Evaluation of mechanisms contributing to valley closure subsidence effects under irregular topographic conditions
Chengguo Zhang
A thesis in fulfilment of the requirements for the degree of
Doctor of Philosophy
School of Mining Engineering
Faculty of Engineering
August 2014 Abstract
Valley closure subsidence has been observed for decades in Australia and overseas where underground extractions have occurred beneath or in close proximity to valleys and other forms of irregular topographies. Valley closure is referred to as the inward movements of the valley sides towards the valley centreline. Upsidence is the reduction in subsidence or relative upward movement at the bottom of the valley. Due to the complexity of the local geology and interplay between several geological, topographical and mining factors, the underlying mechanisms that actually cause this behaviour are not fully understood.
Numerical investigations are conducted to investigate stress related failure mechanisms that may contribute the observed valley closure subsidence. Numerical models are developed using the Distinct Element Code, UDEC and 3DEC, based on assessments of the geological conditions in the Southern Coalfield, New South Wales, Australia. The numerical models are then systematically calibrated and validated against field observations and empirical incremental profiles. The positive correlations indicate that the models are capable of replicating the features of mining induced subsidence and horizontal movements. The calibrated models are subsequently extended to develop several hypothetical scenarios (two-dimensional and three-dimensional) to quantify the effects of the geological, topographical and mining factors on valley closure subsidence and their inter-relationships. The numerical modelling predictions of valley closure subsidence are consistent with field observations. The results of this research indicate that the major influences on valley closure subsidence are longwall locations with respect to valley and the horizontal compressive stress. A stress arching concept is proposed and the redistribution of the horizontal compressive stress within the valley
I results in a strong pushing effect on valley walls, leading to pronounced bedding plane shearing that contributes to greater closure values. The roles of valley geometric factors, depth of cover and geological features around valley have also been quantitatively identified in this study.
The results have applications in the study of both the underlying mechanisms that lead to this non-conventional subsidence behaviour, and how these should be incorporated in future valley closure subsidence prediction and mitigation of its impacts on natural features.
Keywords subsidence, valley closure, upsidence, numerical modelling, mechanisms, valley closure prediction
II Acknowledgements
There are a number of people without whom this thesis might not have been written, and to whom I am greatly indebted.
First and foremost I would like to express my deepest appreciation to my supervisors
Dr. Rudrajit Mitra, Dr. Joung Oh and Professor Bruce Hebblewhite for their consistent encouragement, support and understanding over the past four years. I have been fortunate to work with them. Dr. Rudrajit Mitra has been supportive since the day I began working on my PhD. He has guided me not only by providing a research assistantship for almost four years, but also by giving his friendship and academic and emotional support through the rough road to finishing my PhD. I am very grateful for his kindness, patience, and motivation. Dr. Joung Oh njoi ed my supervision team during the third year of this thesis, and he has been a strong and encouraging adviser to me throughout the toughest time of my research. He is always approachable to discuss any issues.
I greatly appreciate the support from my co-supervisor Professor Bruce Hebblewhite.
He interviewed me in 2009 back in China and offered me the opportunity to pursue my doctoral degree in UNSW. My special thanks to Prof. Hebblewhite, he made it possible for me to achieve this great chapter in my life.
Many thanks to Dr. Greg Tarrant from Metropolitan Coal Mine for believing in my research and providing invaluable advice and field data access. I am also very grateful to Don Kay, James Barbato and Peter DeBono from Mine Subsidence Engineering
Consultants Pty Ltd. They provided excellent help with data collection and model
III validation, and they have taught me a lot. I would also like to acknowledge the help and assistance of staff and friends in the School of Mining Engineering, UNSW.
It is important that I acknowledge here the love and encouragement of my family. My parents and grandparents have always been there to support me in all my pursuits. To my fiancée Lingyun Guo, who has been by my side throughout this PhD, living every single minute of it, and giving up many things for me to accomplish this PhD. Thank you.
