficult, particularly for the rock mass with lower rating. RMR
65
based friction angle estimates vary by 15 degrees over the studied rating range,
from 30 degrees for RMR 50 to 45 degrees for RMR 80. It appears that RMR
correlation produces overall realistic estimates of the rock mass friction angle.
4.4.4 Rock Mass Cohesion
Fig. 4.3(c) shows the variation of the rock mass cohesion estimates for different
equivalent RMR ratings. As shown in this figure and Fig. 4.4, cohesion estimates
differ quite dramatically between the RMR, GSI and Q systems. RMR based
correlation produces a very narrow range of very low rock mass cohesion
estimates, less than 0.4 MPa, which is comparable to the cohesive strength of
stiff clays. Barton (2002) noted that such estimates give the impression of far too
low a cohesion for hard rock, and was perhaps estimated from experience of coal
measure rocks, from which many RMR case records were derived. Overall, it
appears that RMR based cohesion estimates are overly conservative. The GSI
based correlation produces more reasonable values, roughly an order of
magnitude higher than RMR. The Q correlation shows an exponential trend,
giving a very broad cohesion range, so that an equivalent RMR 80 cohesion
estimate of 52 MPa is more than 30 times higher than its equivalent RMR 50
counterpart and more than 100 times higher than the RMR based cohesion for
the same rating. It appears that the Q correlation tends to overestimate rock
mass cohesion for higher ratings (equivalent RMR>60), where cohesion
estimates approach high values characteristic of very strong intact rock.
4.4.5 Rock Mass Tensile Strength
Rock mass tensile strength estimates for RMR, GSI and Q systems are shown in
Fig. 4.4(d). It should be noted that only the GSI allows direct estimates of rock
mass tensile strength. GSI estimated values are in the range 0.2 to 1.6 MPa, for
a corresponding equivalent RMR of 50 to 80. The estimates of tensile strength
below 0.3 MPa for RMR<60 seems to be very low, GSI estimates for the
remaining range seems to be reasonably realistic. For RMR and Q, the tensile
strength was derived using cohesion and friction values and the Mohr-Coulomb
66
relationship (see Table 4.3). It is well recognized that the Mohr-Coulomb based
tensile strength over predicts the actual values and an adjustment through a
“tensile cut-off” is necessary. As discussed in Section 4.2.1, the rock mass
tensile strength may vary over a wide range depending on the density of
discontinuities and deriving a representative tensile cut-off in such circumstances
can be challenging. Provided that the RMR estimated cohesion is within the kPa
range, for the studied rating values the derived tensile strength is about 300 kPa.
Application of a tensile cut-off to such low values is not particularly meaningful.
In contrast, high values of cohesion for Q result in very high tensile strength
estimates, up to 60 times larger than for RMR. Here application of a tensile cut-
off is required. Two cut-off assumptions were considered, 50% and 90%. It
appears that the assumption of a 90% cut-off yields generally realistic tensile
strength estimates for an equivalent RMR>60.
4.4.6 Summary
In summary, comparison of strength and deformability properties derived from

the RMR, Q and GSI systems for corresponding rating values for the most part
showed a lack of consistency. The following can be concluded from this
comparative study:
RMR based correlations produce generally reasonable estimates of rock

mass deformation modulus and friction angle, but yield extremely low
values of cohesion and consequently very low tensile strength estimates.
GSI based correlations produce generally realistic estimates of rock mass
mechanical properties although the friction angles indicated are higher in
comparison to RMR and Q. Deformation modulus estimates produced by
the GSI correlation appear to be stiffer and less conservative than other
methods.
Q based deformation modulus correlation yields reasonably conservative
estimates. Estimates of rock mass cohesion appear to be very high for
higher Q ratings. Friction angle estimates are solely dependant on
conditions along the discontinuities interfaces, which may result in very
67
conservative values. Using the derived cohesion values the calculated
rock mass tensile strength requires the assumption of a substantial tensile
cut-off to achieve reasonable tensile strength estimates.
4.5 Evaluation of Applicability of Rock Mass Classifications

Derived Equivalent Continuum Properties for Modelling of
Block Caving Induced Surface Subsidence
Analyses presented here evaluate the use of rock mass classification systems as
a source of equivalent continuum properties for integrated FEM/DEM-DFN
modelling of block caving induced surface subsidence. A series of numerical
experiments were carried out to investigate the behaviour of a system comprised
of key discontinuities and an equivalent continuum rock mass derived using
varying assumptions for mechanical properties from rock mass classification
systems.
4.5.1 Modelling Methodology
4.5.1.1 Model Setup
A 4000m x 600m ELFEN model sub-divided into non-fracturing and fracturing

regions was used, see Fig. 4.5. The fracturing region spans up to 1000m and
covers the principal area where fractures may potentially develop and has a
higher mesh resolution (2m sized elements). The non-fracturing region has a
lower discretization density (up to 50m elements) and is required to extend to the
model boundaries to minimize potential boundary effects on simulation results. A
sensitivity analysis of the assumed boundary conditions, illustrated that the
adopted model size and boundaries for the fracturing and non-fracturing regions
are adequate for the current modelling, i.e. the effect of the boundary conditions
on modelling domain is negligible. Flores & Karzulovic (2002) studied a number
of block caving mines and reported typical caved ore block heights of around
200m. In the current study, block caving mining is simulated by undercutting and
full extraction of a block of ore (100m x 100m) located within the fracturing region
at 200m depth. The undercut (100m x 4 m) is developed in stages in 20m
68
increments. A uniform draw of ore is assumed. With respect to simulation of the
ore extraction there are generally two options:
Option 1. Looped (repeated) deletion of the caved material (discrete
elements) within the full length of the undercut level;
Option 2. Uniform material withdrawal by gradual lowering of the undercut

floor, as illustrated in Fig. 4.5.
69
Model geometry
Fracturing zone
Non-fracturing zone
600m
4000m
Model setup Constraints

FracMan 3D model 2D trace plane
fractures Caveability Laubscher’s caveability chart
exported
into ELFEN
Cave development Conceptual model of caving
progression by Duplancic & Brady (1999)
Subsidence limits Mining experience
100m
100m
ore
100m
block
4m undercut model response

20o moving platform evaluation
70o
140m
Fig. 4.5 ELFEN model setup and response evaluation procedure
70
Initial modelling trials showed that use of the first option is complicated by the
fact that simultaneous deletion of the caved fragments in the 4m high undercut
triggered sudden collapse of the overlying rock mass, causing shock loading of
the system and creating numerical noise which complicated interpretation of
surface displacements. Significantly smaller looped deletion region heights could
not be considered as available code functionality only allows deletion of discrete
elements fully enclosed in a specified rectangular region and the minimum
element size assumed for the current modelling was 2m. For the above reason
option two was adopted for the current modelling. Similar methodology was
utilized for physical model test studies of caving by Kvapil (2004) as illustrated in
Fig. 4.6. Isotropic material properties were assumed throughout the model and
the draw was continued until the volume of rock equivalent to the volume of the
ore block is extracted.
Fig. 4.6 Physical model of progressive caving development (adapted after Kvapil, 2004
with permission).
71
Mahtab et al. (1973) noted that the fracture system most favourable for caving
includes a low dipping and two nearly orthogonal steeply dipping joint sets. The
3D FracMan DFN model adopted in the current analysis incorporated one
horizontal and two orthogonal vertical sets with widely spaced and moderately
persistent joints. The joint pattern for the 2D model was derived by assuming a
plane parallel to one of the vertical sets within the 3D DFN model. Joint traces
intersecting this plane were delineated and exported into ELFEN. Imported joint
sets were rotated with respect to the model centre to achieve the desired dip.
The following joint orientations were adopted: one sub-vertical set dipping at 70°
and one orthogonal sub-horizontal set (Fig. 4.5). It should be emphasized that
for the current study all modelling was based on a single FracMan realization.
Generated joints were assumed to represent key discontinuities governing rock
mass response.
Representation of the rock mass strength and deformability in the inter-joint

regions was based on the properties derived from the RMR and Q correlations.
GSI correlations were not used in the current analysis. Hoek et al. (2002) stated
that for studies of problems such as block caving in mines it is recommended that
no attempt should be made to relate the Hoek-Brown and Mohr-Coulomb
parameters. Correlations relating these parameters are valid only for cases
where the zone of failure does not extend to the surface.
Estimates of the RMR and Q system based rock mass properties derived in
Section 4.2.3 were utilized. It was assumed that pre-inserted discontinuities
have no cohesive strength and have a friction angle of 35 degrees. As discussed
in Section 3.5.4 the normal penalty at the pre-inserted fractures interface can
effectively be considered equivalent in magnitude to the joint normal stiffness.
Similarly, the tangential penalty was taken as equivalent to the joint shear
stiffness. An estimate of the joint tangential penalty was derived using a
correlation proposed by Barton (1982), see Fig. 4.7. A corresponding normal
penalty was taken as equal to 10Pt, as recommended in the ELFEN manual
(Rockfield, 2007).
72
100
Ks = τ / δ
10
structures with clay gouge
structures in rock
structures in models
active geological faults
1
Secant
shear 0.2
stiffness 0.1
(GPa/m)
0.01
0.001
Approx.
normal stress
(MPa)
0.0001
10mm 1m 100m 10km 1000km
Length of the Structure (subjected to shear)
Fig. 4.7 Derivation of joints shear stiffness using Barton‟s (1982) correlation
4.5.1.2 Modelling Scenarios
Modelling scenarios and corresponding input parameters are summarized in

Table 4.6. Modelling was carried out in two stages. The first stage focused on
evaluation of the simulated caving response. For each modelling scenario
undercutting of 20x20m, 40x40m and 60x60m blocks and their full or partial
extraction was considered. At the second stage, cave progression and
subsidence development were analyzed, considering a 100x100m block.
Flores & Karzulovic (2002) summarized in-situ stress measurement data from a
number of block caving mines and reported median Kmean of 1.2 with standard
deviation of 0.5. For the current modelling, initial stress conditions assumed
vertical/horizontal orientation of principal stresses, a default value of in-situ stress
ratio K of 1 was adopted.
73
Table 4.6 List of principal modelling scenarios and corresponding input parameters
Input parameters
Rock Mass Discontinuities
degrees
Modelling scenarios
Interface cohesion*, cf, MPa
Normal penalty, Pn, GPa/m

Fracture energy*, Gf Jm-2
MPa
Tangential penalty, Pt,

Deformation modulus,
f,
Dilation*, ψ degrees
t,
degrees
Cohesion, ci, MPa
Density*, ρ kgm-3
Interface friction*,
Tensile strength,
Poisson‟s ratio*,
I
Friction,
E, GPa
GPa/m
RMR80 56 0.33 0.4 45
RMR70 32 0.33 0.35 40
RMR60 18 0.31 0.3 35
RMR50 10 0.29 0.25 30
13
Q 47.5 37 52 45
(70% c.o.**)
(equiv. to
RMR 79) 4.3
37 52 45
(90% c.o.)
4.3
Q 17.5 26 15.1 45
(70% c.o.) 0.25 2600 5 60 0 35 2 0.2
(equiv. to
RMR 70) 1.25
26 15.1 45
(90% c.o.)
1.18
Q 6.2 18 4.7 45
(70% c.o.)
(equiv. to
RMR 60)
0.46
18 4.7 45
(90% c.o.)
0.4
Q 0.67 12 1.6 45
(70% c.o.)
(equiv. to
RMR 50)
0.13
12 1.6 45
(90% c.o.)
* assumed based on estimates for Carbonatite ore at Palabora mine (Rockfield, 2003)
** assuming maximum tensile strength at zero cut off
4.5.2 Constraining Criteria
In order to evaluate if behaviour simulated using the varied strength and

deformability properties sets is realistic, it should be constrained against
observed response. The constraining criteria adopted for the current analysis
are given in Fig. 4.8.
74
The first stage of block caving is undercutting the ore block. The ore must be
undercut beyond a certain hydraulic radius (area/perimeter) to provide continuous
caving. Laubscher‟s caveability chart (Diering & Laubscher, 1987) is perhaps the
most widely accepted measure of rock mass caving susceptibility. Based on a
number of cave mining case studies Laubscher delineated zones of varying rock
mass response to caving depending on rock mass rating and dimensions of the
undercut, as shown in Fig. 4.8(a). This criterion was employed as one of the three
approximate constraints for evaluating how realistic the simulation of rock mass
caveability is.
Block undercutting is followed by continuous ore extraction, which triggers

propagation of the caving front upwards. To date, cave propagation behaviour is
far from being fully understood. The available descriptions of caving
mechanisms are qualitative and largely based on tentative interpretations of the
microseismic data. A conceptual model of cave propagation by Duplancic &
Brady (1999), shown in Fig. 4.8(b), is the most widely cited. It was developed
based on the analysis of seismic monitoring data at early stages of caving at
Northparkes E26 block cave. The conceptual model consists of four main
regions: (1) pseudo-continuous domain, where the rock mass is mainly
undisturbed; (2) seismogenic zone, where seismic events occur due to slip on
joints and brittle failure of rock, caused by changing stress conditions related to
the advancing caving front; (3) zone of discontinuous deformation, where the
rock mass provides minimal support to the overlying strata and rock
disintegration occurs; and (4) caved zone and air gap, where the rock blocks that
have fallen from the cave come to rest, and where an air gap may form if there is
a difference between extraction and caving rates (Fig. 4.5b). The Duplancic &
Brady concept was adopted in this study as a second constraint for assessment
of simulated caving behaviour.
Continuous ore extraction eventually leads to the formation of a surface

subsidence zone. A cave mining experience-based rule of thumb (see Section
2.5.1) was utilized as the third subsidence constraining criterion. It was assumed
75
that major surface deformations should not exceed 45 degree limiting angles
drawn from the bottom of the undercut, as shown in Fig. 4.8(c).
Simulated behaviour that tentatively met all three constraining criteria was
deemed acceptable and corresponding equivalent continuum properties
adequately representative.
(a) (b)
100
stable zone
transition
90 zone
80 caving
zone
Modified Rock Mass Rating (MRMR)
70
60
50
40
30
20
10
0
0 10 20 30 40 50 60 70 80
Hydraulic Radius, m
(c)
45o 45o
Fig. 4.8 Constraints for ELFEN modelling of block caving induced subsidence
(a) Laubscher‟s caveability chart; (b) Conceptual model of cave propagation (Duplancic
& Brady, 1999); (c) Subsidence limits
4.5.3 Modelling Results
Modelling results based on RMR and Q property sets are presented in Figs. 4.9
to 4.16. Figs. 4.9 to 4.11 illustrate simulated rock mass response to caving for
varying undercut dimensions. Fig. 4.12 summarizes simulated caveability
76
response and compares it with Laubscher‟s empirical caveability chart. In this
figure rock mass caveability was evaluated based on the following assumptions:
stable conditions were assumed if the cave could not propagate up to the
block centre;
transitional conditions were assumed if the cave propagated more than a
half, but less than the full block height;
caving conditions were assumed if the cave front exceeds block height.
MRMR estimates were based on the assumption of MRMR 0.9RMR
(Flores & Karzulovic, 2002).
Figs. 4.13 to 4.15 show cave development progression at early stages of ore
extraction. Figs. 4.16 and 4.17 show simulated surface subsidence.
4.5.3.1 Simulations with RMR Properties
As illustrated in Figs. 4.9 and 4.12(a), simulations based on RMR properties

showed largely realistic response to caving, from high caveability at RMR=50 to
nearly stable conditions at RMR=80. Modelling results are in overall agreement
with Laubscher‟s caveability chart, although it appears that simulations based on
RMR properties tend to overestimate caveability of the rock mass. Simulated cave
development behaviour, shown in Fig. 4.13, differs quite significantly from the
Duplancic & Brady concept; an overly soft response is observed, the crown pillar
fractured very rapidly, and a clear distinction between the stable and seismogenic
zones cannot be made. The ellipsoid failure pattern formed during ore extraction
indicates mobilization of significant rock volumes even at early stages of ore
extraction, regardless of the rock mass rating. It should be noted that such failure
is characteristic of soft rock conditions. Further simulation of caving progression
up to subsidence development showed that RMR based properties tend to grossly
overestimate the amount of damage and extent of rock mass deformations caused
by caving (see Fig. 4.16), the only exception is RMR=80.
Overall, none of the considered RMR property sets produced simulated

behaviour agreeing with all three constraining criteria. Excessive fracturing
77
associated with the use of RMR properties can be attributed to the very low
estimates of rock mass tensile strength given by this system. Interestingly the
significant decrease in rock mass deformability, with increasing RMR rating,
restricted the rock mass caveability.
4.5.3.2 Simulations with Q Properties
Two Q based property sets were considered using tensile strength cut-off
assumptions of 90% and 70%.
As follows from Figs. 4.10 and 4.12(b) simulations based on 90% cut-off
assumption tend to underestimate caveability at smaller undercut dimensions
(Hr=20), but overall produce reasonable fits with the empirical data at larger
undercuts (Hr=30), exhibiting a realistic trend of decreasing caveability with
higher rating. According to Fig. 4.10, simulation of cave propagation based on
these properties set does not show a reasonable agreement with the Duplancic &
Brady concept, with the exception of the Q 17.5 case (equivalent RMR70). For
this property set gradual cave propagation was observed. Simulations with lower
Q properties produced rapid, rather sporadic caving. For higher Q values caving
could not be simulated due to apparent cave arrest. Subsidence simulation
based on Q properties with 90% cut-off is shown in Fig. 4.17. A Q of 0.67
(equivalent RMR=50) and a Q of 6.2 (equivalent RMR=60) resulted in excessive
fracturing and overestimated the extent of surface deformations. The Q=17.5
property set produced reasonable subsidence extent predictions.
Simulations based on a 70% cut-off assumption underestimated the caving

response for the studied undercut dimensions, see Figs. 4.11 and 4.12(c). In
terms of cave propagation a Q of 0.67 property set resulted in excessively rapid
caving. A Q of 6.2 showed reasonable cave propagation, and a Q of 17.5 and
47.5 could not be analysed due to limited caving. Subsidence modelling based
on a Q of 0.67 produced an unrealistically large zone of subsidence
deformations, whereas subsidence predictions based on a Q of 6.2 were
generally reasonable, Fig. 4.17.
78
Overall, only two selected Q property sets (Q=17.5 with 90% cut-off and Q=6.2
with 70% cut off) showed a reasonable agreement with all three constraining
criteria. Interestingly, for the Q property sets, an excessively soft response was
observed for lower Q ratings and an excessively stiff response was observed for
higher ratings. Middle range values produced apparently reasonable behaviour.
This can be explained by high variability of tensile strength estimates which varied
from 0.13 MPa (Q=0.67 90% cutoff) to 13 MPa (Q=47.5 70% cutoff). Simulations
exhibiting reasonable behaviour had a tensile strength of about 1.2 MPa.
79
Hr10 Hr20 Hr30
RMR 50
RMR 60
RMR 70
RMR 80
Fig. 4.9 Caving response with RMR equivalent continuum properties
80
Hr10 Hr20 Hr30
Q 0.67
90% c.o.
Q 6.2
90% c.o.
Q 17.5
90% c.o.
assumed stable
Q 47.5
90% c.o.
assumed stable assumed stable
Fig. 4.10 Caving response with Q equivalent continuum properties (90% tensile c.o.)
81
Hr10 Hr20 Hr30
Q 0.67
70% c.o.
Q 6.2
70% c.o.
Q 17.5
70% c.o.
assumed stable
Q 47.5 assumed stable assumed stable assumed stable

70% c.o
Fig. 4.11 Caving response with Q equivalent continuum properties (70% tensile c.o.)
82
(a)
100
stable zone Legend:
transition
90 zone stable
80 caving transitional
zone
caving
70
60
50
40
30
20
10
0
0 10 20 30 40 50 60 70 80
Hydraulic Radius, m
(b) (c)
100 100
stable zone stable zone
transition transition
90 zone 90 zone
80 caving 80 caving
zone zone
70 70
60 60
50 50
40 40
30 30
20 20
10 10
0 0
0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80
Hydraulic Radius, m Hydraulic Radius, m
Fig. 4.12 Caveability assessment using Laubscher‟s chart for studied equivalent continuum
properties
(a) RMR; (b) Q - 90% tensile cut-off; (c) Q - 70% tensile cut-off
83
RMR 50 not shown due to excessive fracturing
end of undercutting 4% ore extraction 8% ore extraction 12% ore extraction
RMR 60
RMR 70
RMR 80
Fig. 4.13 Simulation of cave development progression for RMR equivalent continuum properties.
84
Q 0.67 (90% c.o.)
Q 6.2 (90% c.o.)
Q 17.5 (90% c.o.)
Q 47.5 not shown due to limited caving

Fig. 4.14 Simulation of cave development progression for Q (90% tensile cut-off) equivalent continuum properties.
85
Q 0.67 (70% c.o.)
Q 6.2 (70% c.o.)
Q 17.5 (70% c.o.)

Fig. 4.15 Simulation of cave development progression for Q (70% tensile cut-off) equivalent continuum properties.
86
RMR 50 not shown due to excessive fracturing
RMR 60
RMR 70
RMR 80
Fig. 4.16 Simulation of subsidence development with RMR equivalent continuum properties.
87
90% tensile cut-off 70% tensile cut-off
Q 0.67
Q 6.2
Q 17.5

Fig. 4.17 Simulation of subsidence development with Q equivalent continuum properties.
88
4.5.4 Conclusions
The following conclusions can be drawn from the conducted modelling study:
neither the RMR nor Q rock mass classification systems can be relied
upon as a robust source of rock mass properties for hybrid FEM/DEM
modelling;
very low estimates of rock mass cohesion and tensile strength derived
using RMR system lead to overestimated susceptibility of simulated rock
mass material to tensile failure, making use of RMR properties
unacceptable for modelling of caving mechanisms;
among all considered Q property sets the properties for the midrange Q
ratings, with an assumed tensile strength of about 1.2 MPa, provide the
most realistic representation of rock mass caveability, cave development
progression and subsidence response;
it appears that the Q properties for the ratings equivalent to RMR 60 to 70

offer generally adequate estimates of deformation modulus, cohesion and
friction, and also, give flexibility in regards to the assumption of the tensile
strength, thus leaving room for response calibration.
Overall, the conducted analysis clearly shows a need for further research into
estimation of rock mass equivalent continuum properties. In the author‟s opinion,
future progress in this area can be achieved through use of the synthetic rock
mass modelling methodology (Pierce et al., 2007) employing PFC and DFN
(discussed in Section 4.3.3). Further work is required to ensure that it can
produce realistic material properties for different rock mass scales and evaluate
its sensitivity to varying stochastic realizations.
4.6 Summary
Among the various methodologies for estimating rock mass equivalent continuum
properties, the use of rock mass classification systems is the most common. The
89
analysis carried out in this chapter evaluated the properties derived from the
RMR, Q and GSI rock mass classification systems. It was found that strength
and deformability characteristics derived using different systems for the same
rating values a show general lack of consistency. Moreover, none of the studied
rock mass classification based correlations offered adequate estimates of all
equivalent continuum rock mass properties. In some instances property values
appear to be excessively conservative. The analysis highlighted the need for
careful examination of rock mass classification properties outputted prior to
acceptance in engineering calculations. A numerical modelling study was carried
out to investigate the applicability of the RMR and the Q based properties
estimates in combined FEM/DEM-DFN modelling of surface subsidence
associated with block caving mining. A novel subsidence modelling approach
was adopted and using cave mining experience-based constraining criteria, such
as caveability, cave propagation and major subsidence damage limits, applicability
of varying property sets was evaluated. Generally it was found that neither RMR
nor Q systems can be relied upon as a robust source of properties for subsidence
modelling. Property estimates for middlerange Q values produced the most
realistic simulated behaviour with respect to the considered constraining criteria.
90
CHAPTER 5: CONCEPTUAL MODELLING STUDY OF THE
FACTORS CONTROLLING BLOCK CAVING SUBSIDENCE
DEVELOPMENT
5.1 Introduction
In a complex block caving mining environment, subsidence development is a
result of an interplay between several governing factors. Mining experience
suggests that these factors include geological structure (jointing and faults), rock
mass strength, in-situ stress level, mining depth, volume of extracted material and
varying lithological domains. A survey of the literature shows that publications are
limited to a general, qualitative rather than quantitative, description of the influence
of geological structures on the observed subsidence, see Section 2.4. Such
qualitative observations are useful for initial subsidence analysis, however they
require further validation. With regards to other factors, literature sources highlight
their importance but provide only limited further description. Understandably,
discerning the effect of individual factors from field observations is challenging. To
the author‟s knowledge, until now, no comprehensive attempt has been
undertaken to address the fundamental understanding of block caving subsidence
phenomena. This chapter adopts the novel modelling methodology for subsidence
analysis, outlined in Chapter 4, and through a series of conceptual numerical
experiments investigates subsidence development mechanisms and the relative
significance of the factors governing subsidence development.
5.2 Model Setup and Analysis Strategy

The model setup adopted is as described in Section 4.5.1.1. Modelling input
parameters are given in Table 5.1. It should be noted that the rock mass
properties were based on those derived from the Q rock mass classification
system, corresponding to a Q rating of 6.2. Results of the analysis carried out in
91
Section 4.5 illustrated that this property set, with an assumed tensile cut-off of
70%, produced reasonably realistic caving and subsidence development
responses. These properties were further calibrated (through adjustment of
tensile strength) to achieve a better correlation with the constraining criteria. A
tensile cut-off of 75%, corresponding to 1MPa tensile strength, was adopted for
the current modeling. The property set adopted is given in Table 5.1. The
caving behaviour of a rock mass with the assumed properties corresponds to
MRMR ~ 55-60 (i.e. within a typical block caving range of MRMR 30 to 70).
Table 5.1 Modelling input parameters

Parameter Unit Value Parameter Unit Value
Rock mass Discontinuities
Young‟s Modulus, E GPa 18 Fracture cohesion, cf MPa 0
Poisson‟s ratio, 0.25 Fracture friction, f degrees 35
-3
Density, ρ kgm 2600 Normal penalty, Pn GPa/m 2
Tensile strength, t MPa 1 Tangential penalty, Pt GPa/m 0.2
Fracture energy, Gf Jm-2 60
Cohesion, ci MPa 4.7 Stress level
Friction, i degrees 45 In-situ stress ratio, K 1
Dilation, ψ degrees 5
The conceptual study modelling presented in this chapter involved more than 500
days of continuous run time on a 3.4GHz, 2GB RAM single processor PC, with an
average run time of individual simulations of about 16 days. Due to limitations of
computational processing time the number of simulations was limited to 36.
Emphasis in the conceptual study modelling has been given to the analysis of the
effect of geological discontinuities, widely recognized as a primary control on block
caving induced surface subsidence.
92
5.3 Influence of Jointing
5.3.1 Effect of Joint Orientation
5.3.1.1 Model Description
The effect of joint orientation was evaluated through comparison of simulations

assuming five different combinations, as summarized in Table 5.2.
Table 5.2 Modelling scenarios for analysis of the effect of joint orientation
Scenario Number Joint sets Description Figure

of sets dips, °
Orthogonal sets,
Base Case Two sets 90/0 5.1(a)
vertical/horizontal
Orthogonal sets,
J1 Two sets 80/10 5.1(b)
sub-vertical/sub-horizontal
Orthogonal sets, steeply
J2 Two sets 70/20 5.1(c)
dipping/gently dipping
J3 Two sets 70/0 5.1(d)
dipping/horizontal
J4 Three sets 70/20/90 5.1(e)
dipping/gently dipping/vertical
93
-300 -250 -200 -150 -100 -50 0 50 100 150 200 250 300
(a)
BC W E
0°
90° ore
block
(b)
J1
10°
80°
(c)
J2
20°
70°
(d)
J3
0°
70°
(e)
J4
0°
20°
70°
Fig. 5.1 Assumed fracture orientations for BC (a), J1 (b), J2 (c), J3 (d) and J4 (e) models
94
The Base Case, J1 and J2 models are meant to illustrate how varying orientation
of the same joint pattern affects subsidence development mechanisms and final
subsidence footprint. Models J3 and J4 are based on the J2 model and are used
to evaluate the significance of the change of orientation of the sub-vertical set and
the presence of an additional vertical set, respectively. The Base Case model was
selected as a reference as a combination of vertical and horizontal joints represent
conditions “ideal” for caving.
5.3.1.2 Subsidence Mechanisms
Figs. 5.2 to 5.4 present the mechanism of surface deformation development for
Base Case, J1, J2, J3 and J4 models at 35, 50 and 60% caved ore extraction.
All models show a common subsidence crater formation mechanism which can be
summarized as:
caving/unloading induced fracturing coupled with continuous ore extraction

creates favourable kinematic conditions for detachment of major near
surface rock mass segments adjacent to the caving front;
the detached rock mass segments collapse into the cave through rotational
and/or translation failure; surface expressions of such failure involves
formation and growth of multiple tensile cracks which eventually disappear
as the rock mass disintegrates;
the extreme limits of these detaching segments manifested at the surface

represent the initial subsidence crater walls;
continuous removal of the ore leads to lowering of the rubblized rock within
the crater reducing lateral support to the crater walls promoting further
lateral growth of the subsidence crater through rotational and/or
translational failures of the crater wall segments into the cave.
The described mechanism of subsidence deformation development is in general

agreement with that suggested by Abel & Lee (1980) (see Section 2.2).
95
35% ore extraction 50% ore extraction 60% ore extraction
0°
(a) BC
50m
0°
90°
5°
(b) J1
10°
80°
Legend: rotational failure; translational failure; active rock mass movement; developing rock mass failure
9° deviation of the center of surface depression with respect to the block centre vertical axis
Fig. 5.2 Subsidence crater formation for Base Case (a) and J1(b) models
96
9°
(a) J2
20°
70°
4°
(b) J3
0°
70°
Fig. 5.3 Subsidence crater formation for J2 (a) and J3 (b) models
97
1°
J4
0°
20°
70°
Fig. 5.4 Subsidence crater formation for J4 model
Base Case J2
5% ore extraction
Fig. 5.5 Variation of vertical stress (Pa) contours with caving at 5% ore extraction for Base Case and J2 models
98
It can be inferred from Figs. 5.2 to 5.4 that the direction of cave propagation
toward the surface, the location of the cave breakthrough and the mechanisms of
near surface rock mass failure are all strongly controlled by joint orientation. Fig.
5.5 shows the variation of the vertical stress contours at an early stage of ore
extraction for the Base Case and J2 models. It can be inferred from this figure
that the orientation of the sub-vertical/steeply dipping joint set predetermines the
direction of caving induced rock mass unloading and thus the direction of cave
propagation. Comparing the centres of surface depression at 35% ore extraction
for the Base Case (Fig. 5.2(a)), J1 (Fig. 5.2(b)) and J2 (Fig. 5.3(a)) models, it can
be seen that a rotation of the joint pattern skews the direction of cave
propagation away from the block centre vertical axis and cave propagation is
largely controlled by the steeply inclined set. Rotation of the joint pattern by 10°
moves the centre of surface depression by about 4°, reaching 9° for the J2
model. This trend however may be altered depending on the orientation of the
gently dipping set. Comparing models J2 (Fig. 5.3(a)) and J3 (Fig. 5.3(b)) a
change of inclination of the sub-horizontal set from 20° dip to horizontal shifts the
centre of surface depression closer to the block centre vertical axis by 5°, i.e.
more than 50%. Moreover, comparing models J2 (Fig. 5.3(a)) and J4 (Fig. 5.4) it
is evident that the presence of an additional well defined vertical joint set reduces
the significance of the steeply dipping set, so that the centre of initial surface
depression is nearly aligned with the block centre vertical axis.
Joint orientation controls not only the cave propagation direction but also plays a
significant role in the manner the rock mass is mobilized by caving. In order to
characterize rock mass mobilization development in Figs. 5.2 to 5.4, zones of
active rock mass movement and developing rock mass failure are delineated.
Within the former zone the rock mass is fully disintegrated and the latter zone
indicates the damaged and potentially unstable rock mass. Figs. 5.2(a) and 5.4
show that the effect of the vertical joint set is relatively limited and the extent of
the rock mass mobilized during initial stages of caving and ore extraction is
largely symmetrical, with respect to the ore block centre axis. As evident from
Figs. 5.2(b) and 5.3, simulations with sub-vertical and steeply dipping sets result
99
in a larger extent of mobilized rock mass. Failure asymmetry with respect to the
block centre vertical axis can be clearly observed, a larger failure is evident in a
zone where sub-vertical/steeply dipping joints are inclined towards the cave
(west of the block centre vertical axis) and more limited failure is observed where
these joints dip towards the cave (east of the block centre vertical axis). This
asymmetry can be attributed to the varying mechanisms of failure of the rock
mass which is governed by the inclination of the vertical/steeply dipping joints.
West of the block centre vertical axis, inclination of the joint sets favours rock
mass failure through flexural and block-flexural toppling, coupled with inclined
cave propagation this creates suitable kinematic conditions for toppling of
massive rock mass segments. In the eastwards direction, a sub-vertical/steeply
dipping joint set creates favourable conditions for sliding and, in combination with
an orthogonal joint set promotes slide toe toppling. Such a failure does not
appear to exceed the dip angle of the sub-vertical joint set, hence limiting the
extent of the mobilized rock mass.
5.3.1.3 Subsidence Topography
Final subsidence deformation and resultant surface profiles at 100% ore

extraction for the Base Case, J1, J2, J3 and J4 models are shown in Figs. 5.6
and 5.7, respectively. It can be clearly seen from these figures that the rock
mass deformation and the surface depression formed due to caving vary quite
significantly depending on the assumed joint orientation. Rotation of the joint
pattern shifts the centre of the surface depression, positioned at the block
centre vertical axis for Base Case model, in a direction opposite to that of
surface asymmetry (i.e. eastwards) and also results in a shallower subsidence
crater. Rotation of the jointing pattern by 10° results in a decrease of the
maximum depth of the crater by about 10%. The maximum crater depth was
observed for the model with vertical/horizontal joint sets (Base Case) and the
minimum for the simulation with steeply dipping/horizontal joint sets (J3). Models
with different joint orientation exhibit varying subsidence crater topography.
100
-300 -250 -200 -150 -100 -50 0 50 100 150 200 250 300
(a)
BC Legend:
0° 71° angle
of fracture
initiation
90° 10cm displ. contours
3
vertical
MRV = 28114m 76° horizontal
71°
AI = 0.93
(b)
J1
10°
80°
3
MRV = 30762m
AI = 0.96 70° 73°
(c)
J2
20°
70°
3
MRV = 34990m
AI = 0.72 53° 74°
(d)
J3
0°
70°
3
MRV = 35250m
AI = 0.82 61° 74°
(e)
J4
0°
20°
70°
3
MRV = 30836m
AI = 0.82 59° 72°
Fig. 5.6 Subsidence at 100% ore extraction for BC (a), J1 (b), J2 (c), J3 (d) & J4 (e) models
101
0
-10 Lowest point
Vertical displacements, m
coordinates, m
-20
Base case 0, -55
-30
-40 J1 10, -50
-50 J2 9.4, -44.5
-60 J3 28.6, -41
-70 J4 9.4, -49.6
-80
-350 -250 -150 -50 50 150 250 350
Distance from block centre, m
Fig. 5.7 Surface profiles at the end of ore extraction for BC, J1, J2, J3 and J4 models
For the Base Case model a distinct, nearly symmetrical stepped V-shaped crater
is formed. In contrast, for simulations with inclined joints (J1 to J3) the
subsidence crater is asymmetrical. In the direction of maximum asymmetry (i.e.
westwards) the surface subsides without forming major steps, aside from the
crater wall. It is interesting to note that the addition of the vertical joint set in
model J4 reduced crater asymmetry and resulted in a stepped crater topography.
5.3.1.4 Characterization of Major Displacements
In order to quantify the extent of major surface subsidence deformation a 10cm

displacement threshold is adopted. It is assumed that this threshold limits the
zone of major surface disturbance. Fig. 5.6 shows contours of 10cm vertical and
horizontal displacements at 100% ore extraction for Base Case, J1, J2, J3 and
J4 models. Using these contours we can evaluate the mobilized rock mass
volume (MRV), as indicated in the figure. The maximum span of the major
surface displacement induced by the caving is delineated using angular limits, or
angles of fracture initiation according to subsidence characterization terminology
proposed by van As (2003). Comparing angles limiting major surface
deformations for the presented models, it can be seen that in an eastward
direction from the block centre vertical axis all models show consistently steep
limiting angles ranging from 72° to 76°. In a westward direction, dissimilarity in
the limiting angles between the different models is apparent. The lowest
102
minimum angle of fracture initiation of 53° is observed for model J2 (Fig. 5.6(c))
and the highest angle of 71° for the Base Case model (Fig. 5.6(a)), i.e. rotation of
the joint pattern by 20° results in the increase of subsidence limits in the sub-
vertical set dip direction by about 20%. Interestingly, the initially asymmetrical
subsidence development for model J1 with a 10° joint pattern rotation (see Fig.
5.6(b)) eventually becomes more symmetrical, and only a minor increase in the
limiting angle is observed. It appears that the 80° dip of the sub-vertical set is
insufficient to cause extensive flexural toppling. Model J4 (Fig. 5.6(e)) yields the
second lowest angle of fracture initiation of 61° which is about 10% lower than for
J2 model. This indicates that the vertical joint set, by providing additional planes
of weakness, limits the extent of rock mass mobilized by the caving. It is
interesting to note that initial subsidence development for the J4 model was
nearly symmetrical, as shown in Fig. 5.4. Subsidence asymmetry in the
westwards direction began to develop as the constraining effect of the rubblized
rock was diminished due to continuous ore extraction, allowing block toppling
and sliding of the crater wall segments along the gently dipping joint set.
Comparing models J2 and J3 it can be concluded that decreasing the dip of the
gently dipping set by 20° increases the limiting angle by 10° or about 20%. Such
an influence can tentatively be explained by reduction of the potential for rotation
and sliding towards the cave along the gently dipping joint set.
To characterize subsidence asymmetry a block cave subsidence parameter,

asymmetry index, AI, is introduced. This index is defined as the ratio of the
minimum to maximum angles delineating the extent of major (≥10cm) surface
displacements, as shown in Fig. 5.6. Perfect symmetry corresponds to an
asymmetry index of 1.
In addition to using the limiting angles, shown in Fig. 5.6, the zone of major
surface deformation can be further characterized by its total extent and relative
significance with respect to the vertical axis of the block, as shown in Fig. 5.8.
According to Fig. 5.8(a) and (b), changes in the joint set orientation causes an
increase in the extent of the total major surface deformations by up to 30% and
41% for major vertical and horizontal surface displacements, respectively. For all
103
models the total extent of the major surface horizontal deformation is consistently
larger than or equal to the extent of vertical displacements. According to Fig.
5.8(c) and (d) depending on the joint orientation assumption:
west of the block centre vertical axis the extent of major surface
deformations increases up to a maximum of about 40% and 70% for
vertical and horizontal displacements;
east of the block centre vertical axis only a moderate increase of up to
20% for both vertical and horizontal displacements is observed.
(a) (b)
350 350 350 350
normalized by Base Case, %

308

Total extent of 10cm vertical
Total extent of 10cm horiz.

surface displacements, m

290
300 300 300 300
surface displacements
268 269 269
234 245 235
250 250 250 218 250
207
200 200 200 200
150 150 150 150
141%
133%
129%
130%
100 100 100 100
123%
118%
113%
108%
100%
100%
50 50 50 50
0 0 0 0
BC J1 J2 J3 J4 BC J1 J2 J3 J4
(c) (d)
Extent of 10cm surface vertical dispacements in Extent of 10cm surface horizontal displacements in
relation to block centre, m relation to block centre, m
-250 -200 -150 -100 -50 0 50 100 150 200 250 -250 -200 -150 -100 -50 0 50 100 150 200 250
-112 100% BC -118 100% BC

-123 110% J1 -123 104% J1
-161 144% J2 -201 170% J2
-161 144% J3 -161 136% J3
-132 132% J4 -173 147% J4
100% 95 BC 100% 100
BC
J1 117% 111 J1 116% 112
J2 113% 107 J2 107% 107
J3 114% 108 J3 108% 108
J4 119% 113 J4 117% 117
-300 -200 -100 0 100 200 300 -300 -200 -100 0 100 200 300
Extent of 10cm surface vertical displacements in Extent of 10cm surface horizontal displacements in
relation to block centre, normalized by Base Case, % relation to block centre, normalized by Base Case, %
Fig. 5.8 Subsidence characterization for Base Case, J1, J2, J3 and J4 models
Total extent of 10cm vertical (a) and horiz. (b) surface displacements; extent of 10cm
surface vertical (c) and horiz. (d) displacements in relation to centre axis of the block, in m
Evolution of the zones of major (≥10cm) surface deformation with continuous ore
extraction and the rate of their growth west of the block centre vertical axis for
Base Case, J1, J2, J3 and J4 models is shown in Figs. 5.9 and 5.10,
respectively. From these figures it can be inferred that for all considered models
major subsidence deformation develop in a relatively rapid manner reflecting quick
mobilization of the massive rock mass segments. Fig. 5.10(a) shows that for the
majority of the models, with the exception of model J4, about 90% of the
104
maximum vertical deformations is achieved by 50% ore extraction. Model J4
exhibits a more subtle trend in vertical deformation development which can be
attributed to the previously discussed gradual block toppling failure mechanism.
Horizontal deformation development trends are presented in Fig. 5.10(b), which
indicates that for simulations which involve flexural toppling failure, models J1,
J2, and J3, horizontal displacements generally develop at a rate of up to 80%
greater than the vertical displacements.
(a) BC (b) J1
100 100
90 90
80 80
Ore extraction, %
Ore extraction, %
70 70
60 60
50 50
40 40
30 30
YY YY
20 20
10 XX 10 XX
0 0
-250 -200 -150 -100 -50 0 50 100 150 200 250 -250 -200 -150 -100 -50 0 50 100 150 200 250
Extent of 10cm surface deformations, m Extent of 10cm surface deformations, m
(c) J2 (d) J3
100 100
90 90
80 80
Ore extraction, %
Ore extraction, %
70 70
60 60
50 50
40 40
30 30
YY YY
20 20
10 XX 10 XX
0 0
-250 -200 -150 -100 -50 0 50 100 150 200 250 -250 -200 -150 -100 -50 0 50 100 150 200 250
Extent of 10cm surface deformations, m Extent of 10cm surface deformations, m
(e) J4
100
90
80
Ore extraction, %
70
60
50
40
30
YY
20
10 XX
0
-250 -200 -150 -100 -50 0 50 100 150 200 250
Extent of 10cm surface deformations, m
Fig. 5.9 Evolution of zone of major (≥10cm) vertical (YY) and horizontal (XX) surface
deformations with continuous ore extraction for Base Case (a), J1 (b), J2 (c), J3 (d) and
J4 (e) models
105
(a) (b)
120 120
extent of vertical 10cm surface
extent of horizontal 10cm surface

100 100
displacements, % 80 80
displacements, %
60 60
BC_YY BC_XX
40 J1_YY 40 J1_XX
J2_YY J2_XX
20 J3_YY 20 J3_XX
J4_YY J4_XX
0 0
0 20 40 60 80 100 120 0 20 40 60 80 100 120
Ore extraction, % Ore extraction, %
Fig. 5.10 Rate of growth of 10cm surface displacement zone west of the block centre
vertical axis with continuous ore extraction for Base Case, J1, J2, J3 and J4 models
(a) vertical displacements, (b) horizontal displacements
5.3.1.5 Characterization of Far-Field Displacements
When considering the location of mine infrastructure it is important to appreciate the

magnitude of surface displacements at specific distances from the area of imminent
failure (caving boundary and its immediate vicinity). Fig. 5.11 shows total vertical and
horizontal displacements at the end of ore extraction at 300, 250, 200 and 150m
distances from the block centre for Base Case, J1, J2, J3 and J4 models.

(a) -300 -250 -200 -150 150 200 250 300
0
J3
J2
-0.05
J4
-0.1
-0.15
-0.2
-0.25
J2
J3
-0.3 -0.38 -2.1

(b) 0.9 3.8
Horizontal displacements, m
0.3
J3
J2
0.25
0.2
J4
0.15
0.1
J2
J3
J1
J3
J4
BC
0.05
J2
J2
BC
J1
BC
J4
J3
J1
0
-300 -250 -200 -150 150 200 250 300
Fig. 5.11 Total vertical (a) and horizontal (b) surface displacements at the end of ore
extraction at different distances from block centre for Base Case, J1, J2, J3 and J4 models
106
According to this figure the least amount of surface displacements is exhibited by
the Base Case model (90°/0°), so that only minor horizontal displacements of
about 1cm are observed 100m from the caving boundaries (150m from the block
centre vertical axis). The largest magnitude of displacement are observed for the
J2 (70°/20°) and J3 (70°/0°) models, where 1cm horizontal displacements are
noted as far as 200m west of the caving boundaries. Far-field surface
displacements generally mirror the trends observed for the major surface
deformations, showing strong asymmetry in the dip direction of the sub-
vertical/gently dipping joint sets. Apparently, the magnitude of accumulated
surface displacement as well as its extent will depend on the mechanism of the
rock mass failure induced by caving, which, as discussed earlier is strongly
controlled by the joint orientation. Comparing vertical (Fig. 5.11(a)) and
horizontal (Fig. 5.11(b)) far-field displacements, it can be seen that in the studied
models there is a clear trend of higher far-field horizontal displacements. This
trend is in agreement with the measurements of caving induced surficial
displacements at Lakeshore mine, as reported by Panek (1984).
5.3.1.6 Conclusions
Based on the results of ELFEN modelling, the following can be concluded with
respect to the effect of joint orientation on caving induced surface subsidence
development and the resultant deformation footprint:
1. Well defined, persistent, vertical to steeply dipping joints govern the

direction of cave propagation and the mechanism of near surface rock
mass mobilization. The shallower the dip of these joints the more inclined
from vertical the cave propagation direction is and the more asymmetrical
the surface deformation is with respect to the block centre vertical axis. In
cases where multiple well defined and persistent steeply dipping sets are
present, the steepest set will generally have the predominant influence.
2. Modelling results in part confirm observations of Crane (1929) and Wilson

(1958) who noted that in the absence of major geological discontinuities
subsidence deformations are controlled by the joint dip.
107
3. Major subsidence asymmetry is observed in the dip direction of the sub-
vertical/steeply dipping set. Where joints are inclined towards the cave, the
rock mass fails through a combination of block-flexural and block toppling
and the detachment and sliding of major rock segments, which is in
agreement with the conceptual caving failure model proposed by Hoek
(1974). Where a sub-vertical joint set is dipping into the cave, the surface
deformation direction is controlled by the dip of the sub-vertical joint set. In
this case the rock mass fails predominantly through block toppling and
sliding along the sub-vertical joints.
4. The orientation of well defined, persistent, gently dipping joints influences

the extent of the rock mass mobilized by the failure and the degree of
subsidence asymmetry.
5.3.2 Effect of Joint Persistence
The effect of joint persistence on block caving induced surface subsidence

development was evaluated using the J2 (70°/20°) model, which yielded the
maximum subsidence deformation. Two scenarios were considered:
I. J5, where 25% higher persistence was assumed for all joints
II. J6, where 25% lower persistence was assumed for all joints
A series of Excel based macros were developed to process the FracMan output
file and evenly adjust the DFN based fracture length according to a specified
percentage value.
Figs. 5.12(a) and (b) illustrate surface subsidence deformations at 100% ore
extraction for the J5 and J6 models, respectively. Comparing these models
with model J2 (Fig. 5.6(c)) it can clearly be seen that an increase in joint
persistence by 25% results in a larger extent of the rock mass mobilized by
caving in the dip direction of the steeply dipping set. The minimum limiting
angle delineating major (≥10cm) surface deformations decreases from 53° to
47° degrees. A reduction in joint persistence by 25% limits the extent of rock
mass mobilization and an increase in the limiting angle from 53° to 71°.
108
-300 -250 -200 -150 -100 -50 0 50 100 150 200 250 300
(a)
J5
Legend:
20° angle
47°
of fracture
70° initiation
3 10cm displ. contours
MRV = 42076m
vertical
AI = 0.64
47° 73° horizontal
persistence (+)25%
(b)
J6
20°
70°
3
MRV = 32358m
AI = 0.93
persistence (-)25% 71° 76°
Fig. 5.12 Subsidence at 100% ore extraction for J5 (a) and J6 (b) models
Comparing the extent of major rock mass deformation for models J2, J5 and J6,
Fig. 5.13, shows a well defined trend of an increasing zone of influence with
increasing joint persistence. A 50% change in joint persistence between
models J6 and J5 results in an increase in the total extent of the major vertical
and horizontal surface displacements of about 50% and 60% respectively. This
may be attributed to the varying degree of blockiness of the rock mass
depending on joint persistence. Higher joint persistence contributes to a higher
rock mass blockiness, and given a favourable joint orientation leads to
increased kinematic susceptibility of the rock mass to toppling failure. A
decrease in joint persistence reduces the blockiness of the rock mass and its
kinematic susceptibility to failure. This leads to a lower rock mass mobilization
due to the formation of more rock bridges.
109
(a) (b) 350
350 331 350 350 350
308


300 300 300 300
268
normalized by J2, %
normalized by J2, %
250 222 250 250 222 250
200 200 200 200
150 150 150 150
124%
100 100 100 100
114%
100%
100%
83%
72%
50 50 50 50
0 0 0 0
J6 J2 J5 J6 J2 J5
(c) Extent of 10cm surface vertical dispacements in

(d) Extent of 10cm surface horizontal displacements in
-250 -200 -150 -100 -50 0 50 100 150 200 250 -250 -200 -150 -100 -50 0 50 100 150 200 250
-121 75% J6 -121 60% J6

-161 100% J2 -201 100% J2
-219 136% J5 -237 118% J5
J6 94% 101 J6 94% 101
J2 100% 107 100% 107

J2
J5 105% 112 J5 106% 113
-300 -200 -100 0 100 200 300 -300 -200 -100 0 100 200 300
relation to block centre, normalized by J2, % relation to block centre, normalized by J2, %
Fig. 5.13 Subsidence characterization for J2, J5 and J6 models


(a) -300 -250 -200 -150 150 200 250 300
0
J6
J2
-0.05
-0.1
J5
-0.15
-0.2
J5
-0.25
J5
J2
-0.3 -0.38 -0.47
0.4 0.9 1.6

(b)
0.3
J2
J5
J5
0.25
J5
0.2
0.15
J2
0.1
J6
J6
0.05
J2
J6
J6
0
-300 -250 -200 -150 150 200 250 300
extraction at different distances from block centre for J2, J5 and J6 models
110
According to Fig. 5.14, the larger rock mass mobilization associated with higher
joint persistent leads to significantly larger far-field surface displacements west
of the block centre vertical axis.
The results of the modelling show that joint persistence may have a very
significant effect on surface subsidence induced by block caving. It should
however be recognized that the effect of joint persistence will be a function of the
joint inclination. Considering the fact that models assuming vertical/horizontal
joint sets resulted in nearly symmetrical cave development it is unlikely the higher
joint persistence will significantly affect subsidence development in such settings.
Conversely steeply to gently dipping highly persistent joint sets dipping into the
caving front may create favourable conditions for translational and rotational
failure, reducing considerably the shearing resistance of the rock mass.
5.3.3 Effect of Joint Shear and Normal Stiffness
The effect of joint contact properties was evaluated assuming a 500% higher
normal and shear stiffness properties on pre-inserted fracture interfaces. Two
scenarios were considered based on the Base Case (90°/0°) and J2 (70°/20°)
models:
I. J7 (90°/0°; normal stiffness 10GPa/m, shear stiffness 1GPa/m)

II. J8 (70°/20°; normal stiffness 10GPa/m, shear stiffness 1GPa/m)
Comparing caving induced subsidence deformation for the Base Case (Fig.
5.6(a)) and J7 (Fig. 5.15(a)) models, it can be seen that for the scenarios with
vertical/horizontal joints an increase in joint stiffness resulted in a relatively minor
decrease in subsidence damage. According to Fig. 5.16(a) and (b) the total
extent of major (≥10cm) surface deformations was reduced by up to 9%. No
noticeable difference was observed in the magnitude of far-field surface
displacements, see Fig. 5.17.
111
-300 -250 -200 -150 -100 -50 0 50 100 150 200 250 300
(a)
J7
Legend:
0° MRV = 28043m3 angle
75°
of fracture
AI = 0.99 initiation
90°
joints 10cm displ. contours
stiffnesses vertical
(+)500% 75° 76° horizontal
(b)
J8
20° 3
MRV = 33324m
AI = 0.91
70°
joints
stiffnesses
(+)500% 63° 69°
Fig. 5.15 Subsidence at 100% ore extraction for J7 and J8 model
(a) (b)
350 350 350 350
308

normalized by BC/J2, %
281
300 300 300 300
268
254
250 250 250 218 250
207 202
200 188 200 200 200
150 150 150 150
100 100 100 100
100%
100%
100%
100%
95%
93%
91%
91%
50 50 50 50
0 0 0 0
BC J7 J2 J8 BC J7 J2 J8
(c) (d)
-250 -200 -150 -100 -50 0 50 100 150 200 250 -250 -200 -150 -100 -50 0 50 100 150 200 250
-112 100% BC -118 100% BC

-102 91% J7 -102 86% J7
-161 100% J2 -201 100% J2
-126 78% J8 -153 76% J8
BC 100% 95 BC 100% 100
J7 91% 86 J7 100% 100
J2 100% 107 J2 100% 107
J8 120% 128 J8 120% 128
-300 -200 -100 0 100 200 300 -300 -200 -100 0 100 200 300
relation to block centre, normalized by BC/J2, % relation to block centre, normalized by BC/J2, %
Fig. 5.16 Subsidence characterization for Base Case, J7, J2 and J8 models
112
(a) -300 -250 -200 -150 150 200 250 300
0
J8
J2
J8
-0.05
-0.1
-0.15
-0.2
-0.25
J7
-0.3 -0.38
0.9
(b)
0.3
J2
0.25
0.2
J8
0.15
0.1
J2
BC
0.05
J2
J2
J7
J8
BC
J7
0
-300 -250 -200 -150 150 200 250 300
extraction at different distances from block centre for Base Case, J1, J2, J3 and J4 models
A more pronounced influence of the higher joint stiffnesses is observed for model J8
with inclined joints. As observed in Fig. 5.6(c) and 5.15(b), an increase in joint
stiffness by 500% increases the minimum angle limiting major surface deformations
by about 15%, from 53 to 63 , thus reducing subsidence asymmetry. It is interesting
to note that in the direction of major subsidence asymmetry (westward) the extent of
the horizontal displacements was reduced by a larger margin (7%) than for vertical
displacements, see Fig. 5.16 (c) and (d). Comparison of the far-field surface
displacement magnitudes for models J2 and J7, Fig. 5.17, shows a significant
reduction in surface displacements for the model with higher stiffnesses (J7).
Overall, the results of ELFEN modelling infer that for gently to steeply inclined
joint sets dipping into the caving front, joint contact stiffness properties will affect
the degree of rock mass resistance to failure and therefore influence the resultant
surface subsidence. It appears that this influence is moderate and will
predominantly be controlled by the joint‟s resistance to shear, i.e. by the shear
stiffness and shear strength. For the joint sets with vertical to sub-vertical and
horizontal to sub-horizontal dip the influence of joint stiffnesses on the resultant
subsidence is of a lesser significance.
113
5.4 Influence of Faults
The influence of faults on surface subsidence development was evaluated
through a series of models assuming a fault that dips toward the cave,
considering different fault locations with respect to the block centre vertical axis
and varying the fault inclination. To allow comparison of induced subsidence
these preliminary analyses assumed the contact properties on fault interfaces
to be identical to the contact characteristics of pre-inserted discontinuities
(shown in Table 5.1). Two different jointing conditions, the Base Case (90°/ 0°)
and J2 (70°/ 20°), were employed.
5.4.1 Effect of Fault Location
The effect of fault location on surface subsidence development was evaluated

based on five scenarios, Table 5.3.
Table 5.3 Modelling scenarios for analysis of the effect of fault location
Scenario Joint sets dips, ° Fault dip, ° Fault location, m Figure