IV Table of Contents
ABSTRACT I
ACKNOWLEDGEMENTS III
LIST OF FIGURES VIII
LIST OF TABLES XV
CHAPTER 1. INTRODUCTION 1
1.1 Problem statement 1
1.2 Research objective and scope 4
1.3 Methodology 5
1.4 Thesis outline 6
CHAPTER 2. LITERATURE REVIEW 9
2.1 Introduction 9
2.2 Conventional mine subsidence 9
2.3 Horizontal stresses in the Sydney Basin 15
2.4 Horizontal ground movements 23 2.4.1 Systematic horizontal movements 23 2.4.2 Regional horizontal movements 26 2.4.3 Topography related movements 29
2.5 Previous numerical modelling studies 30 2.5.1 Finite element method 30 2.5.2 Finite difference method 33 2.5.3 Boundary element method 36 2.5.4 Distinct element method 38 2.5.5 Hybrid method 43
2.6 Postulated mechanisms of valley closure and upsidence 48 2.6.1 Valley stress relief 48 2.6.2 Effects of horizontal stress 52 2.6.3 Lateral dilation mechanism 56
V 2.6.4 Movements towards goaf area 59 2.6.5 Rigid block model 60 2.6.6 Other field observations 63
2.7 Conclusions 64
CHAPTER 3. LABORATORY INVESTIGATION OF MATERIAL PROPERTIES 67
3.1 Introduction 67
3.2 Overview of geology in the Southern Coalfield 67
3.3 Sampling of rock specimens 75
3.4 Laboratory testing of rock materials 80 3.4.1 Determination of the uniaxial compressive strength 80 3.4.2 Determination of the tensile strength 86 3.4.3 Determination of the strength in triaxial compression 88
3.5 Conclusions 94
CHAPTER 4. MODEL SETUP AND VALIDATION 96
4.1 Introduction 96
4.2 Modelling technique 96
4.3 Estimation of model input data 97 4.3.1 Mechanical properties of intact rock 97 4.3.2 Properties of rock structures 106 4.3.3 In situ stress 111
4.4 Model validation 112 4.4.1 Overview of mining setting in Metropolitan Colliery 112 4.4.2 Overview of the Incremental Profile Method 116 4.4.3 Single longwall panel excavation 118 4.4.4 Multiple panel excavations 124
4.5 Conclusions 137
VI CHAPTER 5. NUMERICAL INVESTIGATION OF THE MECHANISMS CONTRIBUTING TO VALLEY CLOSURE SUBSIDENCE 139
5.1 Introduction 139
5.2 Development of UDEC models 139
5.3 Valley bulging movements 140
5.4 Effects of geological and geotechnical factors on valley closure subsidence 144 5.4.1 Influence of longwall location relative to valley 145 5.4.2 Influence of horizontal stress 156 5.4.3 Influence of valley sloping angle 162 5.4.4 Influence of cover depth above longwall 168 5.4.5 Influence of valley shape 177 5.4.6 Influence of cross bedding plane degree 182
5.5 Conclusions 188
CHAPTER 6. THREE-DIMENSIONAL NUMERICAL MODELLING OF THE POTENTIAL FACTORS CONTRIBUTING TO VALLEY CLOSURE SUBSIDENCE 190
6.1 Introduction 190
6.2 Model setup 190
6.3 Comparison of 3DEC and UDEC results 194
6.4 Three-dimensional parametric analysis 197 6.4.1 Orientation of mining relative to valley 198 6.4.2 Orientation of the major horizontal stress relative to valley and mining 217 6.4.3 Longwall extractions with different offset distance to valley 221
6.5 Conclusions 226
CHAPTER 7. CONCLUSIONS AND RECOMMENDATIONS 228
7.1 Major conclusions 228
7.2 Contributions of this study 230
7.3 Recommendations for future research 232
REFERENCES 234
VII List of Figures
Figure 1.1: Major water bodies, mining leases and upland swamps in the Southern Coalfield .....1
Figure 1.2: (a) Buckling of near surface strata associated with upsidence; (b) override of bedding slabs caused by shear failures in Waratah Rivulet ...... 2
Figure 2.1: Different types of discontinuous subsidence: (a) crown hole; (b) chimney caving; (c) plug subsidence; (d) solution cavities; (e) block caving; (f) hangingwall caving ...10
Figure 2.2: Typical cross section of a longwall face ...... 11
Figure 2.3: Schematic model of mining induced overburden deformation zones ...... 12
Figure 2.4: Typical section of subsidence trough, illustrating various subsidence parameters ...13
Figure 2.5: Effects of panel width, depth of cover on subsidence ...... 15
Figure 2.6: Sydney Basin Stress Map ...... 17
Figure 2.7: Relationships of stress magnitude and depth in the Sydney Basin and in its sub- regions. σH, σh and σv is the major, intermediate and minor principal stress ...... 19
Figure 2.8: Relationship of horizontal to vertical stress ratio and depth...... 21
Figure 2.9: Horizontal movements due to extraction of Longwall 17 at Tower Colliery ...... 24
Figure 2.10: Horizontal deformation of rock masses above longwall extraction ...... 25
Figure 2.11: Far field horizontal movement database in the Southern Coalfield ...... 28
Figure 2.12: Predicted plastic and tensile zone with FEM ...... 32