F1 50 5.18(a)
F2 90/0 100 5.18(b)
F3 60 150 5.18(c)
F4 100 5.18(d)
70/20
F5 150 5.18(e)
Figs. 5.19 to 5.23 illustrate the mechanisms of surface subsidence development at

35, 50 and 60% ore extraction and Fig. 5.24(a,b,c) shows resultant subsidence
deformations at 100% ore extraction for models employing vertical/horizontal joints
(F1, F2, F3). Comparing these models it can be seen that depending on the
location of the fault the degree of its influence on caving induced surface
subsidence will vary. For the model with a fault located at 50m from the block
centre vertical axis (F1, Figs. 5.19 and 5.22(a)), caving induced unloading quickly
triggers translational failure and full disintegration of the fault hanging wall and a
gradual failure of the fault footwall. By the end of ore extraction the fault is almost
fully consumed by the caving. Observed surface subsidence deformations are
largely symmetrical with respect to the block centre vertical axis.
114
-300 -250 -200 -150 -100 -50 0 50 100 150 200 250 300
(a) F1
fault
0°
90° 50m
60°
(b) F2
0°
90° 100m
(c) F4
20°
70° 60°
(d) F3
0°
90° 150m
(e) F5
20°
60°
70°
(f) F6
0°
90°
(g) F8
20°
45°
70°
(h) F7
0°
90°
(i) F9
20°
75°
70°
Fig. 5.18 Assumed fracture orientations and fault positions for F1 to F9 models
115
35% ore extraction
50m
0°
90°
50% ore extraction
60% ore extraction
Legend: rotational failure; translational failure; fault location prior to failure

active rock mass movement; developing rock mass failure
Fig. 5.19 Subsidence crater formation for F1 model
116
35% ore extraction
50m
0°
90°
50% ore extraction
60% ore extraction

117
35% ore extraction
50m
0°
90°
50% ore extraction
60% ore extraction
Legend: rotational failure;

118
35% ore extraction
50m
20°
70°
50% ore extraction
60% ore extraction

119
35% ore extraction
50m
20°
70°
50% ore extraction
60% ore extraction
Legend: rotational failure; translational failure;

120
-300 -250 -200 -150 -100 -50 0 50 100 150 200 250 300
(a) fault location prior

to caving
F1
Legend:
0° 73° angle
of fracture
90° initiation
10cm displ. contours
3 vertical
MRV = 30154m
AI = 1.0 73° 73° horizontal
(b)
F2
0°
90°
3
MRV = 32207m
AI = 0.80 61° 76°
(c)
F3
0°
90°
3
MRV = 27519m
AI = 0.99 73° 74°
(d)
F4
20°
70°
3
MRV = 34630m
AI = 0.82 61° 74°
(e)
F5
20°
70°
3
MRV = 35602m
AI = 0.80 59° 74°
Fig. 5.24 Subsidence at 100% ore extraction for F1, F2, F3, F4 and F5 model
121
The minimum angle delineating the extent of major ( 10cm) surface
displacements is 73°, which is only 2° higher than for the same model but without
a fault (Base Case, Fig. 5.6(a)). For the model with a fault located 100m from the
block centre vertical axis (F2, Fig. 5.20) a notably different subsidence
development mechanism is observed. Only a minor undercuting of the fault
coupled with caving induced unloading triggered translational failure of major
hanging wall segments along the fault interface, eventually resulting in the
hanging wall “sagging” into the cave. The fault footwall withstood the caving and
sustained only minor damage. Surface subsidence deformations are clearly
asymmetrical in a direction towards the fault. The minimum angle delineating the
extent of major surface displacements is 61°, which is 10° less than for the Base
Case model. A fault positioned outside the caving boundaries, at 150m from the
block centre vertical axis (F3, Fig. 5.21), did not exhibit any significant influence
on surface subsidence development. As seen in Fig. 5.25 the presence of a
steeply dipping fault, in a vertical/horizontal jointed rock mass, located at 50 (F1)
and 150m (F3) from the block centre vertical axis had negligible effect on the
extent of the zone of major surface displacements. In contrast, a fault located at
100m (F2) increased the extent of major vertical and horizontal displacements
zone by about 20%, primarily towards the fault.
Subsidence development mechanisms for F4 and F5 models, in which

steeply/gently dipping (70°/20°) joints were assumed, are illustrated in Figs. 5.22
and 5.23 and show similar observed trends as previously discussed for the F2
and F3 models. Final surface subsidence deformations at 100% ore extraction
for models F4 and F5 are given in Fig. 5.24(d,e). Comparing the models, where
a fault is intersecting the block (F2, F4), it can be noted that the change of joint
orientation did not affect the extent of major surface deformation, which was
limited by the fault. For models, where the fault does not intersect the block (F3,
F5), subsidence deformations were primarily governed by jointing. Comparing
the F3 (Fig. 5.24(c)) and Base Case (Fig. 5.6(a)) models, increased tensile
fracturing can be noted in the hanging wall in the vicinity of the caving boundary
indicating the weakening effect of the fault on the hanging wall rock mass.
122
Comparison of J2 (Fig. 5.6(c)) and F5 (Fig. 5.24(e)) models illustrates the limiting
effect of the fault on rock mass mobilization. It appears that the fault prevented
mobilization of a rock mass in the footwall, increasing the limiting angle from 53°
to 59°. According to Fig. 5.26 the presence of a fault in steeply/gently dipping
(70°/20°) joint settings located at 100 and 150m from the block centre vertical
axis decreased the zone of major surface horizontal displacements by 17% and
13%, respectively, in the direction towards the fault.
Figs. 5.27 and 5.28 illustrate far-field displacements for models based on
vertical/horizontal and inclined joints, respectively. For models with
vertical/horizontal joints, faults generally increased the magnitude and extent of
the far-field displacements. The largest increase was observed for the model
with a fault located 150m from the block centre vertical axis (F3), where
horizontal displacements in excess of 1cm were observed as far as 200m from
the caving boundary, which is twice as far as for the model without a fault (Base
case). For models with inclined joints the opposite trend was observed, the
presence of a fault limited both the magnitude and extent of far-field
displacements. Irrespective of jointing orientation horizontal displacements were
predominantly observed.
Caving induced unloading of the hanging wall results in a formation of a

topographical step where the fault daylights. Fig. 5.29 compares differential XY
displacements along fault surfaces with continuous ore extraction for all studied
models. Depending on the fault location with respect to the block centre
movements at the fault surface may vary significantly. For the models F1, F2
and F4, where a fault intersects the block, movements on the order of metres
were observed, whereas for models F3 and F5, where a fault did not intersect the
block, movements of up to several centimetres were noted. Inclination of the
joints affects these movements, such that larger XY displacements, which
develop more rapidly, are observed for models with inclined joints.
123
350 350 350 350


300 300 300 300
255 258
250 250 250 218 220 220 250
207 202 212
200 200 200 200
150 150 150 150
123%
100 100 100 100
118%
100%
100%
102%
101%
101%
98%
50 50 50 50
0 0 0 0
BC F1 F2 F3 BC F1 F2 F3
(a) (b)
(c) relation to block centre, m (d) relation to block centre, m
-250 -200 -150 -100 -50 0 50 100 150 200 250 -250 -200 -150 -100 -50 0 50 100 150 200 250
-112 100% BC -118 100% BC

-110 98% F1 -110 93% F1
-160 143% F2 -160 136% F2
-112 100% F3 -112 95% F3
BC 100% 95 100% 100
BC
F1 97% 92 F1 110% 110
F2 100% 95 F2 98% 98
F3 105% 100 F3 108% 108
-300 -200 -100 0 100 200 300 -300 -200 -100 0 100 200 300
Fig. 5.25 Subsidence characterization for Base case, F1, F2 and F3

Total extent of major (≥10cm) vertical (a) and horizontal (b) surface displacements in m
and in % of Base case value; extent of 10cm surface vertical (c) and horizontal (d)
displacements in relation to centre axis of the block, in m
350 350 320 350

308
300 275 300 310 300

268 269
normalized by J2, %
normalized by J2, %
250 250 300 250
290
200 200 279 200
280
150 150 268 150
270
100 100 260 100
100%
100%
103%
100%
91%
87%
50 50 250 50
0 0 240 0
(a) J2 F4 F5 (b) J2 F4 F5
(c) Extent of 10cm surface vertical dispacements in

relation to block centre, m
(d) Extent of 10cm surface horizontal displacements in
relation to block centre, m
-250 -200 -150 -100 -50 0 50 100 150 200 250 -250 -200 -150 -100 -50 0 50 100 150 200 250
-161 100% J2 -201 100% J2

-161 100% F4 -160 80% F4
-167 104% F5 -171 85% F5
J2 100% 107 100% 107

J2
F4 101% 108 F4 101% 108
F5 101% 108 F5 101% 108
-300 -200 -100 0 100 200 300 -300 -200 -100 0 100 200 300
Fig. 5.26 Subsidence characterization for J2, F4 and F5

and in % of J2 value; extent of 10cm surface vertical (c) and horizontal (d)
displacements in relation to central axis of the block, in m
124
(a) -300 -250 -200 -150 150 200 250 300
0
F3
F3
-0.05
-0.1
-0.15
-0.2
-0.25
F2
-0.3 -2
1.2
(b)
0.3
F2
0.25
0.2
0.15
F1
0.1
F1
F3
BC
0.05
F3
F2
F3
F1
F3
BC
0
-300 -250 -200 -150 150 200 250 300
extraction at different distances from block centre for Base Case, F1, F2 and F3 models

(a) -300 -250 -200 -150 150 200 250 300
0
J2
-0.05
F4
-0.1
-0.15
-0.2
-0.25
F5
J2
-0.3 -0.4 -3.2

0.8 1.2
0.3
(b)
J2
F5
0.25
0.2
F4
0.15
0.1
J2
F4
0.05
F5
F5
J2
F4
J2
0
-300 -250 -200 -150 150 200 250 300
extraction at different distances from block centre for J2, F4 and F5 models
125
0
(a) -0.02m
displacements, m
-1
0°
Differential XY
-2
90° -2.37m
-3
differential F1
-4 F2 hangingwall
XY displacement disintegrated -4.31m
F3
footwall -5
0 10 20 30 40 50 60 70 80 90 100
Ore extraction, %
hanging
(b) 0
wall -0.07m
displacements, m
-1 20°
Differential XY
-2
70°
-3
F4
-4 -3.73m
F5
-5
0 10 20 30 40 50 60 70 80 90 100
Ore extraction, %
Fig. 5.29 Differential XY displacements for surface points on the fault hanging and
footwalls: (a) F1, F2 and F3; (b) F4 and F5 models
Overall, the ELFEN models demonstrated that the location of the fault may play
an important role in defining the extent of surface subsidence deformation:
1. It appears that steeply dipping faults, daylighting into the cave and located
within an area of imminent caving are likely to be caved and are unlikely to
play any major role in the resultant subsidence.
2. Faults partially intersecting the caving area may create favourable
conditions for failure of the entire hanging wall.
3. Depending on rock mass fabric, faults located in the vicinity of the caving zone
may have a minimal influence or decrease the extent of the area of subsidence
deformation. The former behaviour was observed for horizontal/vertical joint
settings and the latter for orthogonal steeply/gently dipping joints. It should be
recognized that current modelling did consider placing the fault on the east
side of the model where it would be closely aligned with steeply dipping joints.
It appears that in this case the fault could extend the zone of influence.
4. A topographical step in the surface profile is formed where the fault
daylights at the surface. Significant movements should be anticipated if
the fault daylights into the cave.
126
5.4.2 Effect of Fault Inclination
The effect of fault inclination on surface subsidence development was evaluated

based on six modelling scenarios, for the fault partially intersecting the block.
Three different fault inclinations and two different jointing conditions were
considered, as summarized in Table 5.4:
Table 5.4 Modelling scenarios for analysis of the effect of fault inclination
Scenario Joint sets dips, ° Fault dip, ° Figure

F6 45 5.18(f)
F2 90/0 60 5.18(b)
F7 75 5.18(h)
F8 45 5.18(g)
F4 70/20 60 5.18(c)
F9 75 5.18(i)
Figs. 5.20, 5.30, 5.31 illustrate surface subsidence development at 35, 50 and
60% ore extraction, and, Figs. 5.24(b) and 5.34(a,b) show resultant subsidence
deformations at 100% ore extraction for models F2, F6 and F7, assuming
vertical/horizontal joints. Figs. 5.22, 5.32 and 5.33 present surface subsidence
development at 35, 50 and 60% ore extraction and Figs. 5.23(d) and 5.34(c,d)
show the resultant subsidence deformation at 100% ore extraction for models
F4, F8 and F9, assuming steeply/gently dipping joints. Comparing subsidence
deformation development for varying fault inclinations and varying joint set
orientations it should be noted that, for all assumed inclinations, faults affect the
development of subsidence deformation. Irrespective of jointing orientation
caving induced failure is controlled predominantly by the plane of weakness
provided by the fault. Continuous ore extraction leads to full mobilization of the
entire hanging wall and its disintegration into segments. The mode of hanging
wall segmentation appears to be controlled by joint orientation. Failure of the
hanging wall leads to formation of a crater wall along the footwall of the exposed
fault, which is particularly pronounced for the 75° and 60° faults. For the 75° fault
models (F7, F9, Fig. 5.34(b,d)) exposure of a steep footwall by the caving causes
its partial failure, the magnitude of this failure is strongly controlled by the jointing.
127
35% ore extraction
50m
0°
90°
50% ore extraction
60% ore extraction

128
35% ore extraction
50m
0°
90°
50% ore extraction
60% ore extraction

129
35% ore extraction
50m
20°
70°
50% ore extraction
60% ore extraction

130
35% ore extraction
50m
20°
70°
50% ore extraction
60% ore extraction

131
-300 -250 -200 -150 -100 -50 0 50 100 150 200 250 300
(a)
F6
Legend:
0° angle
46°
of fracture
90° initiation
3 vertical
MRV = 40798m 46°
75° horizontal
AI = 0.61
(b) fault location prior
to caving
F7
0°
90°
3
MRV = 29594m
AI = 0.95 75° 71°
(c)
F8
20°
70°
3
MRV = 43319m
AI = 0.70 46° 66°
(d)
F9
20°
70°
3
MRV = 33922m
AI = 0.88 59° 67°
Fig. 5.34 Subsidence at 100% ore extraction for F6, F7, F8 and F9 models
132
Vertical/horizontal jointing contributes to formation of a nearly vertical wall,
whereas inclined jointing favours kinematic instability of major near surface rock
mass blocks. For the 60° faults (F2, F4, Fig. 5.24(b,d)), the moderately inclined
footwall was less exposed and the passive support provided by the muck pile
prevented development of major internal instability. Here it should be noted that
reduction of this support will likely trigger further footwall damage, particularly for
the case with inclined joints. For the 45° faults (F6, F8, Fig. 5.34(a,c)), the
footwall sustained only minor damage. It appears that for the simulated jointing
conditions development of major instability in a 45° footwall slope even with
continuous ore extraction is highly unlikely.
Inclination of the fault significantly alters the extent of the caving influence. For
the 45° and 60° faults, irrespective of the assumed joint set conditions, the extent
of major surface deformation toward the fault was determined by the fault
inclination, so that the angular limits of major ( 10cm) surface displacements are
equal or nearly equal to the fault inclination. For the 75° faults the extent of
major surface deformations is a function of stability of the exposed footwall. For
the model with vertical/horizontal joints the limiting angle is 75 , whereas for the
model with inclined joints it is 59 .
Comparison of the extent of major surface displacements for the models with
vertical/horizontal joints without a fault (Base Case) and with fault dips of 75
(F6), 60 (F2) and 45 (F7) is presented in Fig. 5.35. This figure shows that
faults with inclinations of 60 and 45 extended the total zones of major
displacement by about 20 and 60%, respectively. In the direction towards the
fault, for 60 and 45 dipping faults, the zone of influence was increased by 40
and 120%, respectively, i.e. a decrease in fault inclination by 15 extended the
zone of major surface displacements by 80%. The fault with 75 inclination had
only a minor influence on the observed extent of major surface displacements.
Comparison of the extent of major surface displacements for the models with
inclined joints without a fault (J2) and with a fault of 75 (F9), 60 (F4) and 45
(F8) inclination is given in Fig. 5.36. As follows from this figure for faults with 60
133
and 75 inclination the extent of the zone of major surface displacement towards
the fault was reduced by as much as 50%. The surface outcrop location of the
45 fault coincided approximately with the extent of major displacements for the
model without a fault (see Figs. 5.6(c), 5.34(d)), hence no major influence was
observed. Interestingly models with 45 and 75 dipping faults exhibit increased
zones of influence in an eastward direction from the block centre vertical axis.
(a) (b)
350 331 350 350 350


350

300 300 300 300
255 258
250 250 250 218 222 250
207 204
200 200 200 200
150 150 150 150
161%
161%
125%
100 100 100 100
118%
100%
100%
102%
102%
50 50 50 50
0 0 0 0
BC F7 F2 F6 BC F7 F2 F6
(c) (d)
-250 -200 -150 -100 -50 0 50 100 150 200 250 -250 -200 -150 -100 -50 0 50 100 150 200 250
-112 100% -118 100% BC

BC
-102 91% F7 -102 86% F7
-160 143% F2 -160 136% F2
219% F6 208% F6
-245 95 -245 BC 100% 100
BC 100%
F7 107% 102 F7 120% 120
F2 100% 95 F2 98% 98
F6 91% 86 F6 105% 105
-300 -200 -100 0 100 200 300 -300 -200 -100 0 100 200 300
Fig. 5.35 Subsidence characterization for BC, F2, F6 and F7 models

and in % of Base Case value; extent of 10cm surface vertical (c) and horizontal (d)
Far-field displacements for models with vertical/horizontal and inclined joints are
presented in Figs. 5.37 and 5.38 respectively. It can be inferred from these
figures that, in the direction towards the fault, the extent of the far-field
displacements is a function of fault inclination. A shallower fault inclination
resulted in a larger area mobilized by the caving. Conversely, steeper faults limit
such an area. Within the failing hanging wall higher deformation magnitudes
were observed for models with vertical/horizontal joints.
134
(a) (b)
350 375 350 350 384 350

308 305

300 300 300 300
268 269 268
normalized by J2, %
normalized by J2, %
254
250 250 250 250
200 200 200 200
150 150 150 150
140%
125%
100 100 100 100
100%
100%
100%
95%
99%
87%
50 50 50 50
0 0 0 0
J2 F9 F4 F8 J2 F9 F4 F8
(c) (d)
-250 -200 -150 -100 -50 0 50 100 150 200 250 -250 -200 -150 -100 -50 0 50 100 150 200 250
100% J2 100% J2
-161
-126 51% F9 -245 -169 69% F9
66% F4 -160 65% F4

-161
100% F8 100% F8
-245 100% 107 -245 J2 100% 105
J2
F9 149% 128 F9 130% 136
F4 126% 108 F4 103% 108

F8 151% 130 F8 132% 139
-350 -250 -150 -50 50 150 250 350 -300 -200 -100 0 100 200 300
Fig. 5.36 Subsidence characterization for J2, F4, F8 and F9 models


(a) -300 -250 -200 -150 150 200 250 300
-0.05
-0.1
-0.15
-0.2
-0.25
F2
F6
F6
-0.3 -0.8 -2 -0.8

0.8 1.2 0.8
(b)
0.3
F6
F2
F6
0.25
0.2
0.15
F6
0.1
F7
BC
0.05
F2
F7
BC
0
-300 -250 -200 -150 150 200 250 300
extraction at different distances from block centre for Base Case, F2, F6 and F7 models
135
(a) -300 -250 -200 -150 150 200 250 300
0
J2
-0.05
F4
-0.1
F9
F8
-0.15
-0.2
-0.25
F8
J2
-0.3 -0.4
0.8 0.45
0.3
(b)
J2
F8
0.25
F9
0.2
F8
F4
0.15
0.1
J2
F9
F8
F4
0.05 F9
F4
F8
J2 J2
0
-300 -250 -200 -150 150 200 250 300
extraction at different distances from block centre for J2, F4, F8 and F9 models
Depending on the fault inclination the amount of differential displacement at the

surface outcrop of the fault will vary, higher displacements being observed for
models with steeper faults (see Fig. 5.39).
0
-0.02m
displacements, m
-1 0°
Differential XY
-2
90° -2.37m
-3
F1
differential -4 F2 hangingwall -4.31m
F3 desintegrated
XY displacement -5
footwall 0 10 20 30 40 50 60 70 80 90 100
Ore extraction, %
hanging 0
wall -1.36m
displacements, m
-1 20°
Differential XY
-1.16m
-2
70° hangingwall
-3 desintegrated -2.37m
F7
-4 F2
F6
-5
0 10 20 30 40 50 60 70 80 90 100
Ore extraction, %
Fig. 5.39 Differential XY displacements for surface points on the fault hanging and foot
walls for F2, F6 and F7 models
136
The following can be concluded based on the ELFEN modelling results, with respect
to the effect of fault inclination on block caving induced surface subsidence:
1. Unequivocally, the inclination of the fault partially intersecting the

caving area controls the extent of surface subsidence deformations.
Low dipping faults will extend and steeply dipping faults will decrease
the area of surface subsidence.
2. For faults daylighting into the cave, failure of the hanging wall is likely
inevitable. For the assumed hard rock mass conditions in the current
modelling, the stability of the exposed footwall is dependent on its slope,
the amount of passive support provided by the muck pile and the
orientation and persistence of jointing within the footwall. The presence
of well defined steeply/gently dipping joint set approaching
perpendicular orientation with relation to the fault will increase the
kinematic potential for failure of major near surface footwall segments.
In such circumstances a model combining the fault/jointing system is
extremely important.
Both conclusions are in agreement with observations by Abel & Lee (1980) and
Stacey & Swart (2001).
5.5 Influence of Rock Mass Strength and Deformability

Characteristics
The effects of rock mass strength and deformability characteristics were
evaluated based on three modelling scenarios, using model J2 as a basis:
I. RM1 (70°/20°; rock mass tensile strength reduced by 40%, from 1 MPa to
0.6MPa);
II. RM2 (70°/20°; rock mass tensile strength increased by 40%, from 1 MPa to
1.4MPa);
III. RM3 (70°/20°; rock mass deformation modulus increased by 30%, from
18GPa to 23GPa).
137
The mechanism of surface subsidence development for models RM1 and RM3 was
very similar to model J2, shown in Fig. 5.3(a), exhibiting gradual formation of the
subsidence crater. For model RM2, however, an increase in the tensile strength of
the rock mass reduced rock mass cavability, leading to cave propagation arrest at
about block height. Continuous ore extraction resulted in formation of a significant
expansion void, reaching a height of more than 50m, as illustrated in Fig. 5.40(a). At
55% ore extraction large portions of the crown pillar began to detach and by 57%
ore extraction the crown pillar had fully disintegrated and collapsed into the
expansion void, Fig. 5.40(c). The subsidence crater was formed as a result of a
single failure event rather than gradual cave propagation. Similar behaviour was
observed during a caving physical model test by Kvapil (2004), Fig. 4.6(c,d). It
should be noted that sudden crown pillar collapse can cause a major airblast leading
to dramatic consequences such as that which occurred at Northparkes block cave
mine (Hebblewhite, 2003).
Fig. 5.41 presents subsidence profiles at 100% ore extraction for models RM1,
RM2 and RM3. Comparing Figs. 5.42(a,b) and 5.6(c) it can be noted that a
decrease in rock mass tensile strength by 40% increased subsidence deformation
so that the minimum angle delineating major ( 10cm) surface displacements was
decreased by 3 . An increase in rock mass tensile strength by 40% led to higher
rock mass resistance to mobilization, consequently increasing the minimum
delineating angle by 7 . An increase in the rock mass deformation modulus by
30% had practically no influence on the resultant subsidence deformations.
Comparing the extent of major ( 10cm) surface subsidence displacements for

models J2, RM1, RM2 and RM3, Fig. 5.42, it can be noted that depending on the
decrease or increase in rock mass tensile strength the maximum extent of
subsidence displacement in the westward direction is increased by 7% or
decreased by 16%, respectively. Changes in rock mass deformability did not
result in any noticeable difference in the extent of major surface displacements.
138
(a) 35% (b) 50%
50m
(c) 55% (d) 56%
(e) 57% (f) 60%

Fig. 5.40 Stages of subsidence crater formation at different percentages of ore block
extraction for model RM2
(a) - (b) cave propagation arrest and formation of expansion void;
(b) - (e) crown pillar collapse and formation of subsidence crater
139
-300 -250 -200 -150 -100 -50 0 50 100 150 200 250 300
(a)
RM1
Legend:
20° 50° angle
3 of fracture
MRV = 37512m
70° initiation
AI = 0.72
rock mass vertical
tensile str.
(-)40%
(b)
RM2
20°
3
MRV = 35885m
70°
AI = 0.87
rock mass
tensile str.
(+)40% 60° 69°
(c)
RM3
20°
3
70° MRV = 36479m
AI = 0.73
rock mass
deformability
(+)30% 53° 73°
Fig. 5.41 Subsidence at 100% ore extraction for models RM1, RM2 and RM3
Comparison of the far-field surface displacements for the same models, given in
Fig. 5.43, showed no significant difference at either the extent of surface
displacements or their magnitude.
ELFEN modelling demonstrated that for the assumed variations in rock mass
tensile strength and deformability, the effect on the resultant subsidence
deformations was limited. Modelling demonstrated however that an increase in
rock mass tensile strength may have a profound influence on the manner of cave
propagation and subsidence crater formation mechanism.
140
(a) (b)
350 350 350 343 350
308 314
hsurface displacements, m

Total extent of 10cm horz.
297
283 282
300 300 300 300
268
normalized by J2, %
normalized by J2, %
260
250 250 250 250
200 200 200 200
150 150 150 150
100 100 100 100
106%
111%
100%
105%
100%
102%
97%
96%
50 50 50 50
0 0 0 0
J2 RM1 RM2 RM3 J2 RM1 RM2 RM3
(c) (d)
-250 -200 -150 -100 -50 0 50 100 150 200 250 -250 -200 -150 -100 -50 0 50 100 150 200 250
-161 100% J2 -201 100% J2

-140 87% RM1 -215 107% RM1
-163 101% RM2 -169 84% RM2
-169 105% RM3 -201 100% RM3
J2 100% 107 J2 100% 107
RM1 112% 120 RM1 120% 128
RM2 112% 120 RM2 120% 128
RM3 106% 113 RM3 106% 113
-300 -200 -100 0 100 200 300 -300 -250 -200 -150 -100 -50 0 50 100 150 200 250 300
Fig. 5.42 Subsidence characterization for models J2, RM1, RM2 and RM3