Figure 2.13: (a) closure and upsidence after equilibrium, (b) sliding of bedding plane ...... 35
Figure 2.14: Failure modes of the strata in the weathered zone ...... 36
Figure 2.15: Mining induced displacement pattern of valley ...... 37
Figure 2.16: (a) Total fractures during undermining of gorge, (b) fractures remaining open at completion of mining ...... 40
Figure 2.17: Shearing displacements in gorge region after extraction ...... 41
Figure 2.18: Explicit fracturing and displacement at the base of the gorge ...... 42
Figure 2.19: (a) FLOMEC Mesh and surface displacement contours, (b) vectors of horizontal displacement due to underground extraction ...... 45
VIII Figure 2.20: Simulation of subsidence development using (a) an equivalent continuum approach, (b) a mixed discrete/equivalent approach ...... 46
Figure 2.21: (a) LGM 3-D Model, (b) ANSYS 2-D valley model ...... 47
Figure 2.22: Valley cross-section...... 49
Figure 2.23: Valley stress pattern: A-pre-erosion, B-post erosion ...... 50
Figure 2.24: Measured stress directions by comparing with topography ...... 51
Figure 2.25: Stress measurement plan of Warragamba Dam showing the in situ stress direction ...... 51
Figure 2.26: Cross section of gorge showing in situ stresses in Warragamba Dam ...... 52
Figure 2.27: Redistribution of horizontal stresses due to underground mining beneath a valley 53
Figure 2.28: Notch effect with concentration of stresses ...... 54
Figure 2.29: Valley structures in flat-lying rocks ...... 55
Figure 2.30: Imbalance of dilation forces in sloping terrain ...... 58
Figure 2.31: Assumed horizontal movements within strata layers near cliff line ...... 60
Figure 2.32: Block displacements on curved slope ...... 61
Figure 2.33: a) Valley closure in sagging phase of dynamic wave; b) valley closure in hogging phase ...... 62
Figure 2.34: Model for valley closure and upsidence ...... 63
Figure 2.35: Rock fracturing features associated with valley closure in the Southern Coalfield 64
Figure 3.1: The Sydney-Gunnedah Basin and its coalfields ...... 68
Figure 3.2: Sydney Basin sedimentary stratigraphic sections ...... 69
Figure 3.3: Generalised stratigraphy column of the Southern Coalfield ...... 76
Figure 3.4: Field logs for laboratory tests...... 78
Figure 3.5: (a) Core drilling machine and (b) drying oven...... 79
Figure 3.6: Schematic of MTS rock mechanics testing system ...... 81
Figure 3.7: Uniaxial compression test setup for rock specimen ...... 82
Figure 3.8: Typical failure mode of specimens after uniaxial compression tests ...... 83
Figure 3.9: Stress-strain curve of one of the Coal Cliff Sandstone specimens ...... 85
IX Figure 3.10: Typical testing setup for Brazilian test ...... 87
Figure 3.11: Triaxial test setup in the laboratory ...... 89
Figure 3.12: Curve fitting of Hoek-Brown and Mohr-Coulomb envelope to triaxial test data ....94
Figure 4.1: Scale effect on uniaxial compressive strength of intact rock ...... 99
Figure 4.2: Scale effects relations developed by Yoshinaka and Hoek & Brown ...... 101
Figure 4.3: Relationship of UCS and Young’s Modulus at 50% of UCS, showing engineering classification for intact sedimentary rocks ...... 102
Figure 4.4: Relationship between E and UCS for different rock types ...... 103
Figure 4.5: Joint behaviour under normal and shear loading ...... 109
Figure 4.6: Scale effect on peak shear stiffness ...... 110
Figure 4.7: Location of Metropolitan Colliery ...... 113
Figure 4.8: Layout of Longwall 1 to 18 and D-Line in Metropolitan Colliery ...... 114
Figure 4.9: Location of Longwall 20, monitoring line C Line and 9CW Line ...... 115
Figure 4.10: Primary bolt pattern in Metropolitan Colliery ...... 116
Figure 4.11: Graph for the prediction of maximum subsidence for various extraction conditions using Incremental Profile Method ...... 117
Figure 4.12: Layout of Longwall 20 and Monitoring Line 9C ...... 119
Figure 4.13: UDEC model geometry for simulating extraction of Longwall 20 ...... 120
Figure 4.14: Comparison of modelled and monitored subsidence troughs induced by extraction of Longwall 20 ...... 120
Figure 4.15: Generic UDEC model for single longwall extraction ...... 122
Figure 4.16: Modelled maximum subsidence, compared with the prediction profile developed by the Empirical Incremental Profiles method ...... 124
Figure 4.17: Location of Longwall 11, 12 and the ground surface level ...... 125
Figure 4.18: UDEC model geometry for multiple longwall extractions ...... 126