(a) -300 -250 -200 -150 150 200 250 300
0
RM3
J2
RM1
-0.05
RM1
-0.1
-0.15
RM3
-0.2
-0.25
RM2
J2
-0.3 -0.38 -0.37

0.8 0.67 0.44
0.3
RM2
RM3
J2
(b) 0.25
RM1
0.2
RM1
0.15
RM3
RM2
0.1
J2
RM3
RM2
RM1
RM1
RM2
RM3
0.05
J2
J2
0
-300 -250 -200 -150 150 200 250 300
extraction at different distances from block centre for models J2, RM1, RM2 and RM3
141
5.6 Influence of Stress Environment
The influence of the in-situ stress environment on caving induced surface
subsidence development was evaluated by analyzing the effect of change in the
ratio between horizontal and vertical stresses, K. Default modelling conditions
assume hydrostatic stress, K=1, i.e. horizontal and vertical stress are equal. Two
additional assumptions were considered, K=0.5 (200% higher vertical stress) and
K=2 (200% higher horizontal stress). To appreciate the variation in response in
different geological settings, simulations with vertical/horizontal and
steeply/gently dipping joints sets, based on Base Case and J2 models,
respectively, were considered, as summarized in Table 5.5:
Table 5.5 Modelling scenarios for analysis of the effect of stress environment
Scenario Joint sets dips, ° Stress ratio, K

S1 0.5
90/0
S2 2
S3 0.5
70/20
S4 2
Variation in the assumed K ratio did not have any noticeable affect on
subsidence crater formation mechanism and therefore is not shown here.
Resultant surface subsidence deformations at 100% ore extraction for the S1 to
S4 models are given in Fig. 5.44. Comparing models with vertical/horizontal joint
sets, shown in Figs. 5.6(a) and 5.44(a,b), no significant difference in subsidence
deformations can be observed. The same conclusion can be reached for models
with inclined joints, see Figs. 5.6(c) and 5.44(c,d). In terms of the limiting angles
delineating major ( 10cm) subsidence displacements, generally higher angles
are observed for the lower K, and lower angles for the higher K ratio.
Figs. 5.45 and 5.46 compare the extent of major surface subsidence
displacements for models with vertical/horizontal (BC, S1 and S2) and
steeply/gently dipping (J2, S3 and S4) joint sets, respectively. It is interesting to
note that influence of lower K ratio on the extent of major surface subsidence
displacements is more marked than for a higher K ratio. For the models with
vertical/horizontal joints the total extent of major displacements was reduced by a
142
maximum of 6% for K=0.5 and increased by only 2% for K=2. Similar changes
for simulations with inclined joints were 16% and 5%, for K=0.5 and 2
respectively.
If for major subsidence deformation the effect of varying K is relatively subtle, the
influence of the K ratio on far-field displacements is far more pronounced, as
shown in Figs. 5.47 and 5.48. In comparison with models with the lower K ratio,
simulations with K=2 (S2 and S4) exhibited significantly larger magnitudes (up to
5 times) of horizontal far-field displacements. For the model with
vertical/horizontal joint sets (S2), in both the westward and eastward direction
from the block centre vertical axis, a nearly equal increase in displacements is
observed. In contrast for the model with inclined joint sets (S4), higher
displacements (up to 3 times at 250m from caving boundary) in the westward
direction are observed.
Overall, ELFEN modelling to investigate the influence of in-situ stress

environment on block caving induced subsidence, provides the following
conclusions:
1. In general, lower horizontal in-situ stress results in a smaller area affected

by subsidence deformations.
2. Simulations with different joint patterns did not exhibit major differences in
the effect of varying K ratio on surface subsidence development and the
extent of major subsidence displacements.
3. Higher horizontal stresses (K=2), lead to elevated magnitudes of far-field

horizontal surface displacements. A combination of higher horizontal
stresses and inclined joints promoted higher displacement magnitudes in
the dip direction of a steeply dipping set.
143
-300 -250 -200 -150 -100 -50 0 50 100 150 200 250 300
(a)
S1
Legend:
0°
73° angle
of fracture
90° initiation
3 10cm displ. contours
MRV = 28580m
AI = 0.94 vertical
K=0.5 73° 78° horizontal
(b)
S2
0°
90°
3
MRV = 31487m
AI = 0.95
K=2 71° 75°
(c)
S3
20°
70°
3
MRV = 33165m
AI = 0.74
K=0.5 55° 74°
(d)
S4
20°
70°
3
MRV = 35654m
AI = 0.76
K=2 53° 70°
Fig. 5.44 Subsidence at 100% ore extraction for models S1, S2, S3 and S4
144
350 350 350 350



300 300 300 300
250 250 250 218 222 250
204 207 212 204
200 200 200 200
150 150 150 150
100 100 100 100
102%
102%
100%
100%
99%
94%
50 50 50 50
0 0 0 0
(a) S1 BC S2 (b) S1 BC S2
(c)Extent of 10cm surface vertical dispacements in (d) Extent of 10cm surface horizontal displacements in
-250 -200 -150 -100 -50 0 50 100 150 200 250 -250 -200 -150 -100 -50 0 50 100 150 200 250
-112 100% S1 -112 95% S1
-112 100% BC -118 100% BC
-110 98% S2 -115 97% S2
S1 97% 92 S1 92% 92
95 BC 100% 100
BC 100%
S2 S2 107% 107
107% 102
-300 -200 -100 0 100 200 300 -300 -200 -100 0 100 200 300
Fig. 5.45 Subsidence characterization for models Base Case, S1 and S2

350 350 350 324 350
308

296
279
300 300 300 300
268
normalized by J2, %
normalized by J2, %
250 226 250 250 250

200 200 200 200
150 150 150 150
100 100 100 100
105%
104%
100%
100%
96%
84%
50 50 50 50
0 0 0 0
(a) S3 J2 S4 (b) S3 J2 S4
(c)
Extent of 10cm surface vertical dispacements in (d) Extent of 10cm surface horizontal displacements in
-250 -200 -150 -100 -50 0 50 100 150 200 250 -250 -200 -150 -100 -50 0 50 100 150 200 250
-118 73% S3 -189 94% S3
-161 100% J2 -201 100% J2
97% S4 -201 100% S4

-156
S3 101% 108 S3 100% 107
J2 100% 107 J2 100% 107
S4 S4 115% 123
115% 123
-300 -200 -100 0 100 200 300 -300 -200 -100 0 100 200 300
Fig. 5.46 Subsidence characterization for models J2, S3 and S4

145
0.3
0.25
0.2
0.15
S2
0.1
S2
S2
S2
S2
S2
BC
S2
0.05 S2
S1
BC
0
-300 -250 -200 -150 150 200 250 300
Fig. 5.47 Total horizontal surface displacements at the end of ore extraction at different
distances from block centre for models Base Case, S1 and S2

(a) -300 -250 -200 -150 150 200 250 300
0
J2
S4
-0.05
-0.1
-0.15
S3
-0.2
-0.25
S4
J2
-0.3 0.9 3.8

0.8 0.7
0.3
J2
S4
(b)
S3
0.25
0.2
0.15
S4
0.1
J2
S4
S4
S3
S4
S4
S4
S3
0.05
J2
J2
0 S4
-300 -250 -200 -150 150 200 250 300
extraction at different distances from the block centre for models J2, S3 and S4
146
5.7 Influence of Varying Lithological Domains
Natural rock masses are highly heterogeneous and consist of domains of varying
competency and discontinuity networks. The default modelling conditions
assumed identical properties throughout the simulated rock mass and did not
account for possible variation in discontinuity orientation between the ore and
surrounding host rock. This section investigates the influence of varying rock
mass strength and jointing pattern, between the ore body and the surrounding
host rock, on block caving induced surface subsidence.
5.7.1 Effect of Varying Strength between Ore and Host Rock
The effect of variation in rock mass strength was analyzed by assuming a stronger
host rock and increasing its tensile strength by 40% (from 1MPa to 1.4MPa). Rock
mass conditions with and without a caprock, as illustrated in Fig. 5.49, were
considered.
-300 -250 -200 -150 -100 -50 0 50 100 150 200 250 300
(a) GD1
0° stronger
host rock
90°
(c) GD3
20°
ore
70°
(b) GD2
0° stronger stronger
ore host rock
host rock
90°
(d) GD4
20°
extracted
70° ore block
Fig. 5.49 Assumed geometries for models GD1 to GD4
147
For each assumed rock mass strength condition assumption two modelling
scenarios with varying joint orientation (as in Base Case and J2 models) were
considered, as summarized in Table 5.6:
Table 5.6 Modelling scenarios for analysis of the effect of varying strength between ore
and host rock
Scenario Joint sets dips, ° Caprock Figure

GD1 + 5.49(a)
90/0
GD2 - 5.49(b)
GD3 + 5.49(c)
70/20
GD4 - 5.49(d)
Figs. 5.50 and 5.51 illustrate subsidence crater development with continuous ore
extraction for the above models. Comparing simulations with vertical/horizontal
joints (GD1, GD2) it can be seen that the presence of a stronger caprock delayed
formation of a subsidence crater. As shown in Fig. 5.50(a), cave propagation
arrested just above the boundary of the ore block. With continuous ore extraction
an expansion void was formed, reaching a maximum height of about 30m at 45%
ore extraction. At this point the crown pillar began to fracture and rapidly
disintegrate, in a similar fashion as for RM2 model (Fig. 5.40). In contrast, it is
evident from Fig. 5.50(b) that for a model without a stronger caprock the
subsidence crater developed gradually. Comparing this model with the Base Case,
shown in Fig. 5.2(a), it can be inferred that the stronger host rock limited the extent
of fracturing damage outside the ore body boundaries and resulted in an initial
subsidence crater with nearly vertical walls. It is worth noting that rapid crown pillar
collapse for the model with a caprock led to increased fracturing damage in the
near surface rock mass. Comparing subsidence crater development for
simulations with inclined joints (GD3, GD4) and the reference model J2, Fig. 5.3(a),
it appears that the stronger host rock significantly limited the amount of rock mass
mobilized by the failure; cave propagation was largely vertical, although some
preferential fracturing along the steeply dipping set was apparent. Contrasting
models GD3 (Fig. 5.51(a)) and GD4 (Fig. 5.51(b)) it is apparent that the presence
of a stronger caprock reduced the fracturing damage in the rock mass westward
148
from the block centre vertical axis and prevented development of large scale rock
mass failure.
(a) GD1 (b) GD2
35% 35%
50m
50% 50%
60% 60%

Fig. 5.50 Subsidence crater formation for models (a) GD1 with a caprock and (b) GD2
without a caprock
149
(a) GD3 (b) GD4
35% 35%
50m
50% 50%
60% 60%

Fig. 5.51 Subsidence crater formation for models (a) GD3 with a caprock and (b) GD4
without a caprock
150
Resultant surface subsidence at 100% ore extraction for all lithological domain
models is given in Fig. 5.52.
-300 -250 -200 -150 -100 -50 0 50 100 150 200 250 300
(a)
GD1
Legend:
caprock
0° 73° angle
of fracture
initiation
90° 10cm displ. contours
MRV = 27888m
3 vertical
(b)
GD2
0°
90°
3
MRV = 27095m
AI = 0.96 73° 76°
(c)
GD3
caprock
20°
70°
3
MRV = 28661m
AI = 0.92 75° 69°
(c)
GD4
20°
70°
3
MRV = 32862m
AI = 0.85 63° 74°
Fig. 5.52 Subsidence at 100% ore extraction for models (a) GD1, (b) GD2, (c) GD3 and
(d) GD4
151
For the simulations with vertical/horizontal joints, surface subsidence
deformations are nearly symmetrical with respect to the block centre vertical axis.
In contrast to the Base Case model, see Fig. 5.6(a), the stronger host rock
resulted in slightly steeper angles delineating major ( 10cm) surface
deformations (to a maximum of 3°). Assumption of a caprock led to only a minor
1° increase in delineating angles. For the simulations with inclined joints, the
influence of a higher host rock strength was quite dramatic. Models with a
caprock showed higher asymmetry in the eastward direction. The delineating
angle in the westward direction was 75°, which is 24° steeper than for model J2
(Fig. 5.6(c)). The corresponding angle for the model without a caprock was 63°,
or 12° steeper than for J2 model.
As shown in Fig. 5.53, for models with vertical/horizontal joints the presence of a
stronger host rock reduced the total extent of major surface displacements by a
negligible amount. Compared to the Base Case model a maximum reduction of
6% was observed for the model with caprock (GD1) and only a 2% reduction was
exhibited by the model without a caprock (GD2). According to Fig. 5.54, for
models with inclined joints, the maximum reduction in the extent of major surface
displacements of 35% was observed for the case with a caprock (GD3), whereas
for the case without a caprock (GD4), the reduction was 15%. Notably, for both
models the extent of surface subsidence was reduced primarily in the direction of
the joint controlled subsidence asymmetry, suggesting a decreasing rock mass
susceptibility to toppling failure with increased rock mass tensile strength. An
increase in host rock mass tensile strength by 40% caused a reduction in the
extent of major surface displacements in the westward direction by a maximum
of 49% for the model with and 24% for the model without a caprock.
Far-field displacements for GD1 and GD2 did not differ from the Base Case
values and therefore are not shown here. Far-field displacements for GD3 and
GD4 models are given in Fig. 5.55, it can be seen that the stronger host rock
moderately decreased the displacement magnitudes.
152
350 350 350 350




300 300 300 300
250 250 250 218 250
207 204 207 214
195
200 200 200 200
150 150 150 150
100 100 100 100
100%
100%
99%
98%
94%
95%
50 50 50 50
0 0 0 0
(a) BC GD1 GD2 (b) BC GD1 GD2
(c) Extent of 10cm surface vertical dispacements in (d) Extent of 10cm surface horizontal displacements in
-250 -200 -150 -100 -50 0 50 100 150 200 250 -250 -200 -150 -100 -50 0 50 100 150 200 250
-112 100% BC -118 100% BC
-109 97% GD1 -109 92% GD1
-110 98% GD2 -110 93% GD2
95 BC 100% 100
BC 100%
GD1 91% 86 GD1 98% 98
GD2 GD2 104% 104

99% 94
-300 -200 -100 0 100 200 300 -300 -200 -100 0 100 200 300
Fig. 5.53 Subsidence characterization for models Base Case, GD1 and GD2
350 350 350 350
308

300 300 300 300
268
normalized by J2, %
normalized by J2, %
260 261
250 250 250 250
193 200
200 200 200 200
150 150 150 150
100 100 100 100
100%
100%
97%
85%
72%
50 50 50 50
65%
0 0 0 0
(a) J2 GD3 GD4 (b) J2 GD3 GD4
(c) Extent of 10cm surface vertical dispacements in (d) Extent of 10cm surface horizontal displacements in
-250 -200 -150 -100 -50 0 50 100 150 200 250 -250 -200 -150 -100 -50 0 50 100 150 200 250
-161 100% J2 -201 100% J2
-100 62% GD3 -102 51% GD3
-152 94% GD4 -152 76% GD4
107 J2 100% 107

J2 100%
GD3 87% 93 GD3 92% 98
GD4 GD4 102% 109

101% 108
-300 -200 -100 0 100 200 300 -300 -200 -100 0 100 200 300
Fig. 5.54 Subsidence characterization for models J2, GD3 and GD4
153
(a) -300 -250 -200 -150 150 200 250 300
0
J2
-0.05
-0.1
GD4
-0.15
-0.2
-0.25
J2
-0.3 0.4
0.8
0.3
J2
(b)
0.25
0.2
0.15
GD4
GD3
0.1
J2
GD4
GD3
GD4
GD4
GD3
GD3
0.05
J2
J2
0
-300 -250 -200 -150 150 200 250 300
extraction at different distances from block centre for models J2, GD3 and GD4
Overall, with regards to the influence of varying rock mass strength on caving
induced surface subsidence, the ELFEN results indicate the following:
1. An increase in the strength of the rock mass surrounding the excavated

block reduces the subsidence deformation. The magnitude of reduction in
subsidence deformations is highly dependent on assumed rock mass
jointing conditions. For the assumed increase in host rock tensile strength
simulations with vertical/horizontal joints, results indicated a generally
minor influence on the resultant surface subsidence. Simulations with
inclined joints showed a marked decrease in subsidence deformation
primarily in the dip direction of the steeply inclined joint set.
2. The presence of a stronger caprock may cause a temporary arrest in the

cave propagation followed by a rapid crown pillar collapse.
5.7.2 Effect of Varying Joint Orientation within the Ore and Host Rock
The influence of varying jointing was investigated by assuming that the ore body
extends to the surface and has a different joint orientation to the surrounding host
154
rock. Here it should be recognized that in natural rock masses the contact
between varying discontinuities patterns is not well defined. Typically there is a
transition zone, where patterns overlap. The extent of such a transition zone,
depends on a variety of factors primarily related to tectonic history. For the
preliminary conceptual analysis an idealized condition with minor discontinuity set
overlap was adopted, as shown in Fig. 5.56.
(a) GD5
host rock
0°
90°
ore
20°
70°
(b) GD6
host rock
20°
70°
ore
0°
90°
Fig. 5.56 Assumed modelling geometries for models (a) GD5 and (b) GD6
Two modelling scenarios were considered, as summarized in Table 5.7:
Table 5.7 Modelling scenarios for analysis of the effect of varying joint orientation within
the ore and host rock
Scenario Host rock Ore body Figure

joint sets dips, ° joint sets dips, °
GD5 90/0 70/20 5.56(a)
GD6 70/20 90/0 5.56(b)
Subsidence crater development for the above models is illustrated in Fig. 5.57. It
can be inferred from Fig. 5.57(a) that inclined joints within the ore initially
promoted skewed cave propagation in an eastward direction, terminating against
the host rock with vertical/horizontal joints and then became nearly vertical. As a
155
result of inclined cave propagation, a major rock mass segment overhanging the
caved block was formed and eventually toppled into the cave. As follows from
Fig. 5.57(b), for the ore with vertical/horizontal joints cave propagation was
generally vertical, although inclined joints in the host rock caused rock mass
mobilization and fracturing damage that resembles the response of model J2
(Fig. 5.3(a)).
The resultant surface subsidence deformation and subsidence displacement

characterization for models GD5 and GD6 are presented in Figs. 5.58 and 5.59,
respectively. Comparing subsidence deformations for model GD5 (Fig. 5.58(a))
and the Base Case (Fig. 5.6(a)) it can be noted that inclined orientation of the
joints within the ore surrounded by a host rock with vertical/horizontal joints had
practically no influence on caving induced subsidence, the angles delineating
major ( 10cm) surface displacements being identical. As apparent from Fig.
5.59, the total extent of major subsidence displacement for model GD5 was
changed by a maximum of 3%. Comparison of models GD6 (Fig. 5.58(b)) and J2
(Fig. 5.6(c)) shows that the assumption of vertical/horizontal joints for the ore
resulted in a decrease in rock mass fracturing damage and steepened the
delineating angles by up to 5°. According to Fig. 5.59, the total extent of
maximum major surface displacements was reduced by 9%, with larger
reductions in deformations being observed in the eastward direction.
Fig. 5.60 compares far-field surface displacements for the Base case, GD5, J2
and GD6 models. GD5 exhibited a minor increase and GD6 showed a minor
decrease in the far field displacements.
Overall, the ELFEN modelling performed showed that the host rock jointing
pattern has an important influence on caving induced surface subsidence
development.
156
(a) GD5 (b) GD6
35% 35%
50m
50% 50%
60% 60%
Legend: rotational failure; translational failure

Fig. 5.57 Subsidence crater formation for models GD5 and GD6
157
-300 -250 -200 -150 -100 -50 0 50 100 150 200 250 300
(a)
GD5
Legend:
host
rock angle
73°
90°/0° of fracture
initiation
ore 10cm displ. contours
70°/20° vertical
3
MRV = 28415m
AI = 0.91
(b)
GD6
host
rock
70°/20°
ore
90°/0°
3
MRV = 33531m
AI = 0.70 55° 79°
Fig. 5.58 Subsidence at 100% ore extraction for models GD5 and GD6
350 350 350 350


308
300 300 300 279 300

268
250 218 250 250 250
207 203 218 212
200 200 200 200
150 150 150 150
100 100 100 100
100%
100%
98%
97%
100%
100%
91%
81%
50 50 50 50
0 0 0 0
(a) BC GD5 J2 GD6 (b) BC GD5 J2 GD6
(c)
Extent of 10cm surface vertical dispacements in (d) Extent of 10cm surface horizontal displacements in
-250 -200 -150 -100 -50 0 50 100 150 200 250 -250 -200 -150 -100 -50 0 50 100 150 200 250
-112 100% BC -118 100% BC

-109 97% GD5 -118 100% GD5
-161 100% J2 -201 100% J2
-128 80% GD6 -189 94% GD6
BC 100% 95 BC 100% 100
GD5 99% 94 GD5 94% 94
J2 100% 107 J2 100% 107
GD6 84% 90 GD6 84% 90
-300 -200 -100 0 100 200 300 -300 -250 -200 -150 -100 -50 0 50 100 150 200 250 300
relation to block centre, normalized by BC/J2, % relation to block centre, normalized by BC/J2, %
Fig. 5.59 Subsidence characterization for models Base Case, J2, GD5 and GD6
158
(a) -300 -250 -200 -150 150 200 250 300
0
GD6
J2
-0.05
GD6
-0.1
-0.15
-0.2
-0.25
J2
-0.3 -0.38
-0.8
0.3
J2
(b) 0.25
GD6
0.2
0.15
GD6
0.1
J2
GD5
GD6
GD5
GD6
BC
0.05
J2
BC
J2
0
-300 -250 -200 -150 150 200 250 300
extraction at different distances from block centre for models Base Case, J2, GD5 and GD6
159
5.8 Influence of Block Depth and Excavation Volume
This section investigates the influence of depth of the extracted ore block and the
volume of the extracted ore on surface subsidence development. The default
model setup assumed a 100m×100m block located at 200m depth, and the
simulations were run up to 100% block volume extraction. Here two scenarios
are considered with the block located at 300m depth, and jointing corresponding
to Base Case and J2 models was assumed:
I. BD1 (90°/0°; 300m block depth)
II. BD2 (70°/20°; 300m block depth)
These scenarios were run up to 100 and 150% block volume extraction. In addition,
scenarios Base Case and J2 were run up to 150% block volume extraction.
Fig. 5.61 illustrates formation of a surface subsidence crater with continuous ore
extraction for models BD1 and BD2. For both models at earlier stages of ore
extraction (up to 30%), caving induced fracturing damage closely resembles an
ellipsoid, with a more slender shape observed for the model with vertical/horizontal
joints (BD1). This caving shape corresponds to the limit ellipsoid defined by
Janelid & Kvapil (1966) based on material draw studies. According to these
authors, the limit ellipsoid contains the zone of broken material that has moved and
expanded under gravity to fill the volume created by draw. Similar ellipsoids can
be delineated for the corresponding Base Case (Fig. 5.2(a)) and J2 (Fig. 5.3(a))
models with a shallower located block. Bétournay (2002) stated that complete
failure of the crown pillar should be anticipated if the draw ellipsoid intersects the
surface, this was actually the case for the studied models. As follows from Fig.
5.61, for the simulation with vertical/horizontal joints the long axis of the ellipsoid
was vertical, with rotation of the pattern by 20° skewing this axis by 7°. Continuous
ore extraction led to a limited, largely symmetrical mobilization of the near surface
rock mass for model BD1, whereas for model BD2 extended and asymmetrical
rock mass mobilization was observed. It appears that for model BD2 major rock
mass mobilization was initiated between 30-40% ore extraction when full
disintegration of the crown pillar occurred.
160
(a) BD1 (b) BD2 7°
30% ore extraction
50m 50m
0° 20°
90° 70°
40% ore extraction
50% ore extraction
60% ore extraction
70% ore extraction
Fig. 5.61 Subsidence crater formation for models BD1 and BD2
161
Fig. 5.62 shows the resultant surface subsidence for model BD1 at 100 and
150% ore extraction. The increase in surface subsidence damage due to an
additional 50% ore extraction was rather minor, the angles delineating major (≥
10cm) subsidence deformations decreasing by 2 to 5 degrees. A comparable
delineating angle can be observed when contrasting subsidence deformation at
corresponding percentages of ore extraction for the Base Case model, shown in
Figs. 5.6(a) and 5.63, with the ore block positioned 100m shallower. A similar
minor increase in subsidence due to additional ore extraction was noted for the
models with inclined joints, BD2 and J2, presented in Figs. 5.64, 5.6(c) and 5.65.
-300 -250 -200 -150 -100 -50 0 50 100 150 200 250 300
(a)
BD1
100%
0°
90° Legend:
74° angle
of fracture
initiation
3 vertical
MRV = 45887m
(b)
BD1
150%
0°
90°
3
MRV = 49966m
AI = 0.91 69° 76°
Fig. 5.62 Subsidence at 100% and 150% for BD1 model
162
(c)
BC
150%
0°
3
90° MRV = 30424m
AI = 0.91 67° 74°
Fig. 5.63 Subsidence at 150% ore extraction for Base Case model
-300 -250 -200 -150 -100 -50 0 50 100 150 200 250 300
(a)
BD2
100%
20°
70° Legend:
71° angle
of fracture
initiation
3 vertical
MRV = 68476m
(b)
BD2
150%
20°
70°
3
MRV = 70465m
AI = 0.80 55° 69°
Fig. 5.64 Subsidence at 100% (a) and 150% (b) for BD2 model
163
J2
150%
20°
70°
3
MRV = 38847m
AI = 0.76 53° 70°
Fig. 5.65 Subsidence at 150% ore extraction for J2 model
As illustrated in Fig. 5.66 for the simulations with vertical/horizontal joints, an

increase in block depth led to steeper delineating angles, where placing the block
100m deeper reduced the angles by up to 3°. Comparison of models with
inclined joints shows a more complex response. Westward from the block centre
vertical axis the delineating angle was decreased by 2°, whereas in an eastward
direction it was increased by 3°, so that the delineating angle approached the
inclination of the joints dipping into the cave.
Figs. 5.67 and 5.68, respectively, present surface profiles for models BC, BD1
and J2, BD2 at 100 and 150% ore extraction respectively. Contrasting the
maximum crater depth at 100% ore extraction for models with vertical/horizontal
and inclined joints, respectively a 37% and 50% decrease with 100m block
deepening is observed. Comparing change in the maximum crater depth with an
additional 50% ore extraction, it can be noted that for simulations with
vertical/horizontal joints a 20.3m or 36% increase in depth was observed for the
model with the ore block positioned at 200m depth (BC) and a 22.3m or 64%
increase for the block positioned at 300m depth (BD1). For the simulations with
inclined joints the corresponding changes were 32.5m or 73% (J2) and 25.6m or
116% (BD2). It can be inferred that with increase in the ore extraction volume
the subsidence crater deepens more rapidly for simulations with deeper located
blocks.
164
74° 78° BD1
0
(a) -10
300m (3H)
Lowest point
coordinates, m
71° 76° BC -20
BC 100% 0, -55
-30
-40 BC 150% 0.9, -75.3