Figure 4.19: Chain pillar failure in the UDEC model ...... 127
Figure 4.20: Phase2 model setup for extraction of Longwall 11 and 12 ...... 128
Figure 4.21: Chain pillar failure and stress contour from Phase2 modelling ...... 128
Figure 4.22: Pillar failure subsidence ...... 130
X Figure 4.23: Pillar compression concept for subsidence ...... 131
Figure 4.24: Contour of vertical stress around chain pillar in the UDEC model ...... 131
Figure 4.25: Comparison of the modelled subsidence troughs to those measured in field ...... 134
Figure 4.26: Comparison of horizontal displacement predicted by UDEC to those measured in field ...... 134
Figure 4.27: Comparison of modelled incremental maximum subsidence to the prediction profiles developed by the Empirical Incremental Profiles with different modelled width to depth ratios ...... 137
Figure 5.1: Vertical section of a river valley in the Southern Coalfield ...... 140
Figure 5.2: Bulging effect at the base of a valley ...... 141
Figure 5.3: Valley base failure mechanisms ...... 141
Figure 5.4: Displacement vectors showing valley bulging ...... 142
Figure 5.5: Horizontal stress contours in the valley ...... 143
Figure 5.6: Yield, tensile and shearing failure for the valley ...... 143
Figure 5.7: Sketch showing the offset distance from longwall to valley ...... 146
Figure 5.8: Locations of single longwall panel with varying offset distance to valley, showing mining induced failure zones ...... 147
Figure 5.9: Valley closure for single longwall extraction ...... 148
Figure 5.10: Smax in the valley for single longwall extraction ...... 148
Figure 5.11: Failure zones for sequential extraction of 6 panels towards the valley, showing the longwall layouts in regards to the valley ...... 150
Figure 5.12: Valley closure for multiple longwall mining towards valley ...... 151
Figure 5.13: Smax in valley for multiple longwall mining towards valley...... 152
Figure 5.14: Valley closure for multiple longwall mining away from valley ...... 152
Figure 5.15: Smax in valley for multiple longwall mining away from valley ...... 153
Figure 5.16: Longwall layouts in the Metropolitan Colliery ...... 154
Figure 5.17: Comparison of modelled (blue) and field monitored (red) valley closure when mining towards valley ...... 154
XI Figure 5.18: Comparison of modelled (blue) and field monitored (red) valley closure when mining away from valley ...... 155
Figure 5.19: Valley closure and Smax in valley for different horizontal stresses ...... 157
Figure 5.20: Horizontal stress contour for different longwall extractions: (a) mining with the angle of draw of 33°, (b) mining of the critical longwall and (c) mining directly beneath the valley ...... 159
Figure 5.21: Shear displacements and horizontal stress contours in the model ...... 160
Figure 5.22: Yield, tensile and shearing failure for different Sigma H/ Sigma V ratios: (a) 1, (b) 2 and (c) 3 ...... 162
Figure 5.23: Valley closure and Smax in valley for different valley sloping angles ...... 163
Figure 5.24: Shearing on the valley walls for different valley sloping angles ...... 164
Figure 5.25: Contour of the shear displacement around the weak bedding plane ...... 165
Figure 5.26: Horizontal stress contours for different valley sloping angles ...... 167
Figure 5.27: Horizontal stress at a -70m monitoring line for different valley sloping angles. ..168
Figure 5.28: Model grid for different cover depths: (a) 472 m, (b) 408 m, (c) 326 m and (d) 272 m, with location of the longwall marked in red ...... 169
Figure 5.29: Relationship between valley closure and panel width/depth ratio ...... 170
Figure 5.30: Mining induced horizontal stress contours for varying panel width/depth ratios: (a) 0.35, (b) 0.4, (c) 0.5 and (d) 0.6...... 173
Figure 5.31: Recorded horizontal compressive stresses at the valley base ...... 174
Figure 5.32: Model setup with uniform blocks in the overburden ...... 175
Figure 5.33: Comparison of valley closure for uniform block and multiple formations model 175
Figure 5.34: Modelling results in comparison with filed observations ...... 176
Figure 5.35: U-shaped and V-shaped valley ...... 178
Figure 5.36: (a) Sandstone valley model and (b) shale valley model ...... 179
Figure 5.37: Comparison of the horizontal displacement for the sandstone (U-shaped) and shale (V-shaped) scenario ...... 180
Figure 5.38: Comparison of the subsidence for the sandstone (U-shaped) and shale (V-shaped) scenario ...... 181
Figure 5.39: Cross bedded sandstone units ...... 182