0°
-50 BD1 100% 0.9, -34.7
200m
(2H) -60
90° BD1 150% 0.9, -57
100m
(H) -70
-80
-350 -250 -150 -50 50 150 250 350
100m (W) Distance from block centre, m
55° 71° BD2 Fig. 5.67 Surface profiles at the end of ore extraction for BC (100%),
BC (150%), BD1 (100%) and BD1 (150%) models
(b) 300m
0
53° 74° J2 -10
Lowest point
coordinates, m
-20
J2 100% 9.4, -44.5
-30
20°
200m -40 J2 150% 0, -77
70° -50 BD2 100% 10.3, -22
-60
BD2 150% 0.9, -47.6
-70
-80
Fig. 5.66 Comparison of surface profiles and angles limiting -350 -250 -150 -50 50 150 250 350
major surface deformations for models with varying block Distance from block centre, m
depth and joint inclination at 100% ore extraction
(a) BC and BD1, (b) J2 and BD2 Fig. 5.68 Surface profiles at the end of ore extraction for J2 (100%),
J2 (150%), BD2 (100%) and BD2 (150%) models
165
Comparative characterization of major subsidence deformation for models BC,
BD1 and, J2 and BD2 is given in Figs. 5.69 and 5.70, respectively. For the
simulations with vertical/horizontal joints (BC, BD1) an additional 50% ore
extraction irrespective of the block depth caused an increase of about 15% in the
total extent of the zone of major surface displacements. For simulations with
inclined joints (J2, BD2) a more moderate increase of <5% was observed. It is
interesting to note that in the direction of major asymmetry the increase in the
extent of major displacements was minimal. This indicates that the limits of rock
mass mobilization limits in the dip direction of the sub-vertical joint set were
largely reached at early stages of ore extraction. Subsidence asymmetry was in
fact decreased by about 20% due to failure developing in an eastward direction,
where inclined joints are dipping into the cave. It appears that reduction in the
muck pile support due to ore extraction induced rock mass failure through rock
bridge shearing and sliding along steeply inclined joints dipping into the cave.
Increase in block depth by 100m led to increases in the total extent of major
surface displacements of up to 15 and 34 % for models with vertical/horizontal
(BD1) and inclined (BD2) joints, respectively. Subsidence asymmetry remained
virtually unchanged for model BD1 and decreased by about 30% for model BD2.
Far-field displacements for the studied models are given in Figs. 5.71 and 5.72.
Greater block depth, as well as a larger volume of extracted material, led to an
increased zone of subsidence displacements. A sharp increase in displacement
values (more than an order of magnitude) is observed for simulations with a
300m deep block. For all considered simulations, extraction of an additional 50%
of ore led to a relatively moderate increase in surface displacements and
extended the zone of 1cm displacements by up to 50m. Surface displacements
in excess of 1cm were observed as far as 200m from the caving boundary (250m
from the block centre vertical axis) for model BD1, and 300m for model BD2.
166
350 350 350 350



291
300 300 300 300
252 242 252
250 234 228 250 250 250
207 218
200 200 200 200
150 150 150 150
100 100 100 100
115%
113%
111%
111%
100%
100%
100%
100%
50 50 50 50
0 0 0 0
BC BC BD1 BD1 BC BC BD1 BD1
(a) 150% 150% (b) 150% 150%
(c) relation to block centre, m
(d) relation to block centre, m
-250 -200 -150 -100 -50 0 50 100 150 200 250 -250 -200 -150 -100 -50 0 50 100 150 200 250
-112 100% BC -118 100% BC

-126 113% BC 150% -134 114% BC 150%
-128 100% BD1 -136 100% BD1
-136 106% BD1 150% -167 123% BD1 150%
BC 100% 95 BC 100% 100
BC 150% 114% 108 BC 150% 108% 108
BD1 100% 100 BD1 100% 116
BD1 150% 116% 116 BD1 150% 107% 124
-300 -200 -100 0 100 200 300 -300 -250 -200 -150 -100 -50 0 50 100 150 200 250 300
relation to block centre, normalized by BC / BD1, % relation to block centre, normalized by BC / BD1, %
Fig. 5.69 Subsidence characterization for models BC (100%), BC (150%), BD1 (100%)
and BD1 (150%)
350 354 383 350 350 350
412 425
normalized by J2 / BD2, %

normalized by J2 / BD2, %
292
300 300 340 300
268
250 250 330 324 250
200 200 320 200
308
150 150 310 150
100 100 300 100
109%
108%
105%
100%
100%
100%
100%
103%
50 50 290 50
0 0 280 0
J2 J2 BD2 BD2 J2 J2 BD2 BD2
(a) 150% 150% (b) 150% 150%
(c) relation to block centre, m
(d) relation to block centre, m
-250 -200 -150 -100 -50 0 50 100 150 200 250 -250 -200 -150 -100 -50 0 50 100 150 200 250
-161 100% J2 -201 100% J2

-169 105% J2 150% -201 100% J2 150%
-223 100% BD2 -258 100% BD2
-229 103% BD2 150% -260 101% BD2 150%
J2 100% 107 J2 100% 107
J2 150% 115% 123 J2 150% 115% 123
BD2 100% 131 BD2 100% 154
BD2 150% 118% 154 BD2 150% 107% 165
-300 -200 -100 0 100 200 300 -300 -250 -200 -150 -100 -50 0 50 100 150 200 250 300
relation to block centre, normalized by J2 / BD2, % relation to block centre, normalized by J2 / BD2, %
Fig. 5.70 Subsidence characterization models J2 (100%), J2 (150%), BD2 (100%) and
BD2 (150%)
167
(a) -300 -250 -200 -150 150 200 250 300
0
BD1
BD1 150%
-0.05
-0.1
-0.15
BD1 150%
-0.2
-0.25
BD1
-0.3 -0.38 -0.54
0.3
(b)
0.25
BD1 150%
0.2
BD1
BD1 150%
BD1 150%
BD1 150%
0.15
BD1 150%
BC 150%
BC 150%
BC 150%
BC 150%
0.1
BD1
BD1
BD1
BD1
BC
0.05
BC
0
-300 -250 -200 -150 150 200 250 300
extraction at different distances from block centre for BC (100%), BC (150%), BD1 (100%)
and BD1 (150%) models

(a) -300 -250 -200 -150 150 200 250 300
0
J2
J2 150%
BD2
BD2 150%
-0.05
-0.1
-0.15
BD2 150%
BD2 150%
-0.2
J2 150%
BD2
-0.25
BD2
J2
-0.3 -0.6 -0.4 -1.8

-0.6 -1.3 -0.8 -1.8 -0.6 -6.4
0.3
BD2
J2
J2 150%
BD2 150%
BD2
(b)
BD2 150%
0.25
BD2 150%
0.2
BD2 150%
BD2 150%
J2 150%
BD2 150%
BD2
0.15
J2 150%
J2 150%
J2 150%
J2 150%
BD2
0.1
J2
BD2
BD2
0.05
J2
J2
0
-300 -250 -200 -150 150 200 250 300
extraction at different distances from block centre for J2 (100%), J2 (150%), BD2 (100%)
and BD2 (150%) models
168
Overall, the conducted ELFEN modelling showed that the depth of the ore block
and the volume of extracted material may influence caving induced surface
subsidence development. The assumed increase in ore block depth and higher
volume of extracted material generally led to larger subsidence deformations. The
key modelling observations can be summarized as follows:
1. An increase in the block depth from 2H to 3H did not affect the joint
orientation controlled mechanism of subsidence deformations development.
2. Simulation with vertical/horizontal joints exhibited a moderate (<0.35H)

increase in the total extent of the zone of major (≥10 cm) surface
displacements, whereas for the case with inclined joints the increase in the
total extent of major subsidence displacements was significant (> 0.5H).
3. For both assumed jointing conditions, excavation of a deeper block led to

a significant increase in extent and magnitudes of far-field surface
displacements.
4. Extraction of an additional 50% of ore block volume did not result in any
major increase in the subsidence damage, leading to relatively moderate
changes in the total extent of major subsidence deformation, with larger
accumulated damage for the case with vertical/horizontal joints.
5.9 Results Synthesis

The preceding sections presented the results of the conceptual modelling study and
outlined key modelling observations and conclusions with respect to the influence of
the individual factors. Here results of the entire conceptual study are synthesized.
Comparative summary of all conceptual simulations is given in Figs. 5.73 to 5.76,

which compare, accordingly, the total extent of major (≥10cm) subsidence
displacements, extent of major surface displacements with respect to the caving
boundaries, minimum angle of fracture initiation and asymmetry index, and
mobilized rock mass volume. A preliminary statistical summary of conceptual
modelling results is given in Appendix A.
169
Based on analysis of the conceptual modelling results a preliminary classification
of block caving induced surface subsidence for cases, where no major geological
discontinuities intersect the cave, is proposed, as shown in Table 5.2. This
classification is supplemented by a preliminary classification of the influence of
major geological discontinuities on surface subsidence, given in Table 5.8. It
should be emphasized that these classifications are based on the modelling that
assumed a rock mass corresponding to ~ MRMR 50-60, uniform ore extraction
and block depth 2H (where H is block height). The relative significance of
individual parameters is summarized in the influence assessment matrix, shown
in Table 5.9. Recognizing the fact that this matrix was developed based on
simulated variations of specific parameters it is believed that it captures overall
behavioural trends.
The proposed preliminary subsidence assessment classifications and the

parametric influence matrix are meant to aid practical engineering assessment of
potential block caving subsidence at pre-feasibility and design stages. It cannot
be over-emphasized that such assessments must be tailored to and take full
consideration of specific mine site conditions.
170
Fig. 5.73 Comparative summary of modelling results - total extent of major (≥10cm) surface displacements
171
Fig. 5.74 Comparative
summary of modelling
results - major (≥10cm)
asymmetry with respect to
the caving boundaries
172
Fig. 5.75 Comparative summary of modelling results - minimum angles delineating the extent of major (≥10cm) surface displacements,
asymmetry index
173
Fig. 5.76 Comparative summary of modelling results – volume of the rock mass mobilized by the caving
174
Table 5.8. Preliminary classification of caving induced surface subsidence for cases with no major geological discontinuities
Subsidence Typical subsidence deformations Associated Min Asymm. Disturbed rock
category rock mass conditions angle of index mass volume in
fracture % of ore
initiation extracted
I. Moderate < 2.5H Primary controls: > 70° 0.9 -1 < 220
intact combination of
rock mass
highly
vertical/sub-vertical and
disturbed to 2H horizontal/sub-horizontal
disturbed
rubblized persistent joint sets,
rock mass
rock mass stronger host rock
Contributing factors:
W=H
low horizontal stress
II. Significant < 3H Primary controls: 55° - 70° 0.7 - 0.9 220 - 250
combination of steeply
and low dipping
persistent joint sets,
higher joints stiffness
III. Extensive > 3.5H Primary controls: < 55° < 0.7 >250
combination of steeply
and low dipping highly
persistent joint sets
high horizontal stress
175
Table 5.9. Preliminary classification of the influence of major geological discontinuities on caving induced surface subsidence
Degree of Typical subsidence deformations Description

influence
I. Low to intact I(a) fault located at distances exceeding 0.5H from the
Moderate rock mass
highly
caving boundary
disturbed to 2H fault may act as a displacement barrier limiting rock mass
rubblized
fault rock mass disturbed movements in far-field
rock mass
W=H
I(a) I(b) more than 2/3 of the fault near surface segment is
located within caving zone
fault may affect caving mechanism
fault
I(b)
II. Significant to II(a) steeply inclined (80 - 60) faults intersecting caving
Extensive boundary
fault
II(b) moderately inclined (60 - 30) faults intersecting caving
II(a) boundary
major in both cases extent of surface subsidence and subsidence

block
asymmetry will be governed by fault inclination
fault
II(b)
176
Table 5.10. Initial influence assessment matrix of factors contributing to block caving induced surface subsidence
Factors Specific conditions Potential degree of influence (”+” increase, “-“ decrease) Potential worst case
combinations
extent of major footprint asymmetry extent of far-field
disturbances
1. Rock mass 1(a) steeply dipping persistent sets High (+) High (+) High (+) 3(a), 4(b), 6(a)
jointing
1(b) vertical/sub-vertical persist. sets High (-) High (-) High (-)
1 1 1
2. Major 2(a) faults intersecting caving High (+/-) High (+/-) High (+/-) 3(a), 4(b), 6(a)
geological boundary
discontinuities 1 1 1
2(b) faults located in immediate vicinity Moderate (+/-) Moderate (+/-) Moderate (+/-) 3(a), 4(b), 6(a)
from the caving boundary
1 1 1
2(c) faults located at distances Low (+/-) Low (+/-) Low (+/-)
exceeding half of block height
3. Rock mass 3(a) lower strength Moderate (+) Low Low 1(a), 6(a)
tensile strength
3(b) higher strength Moderate (-) Low Low
4. Stress level 4(a) K<1 Low (-) Low (-) Moderate (-)
4(b) K>1 Low (+) Low (+) High (+) 1(a), 6(a)
5. Varying 5(a) stronger caprock

2 2 2
lithological Low to High (-) Low to High (-) Low to Moderate (-)
domains 5(b) stronger host rock
2
6. Block depth 6(a) block depth ≤2H (block height) Moderate (+) Low (+) High (-) 1(a), 2(a), 2(b), 3(a),
2
4(a)
6(b) block depth >2H Moderate (-) Low (-) High (+)
7. Extraction 7(a) additional 50% Low (+) Low (+) Low (+) 2(a), 3(a)
volume
1 2
function of fault inclination; function of prevailing joint sets inclination
177
5.10 Summary
The conducted modelling demonstrated the significant potential benefit of the
proposed FEM/DEM-DFN approach to block caving subsidence analysis. A series
of numerical experiments highlighted the importance of joint set orientation and
persistence, fault location and inclination in determining the subsidence
development mechanisms and their governing role in defining the degree of
surface subsidence asymmetry. The modelling results correlated reasonably
well with the available field observations. New insights were gained into the
complex mechanism of caving induced rock mass deformations and subsequent
subsidence development. This allowed development of preliminary block caving
subsidence classifications and a parametric influence assessment matrix aimed
to aid practical engineering analysis.
178
CHAPTER 6: BLOCK CAVING INDUCED INSTABILITY IN
LARGE OPEN PIT SLOPES
6.1 Introduction
Low cost and high efficiency are making block caving an attractive option for the
continuation of mining activities at large open pit operations that otherwise are
approaching their economic limits. A few such projects have been implemented
and many more are being planned, including, but not limited to, the world largest
open pit mines, Bingham Canyon (USA) and Chuiquicamata (Chile). Recent
implementation of block caving at a large open pit mine at Palabora (South
Africa) illustrates that transition projects can be successfully carried out and
achieve targeted ore output. At the same time however, this mine encountered a
series of complex geotechnical issues, including major surface subsidence
accompanied by a massive failure of the North pit wall. This development
highlighted the need for a better understanding of the complex response of pit
slopes to caving. This chapter examines the mechanisms leading to block
caving induced failure of large open pit slopes, focusing on step-path driven
failure and also presents preliminary FEM/DEM-DFN based modelling of the
Palabora mine failure.
6.2 Transition from Open Pit to Block Cave Mining

Stacey & Terbrugge (2000) noted that in an optimum open pit operation, pit
slopes will have been designed to be very close to the limit of their stability.
Therefore the transition operation should be designed with full consideration of
the high sensitivity of large open pit slopes to caving induced disturbances.
These authors also reported a case study where pit slope design was flattened
by 5 degrees to account for a future block cave. From the open pit mining
perspective, often, the original planning did not consider future underground
179
mining, leaving little scope for steepening of slopes to avoid impact to mining
infrastructure, particularly where located in the vicinity of the pit rim. At the same
time, caving operations require high upfront investments and are very inflexible
once commenced. Considering these circumstances reliable assessment of the
interaction between the developing cave and the existing open pit becomes
important for successful adaptation of block cave mining.
According to Eberhardt et al. (2007) the rock engineering interactions involved

with transition projects are highly complex. On surface, pit wall slopes frequently
exceed heights of several hundred metres and the potential for deep-seated,
stress-controlled rock slope failures is becoming more of an issue compared to
bench-scale, structurally-controlled wedge failures. Block caving by its very
design results in an almost immediate response of the rock mass leading to
deformation and surface subsidence. Beck & Pfitzner (2008) emphasized the
forecasting and characterization of the underground - slope interaction as one of
the most challenging tasks in rock mechanics. Unfortunately, to the author‟s
knowledge, slope stability issues related to open pit - caving interaction have to
date received relatively limited attention in the published literature.
As discussed in Section 5.2.3.2, Flores & Karzulovic (2004) performed a detailed

analysis of the subsidence associated with open pit caving interaction, using the
continuum code FLAC2D and a limit equilibrium technique. Eberhardt et al.
(2007) compared application of FLAC2D and UDEC for the analysis of block
caving induced slope deformations. The results demonstrated that both the
magnitude and shape of the subsidence profile modelled can vary as a function
of modelling approach (continuum vs. discontinuum), constitutive model (elastic
vs. elasto-plastic), and geometry of the discontinuity network. The authors
indicated that one significant limitation of conventional continuum and
discontinuum numerical analyses is their inability to explicitly account for brittle
fracture processes, and their subsequent role in underground-surface mine
interactions.
180
Beck & Pfitzner (2008) suggest that the interaction between the developing cave
and the existing mine operation during cave propagation, breakthrough and draw
down need to be simulated so that the transition can be properly planned, and so
that the risks and effects of the new block caves can be properly appreciated.
These authors provide example applications of the three dimensional continuum
code ABAQUS (Simulia, 2007) to the analysis of interaction between open pit
and block cave and two neighbouring block caves. They proposed a set of
milestones to assess caving induced interaction and employed dissipated plastic
energy and plastic strain as interaction indicators.
Elmo et al. (2007b, 2008) adopted a FEM/DEM-DFN modelling methodology for

open pit - block cave interaction modelling. A series of conceptual models were
run investigating the effect of joint orientation, stress ratio and rock mass strength
on caving induced slope instability. It was found that the joint orientation may
have a defining role in caving induced slope failure. This agrees with the findings
of Salim & Stacey (2006), whose numerical modelling study has shown that
variability in the geometry of the jointing can have a major effect on the slope
behaviour, and on geometry and extent of the volume of collapse. It should be
recognized that use of large scale continuum modelling where jointing can only
be accounted for implicitly may not be applicable in all cases. Research by Elmo
et al. (2007b, 2008) illustrated that use of the FEM/DEM-DFN methodology may
provide valuable insights into complex interaction behaviour.
6.3 Characteristic Slope Failure Mechanisms in Large Open Pits

According to Franz et al. (2007) large scale rock slope failure mechanisms are not
completely understood, and may often comprise a number of different mechanisms.
As summarized by Baczynski (2000) high rock slope failures may include:
sliding on one or more major geological discontinuities (planar,
tetrahedral, active-passive wedges);
sliding along circular or quasi-circular failure paths through a highly
fractured or weak rock mass, or across rock mass fabric (rotational
failure);
181
toppling; and
composite modes, involving two or more of the above mechanisms.
Piteau & Martin (1982) stated that failure mechanisms are easiest to assess if they
can be represented two dimensionally, so that the failure surface is assumed to be
sub-parallel to the strike of the slope allowing analysis to be carried out for a unit
width of the slope. Two dimensional failure assumes the presence of lateral release
surfaces which do not provide any resistance to failure. Two-dimensional failures
which involve a single discontinuity or a single set of discontinuities are planar
failures and toppling. Rock mass failure encompassing rotational shear failure
develops when the slope is sufficiently high and steep so that the shear stress of the
rock mass is exceeded due to high stresses in the slope. It should be emphasized
that this type of failure is more characteristic of weaker materials and development
of a circular failure in strong rock is uncommon. Wedge failure is a three
dimensional phenomenon that develops when two intersecting discontinuities form a
tetrahedral block which can slide out of the rock slope. Stacey (2007) indicated that
failure mechanisms in high, hard rock slopes are much more complex than planar,
wedge and circular shear failure surface, and toppling. Progressive failure in hard
rock slopes involves initiation and progression of failure along existing weakness
planes, and initiation and progression of failure in intact rock, i.e. step-path failure
mechanism. The mechanisms described above are illustrated in Fig. 6.1. Additional
information on planar, toppling, sliding and wedge failure mechanisms in rock slopes
can be found in Wyllie & Mah (2004).
This chapter will focus on step-path failure where the contribution of intact rock
bridges to slope stability can be quite significant. As reported by Piteau & Martin
(1982) analysis carried out by Martin (1978) has shown that for slopes greater than
300m high in moderately hard rock, the occurrence of less than 10% rock bridges
(vertical spacing of rock bridges as a % of total discontinuity length) along a
prospective failure surface would provide enough resistance to shear failure to
achieve limit equilibrium.
182
Planar failure Toppling failure
Rotational shear Wedge failure

failure
Step-path failure
Fig. 6.1 Slope failure modes (based on Sjöberg, 1999)
Jennings (1970) pioneered detailed step-path analyses of rock slopes with

development of a limit equilibrium approach that incorporated shear failure along
joints, shear through intact rock and tensile failure of rock bridges. To consider
continuity of jointing, coefficients of continuity and discontinuity were introduced.
The Factor of Safety (FS) was calculated based on an apparent cohesion and
friction angle on an assumed mean failure plane. Jennings (1970) also analyzed the
various modes of step-path development from upper to lower joints and the
existence of both stepped surfaces in the dip and strike directions. The work by
Jennings was further extended by Jaeger (1971). Fig. 6.2 compares the limit
equilibrium solutions for planar step path failure.
183
where: A - surface area of failure; W - sliding rock mass weight;
α - slope angle of failure; c and υ - cohesion and friction
along failure surface; Fτ and Fn - tangential and normal
forces
where: k - coefficient of continuity; ci and υi - cohesion and friction

of intact rock; cj and υj - joint cohesion and friction
where: To - intact tensile strength; β - angle of discontinuity in

step-path
Fig. 6.2 Comparison of limit equilibrium analysis of planar and step-path failure
(modified after Eberhardt et al., 2004a, with permission)
Among recent developments in limit-equilibrium based solutions of step-path failure,

work by Baczynski (2000) is of particular interest. This author, based on the
research of McMahon (1979) and Read & Lye (1984) developed the STEPSIM4
code that can evaluate step-path development using probabilistic analysis. The
STEPSIM4 “step-path” method provides an avenue for assessing the statistical
shear strength along potential two-dimensional failure paths through rock masses.
A potential step-path failure surface is evaluated along a system of ground condition
“cells”. Where each “cell” is statistically associated with failure through one of
several failure modes including sliding along adverse joints, stepping up along
steeper dipping joints and direct shear through intact rock bridges. Repeating the
simulation for a large number of potential failure paths (> 2000), STEPSIM4
184
provides a statistical distribution of shear strength along critical step paths. This tool
provides a valuable and logical approach, however, it does not specifically consider
the increased importance of the stress field and deformation processes in high rock
slopes (Franz et al., 2007). Moreover, it does not consider explicitly intact rock
fracture mechanisms.
According to Stead et al. (2007) the role of brittle fracture modelling in rock slope
instability both in engineered and natural slopes is the subject of considerable on-
going research. Impetus for this work was originally derived from the failure of high
mountain slopes, however the increasing number of large open pits with projected
depths of 1 km or more has become a major driver for understanding intact rock
fracture in rock slope environments.
Recognizing the significance of the diversity of roles and scale of brittle fracture
within rock slopes, Stead et al. (2007) proposed a classification of brittle fracture
processes in rock slopes in terms of primary, secondary and tertiary processes:
Primary brittle fracture processes occur prior to onset of failure and
include (i) propagation of failure surfaces through fracture tip growth, (ii)
coalescence of fractures and failure of intact rock bridges and (iii) shearing
along discontinuities involving removal of asperities.
Secondary brittle fracture processes occur following the onset of failure
and involve (i) development of rear and lateral release surfaces leading
toward global slope failure and (ii) internal deformation, fracturing and
dilation of the rock slope mass associated with translational failure,
toppling or multiple complex interacting mechanisms.
Tertiary brittle fracture processes are associated with the final stages of
slope failure involved the comminution of the rock mass associated with
transport leading up to debris deposition.
Stead et al. (2007) emphasized the particular importance of simulation of the first
two processes for large scale failures. Simulation of rock comminution may involve
significant run-times and therefore may not be practical in all cases.
185
Numerical modelling studies that addressed primary and in some cases
secondary and tertiary brittle fracture processes in slopes include Scavia (1990,
1995), Scavia & Castelli (1996), Muller & Martel (2000), Stacey et al. (2003),
Wang et al. (2003), Eberhardt et al. (2004a,b), Stacey (2006), Stead & Coggan
(2006), Elmo et al. (2007b), Karami et al. (2007), Franz et al. (2007), Eberhardt
(2008) and Yan (2008).
Stead et al. (2007) reviewed the state-of-the art in brittle fracture modelling as
applied to both large natural and open pit slopes and illustrated several example
applications of fracture mechanics based FEM/DEM modelling, including a pit slope
proximity problem, as shown in Fig. 6.3. It has been shown that FEM/DEM
modelling may provide valuable insight into slope failure mechanisms. A particular
importance of the integration of FEM/DEM and DFN was emphasized.
Fig. 6.3 The pit-slope proximity problem: preliminary brittle fracture analysis (adapted after
Stead et al., 2007, with permission)
186
6.4 Conceptual Study of Block Caving Induced Step-path Driven
Failure in Large Open Pit Slope
6.4.1 Modelling Methodology
To investigate step-path development mechanisms during block caving - open pit

interaction a series of conceptual ELFEN models were run.
As illustrated in Fig. 6.4, a simplified 2D model geometry of a 750m deep open pit
with 50o slopes and the caving operation located 400m beneath the pit was adopted.
It was assumed that the rock mass fabric forms a potential failure surface consisting
of non-coplanar step path joints dipping into the cave and intact rock bridges with a
vertical spacing of 15m. Cave mining induced development of step-path failure was
analyzed. For the purpose of the present analysis, fracturing was allowed only within
the areas of rock bridges and in the cave itself. A fine mesh was adopted for the
fracturing regions, i.e. 2m within the cave and 0.7m for the rock bridges. Modelling
adopts the same ore extraction methodology as described in Section 4.5.1.1 and
employs the calibrated equivalent continuum properties, same as in Chapter 5, for
the cave and intact rock properties for the rock bridges. GSI based equivalent
continuum properties were adopted for area adjacent to the caving footprint, e.g.
open pit slopes. Modelling input parameters are presented in Table 6.1.
Conceptual modelling was focused on the analysis of the effect of varying:

rock bridge strength;
joint cohesion;
number of rock bridges.
Table 6.2 shows the model runs undertaken. To capture the full picture of the
interaction mechanisms a “total interaction” analysis was adopted relating cave
propagation, stress redistribution in the crown pillar with step-path failure
development and surface subsidence. History points (see Fig. 6.4) were placed at:
the footwall and hanging wall edges of the surface outcrop of the
discontinuities, where differential XY displacements (similar to Fig. 5.29)
were monitored;
187
the centres of rock bridges, where variation of shear stress was tracked;
50m below the pit bottom, where variation of vertical stress was tracked.
Equivalent
continuum
60o
2200m
4000m
75m Fracturing
regions history point
10 excavation stages History point
RB600
750m
RB300 RB300 – rock

bridge located
60o 50o 300m away
from the cave
footprint
400m
300m 300m
Fig. 6.4 Typical model geometry for simulation of block caving induced step-path failure
(cases with two rock bridges shown).
188
Table 6.1 Modelling input parameters used in conceptual modelling
Value
Parameter Unit Open pit slopes1 Intact rock
Ore
GSI 70 bridge
Rock properties
Young‟s Modulus, E GPa 18 44 60
Poisson‟s ratio, 0.25 0.25 0.25
-3
Density, ρ kgm 2600 2600 2600
Tensile strength, t MPa 1 0.88 10
-2
Fracture energy, Gf Jm 60 60 60
Cohesion, ci MPa 4.7 6.6 20
Friction, i degrees 45 45 50
Dilation, ψ degrees 5 5 5
Discontinuities
Fracture cohesion, cf MPa 0
Fracture friction, f degrees 35
Normal penalty, Pn GPa/m 2
Tangential penalty, Pt GPa/m 0.2
Stress level
In-situ stress ratio, K 1
1
GSI based properties were established using the RocLab v1.031 program (Rocscience Inc., 2007), assuming
mi=15, D=0 and intact rock properties σci=127MPa, Ei=60GPa
Table 6.2 Modelling scenarios
Number of rock Step-path

Rock bridges tensile
Scenario bridges / rock discontinuities
strength, MPa
bridges % cohesion, MPa
M1 2/4 10 0
M2 2/4 15 0
M3 2/4 10 0.5
M4 3/6 10 0
M5 4/8 10 0
189
6.4.2 Modelling Results
Block caving and associated development of step-path failure in the open pit for
model M1 is illustrated in Fig. 6.5. The open pit is stable prior to caving, after which
progressive caving initiates stress redistribution within the slope hence triggering
failure of the rock bridges. The failure is initiated at the rock bridge RB600 located
furthest from the caving footprint and with some delay steps through the second
rock bridge RB300. Fig. 6.6 shows the concentration of the tensile stresses in the
rock bridges prior to fracturing and Fig. 6.7 illustrates characteristic development of
fracturing within the rock bridges. The rock bridges fail in shear, where initially the
formation of en-echelon fracturing is observed parallel to the orientation of the
major principal stress, followed by the coalescence of these fractures to form a
shear failure plane. A combined stress/displacement analysis of caving/open pit
interaction is shown in Fig. 6.8. The lower graph describes the rock mass response
below the pit bottom and shows cave propagation expressed in crown pillar
thickness and the change in vertical stress with continuous caving at a history point
located 50m below the pit bottom. The upper graph illustrates the response in the
open pit slope, showing change in shear stress in the rock bridges (normalized by
the value at the end of pit excavation before caving is initiated) and differential
displacements at the surface outcrop.
Fig. 6.8 clearly shows that the surface subsidence in the open pit wall is directly
related to caving. When the thickness of the crown pillar is reduced down to
approximately 175m, a rapid destressing in the crown pillar is initiated; this causes an
unloading of the pit slope toe allowing movements within the slope, which trigger
failure of the RB600 rock bridge. Interestingly, this rock bridge fails at a very low level
of destressing, i.e at about a 3% vertical stress decrease (in relation to the stress
level at the end of pit excavation). It can clearly be seen that on failure of the rock
bridge RB600 the shear stress in the RB300 rock bridge rapidly increases and this
rock bridge then fails at about 16% vertical stress decrease in the crown pillar.
Failure of the RB600 rock bridge is associated with a distinctive drop in the
differential displacements at the location of the surface outcrop of step-path
discontinuities. Failure of the lower rock bridge occurs when accumulated differential
190
displacements reach about 3cm. It should be noted that the slope of the surface
displacement follows quite closely the variation in the crown pillar vertical stress.
end of pit excavation progressive caving
RB600 failure RB300 failure
RB600
RB300
remaining crown
pillar thickness
Fig. 6.5 Block caving induced step-path failure in large open pit slope (model M1)
191
Fig. 6.6 Maximum principal stress contours (Pa) - tensile stress concentrations (red) in rock
bridges prior to failure
Fig. 6.7 Typical rock bridge failure development
192
ΔXY displ. at surface outcrop, m
15 0
Norm. shear stress XY, MPa

RB600
10 -0.05
RB300
5 differential XY displ. at -0.1

surface outcrop
0 -0.15
20 22 24 26 28 30 32 34 36
-5 simulation time, num.sec -0.2

0 0
Crown pillar thickness, m

Vertical stress YY, MPa
50
end of pit excavation
σyy (50m below pit bottom)

100
-5 crown pillar thickness, m
150
200
250
-10
300
350
-15 400
Fig. 6.8 Stress/displacement analysis of caving - open pit slope interaction (rock bridges
failure, model M1)
Model M2, in which the tensile strength of the rock bridges was increased by
50% shows generally a similar step-path failure development as in Model M1,
Fig. 6.5. Comparing Figs. 6.8 and 6.9, it should be noted that increasing the rock
bridge tensile strength did not result in a significant change in the simulated
response, showing only slightly more brittle behaviour, i.e. the failure of the upper
and lower rock bridge was initiated at about 2 and 12% of crown pillar
destressing, respectively. It appears that for the given step-path arrangement of
joints the magnitude of the tensile strength increase was not sufficient to alter the
overall failure process. Contrasting models M1 and M3 (Figs. 6.8 and 6.10) it is
evident that even assumption of a moderate cohesion of 0.5MPa along the
discontinuity surfaces can affect the step-path failure development quite significantly.
Failure of the upper and lower rock bridges occurs nearly simultaneously at about
18% crown pillar destressing. It appears that cohesion on the discontinuities
provides a load distribution link between the rock bridges and therefore they react to
the slope unloading more as a system rather than individually.
193
15 0.00

RB600
10 -0.05
RB300
5 differential XY displ. at -0.10

surface outcrop
0 -0.15
20 22 24 26 28 30 32 34 36

0 0

50
σyy (50m below pit bottom) 100
-5 crown pillar thickness, m 150
200
250
-10
300
350
-15 400
Fig. 6.9 Stress/displacement analysis of caving - open pit slope interaction (rock bridge
failure, model M2)

15 0.00
RB600
10 -0.05
RB300
5 differential XY displ. at surface -0.10

outcrop
0 -0.15
20 22 24 26 28 30 32 34 36

0 0
50

-5
crown pillar thickness, m 150
200
250
-10
300
350
-15 400
Fig. 6.10 Stress/displacement analysis of caving - open pit slope interaction (rock bridge
failure, model M3)
194
Models M4 and M5 assumed three and four rock bridges, respectively, as shown
in Fig. 6.11 and 6.12. For model M4, failure initiated in the upper portion of the
slope where a single rock bridge (RB600) provided limited shearing resistance,
then the failure subsequently stepped through the rock bridges located in the
lower portion of the slope (RB300 and RB150). For model M5 failure started in
the middle of the slope (RB300), stepped down to the lowest rock bridge (RB150)
and stepped up along the upper rock bridges (RB450 and RB600). It appears
that such complex step-path development patterns can be explained as follows:
the RB300 rock bridge is experiencing the highest concentrated loading from the
portion of the slope undercut by the step-path joints, and given that the shearing
resistance is nearly evenly distributed between the upper and lower portions of
the slope, the failure must be initiated at the rock bridge experiencing the
maximum loading. Subsequent failure of the lower rock bridge is related to
continuous toe unloading and hence reduced shearing resistance in the lower
portion of the slope. The upper portion of the slope is then effectively pulled
downwards by the weight of the failing slope.
As follows from Fig. 6.13, the step-path failure for simulation M4 initiates at 87%
and fully develops by 90% of crown pillar destressing, RB600 and RB300 rock
bridges fail nearly simultaneously, and the RB150 rock bridge fails with some
delay. The latter is associated with accumulated differential displacements of the
sliding block at surface of about 5cm and the rapid build up of shear stresses
within the rock bridge. According to Fig. 6.14, the step-path failure for model M5,
with four rock bridges, was initiated at 95% crown pillar destressing and
continued crown pillar collapse (see Fig. 6.12). For this model no substantial
surface subsidence was observed prior to the failure of the last rock bridge
(RB600).
195
RB600
RB300
RB150
Fig. 6.11 Block caving induced rock step-path failure in large open pit slope (model M4)
196
RB300
RB150
RB600
RB450
Fig. 6.12 Block caving induced step-path failure in large open pit slope (model M5)
197
15 0.00

RB600 RB150
10 -0.05
RB300
RB150
5 -0.10
differential XY displ. at
surface outcrop
RB300
0 -0.15
20 22 24 26 28 30 32 34 36

0 0

50
100
σyy (50m below pit bottom)
-5
200
250
-10
RB600 failure
RB300 failure
RB150 failure
300
350
-15 400
failure, model M4)

15 0
RB600
RB450
10 -0.05
RB300 RB150
RB150 RB300 RB450
5 -0.1
differential XY displ.
at surface outcrop RB600
0 -0.15
20 22 24 26 28 30 32 34 36
-5 simulation time, num. sec -0.2

0 0
50

-5
200
250
RB300 failure
RB150 failure
RB450 failure
RB600 failure
-10
300
350
-15 400
failure, model M5)
198
Fig. 6.15 summarizes the interrelationship between block caving, expressed in
reduction of thickness and destressing of the crown pillar, and the step-path failure
response (expressed as failure of the first and last rock bridges) based on models
with different percentages of rock bridges: 4 (M1), 6 (M4) and 8% (M5). The
percentage of the rock bridges relates the sum of the rock bridges spacing to the
total length of step-path forming discontinuities. It is evident that with increase in
percentage of the rock bridges along the step-path failure surface the degree of
crown pillar destressing needed to mobilize the failure increases. It appears that
the simulation with 8% rock bridges approaches the limiting equilibrium condition,
the failure development becomes less sensitive to crown pillar behaviour and is not
fully realized until complete crown pillar collapse. This agrees well with the
equilibrium threshold of 10% proposed by Martin (1978), as discussed in Section
6.3.
100 0
crown pillar:
destressing, %
Crown pillar destressing, %

Remaining crown pillar
80 thickness, % -20
step-path failure:
thickness , %
60 first rock bridge failure -40

last rock bridge failure
40 -60
20 -80
0 -100
2 4 6 8 10
% rock bridges
Fig. 6.15 Development of step-path failure in the open pit slope during caving mining
with relation to crown pillar geometry and stress level for simulations with different % of
rock bridges in step-path failure surface
Fig. 6.16, which compares variation of vertical stress in the crown pillar (at 50m
depth below pit bottom) for studied scenarios, indicates that there is an
interrelation between the stress level in the crown pillar and the number of rock
199
bridges in the step-path failure surface. Simulation with more than two rock
bridges (M4, M5) exhibit lower stress levels in the crown pillar at the end of the pit
excavation, as well as a quite different stress unloading behaviour during caving.
Generally, a larger number of rock bridges is associated with lower stresses in the
crown pillar. Fewer rock bridges result in more rapid stress changes during
unloading. It appears that higher shearing resistance in the slope related to a
higher number of rock bridges reduces the active pressure of the slope onto the
crown pillar and vice versa. This may have implications on the manner of crown
pillar collapse. A weaker slope may impose higher stress in the crown pillar which
may in turn delay the cave propagation and therefore increase a risk of rapid
crown pillar collapse. Evidently the mechanisms of large open pit slope - caving
interactions are highly complex. The open pit rock mass competency may
influence the crown pillar response and affect cave propagation behaviour and in
turn the caving induced unloading of the open pit influences open pit slope
stability.
Simulation time, num.sec

0
20 22 24 26 28 30 32 34 36
M1 M2 M3 four rock
-5 bridges
M4 M5 two rock bridges
three rock
bridges
-10
-15
Fig. 6.16 Variation of vertical stress in the crown pillar (50m below pit bottom) for models
M1-M5
6.4.3 Conclusions
Conceptual ELFEN modelling has illustrated the potential application of

FEM/DEM modelling to the analysis of primary fracture processes within an open
pit slope during block caving. The proposed “total interaction” analysis approach
has allowed an improved understanding of the interaction mechanisms and the
200
establishment of correlations between percentage of rock bridges within the step-
path failure surface and caving induced deformations. The following section
applies a combined FEM/DEM-DFN methodology to the analysis of primary and
secondary brittle fracture processes associated with the block caving induced
open pit wall failure at Palabora mine.
6.5 Preliminary Modelling of Block Caving Induced Failure of

the North Wall, Palabora mine
6.5.1 Problem Description
6.5.1.1 Background Information
Palabora mine, located in Limpopo province of South Africa is one of the

steepest and deepest large open pits in the world. Open pit mining at Palabora
commenced in 1966 at a rate of 30,000 tonnes per day (tpd) increasing to 82,000
tpd prior to closure in 2002. In total about 960 Mt of ore and 1,300 Mt of waste
were mined. Surface dimensions of the oval shaped open pit are near 1650 m in
the north-south direction and about 1950 m in the east-west direction. The pit is
approximately 800 m deep with interramp slope angles ranging from 37° in the
upper weathered lithologies to about 58° in the competent constrained ground
toward the base of the pit (Moss et al., 2006; Piteau Associates, 2005).
In 1996 a feasibility study was completed for a block cave targeting a block
height of around 500m and an ore reserve in excess of 220 Mt of carbonatite ore
at 0.7% copper. Target production was 30,000 tpd which translates into a life of
mine of about 23 years (Pretorius & Ngidi, 2008). Upon completion of major
open pit operations, a cave with a footprint of 150 to 300 m north-south and
about 700 m east-west was initiated approximately 400 m below the pit floor, as
shown in Fig. 6.17. For a short period of time, simultaneous open pit scavenging
and caving mining activities were in place (Piteau Associates, 2005).
201
Fig. 6.17 3D view of Palabora pit and cave mine (adapted after Brummer et al., 2005, with
permission)
6.5.1.2 Geological Settings
The Phalaborwa Igneous Complex, 8 km long and 3.2 km wide, consists of a

succession of subvertical pipe-like bodies of alkaline and ultramafic rocks which
have intruded the surrounding Archean granite. The Palabora copper orebody
occurs in the Loolekop pipe near the center of the complex. In plan view it
represents an elliptical zone some 1.4 km long and 0.8 km wide elongated in an
east-west direction. This pipe is a composite vertical intrusion with an elliptical
interbanded configuration in which the component rock types were emplaced in
the pyroxenite host. Micaceous pyroxenite, the first intrusion, was in turn
intruded by foskorite and banded carbonatite. Late stage fracturing and forceful
intrusion resulted in emplacement of a transgressive carbonatite body at the
centre of the pipe. Most copper, iron and other ore mineralization occurs within
the foskorite and carbonatite (Piteau Associates, 1980). Detailed descriptions of
the regional and local geology can be found in Hanekom et al. (1965).
Fig. 6.18 illustrates the geological units encompassed by the Palabora open pit
boundaries as well as the pit slope geometry. A brief description of the rock units
present in the North wall is given in Table 6.3.
202
Table 6.3 Description of rock units present in the North Wall
(based on FLUOR, 1994; Piteau Associates, 2005)
UCS,
Rock type Description RMR
MPa
Carbonatite Forms the lower benches of the pit wall, 127 61

primarily comprised of magnesium calcite with
variable amount of magnetite and accessory
minerals
Foskorite Positioned in the lower middle part of the pit 90 56

wall and comprised of serpentinized olivine,
magnetite, apatite and some phlogopite
Micaceous Occupies the upper half of the pit wall, and 86 59

pyroxenite comprises mainly of diasporite, phlogopite and
accessory minerals
Structural geology at the site comprises a number of sub-vertical dolerite dykes

trending north-easterly and intersecting the complex and four major faults, as
shown in Fig. 6.19. As reported by Piteau Associates (2005) the detailed
mapping carried out at the site found eight pronounced discontinuity sets in the
pit, of which three steeply dipping sets with dip/dip direction 80o/320o, 82o/270o
and 85o/020o are present throughout the pit. In the upper portion of the pit wall a
more representative orientation of the 85o/020o set is 80o/225o (Piteau Associates,
2005). Mapping data presented by Martin et al. (1986) indicates the presence of a
sub-horizontal set 014o/07o in carbonatite, foskorite and micaceous pyroxenite.
Piteau Associates (1980) showed that the joint sets at Palabora are reasonably
consistent with depth over the mapped area and noted that joints in foskorite and
pyroxenite, which are located at increasing distance from the centre of the
orebody, have more diffuse populations and lower intensities for the peak values
compared to the carbonatite rocks.
203
Fig. 6.18 General geology and pit slope geometry at Palabora mine (adapted after Moss et
al., 2005, with permission)
Legend:
MPY - micaceous
pyroxenite
FOSK - f oskorite
GLM - glimmerite
FEN - f enite
CARB - carbonatite
DOL - dolerite
Fig. 6.19 Major geological structures at Palabora mine (based on data provided by Palabora
Mining Company Limited)
204
6.5.1.3 North Wall Failure
As noted by Brummer et al. (2005), concurrent with the cave breakthrough into the
pit floor in late 2003 and early 2004, failure of major portion of the North wall became
apparent. Moss et al. (2006) described the open pit failure development as follows:
“Movement of all pit walls increased substantially upon cave

breakthrough into the bottom of the pit. The greatest amount
occurred in the North Wall where cumulative movements of in
excess of 1.5m were measured. The first indication of a major
problem, however, was a bench failure adjacent to one of the pit
sumps. This was followed by the discovery of large cracks some
250m back from the pit rim (note: it is not known if the cracking
occurred before or after the initial bench scale failures as only once
the failure occurred was a survey made of the dense bush that
surrounds that potion of the pit). The failure grew in size until after
a period of about 18 months it encompassed a major section of the
North Wall with the crest some 50 m back from the pit rim and the
toe somewhere near the original pit floor. The failure dimensions
were some 800 m high by 300m along the wall.”
Photographs of failure development are given in Appendix C. The failure impacted

several segments of the mine‟s infrastructure that had to be relocated including
access and haul roads, tailings, water and power lines, water reservoirs and a
railway line. Pretorius & Ngidi (2008) note that an estimated 130 million tons of
waste material failed into the open pit and that the failure resulted in a reduction in
life of the mine with potential loss of reserves of up to 30%. A plan view of the
failure, as well as the major joint sets are shown in Fig. 6.20. It can be seen that
the failure boundaries are generally defined by the major discontinuity sets.
As indicated by Pretorius & Ngidi (2008) modelling of the open pit - cave interaction
carried out by external consultants prior to the failure event concluded that pit walls
are to be stable above approximately the middle of the pit depth. Following the
failure, a back analysis was carried out by Piteau Associates (2005) using the limit
205
equilibrium package SLIDE (RocScience, 2007), and by Itasca (2005) using 3DEC
simulations (Brummer et al., 2005). The limit equilibrium study did not explain the
mechanisms leading to the North Wall failure, it was recognized that this type of
analysis has a limited ability to simulate rock mass deformations due to caving.
Brummer et al.‟s (2005) 3DEC models were constructed in order to calibrate the
properties of the rock mass with the monitored displacement, to match the failure
mode of the North wall, and to predict the likely long-term stability of the pit walls.
This study concluded that movement and deterioration of the North Wall was directly
linked to the block cave mining. Itasca (2005) concluded that: “The stability of the
North wall is controlled by joint sets. The single on-site estimated joint set of
75o/250o (dip/dip direction) produces a failure mode that matches the failure zone. A
more detailed model based on the two mapped sets 80o/140o and 80o/225o also
matches the failed zone”. Brummer et al. (2005) indicated that the deep seated
failure of the North Wall is possible if the ore is to be extracted from the caving area.
Fig. 6.20 Plan view of North Wall failure at Palabora mine
206
Assessing the possible role of major geological structures in the failure, Brummer
et al. (2005) stated that there were no combinations of major structures that
would delineate a slope failure with a dip or plunge flatter than about 66°, with
most structural combinations having a dip or plunge of at least 75°. There are no
combinations of major structures that dip or plunge to the south.
Analysis of the North Wall displacements carried out by Piteau Associates (2005)
showed that the displacements within some areas of the main zone of instability
have approached the plunge of the line of intersection of two discontinuity sets
that appear to control the stability of the North Wall. At the same time, the line of
intersection at the northern limit of the failure zone does not appear to intersect
the block cave footprint. This indicates that North Wall failure probably involved
a complex mechanism, primarily governed by the dominant rock mass fabric with
elements of brittle intact rock fracture and step-path failure.
6.5.2 General Approach in the Current Modelling Analysis
The modelling presented here is not intended to be a rigorous back analysis of

the Palabora failure and makes no attempt to exactly match the observed
deformations and associated pit slope displacements. Instead a conceptual
modelling approach is adopted which is founded on the engineering judgement
based assessment of the site geotechnical conditions with a full consideration of
the limited data available. The analysis focuses on understanding the general
principles of open pit - caving interaction and associated failure mechanisms
using the Palabora geometry as input.
Analysis of the extent of the Palabora North wall failure indicates that the failure
boundaries are largely defined by three joint sets mapped at the site. It appears
that a combination of 80o/320o and 82o/270o sets contributed to the formation of
the lateral failure release and 80o/225o (85o/020o) influenced the formation of the
rear release surface. The depth of the failure surface and the mechanism of
failure development remain uncertain. As previously stated step-path failure was
likely a major factor in the formation of the basal failure surface.
207
Notwithstanding the complexity and inherent 3D nature of the North Wall
deformations, it is believed that utilization of FEM/DEM-DFN technique even in
2D may help to provide better understanding of the failure mechanics. The
analysis presented here assumes that lateral release conditions exist a priori and
investigates formation of the basal failure plane.
6.5.3 Model Setup
Based on the interpretation of jointing data (from detailed line mapping) reported
by Piteau Associates (1980) and Martin et al. (1986) a preliminary FracMan DFN
model of part of the North Wall situated within the failure zone, corresponding to
cross-section A-A in Fig. 6.20, was generated, shown in Fig. 6.21. It was
assumed that the density of the jointing decreases away from the orebody. DFN
model input parameters are given in Appendix B. Here it should be emphasized
that the DFN model presented in Fig. 6.22 is preliminary and is based on the best
estimate of the actual jointing conditions. Ongoing work based on
photogrammetry is being conducted to refine the DFN. Due to limited available
data, it was not possible to carry out a comprehensive DFN validation analysis.
The DFN model was imported into the ELFEN model, shown in Fig. 6.22.
The mesh resolution was optimized with respect to the computing resources
available, resulting in a 2m mesh within the caving boundaries and a graded mesh
of up to 5m in the open pit slope. The modelling adopts the same ore extraction
methodology as described in Section 4.5.1.1. As evident from Table 6.3 all three
rock mass domains represented in the North Wall have a very similar rock mass
rating. For the purpose of the current analysis it was assumed that the rock mass
has uniform characteristics which are based on calibrated material properties as in
Chapter 5. To account for decreasing joint density away from the orebody, tensile
strength in the foskorite and micaceous pyroxenite was increased. Two cases
were considered:
Model P1 with a tensile strength increase of 100 and 150% in the foskorite
and micaceous pyroxenite, respectively; and
208
Model P2 with a similar tensile strength increase of 150 and 200%.
The input parameters adopted for the modelling are given in Table 6.4.
(a)
(b)
Fig. 6.21 Preliminary DFN model of Palabora mine North Wall section
(a) 3D DFN model; (b) fracture traces on traceplane1
1-
assistance of Dr. Davide Elmo with generation of DFN model is gratefully acknowledged
209
3300m
9000m
16 excavation stages
800m
Micaceous Pyroxenite
Foskorite
Carbonatite
400m
Fig. 6.22 ELFEN model of Palabora mine NW-SE section (section A-A in Fig. 6.20).
200m
210
Table 6.4 Modelling input parameters for preliminary Palabora failure simulation
Value
Parameter Unit Micaceous
Foksorite Carbonotite
Pyroxenite
Rock properties
Young‟s Modulus, E GPa 18 18 18
Poisson‟s ratio, 0.25 0.25 0.25
-3
Density, ρ kgm 2600 2600 2600
2 (P1) 1.5 (P1) 1 (P1)
Tensile strength, t MPa
2.5 (P2) 2 (P2) 1 (P2)
Fracture energy, Gf Jm-2 60 60 60
Cohesion, ci MPa 4.7 4.7 4.7
Friction, i degrees 45 45 45
Dilation, ψ degrees 5 5 5
Discontinuities
Fracture cohesion, cf MPa 0
Fracture friction, f degrees 35
Normal penalty, Pn GPa/m 2
Tangential penalty, Pt GPa/m 0.2
Stress level
In-situ stress ratio, K 1
6.5.4 Modelling Results and Discussion
Figs. 6.23 and 6.24 illustrate caving induced slope deformations for model P1 at
cave breakthrough and at 40% ore extraction. It can be seen that at cave
breakthrough slope failure is initiated. Caving induced slope unloading led to
mobilization of the lower portion of the slope where significant fracturing is
observed, this agrees well with the field observations. Fig. C1 (Appendix C)
illustrates caving induced failure along several benches at estimated cave
breakthrough. Initial step-path fracturing in the upper portion of the slope, as well
as formation of the tensile cracks in the open pit slope and at the pit crest, was
also observed. The basal failure surface that encompasses the entire pit slope
did not fully develop until about 40% ore extraction. As shown in Fig. 6.24 this
failure surface is step-path driven and is strongly defined by the rock mass fabric.
211
The failure outcrop at the pit crest over predicts the location of the actual failure,
by about 50m. It should be noted that this is not a significant margin given the
scale and complexity of the problem. The portion of the slope defined by the
failure surface is split into three major segments: the lower segment mobilized at
the cave breakthrough fully lost its structural coherence, while the upper
segments sustained generally minor damage.
Caving induced slope deformations at cave breakthrough and at 40% ore

extraction for model P2 are given in Figs. 6.25 and 6.26, respectively. In contrast
to model P1, here full scale slope failure did not materialize. The lower portion of
the slope mobilized at cave breakthrough, continuing to disintegrate and unravel
with ore extraction. Only minor fracturing within the slope, insufficient to form a
failure surface, was observed. This highlights the sensitivity of the modelling
outcome to the assumption of the pit slope rock mass strength.
Due to very long run-times the modelling was terminated at about 80% ore
extraction, with no changes in the observed deformation trends. In general,
models P1 and P2 illustrated quite different outcomes which could lead to very
different implications in terms of mine planning. The uncertainty in the modelling
input parameters and overall limited understanding of the strength of the rock
mass at a large open pit scale pose an important dilemma for decision makers.
Overall the conducted analysis of North Wall jointing conditions, observed

deformations and the conducted modelling suggest that failure of the North Wall
was largely governed by rock mass fabric. The in-situ conditions provided the
means to enable the formation of lateral and rear release surfaces as well as
formation of a step-path driven failure surface. The analysis showed that when
based on sound engineering judgement, even with limited data, FEM/DEM-DFN
modelling can contribute to the development of understanding of complex failure
mechanisms related to open pit - caving interaction. It appears that FEM/DEM-
DFN technique can be successfully employed for analysis of practical interaction
problems.
212
98 m
approximate f ailure approximate f ailure
crest location crest location
Fig. 6.23 Pit slope deformation at cave breakthrough for Fig. 6.24 Pit slope deformation at 40% ore extraction for model
model P1 P1
213
approximate f ailure approximate f ailure
crest location crest location
Fig. 6.25 Pit slope deformation at cave breakthrough for Fig. 6.26 Pit slope deformation at 40% ore extraction for model
model P2 P2
214
6.5.5 Conclusions
Transition from open pit to underground mining at Palabora mine presents an

important example of the pit wall instability triggered by caving operations. Using a
combined FEM/DEM-DFN modelling approach it was possible to investigate the
formation of the basal failure surface within the open pit slope as a direct result of
caving. The modelling highlighted the importance of open pit slope strength and its
influence on caving induced slope response. It appears that in open pit - caving
transition projects, reliable estimates of open pit slope strength is equally important
to the assessment of the role of major geological structures. It is however realized
that establishing geomechanical parameters for large open pits remains a challenge
yet to be resolved.
In the author‟s opinion the modelling of complex interaction problems requires

careful consideration of site specific conditions, including, but not limited to, rock
mass strength, rock mass fabric and loading/unloading conditions. Given the
uncertainties associated with the development of major transition projects a
range of possible conditions should be considered, including varying
assumptions of rock mass strength, variability of jointing orientation and
persistence, as well as different possibilities of caving development (slope
unloading conditions). Considering the importance of the implications of design
errors the additional effort to carry out detailed numerical modelling analysis
using techniques capable of capturing the problem complexity is without doubt
justifiable. Such exercise may be time consuming given the current reliance on
single processor computing capabilities. It is however anticipated that further
code improvements and use of computing clusters will enable comprehensive
modelling analysis of interaction problems in a timeframe acceptable for practical
engineering design.
6.6 Summary
This chapter investigated the interrelation between caving mining and open pit slope
stability. The proposed “total interaction” analysis related the destressing of the
215
crown pillar due to caving with unloading induced failure within the slope and the
resultant subsidence at the surface. Analysis indicates that there is a threshold of
critical intact rock bridge percentage along the step-path failure plane that may
ensure stability of the open pit throughout caving operations. A preliminary analysis
of the Palabora mine case study indicates that the formation of a step-path driven
basal failure surface within the pit slope rock mass fabric is plausible. Modelling
highlighted the sensitivity of the results to the assumed rock slope tensile strength.
Notwithstanding, this research has demonstrated that comprehensive FEM/DEM-
DFN modelling provides a promising technique for the analysis of highly complex
rock engineering problems. This is particularly encouraging in light of the future
need for reliable design tools for underground – open pit transition projects.
216
CHAPTER 7: CONCLUSIONS AND RECOMMENDATIONS
FOR FURTHER RESEARCH
7.1 Conclusions
Increasing use of block caving mining methods and ever increasing ore
extraction volumes are posing the questions of the effect of caving mining at the
surface and the implications of surface subsidence on mining activities, the
environment and socio-economics. The assessment of caving subsidence
phenomena is one of the most challenging tasks in mining geomechanics.
7.1.1 Current State of Knowledge of Block Caving Induced Subsidence
A comprehensive literature review carried out as part of the current study has shown
that our knowledge of rock mass behaviour leading to surface subsidence in block
caving settings is rather tentative and primarily founded on empirical, mostly
qualitative, observations. This is perhaps not surprising given the scale of the block
caving problem, the complexity of rock mass response and the multitude of factors
affecting subsidence. The literature focuses on the importance of the effect of
geological discontinuities in subsidence development, stopping short of elaborating
on the effect of other factors. Available methods of subsidence analysis include
empirical, analytical and numerical approaches. Empirical methods are not
particularly reliable. Analytical approaches are restrictive, and being based
principally on Hoek‟s (1974) assumed failure mechanism, are able to provide only
estimates for the angle of break. Numerical approaches being inherently more
flexible and sophisticated offer an opportunity to improve our understanding of block
caving subsidence phenomena and increased accuracy of subsidence predictions.
However, previous modelling studies have been largely directed toward providing
subsidence predictions for a particular site. The modelling study by Flores &
Karzulovic (2004) was the first attempt after Laubscher (2000) to provide non-site
217
specific guidance for subsidence analysis. It should be noted that their modelling
was limited to the case with an open pit and no consideration was given to the
effects of significant factors including rock structure. It appears that to date no
comprehensive attempt has been undertaken to evaluate the general principles
characterizing surface subsidence development in block caving settings and the
predominant factors governing subsidence phenomena.
7.1.2 FEM/DEM-DFN Approach to Subsidence Analysis
Block caving subsidence is a product of a complex rock mass response to

caving. This response comprises massive brittle fracture driven failure of the
rock mass both in tension and compression, along existing discontinuities and
through intact rock bridges. Moreover, block caving subsidence development
almost invariably involves complex kinematic mechanisms. It appears that
realistic simulation of the subsidence phenomena necessitates consideration of
fracture mechanics principles for brittle fracturing simulation and a blend of
continuum and discontinuum approaches to capture the complex failure
mechanisms. The current study adopted a state-of-the-art hybrid continuum-
discontinuum approach based on finite/discrete element method (Munjiza et al.,
1995) and incorporating fracture mechanics principles, implemented in the
proprietary code ELFEN (Rockfield Software Ltd., 2006). This code has been
extensively validated and widely applied to the analysis of a variety of rock
engineering problems.
Vyazmensky et al. (2007) evaluated various approaches to rock mass

representation within a FEM/DEM framework and concluded that explicit
consideration of discontinuities in block caving subsidence modelling is essential.
Therefore, the mixed discrete fracture network/equivalent continuum approach to
rock mass representation was utilized, where the rock mass is represented as an
assembly of spaced discontinuities and intervening regions with reduced intact
properties, the latter being scaled to incorporate the weakening effect of smaller
scale discontinuities not explicitly included in the model. The geologically sound
representation of discontinuities was achieved by employing DFN modelling,
218
using the proprietary code FracMan (Golder, 2005) and exporting the DFN output
into ELFEN.
7.1.3 Modelling Input Parameters
One of the key aspects in the numerical modelling of rock engineering problems
is establishing representative rock mass material strength and deformability
characteristics. As part of the current study the properties output from rock mass
classification systems were critically examined by comparison of RMR, GSI and
Q based properties for a fictitious dry, hard rock mass. In summary, comparison
of strength and deformability properties derived from the studied systems for
corresponding rating values for the most part showed a lack of consistency (see
Fig. 4.3 and 4.4). Generally, none of the studied rock mass classification based
correlations offered adequate estimates of all of the required equivalent
continuum rock mass properties. Output from these classifications should
undergo critical evaluation before being accepted in any analysis.
A novel block caving subsidence modelling methodology was devised and the
applicability of the use of RMR and Q rock mass classification systems as a
source of equivalent continuum properties for integrated FEM/DEM-DFN
modelling was analyzed. A series of numerical experiments were carried out to
investigate the behaviour of a system comprised of key discontinuities and an
equivalent continuum rock mass derived using varying assumptions for
mechanical properties from rock mass classification systems. A novel properties
assessment/calibration procedure that incorporates use of response constraining
criteria based on Laubscher‟s caveability chart (Diering & Laubscher, 1987), the
cave propagation concept of Duplancic & Brady (1999) and caving subsidence
experience was utilized.
The following are the key conclusions from the comparative analysis performed:
neither the RMR nor Q rock mass classification systems can be relied
upon as a robust source of rock mass properties for hybrid FEM/DEM
modelling;
219
very low estimates of rock mass cohesion and tensile strength derived
using RMR system led to overestimated susceptibility of simulated rock
mass material to tensile failure, making use of RMR properties
unacceptable for modelling of caving mechanisms;
among all considered Q property sets, the properties for the midrange Q
ratings, provide the most realistic representation of rock mass caveability,
cave development progression and subsidence response;
it appears that Q properties for the ratings equivalent to RMR 60 to 70
offer generally adequate estimates of deformation modulus, cohesion and
friction, and, also, give flexibility in regards to the assumption of the tensile
strength, thus leaving room for response calibration.
7.1.4 Factors Governing Block Caving Subsidence Development
In a complex block caving mining environment subsidence development is a

result of interplay of several governing factors. Mining experience suggests that
among such factors are geological structure (jointing and faults), rock mass
strength, in-situ stress level, mining depth, volume of extracted material and
varying lithological domains. A survey of the literature has shown that
publications are limited to a general, qualitative rather than quantitative,
description of the influence of geological structures on the observed subsidence.
Such qualitative observations are useful for initial subsidence analysis, however
they require further validation. With regards to other factors, literature sources
highlight their importance but provide only limited further description.
Understandably, discerning the effect of individual factors from field observations
is challenging. Adopting the novel modelling methodology for subsidence
analysis, through more than 30 extensive conceptual numerical experiments,
subsidence development mechanisms and the relative significance of the factors
governing subsidence were investigated.
The adopted modelling methodology allowed simulation of the subsidence

deformation mechanism, from caving initiation to the final subsidence, agreeing
closely with the observations based conceptual model of subsidence
220
development proposed by Abel & Lee (1980). This reinforced the validity of the
adopted approach for block caving subsidence analysis.
New valuable insights were gained into the complex mechanism of caving
induced rock mass deformations and subsequent subsidence development.
Conducted analyses clearly demonstrated the importance of joint set orientation
and persistence, and fault location and inclination, in determining the subsidence
development mechanisms and their governing role in defining the degree of
surface subsidence asymmetry. The modelling results correlated reasonably
well with the available field observations. Based on the analysis of conceptual
modelling study results a preliminary classification of block caving induced
surface subsidence for cases, where no major geological discontinuities intersect
the cave, is proposed, as shown in Table 5.8. This classification is
supplemented by a preliminary classification of the influence of major geological
discontinuities on surface subsidence, given in Table 5.9. The relative
significance of individual parameters is summarized in the influence assessment
matrix, shown in Table 5.10. The proposed subsidence assessment
classifications and the parametric influence matrix are meant to aid practical
engineering assessment of potential block caving subsidence at pre-feasibility
and design stages, and represent an important step towards better
understanding and quantifying block caving induced surface subsidence. It
cannot be over-emphasized that such assessments must be tailored to, and take
full consideration of, specific mine site conditions and should be based on sound
engineering judgement.
7.1.5 Large Open Pit - Block Caving Interaction
Brittle fracturing processes are one of the primary controls on slope deformation
development in large open pits. The FEM/DEM modelling approach was utilized
in the analysis of primary fracture process (using terminology of Stead et al.,
2007) associated with block caving induced step-path failure development in
large open pit slopes. The proposed “total interaction” analysis allowed relating
the destressing of the crown pillar due to caving to the development of unloading-
221
induced failure within the slope and the resultant subsidence at the surface.
Analysis indicated that there is a threshold or critical intact rock bridge
percentage along step-path failure planes that may ensure stability of an open pit
throughout caving operations.
Transition from open pit to underground mining at Palabora mine presents an

important example of the pit wall instability triggered by caving operation. Using
a combined FEM/DEM-DFN modelling approach it was possible to investigate
the formation of the basal failure surface within an open pit slope as a direct
result of caving. The modelling of Palabora highlighted the importance of rock
mass tensile strength and its influence on caving induced slope response.
7.2 Key Scientific Contributions

The following are the key contributions of this study:
A new FEM/DEM-DFN modelling approach was developed and

successfully applied to block caving subsidence and caving - large open
pit interaction analysis. This methodology allows physically realistic
simulation of the entire caving process from caving initiation to final
subsidence deformations.
Limitations of the rock mass classifications properties output were

highlighted and a procedure for calibrating rock mass classifications based
properties for FEM/DEM-DFN subsidence analysis was devised.
Through a comprehensive conceptual numerical modelling analysis major

advances were gained in our understanding of the general principles of
block caving induced subsidence development and the role of major
contributing factors. This included proposals for subsidence
characterization through an asymmetry index and the mobilized rock mass
volume. Based on the modelling results a preliminary quantitative
classification of block caving induced surface subsidence was proposed
and the role of major geological structures was classified. In addition, an
initial qualitative influence assessment matrix was proposed that
222
summarizes the effect of major factors contributing to subsidence
development.
The principles of step-path failure development in large open-pit - caving

mining environment were investigated using a proposed “total interaction”
approach to modelling data interpretation. Indications of a critical rock
bridge percentage threshold in step-path failure development were found.
Applicability of the FEM/DEM-DFN modelling for practical engineering

analysis was demonstrated in the preliminary simulation of the Palabora
mine failure.
7.3 Recommendations for Further Research
7.3.1 Equivalent Continuum Rock Mass Properties
The analysis conducted clearly shows a need for further research into estimation
of rock mass equivalent continuum properties. In the author‟s opinion future
progress in this area can be achieved through use of a synthetic rock mass
modelling methodology employing either PFC or ELFEN and DFN techniques.
Further work is required to ensure that it can produce realistic material properties
for different rock mass scales and to evaluate its sensitivity to varying stochastic
realizations.
7.3.2 Modelling of 3D Aspects of Block Caving Subsidence Development
Current modelling was focused on 2D analysis. The actual rock mass conditions
however are three dimensional and more representative modelling should consider
3D effects. Of a particular interest are the 3D effects of rock mass fabric and 3D
slope stability in open-pit - block caving environment.
7.3.3 ELFEN Code Enhancements
The ELFEN code has shown strong capabilities to simulate highly complex rock
engineering problems. Further code improvements should make it more accessible
223
for practical engineering analysis. The following code improvements are
recommended:
I. ELFEN modelling carried out during present study demonstrated that the run-
times for simulations of caving induced deformations are often prohibitive for
practical engineering analysis. Adaptation of the code to support 64-bit and
parallel computing will allow affordable modelling of larger and more complex
problems in 2D and 3D space.
II. Another important improvement can be made in the inter-element fracturing

routine which is currently not robust (as discussed in Section 3.4.3), this
could allow consideration of coarser mesh further reducing simulations run-
time.
III. More realistic brittle fracture simulation can be achieved through

development of constitutive relationships that in addition to mode I support
mode II and mixed mode I-II fracturing in both 2D and 3D.
7.3.4 In-situ Subsidence Characterization
Literature reviews indicated very limited availability of factual (i.e. measured)

subsidence data at block cave mines. Such data is essential for constraining and
validating modelling studies. It is prudent to summarize the available information
from the current block caving mines and carry out a comprehensive subsidence
monitoring study using a combination of InSAR, LiDAR and photogrammetric
techniques, as well as traditional surveying methods and geotechnical
instrumentation at several block caving mines. Some elements of this work are
already being undertaken at SFU and UBC and it is hoped that the collected data
will further reinforce the validity of the modelling findings.
224
REFERENCE LIST
ABAQUS. 2007. Simulia. http://www.simulia.com

Abel, J.F.& Lee, T.F. 1980. Subsidence potential in shale and crystalline rocks. U.S.
Geological Survey. Open File Report 80-1072. 52 pp.
ASTM D4395-04. Standard test method for determining the in-situ modulus of
deformation of rock mass using the flexible plate loading method. ASTM
International. West Conshohocken, PA. www.astm.org.
Atkinson, B.K. (Ed.). 1987. Fracture mechanics of rocks. Academic Press. San Diego.
Calif. 548pp.
Backers, T., Stephansson, O., Moeck, I., Holl, H-G, Huenges, E. 2006. Numerical
borehole breakout analysis using FRACOD2D. In: EUROCK 2006. Proc. of
European ISRM Symposium. Liege. Belgium. Taylor & Francis Group, London.
pp. 683-688.
Baczynski, N.R.P. 2000. STEPSIM4 “step-path” method for slope risks. In: GEOENG2000.
Proc. International Conference on Geomechanical and Geological Engineering.
Melbourne. Technomic Publishing Co. Lancaster. Vol.2: pp. 86-92.
Bandis, C.S. 1993. Engineering properties and characterisation of rock discontinuities.
In: Comprehensive rock engineering. Pergamon. Vol.1. pp. 155-183.
Barton, N. 2007. Future directions for rock mass classification and characterization –
Towards a cross-disciplinary approach. In: Proc. 1st Canada-U.S. Rock
Mechanics Symposium. Vancouver. Canada. Taylor & Francis Group Ltd.
London. Vol.1. 179-189.
Barton, N. 2002. Some new Q-value correlations to assist in site characterization and
tunnel design. Int. J. Rock Mech. & Min. Sci. 39(1): 185-216.
Barton, N. 1982. Shear strength investigations for surface mining. In: Proc. 3rd
International Conference on Stability in Surface Mining. Vancouver. Canada,
Society of Mining Engineers, AIME: New York. pp. 171-196.
Barton, N., Lien, R. & Lunde, J. 1974. Engineering classification of rock masses for the
design of tunnel support. Rock Mech. 6: 189-239.
Beck, D. & Pfitzner, M. 2008. Interaction between deep block caves and and existing,
overlying caves or large open pits. In: Proc. 5th International Conference and
Exhibition on Mass Mining (MassMin2008). Luleå. Sweden. Luleå University of
Technology Press. Luleå. Sweden. pp. 381-391.
Becker, A.A. 1992. The boundary element method in engineering. McGraw-Hill. 348 pp.
225
Betournay, M.C. 2002. Underground mining and its surface effects. Fourth biennial
abandoned underground mine workshop. Report MMSL 02-021(TR). Web
publication. http://www.fhwa.dot.gov
Bieniawski, Z.T. 1989. Engineering Rock Mass. Classifications. Wiley. New York. 251 pp.
Bieniawski, Z.T. 1976. Rock mass classification of jointed rock masses. Exploration for
Rock Engineering. In: Proc. Symp. Expl. Rock. Eng. Johannesburg. Balkema.
Cape Town. Vol.1. pp. 97-106.
Blodgett, S. & Kuipers, J. 2002. Underground hard rock mining: subsidence and
hydrologic environmental impacts. Centre for Science in Public Participation.
Web publication. http://www.csp2.org
Brady, B.H.G. & Brown, E.T. 2004. Rock mechanics for underground mining. 3rd edition.
Kluwer Academic Publishers. 626 pp.
Brown, E.T. 2003. Block caving geomechanics. The University of Queensland. 516 pp.
Brown, E.T. & Fergusson, G.A. 1979. Prediction of progressive hangingwall caving,
Gath‟s mine, Rhodesia. Trans. Instn. Min. Metall. 88: A92-A105.
Brummer, R. K., Li, H. & Moss, A. 2006. The transition from open pit to underground
mining. An unusual failure mechanism at Palabora. In: Proc. of International
Symposium on Stability of Rock Slopes in Open Pit Mining and Civil Engineering.
Cape Town. South Africa. The South African Institute of Mining and Metallurgy.
Symposium Series S44. pp. 411-420.
Cai, M. & Kaiser, P.K. 2004. Numerical simulation of the Brazilian test and the tensile
strength of anisotropic rocks and rocks with pre-existing cracks. Int. J. Rock.
Mech. Min. Sci. 41(1): 478-483.
Cai, M., Kaiser, P.K., Uno, H., Tasaka, Y. & Minami, M. 2004. Estimation of rock mass
deformation modulus and strength of jointed hard rock masses using the GSI
system. Int. J. Rock Mech. and Min. Sci. 41: 3-19.
Cavieres, P., Gaete, S., Lorig, L. & Gómez, P. 2003. Three-dimensional analysis of
fracturing limits induced by large scale underground mining at El Teniente mine.
In: Proc. 39th U.S. Rock Mechanics Symposium. Cambridge, USA. Verlag
Gluckauf GmbH, Essen. pp. 893-900.
Christianson, M.C., Board, M.P. & Rigby, D.B. 2006. UDEC simulation of triaxial testing
of lithophysal tuff. In: Golden Rocks 2006. Proc. 41st U.S. Rock Mechanics
Symposium. ARMA. Paper No. 06-968.
Choi, S.K. 1992. Application of the distinct element method for rock mechanics
problems. Engineering Computations. 9(2): 225-233.
Clark, I.H. 2006. Simulation of rockmass strength using ubiquitous joints. In: Proc. 4th
International FLAC Symposium on Numerical Modelling in Geomechanics.
Minneapolis, Minnesota: Itasca Consulting Group, Inc. Paper No. 08-07.
Cividini, A. 1993. Constitutive behaviour and numerical modelling. In: Comprehensive
rock engineering. Pergamon. Vol. 1. pp. 395-426.
226
Coggan, J.S. & Stead, D. 2005. Numerical modelling of the effects of weak mudstone on
tunnel roof behaviour. In: Proc. 58th Canadian Geotechnical Conference.
Saskatoon. Canada. Paper GS502.
Coggan, J.S., Pine, R.J., Stead, D. & Rance, J. 2003. Numerical modelling of brittle rock
failure using a combined finite-discrete element approach: implications for rock
engineering design. In: Proc. ISRM 2003. South African Institute of Mining and
Metallurgy. SAIMM Symposium Series S33. pp. 211-218.
Crane, W.R. 1929. Subsidence and ground movement the copper and iron mines of the
Upper Peninsula, Michigan” USBM Bulletin 285. 66 pp.
Crouch, S.L. & Starfield, A.M. 1983. Boundary element methods in solid mechanics.
London: George Allen & Unwin. 322 pp.
Cundall, P.A. 2008. Recent advances in numerical modelling for large-scale mining
projects. ACG News. 30: 1-7.
Davis, J.L. 1986. Finite difference methods in dynamics of continuous media. Macmillan
Publishing Company. 238 pp.
Dershowitz, W., LaPointe, P., Eiben, T. & Wei, L. 1998. Integration of discrete fracture
network methods with conventional approaches. Society of Petroleum Engineers.
Paper 49069.
Dershowitz, W., Wallmann, P., Shuttle, D. & Follin, S., 1996. Canister and far-field
demonstration of the discrete fracture analysis approach for performance
assessment. Stockholm, Sweden. SKB Progress Report U-96-41.
Dershowitz, W. 1992. The role of the Stripa Phase 3 Project in the development of
practical discrete fracture modelling technology. In: Proc. 4th International
NEA/SKB Symposium, Stockholm. OECD/NEA. Paris. pp. 237-258.
Dershowitz, W. & Einstein, H.H. 1987. Characterizing rock joint geometry with joint
system models. Rock Mech. & Rock Eng. 21: 21-51.
Diering, J.A.C. & Laubscher, D.H. 1987. Practical approach to the numerical stress
analysis of mass mining operations. Trans. Instn. Min. Met. 96: A179-188.
Duplancic, P. & Brady, B.H. 1999. Characterisation of caving mechanisms by analysis of
seismicity and rock stress. In: Proc. 9th International Congress on Rock
Mechanics, Paris. Balkema. Vol.2, pp. 1049-1054.
Eberhardt, E. 2008. The role of advanced numerical methods and geotechnical field
measurements in understanding complex deep-seated rock slope failure
mechanisms. Canadian Geotechnical Journal. 45(4): 484-510.
Eberhardt, E., Stead, D., Elmo, D., Dunbar, S., Scoble, M., Moss, A., Sturzenegger, M.,
Vyazmensky, A., Tollenaar, R., & Eissa, H. 2007. Transition from surface to
underground mining: Integrated mapping, monitoring and modeling data to better
understand complex rock mass interaction. In: Proc. International Symposium on
Rock Slope Stability in Open Pit and Civil Engineering. Perth. Australia.
Australian Centre for Geomechanics. pp. 321-332.
227
Eberhardt, E., Stead, D. & Coggan, J.S. 2004a. Numerical analysis of initiation and
progressive failure in natural rock slopes - the 1991 Randa rockslide. Int. J. Rock
Mech. Min. Sci. 41: 69-87.
Eberhardt, E., Stead, D., Karami, A. & Coggan, J.S. 2004b. Numerical analysis of brittle
fracture propagation and step-path failure in massive rock slopes. In: Proc. 57th
Canadian Geotechnical Conference. Quebec City. Canada. Paper G13-154:1-8.
Edelbro, C., Sjöberg, J. & Nordlund, E. 2007. A quantitative comparison of strength
criteria for hard rock masses. Tunnelling and Underground Space Technology.
22(1): 57-68.
Elmo, D., Vyazmensky, A., Stead, D. & Rance, J.R. 2008. Numerical analysis of pit wall
deformation induced by block-caving mining: a combined FEM/DEM - DFN
synthetic rock mass approach. In: Proc. 5th International Conference and
Exhibition on Mass Mining (MassMin2008). Luleå. Sweden. Luleå University of
Technology Press. Luleå. Sweden. pp. 1073-1082.
Elmo, D., O‟Connor, C., Vyazmensky, A., Stead, D., Dunbar, S., Eberhardt, E., Scoble,
M. & Moss, A. 2007a. Integrated modelling of subsidence mechanisms and
impacts due to mine caving. CIM Energy & Mines Conference. Montreal. Online
publications. http://www.infomine.com
Elmo, D., Vyazmensky, A., Stead, D. & Rance, J.R. 2007b. A hybrid FEM/DEM
approach to model the interaction between open pit and underground block
caving mining. In: Proc. 1st Canada-U.S. Rock Mechanics Symposium.
Vancouver. Canada. Taylor & Francis Group Ltd. London. Vol.2. pp. 1287-1294.
Elmo, D. 2006. Evaluation of a hybrid FEM/DEM approach for determination of rock
mass strength using a combination of discontinuity mapping and fracture
mechanics modelling, with particular emphasis on modelling of jointed pillars.
PhD Thesis. Camborne School of Mines. University of Exeter. U.K.
Elmo, D., Pine R.J. & Coggan, J.S. 2005. Characterisation of rock mass strength using a
combination of discontinuity mapping and fracture mechanics modelling. In: Proc.
40th U.S. Rock Mechanics Symposium. Anchorage. USA. ARMA/USRMS. Paper
05-733.
Esci, L. & Dutko, M. 2003. Large scale fracturing and rock flow using discrete element
method, 2D application for block caving. In: Proc. Numerical Methods in
Continuum Mechanics, Žilina, Slovak Republic. Abstract. 3 pp.
Fairhurst, C. 2006. Keynote lecture, ARMA, 41st US Symposium on Rock Mechanics,
Golden, Colorado.
Fang, Z. & Harrison, J.P. 2002. Development of a local degradation approach to the
modelling of brittle fracture in heterogeneous rocks. Int. J. Rock Mech. & Min.
Sci. 39: 443-457.
Flores, G. & Karzulovic, A. 2004. Geotechnical guidelines for a transition from open pit
to undeground mining. Subsidence. ICSII. Task 4. Technical report. 78 pp.
Flores, G. & Karzulovic, A. 2002. Geotechnical guidelines for a transition from open pit to
undeground mining. Benchmarking Report for ICSII. Task 4. 91 pp.
228
FLUOR Ltd. 1994. Undeground mining project. Feasibility report. Vol. IV. Mine. 139p.
Franz, J., Cai, Y. & Hebblewhite, B. 2007. Numerical modelling of composite large scale
rock slope failure mechanisms dominated by major geological structures. In:
proc. 11th Congress of the International Society for Rock Mechanics. Lisbon.
Taylor & Francis Group. London. Vol.1. pp. 633-636.
Gilbride, L.J., Free, K.S. & Kehrman, R. 2005. Modelling block cave subsidence at the
Molycorp, Inc., Questa mine. In: Proceedings 40th U.S. Rock Mechanics
Symposium. Anchorage. USA. ARMA/USRMS. Paper 05-881.
Golder Associates. 2007. FracMan. Version 6.54. www.fracman.golder.com.
Golder Associates. 1989. NFOLD. Program user manual.
Griffith, A.A. 1921. The phenomena of rupture and flow in solids. Philos. Trans. Roy.
Soc. London Ser. A. 221: 163-198.
Hajiabdolmajid, V., Kaiser, P.K. & Martin, C.D. 2002. Modelling brittle failure of rock. Int.
J. Rock Mech. Min. Sci. 39: 731-741.
Hanekom, H. J., van Staden, C. M., Smit, P. J. & Pike. D. R. 1965. The geology of the
Palabora igneous complex. South African Geol. Survey. Mem 54. 179 pp.
Hartman, H.L. (ed.) 1992. SME mining engineering handbook. Society for Mining,
Metallurgy and Exploration. 2394pp.
Hebblewhite, B.K. 2003. Northparkes Findings - The implications for geotechnical
professionals in the mining industry. The First Australasian Ground Control in
Mining Conference. Online publication. http://www.mining.unsw.edu.au
Hellewell, F.G. 1988. The influence of faulting on ground movement due to coal mining:
the U.K. and European experience. Mining Engineer. 147: 334-337.
Herdocia, A. 1991. Hanging wall stability of sublevel caving mines in Sweden. PhD
thesis. Luleå University of Technology.
Heslop, T.G. & Laubscher, D.H. 1981. Draw control in caving operations on South
African chrysotile asbestos mines. In: Design and Operation of Caving and
Sublevel Stoping Mines. Stewart R.D (editor). SME-AIME. New York. pp. 775-
776.
Hoek, E. & Diederichs, M.S. 2006. Empirical estimation of rock mass modulus. Int. J.
Rock. Mech. Min. Sci. 43: 203–215.
Hoek, E., Carranza-Torres, C.T., & Corkum, B. 2002. Hoek-Brown failure criterion: 2002
edition. In: Proc. of the North American Rock Mechanics Society Meeting.
Toronto. Canada. University of Toronto Press. Toronto. pp. 267-274.
Hoek, E. & Brown, E.T. 1997. Practical estimates or rock mass strength. Int. J. Rock
mech. Min. Sci. & Geomech. Abstr. 34(8): 1165-1186.
Hoek, E., Kaiser, P.K. & Bawden, W.F. 1995. Support of underground excavations in
hard rock. Balkema. Rotterdam. 240 pp.
Hoek, E. 1994. Strength of rock and rock masses. ISRM News Journal. 2(2): 4-16.
229
Hoek, E., Grabinsky, M.W. & Diederichs, M.S. 1991. Numerical modelling for
underground excavation design. Transactions of the Institution of Mining and
Metallurgy (Section A: Mining Industry), 100, pp. A22–A30.
Hoek, E. 1974. Progressive caving induced by mining an inclined orebody. Trans Instn
Min. Metall. 83: A133-A139.
Hudson, J.A. & Harrison, J.P. 1997. Engineering rock mechanics – an introduction to the
principles. Elsevier Science & Technology. Oxford. 456 pp.
Irwin, G.R. 1957. Analysis stresses and strains near the end of a crack traversing a
plate. Journal of Applied Mechanics. 24: 109-114.
ISRM Commission on testing methods. 1988. Suggested methods for determining the
fracture toughness of rock. Int. J. Rock mech. Min. Sci. & Geomech. Abstr. 25:
71-97.
Itasca. 2007. FLAC2D, FLAC3D, 3DEC, UDEC, PFC2D and PFC3D codes. Itasca
Consulting Group Ltd. http://www.itascacg.com
Itasca. 2005. Analysis of pit wall failure mechanism and assessment of long-term
stability of pit walls. Palabora mine. Technical report. 78 pp.
Jaeger, J.C. 1971. Friction of rocks and stability of rock slopes. 11th Rankine lecture.
Geotechnique. 21(2): 97–134.
Jaeger, J.C. 1960. Shear fracture of anisotropic rocks. Geol. Mag. 97: 65–72.
Janelid, I. & Kvapil, R. 1966. Sublevel caving. Int. J. Rock Mech. Sci., 3, pp. 129-153.
Jennings, J.E. 1970. A mathematical theory for the calculation of the stability of slopes in
open cast mines. In: Proc. Planning of Open Pit Mines. Johannesburg. South
Africa. Balkema. Cape Town. pp. 87-102.
Jing, L. & Stephansson, O. 2007. Fundamentals of discrete element methods for rock
engineering. Elsevier Publishers. Oxford, UK. 450 pp.
Jing, L. 2003. A review of techniques, advances and outstanding issues in numerical
modelling for rock mechanics and rock engineering. Int. J. Rock Mech. Min. Sci.
40: 283-353.
Jing, L. 1998. Formulation of discontinuous deformation analysis (DDA) - an implicit
discrete element model for block systems. Eng. Geol. 49: 371-381.
Karami, A. & Stead, D. 2008. Asperity degradation and damage in the direct shear test:
a hybrid DEM/FEM approach. Rock Mechanics and Rock Engineering. 41(2):
229-266.
Karami, A., Beddoes, R. & Greer, S. 2007. Numerical assessment of step-path failure of
Northwest Wall of A154 Pit, Diavik Diamond mines. In: Proc. International
Symposium on Rock Slope Stability in Open Pit and Civil Engineering. Perth.
Australia. Australian Centre for Geomechanics. pp. 293-306.
230
Karzulovic, A, Cavieres, P & Pardo, C. 1999. Caving subsidence at El Teniente Mine (in
Spanish). In: Proc. Symp. Min. Engng. Santiago. Chile. Dept. Min. Engng, Univ.
Santiago.
Karzulovic, A. 1990. Evaluation of angle of break to define the subsidence crater of Rio
Blanco mine‟s panel III. Technical Report. Andina Division, CODELCO-Chile.
Kemeny, J. 2005. Time-dependent drift degradation due to the progressive failure of
rock bridges along discontinuities. Int. J. Rock. Mech. Min. Sci. 42: 35-46.
Klerck, P.A., Sellers, E.J. & Owen, D.R.J. 2004. Discrete fracture in quasi-brittle
materials under compressive and tensile stress states. Comput. Methods Appl.
Mech. Engrg. 193: 3035-3056.
Klerck, P.A. 2000. The finite element modelling of discrete fracture in quasi-brittle
materials. PhD thesis. University of Swansea. UK.
Kratzsch, H. 1983. Mining Subsidence Engineering. Trans. Springer-Verlag. New York.
543 pp.
Kvapil, R. 2004. Gravity flow in sublevel and panel caving – a common sense approach.
Luleå University of technology Press. 152 pp.
Kvapil, R., Baeza, L., Rosenthal, J. & Flores G. 1989. Block caving at El Teniente mine,
Chile. Trans. Instn Min. Metall. 98: 43-56.
Laubscher, D.H. 2000. Block caving manual. Report for International Caving Study,
JKMRC and Itasca Consulting Group, Inc.
Laubscher, D.H. 1994. Cave mining: state of the art. Inst. Min. Metall. 94: 279-293.
Laubscher, D.H. 1981. Selection of mass underground mining methods. In: Design and
Operation of Caving and Sublevel Stoping Mines. Stewart R.D (editor). SME-
AIME. New York. pp. 22-38.
Liu H. 2003. Numerical modelling of the fracture process under mechanical loading. PhD
Thesis. Dept. of Civil and Mining Engineering. Lulea University of Technology.
Long, J.C.S., Remer, J.S., Wilson, C.R. & Witherspoon, P.A. 1982. Porous media
equivalents for networks of discontinuous fractures. Water resources research.
18(2): 645-658.
Lupo, J. 1998. Large-scale surface disturbances resulting from underground mass
mining. Int. J. Rock Mech & Min. Sci. Abstract. 35(4/5): 399.
Lupo, J. 1996. Evaluation of deformations resulting from mass mining of an inclined
orebody. PhD thesis. Colorado School of Mines. Golden. USA.
Mahtab, M.A., Bolstad, D.D. & Kendorski, F.S. 1973. Analysis of the geometry of
fractures in San Manuel copper mine, Arizona. Bureau of Mines. Technical report
RI 7715.
Martin, D.C., Steenkamp, N.S.L. & Lill, J.W. 1986. Application of statistical analysis
technique for design of high rock slopes at Palabora mine, South Africa. Mining
Latin America. pp. 241-255.
231
Martin, D.C. 1978. The influence of fabric geometry and fabric history on the stability of
rock slopes. MSc dissertation. 99 pp. Imperial College. London. Unpublished.
(reported by Piteau & Martin, 1982).
Marinos, V., Marinos, P. & Hoek, E., 2005. The geological strength index: applications
and limitations. Bulletin of Engineering Geology and the Environment. 64: 55-65.
Mas Ivars, D., Deisman, N., Pierce, M. & Fairhurst, C. The synthetic rock mass approach
- a step forward in the characterization of jointed rock masses. In: Proc. 1st
Canada-U.S. Rock Mechanics Symposium. Vancouver. Canada. Taylor &
Francis Group. London. Vol.1. pp. 485-490.
McIntosh Engineering Ltd. 2003. Hard rock miners handbook. Rules of thumb. 3rd
edition. On-line publication. http://www.mcintoshengineering.com
McMahon, B.K. 1979. Report to Bougainville Copper Limited on slope design studies,
Pan Hill. McMahon, Burgess and Yeates, Sydney.
Milne, D., Hadjigeorgiou, J. & Pakalnis, R. 1998 Rock mass characterization for
underground hard rock mines. Tunnelling and Underground Space Technology.
13(4): 383-391.
Mine Modelling Pty Ltd. 2006. MAP3D. www.map3d.com.
Moss, A., Diachenko, S. & Townsend, P. 2006. Interaction between the block cave and
the pit slopes at Palabora mine. In: International Symposium on Stability of Rock
Slopes in Open Pit Mining and Civil Engineering. Cape Town. South Africa. The
South African Institute of Mining and Metallurgy. Symposium Series S44. pp 399-
409.
Muller, J.R. & Martel, S.J. 2000. Numerical models of translational landslide rupture
surface growth. Pure and Applied Geophysics. 157: 1009-1038.
Munjiza, A. 2004. The combined finite-discrete element method. J. Wiley & Sons.
Chichester. 348 pp.
Munjiza, A. & Latham, J.P. 2004. VGW - virtual geoscience workbench. Internal report.
Department of Engineering. Queen Mary, University of London. 212 pp.
Munjiza, A., Andrews, K.R.F. & White, J.K. 1999. Combined single and smeared crack
model in combined finite-discrete element analysis. Int. J. Numer. Meth. Engng.
44: 41-57.
Munjiza, A., Owen, D.R.J. & Bicanic, N. 1995. A combined finite-discrete element
method in transient dynamics of fracturing solids. Engineering Computations. 12:
145-174.
National Coal Board. 1975. Subsidence Engineers Handbook. 2nd (rev.) ednt. National
Coal Board Mining Dept. London. 111 pp.
Owen D.R.J, Feng Y.T., de Souza Neto, E.A., Cottrell, M.G., Wang, F., Andrade Pires,
F.M. & Yu, J. 2004a. The modelling of multi-fracturing solids and particulate
media. Int. Jour. Num. Meth. Eng. 60: 317-339.
232
Owen D.R.J, Pires F.M., De Souza Neto E.A. & Feng Y.T. 2004b. Continuous/discrete
strategies for the modelling of fracturing solids. Publication of the Civil &
Computational Eng. Centre, University of Wales - Swansea.
Owen, D.R.J. & Feng, Y.T. 2001. Parallelised finite/discrete element simulation of multi-
fracture solids and discrete systems. Engineering Computations, 18(3/4): 557-
576.
Panek, L.A. 1984. Subsidence in undercut - cave operations, subsidence resulting from
limited extraction of two neighboring cave operations. In: Geomechanical
Applications in Hard Rock Mining. (ed. Pariseau, W.G.) pp 225-240.
Parker, J. 1978. What can be learned from surface subsidence. E&MJ.
Peng, S.S. 1992. Surface subsidence engineering. Society for Mining, Metallurgy, and
Exploration, Inc. Littleton. USA. 161 pp.
Pierce, M., Cundall, P., Potyondy, D. & Mas Ivars, D. 2007. A synthetic rock mass model
for jointed rock. In: Proc. 1st Canada-U.S. Rock Mechanics Symposium.
Vancouver. Canada. Taylor & Francis Group. London. Vol.2. pp. 341-349.
Pine, R.J., Owen, D.R.J., Coggan, J.S. & Rance, J.M. 2007. A new discrete modelling
approach for rock masses. Geotechnique. 57(9): 757-766.
Pine, R.J., Coggan, J.S., Flynn, Z.N. & Elmo, D. 2006. The development of a new
numerical modelling approach for naturally fractured rock masses. Rock Mech.
Rock Engng. 39(5): 395-419.
Piteau Associates Engineering Ltd. 2005. Assessment of pit wall instability and slopes
displacement as a result of interaction between the open pit and underground
mine. Technical report. 96 pp.
Piteau Associates Engineering Ltd. 1980. Report on slope stability analysis and design
of slopes in the copper open pit. Technical report. 150 pp.
Piteau, D.R. & Martin, D.C. 1982. Mechanics of rock slope failure. In: Stability in Surface
Mining. Vol.3. SME. New York. pp. 113-166.
Pretorius, D. & Ngidi, S. 2008. Cave management ensuring optimal life of mine at
Palabora. In: Proc. 5th International Conference and Exhibition on Mass Mining.
Luleå. Sweden. Luleå University of Technology Press. Luleå. Sweden. pp. 63-71.
Potyondy, D.O. & Cundall, P.A. 2004. A bonded particle model for rock. Int. J. Rock
Mech. & Min. Sci. 41(8): 1329-1364.
Ramamurthy, T. 2004. A geo-engineering classification for rocks and rock masses. Int.
J. Rock. Mech. Min. Sci. 41(1): 89-101.
Rance, J.M., van As, A., Owen, D.R.J., Feng, Y.T., Pine, R.J. & Coggan, J. 2007.
Computational modelling of multiple fragmentation in rock masses with
application to block caving. In: Proc. 1st Canada-U.S. Rock Mechanics
Symposium. Vancouver. Canada. Taylor & Francis Group. London. Vol.1. pp.
477-484.
233
Read, J.R.L. & Lye, G.N. 1984. Pit slope design methods: Bougainville copper open cut.
In: Proc. 5th International Congress on Rock Mechanics, Melbourne. pp. C93-C98.
Reyes-Montes, J.M., Pettitt, W.S. & Young, R.P. 2007. Validation of a synthetic rock
mass model using excavation induced microseismicity. In: Proc. 1st Canada-U.S.
Rock Mechanics Symposium. Vancouver. Canada. Taylor & Francis Group.
London. Vol.1. pp. 365-369.
Rockfield. 2007. Rockfield Technology Ltd. Swansea. UK. http://www.rockfield.co.uk
Rockfield. 2003. Development of a block caving method using discrete elements.
Technical report PFR/1188. Rockfield Technology Ltd. Swansea. UK.
http://www.rockfield.co.uk
Rocscience. 2007. Phase2, RocLab, SLIDE. Rocscience Inc. Toronto.
http://www.rocscience.com
Rogers, S.F., Kennard, D.K., Dershowitz, W.S & van As, A. 2007. Characterising the in-
situ fragmentation of a fractured rock mass using a discrete fracture network
approach. In: Proc. 1st Canada-U.S. Rock Mechanics Symposium. Vancouver.
Canada. Taylor & Francis Group. London. Vol.1. pp. 137-143.
Rogers, S.F., Moffitt, K.M. & Kennard, D.T. 2006. Probabilistic slope and tunnel block
stability analysis using realistic fracture network models. In: Proc. 41st US
Symposium on Rock Mechanics. Golden. USA. ARMA/USRMS. Paper 06-1052.
Salim, A. & Stacey, T.R. 2006. Unstable rock slope behaviour in discontinuous rock
masses. In: Proc. Symp. SANIRE 2006. S, Afr. National Institute of Rock
Engineering. Rustenburg. pp 30-45.
Scavia, C. & Castelli, M. 1996. Analysis of the propagation of natural discontinuities in
rock bridges. In: Proc. Eurock96. Turin. pp. 445-451.
Scavia, C. 1995. A method for the study of crack propagation in rock structures.
Geotechnique. 45: 447-463.
Scavia, C. 1990. Fracture mechanics approach to stability analysis of rock slopes.
Engineering Fracture Mechanics. 35: 899-910.
Scholz, C.H. 2002. The mechanics of earthquakes and faulting, 2nd ed. Cambridge
University Press. Cambridge. 508 pp.
Serafim, J.L. & Pereira, J.P. 1983. Considerations on the geomechanical classification of
Bieniawski. In: Proc. Int. Symp. Engineering Geology and Underground
Construction. Lisbon. Theme II. Vol.1 pp II.33 – II.42.
Singh, U.K., Stephansson, O.J. & Herdocia, A. 1993. Simulation of progressive failure in
hangingwall and foot wall for mining with sublevel caving. Trans. Instn. Min.
Metall. 102: A188-A194.
Sjöberg, J. 1999. Analysis of large scale rock slopes. Doctoral Thesis. Luleå University
of Technology. Luleå. Sweden.
234
Society of Mining Metallurgy and Exploration (SME). 1992. Mining engineering
handbook. (ed. Hartman, H.L.). Society for Mining, Metallurgy and Exploration
Inc. 2394 pp.
Stacey, T.R. 2007. Slope stability in high stress and hard rock conditions. Keynote
address. In: Proc. Int. Symp. On Rock Slope Stability in Open Pit Mining and Civil
Engineering .Perth. Australian Centre for Geomechanics. pp 187-200.
Stacey, T.R. 2006. Considerations of failure mechanisms associated with rock slop
instability and consequences for stability analysis. J.S. Afr. Inst. Min Metall. 106:
485-93.
Stacey, T.R., Xianbin, Y., Armstrong, R. & Keyter, G.J. 2003. New slope stability
considerations for deep open pit mines. J.S. Afr. Inst. Min Metall. 103: 373-389.
Stacey, T R & Swart, A.H. 2001. Practical rock engineering practice for shallow and
opencast mines. SIMRAC. The Safety in Mines Research Advisory Committee.
66 pp.
Stacey, T.R. & Terbrugge, P.J. 2000. Open pit to underground - transition and
interaction. In: Proc. MassMin 2000. Brisbane. AusIMM. pp. 97-104.
Staub I., Fredriksson A. & Outters N. 2002. Strategy for a rock mechanics site
descriptive model development and testing of the theoretical approach.. Golder
Associates AB. SKB Report R-02-02.
Stead, D., Coggan, J.S., Elmo, D. & Yan, M. 2007. Modelling brittle fracture in rock
slopes: experience gained and lessons learned. In: Proc. Int. Symp. On Rock
Slope Stability in Open Pit Mining and Civil Engineering .Perth. Australian Centre
for Geomechanics. pp. 239-252.
Stead, D. & Coggan, J.C. 2006. Numerical modelling of rock slopes using a total slope
failure approach. In: Landslide from Massive Rock Slope Failure. (eds. Evans, S.,
Hermans, R., & Strom, A.) Springer, Dordrecht, Netherlands. pp. 131-142.
Stead, D., Coggan, J.S. & Eberhardt, E. 2004. Realistic simulation of rock slope failure
mechanisms: The need to incorporate principles of fracture mechanics.
SINOROCK 2004: Special Issue of Int. Journal of Rock Mechanics. 41(3): 6pp.
Stefanizzi, S. 2007. Numerical modelling of strain driven fractures in around tunnels in
layered materials. Ph.D. thesis. Politechnico di Torino. 177 pp.
Stefanizzi, S., Barla G. & Kaiser, P. 2007. Numerical modelling of strain driven fractures
in around tunnels in layered rock masses. In: Proc. 11th ISRM congress. Lisbon.
Portugal. pp. 971-974.
Stille, H. & Palmstrom, A. 2003. Classification as a tool in rock engineering. Tunnelling
and Underground Space Technology. 18(4): 331-345.
Szymakowski, J. 2007. Shear testing of soft rock masses. Geotechnical Testing Journal.
30(2): 133-140.
van As, A. 2003. “Subsidence Definitions for Block Caving Mines”. Technical report. Rio
Tinto. 48 pp.
235
Villegas, T. & Nordlund, E. 2008a. Numerical analysis of the hangingwall failure at the
Kiirunavaara mine. In: Proc. 5th International Conference and Exhibition on Mass
Mining (MassMin2008). Luleå. Sweden. Luleå University of Technology Press.
Luleå. Sweden. pp. 1073-1082.
Villegas, T. & Nordlund, E. 2008b. Numerical simulation of the hangingwall subsidence
using PFC2D. In: Proc. 5th International Conference and Exhibition on Mass
Mining (MassMin2008). Luleå, Sweden. pp .907-916.
Vyazmensky, A., Elmo, D., Stead, D. & Rance, J.R. 2007. Combined finite-discrete
element modelling of surface subsidence associated with block caving mining. In:
Proc. 1st Canada-U.S. Rock Mechanics Symposium. Vancouver. Canada. Taylor
& Francis Group. London. Vol.1. pp. 467-475.
Wei Z.Q. 1988. A fundamental study of the deformability of rock masses. Ph.D. thesis.
Imperial College. University of London. 268 pp.
Wang, C., Tannant, D.D. & Lilly, P.A. 2003. Numerical analysis of the stability of heavily
jointed rock slope using PFC2D. Int. J. Rock Mech. and Min. Sci. 40: 415-424.
Whittaker, B.N., Singh, R.N., Sun, G. 1992. Rock fracture mechanics: principles, design,
and applications. Elsevier. 570 pages.
Whittaker, B.N. & Reddish, D.J. 1989. Subsidence - occurrence, prediction and control.
Elsevier. 528 pp.
Wilson, E.D. 1958. Geologic factors related to block caving at San Manuel copper mine,
Pinal county, Arizona. Bureau of Mines. Rept. of Inv. 5336. 40 pp.
Wyllie, D.C. & Mah, C.W. 2004. Rock Slope Engineering: Civil and Mining. 4th ed. Spon
Press. London. 431 pp.
Woodruff, S. 1966. Methods of working coal and metal mines. Pergamon Press. 1545
pp.
Yan, M. 2008. Numerical modelling of brittle fracture and step-path failure: from
laboratory to rock slope scale. PhD thesis. Simon Fraser University. Vancouver.
Canada.
Yu, J. 1999. A contact interaction framework for numerical simulation of multi-body
problems and aspects of damage and fracture for brittle materials. PhD thesis.
University of Wales. Swansea. UK.
Zienkiewicz, O.C., Taylor, R.L. & Zhu, J.Z. 2005. The finite element method: its basics
and fundamentals. 6th ed. Butterworth-Heinemann. 752 pp.
Zipf, R. K. 1992. MULSIM/NL. Theoretical and programmer‟s manual. BuMines IC 9322.
236
APPENDIX A. PRELIMINARY STATISTICAL SUMMARY OF
CONCEPTUAL STUDY MODELLING RESULTS
0.45
0.4
0.35
0.3
Probability
0.25
0.2
0.15
0.1
0.05
0
175 200 225 250 275 300 325 350 375 400 425
Extent of major (≥10cm) vertical displacements, m
Fig. A.1 Probability (relative frequency) of the caving induced total extent of major
(≥10cm) vertical surface displacements based on 37 model runs.
0.45
0.4
0.35
0.3
Probability
0.25
0.2
0.15
0.1
0.05
0
175 200 225 250 275 300 325 350 375 400 425 450
Extent of major (≥10cm) horizontal displacements, m
Fig. A.2 Probability of the caving induced total extent of major (≥10cm) horizontal
surface displacements.
237
0.45
0.4
0.35
0.3
Probability
0.25
0.2
0.15
0.1
0.05
0
250 275 300 325 350 375 400 425 450 475 500 525 550 575 600
Mobilized rock mass volume normalized by
extracted ore volume, %
Fig. A.3 Probability of rock mass volume mobilized by caving, in percent of extracted ore
volume.
0.45
0.4
0.35
0.3
Probability
0.25
0.2
0.15
0.1
0.05
0
45 50 55 60 65 70 75 80
Minimum angles delineating the extent of major (≥10cm)
surface displacements, degrees
Fig. A.4 Probability of minimum angles delineating the extent of caving induced major
surface displacements.
0.45
0.4
0.35
0.3
Probability
0.25
0.2
0.15
0.1
0.05
0
0.65 0.7 0.75 0.8 0.85 0.9 0.95 1
Assymetry Index
Fig. A.5 Probability of subsidence asymmetry index.
238
APPENDIX B. DFN PARAMETERS FOR PALABORA
MODEL
Table B.1 Description of rock units present in the North Wall
Joint Set Dip Fracture Intensity, Fracture length,

Geological domain
direction / dip,◦ P10 m
160/07 0.1 10
Carbonatite
130/90 0.1 10
160/07 0.06 10
Foskorite 020/85 0.03 10
225/85 0.03 10
160/07 0.05 10
Micaceous
020/80 0.025 15
Pyroxenite
225/80 0.025 15
239
APPENDIX C. DEVELOPMENT OF NORTH WALL
FAILURE AT PALABORA MINE (PHOTOS)
Photographs presented in this Appendix are a courtesy of Rio Tinto Ltd and are
adapted with permission
Fig. C.1 Bench failure at the bottom of the pit - July 2004.
240
Fig. C.2 Developing failure of the North Wall - October 7th, 2004.
Fig. C.3 Failed North Wall – late May, 2005.
241

Numerical Modelling of Surface Subsidence Associated With Block Cave Mining Using A Finite Element / Discrete Element Approach

Transféré par

Informations du document

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Numerical Modelling of Surface Subsidence Associated With Block Cave Mining Using A Finite Element / Discrete Element Approach

Transféré par

Droits d'auteur :

Formats disponibles

NUMERICAL MODELLING OF SURFACE SUBSIDENCE

ASSOCIATED WITH BLOCK CAVE MINING USING A

Master of Applied Science, University of British Columbia, 2005

THESIS SUBMITTED IN PARTIAL FULFILLMENT OF

© Alexander Vyazmensky 2008

SIMON FRASER UNIVERSITY

All rights reserved. This work may not be

Through the use of an integrated FEM/DEM-DFN modelling technique this thesis

This research represents a valuable contribution to block caving geomechanics

Keywords: rock mechanics; numerical modelling; block cave mining; block

It is my pleasure to acknowledge the roles of several individuals who were

I would also like to acknowledge helpful suggestions from my committee members:

These acknowledgements would not be complete without mentioning my research

My deepest appreciation belongs to my family for their patience and understanding.

With regards to numerous questions about my future academic endeavours from

Approval ...................................................................Error! Bookmark not defined.

3.4.3 Continuum Degradation through Fracturing .................................. 41

5.3.3 Effect of Joint Shear and Normal Stiffness ................................. 111

7.3.4 In-situ Subsidence Characterization ........................................... 224

Fig. 1.1 Typical block caving mine layout. .......................................................... 2

Fig. 3.6 Comparison of contacts interaction handling in DEM and

Fig. 5.23 Subsidence crater formation for F5 model ......................................... 120

Fig. 6.9 Stress/displacement analysis of caving - open pit slope

Fig. A.3 Probability of rock mass volume mobilized by caving, in percent

Table 2.1 Degree of influence of various factors affecting surface

Table 5.8. Preliminary classification of caving induced surface subsidence

1.1 Block Caving Mining and Associated Surface Subsidence

Fig. 1.1 Typical block caving mine layout.

Fig. 1.3 Block caving mining and associated surface

Note: Figs. 1.1 - 1.3 are block caving animation screenshots

1.2 Research Objectives

This thesis has three principal research objectives:

1. Introduce a new methodology for the numerical analysis of surface

2. Improve understanding of subsidence phenomena and rock mass

3. Investigate mechanisms of block caving induced instability in large open pit

Chapter 3 discusses the available modelling approaches for simulation of block

Chapter 4 addresses the question of selection of equivalent rock mass properties

Chapter 5 gains fundamental understanding of the block caving subsidence

Chapter 6 addresses the problem of slope instability/subsidence in a large

Finally, Chapter 7, contains research conclusions, outlines key scientific

Underground mining induced surface subsidence has been discussed extensively

The objective of this chapter is to summarize the current state of knowledge on

stages in surface subsidence development and associated rock mass

2.2 Stages of Subsidence Development and Associated Rock

1. Collapse of rock progresses upward from the extraction level as ore is

3. Further extraction of caved ore from below results in increased

4. Continued extraction of ore will result in breaching of the surface. The

a. Caving initiation b. Cave advance towards surface

c. Formation of surface depression d. Cave breakthrough and crater

Fig. 2.1 Development of surface subsidence induced by block caving

The caved rock zone corresponds to a zone of active cave movement. It is

The large scale cracking or fractured zone is characterized by an irregular broken

continuous subsidence zone Stable Zone

caved rock or crater zone

Fig. 2.3 Example of surface subsidence zonation at Northparkes mine

The angle of subsidence defines the limits of surface deformations. It should be

2.4 Factors Governing Subsidence Development

Degree of influence on surface subsidence

It should be emphasized that the published literature sources, surveyed as part

According to van As (2003) many observations of the influence of discontinuities

UNIFORM ROCK MASS subsidence

Fig. 2.5 Beneficial and detrimental effects of major weakness planes

2.5 A Block Caving Subsidence Analysis Toolbox