XII Figure 5.40: UDEC model with cross bedding plane angle of (a) 5° and (b)10° ...... 183
Figure 5.41: Valley closure for different degree of bedding plane ...... 185
Figure 5.42: Smax in valley for different degree of bedding plane ...... 185
Figure 5.43: Horizontal stress contour around the valley for different bedding plane dipping angle ...... 186
Figure 5.44: Plastic zones and shearing within the valley for different bedding plane dipping angle ...... 187
Figure 6.1: 3DEC full model mesh in different formations ...... 191
Figure 6.2: An example of the model equilibrium process, showing the displacement contour with (a) the unbalanced force, and (b) vertical displacements of three monitoring points ...... 193
Figure 6.3: Model grid system for: (a) UDEC model, (b) 3DEC slice model and (c) 3DEC full model ...... 195
Figure 6.4: Comparison of horizontal movements for the three scenarios at ground level ...... 196
Figure 6.5: Comparison of subsidence profiles for the three scenarios ...... 196
Figure 6.6: Schematic view of longwall position relative to valley ...... 199
Figure 6.7: The position of longwall with respect to the valley: (a) mining parallel to valley and (b) mining perpendicular to valley ...... 200
Figure 6.8: Plot of the horizontal displacement contour for mining (a) parallel to valley and (b) perpendicular to valley ...... 201
Figure 6.9: Plot of the horizontal displacement contour in the flat terrain for mining (a) parallel to valley and (b) perpendicular to valley ...... 202
Figure 6.10: Mining induced horizontal movements when the longwall is directed along the valley ...... 204
Figure 6.11: Mining induced subsidence profiles when the longwall is directed along the valley ...... 204
Figure 6.12: Mining induced horizontal movements when the longwall is directed perpendicular to the valley ...... 205
Figure 6.13: Mining induced subsidence profiles when the longwall is directed perpendicular to the valley ...... 205
Figure 6.14: Definition of angle between valley and longwall ...... 207
XIII Figure 6.15: Field observed and predicted valley closure for different alignment angles between valley and longwall ...... 207
Figure 6.16: Major horizontal stress distribution for mining parallel to valley ...... 209
Figure 6.17: Major horizontal stress distribution for mining perpendicular to valley ...... 209
Figure 6.18: Shearing displacement contour in the valley base for mining parallel to valley ...210
Figure 6.19: Shearing displacement contour in the valley base for mining normal to valley ....210
Figure 6.20: Location of the stress monitoring point ...... 212
Figure 6.21: Monitored horizontal stress in the base of the valley when mining parallel to the valley ...... 213
Figure 6.22: Monitored horizontal stress changes in the near surface sandstone ...... 214
Figure 6.23: Location of the stress cell and longwalls ...... 215
Figure 6.24: Summary of horizontal compressive stress changes at the valley floor ...... 215
Figure 6.25: Monitored principal horizontal stress in the base of the valley when mining perpendicular to the valley ...... 216
Figure 6.26: Plot of the horizontal displacement contour for mining (a) parallel to valley and (b) perpendicular to valley, when the major horizontal stress directed along the valley ...... 218
Figure 6.27: Comparison of horizontal displacements for different S1 orientations ...... 219
Figure 6.28: Comparison of subsidence profiles for different S1 orientations ...... 220
Figure 6.29: Summary of the trend of valley closure and subsidence with different longwall locations, when S1 was perpendicular to valley ...... 223
Figure 6.30: Comparison of the trend of valley closure and subsidence with varying S1 orientations, for different longwall locations ...... 225
XIV List of Tables
Table 2.1: Measured principal stresses in the Southern Coalfield ...... 18
Table 3.1: Subdivision of the sequence in the Southern Coalfield ...... 69
Table 3.2: Stratigraphic horizons from PM03 ...... 76
Table 3.3: Results of uniaxial compressive strength tests...... 84
Table 3.4: Calculated Young’s modulus and Poisson’s ratio ...... 85
Table 3.5: Calculated Young’s modulus, Poisson’s ratio, bulk modulus and shear modulus...... 86
Table 3.6: Results of the Brazilian test ...... 88
Table 3.7: Results of the Triaxial Compression tests for different rock types ...... 91
Table 4.1: Stratigraphic columns for numerical modelling ...... 97
Table 4.2 Expression of relationship between E and UCS ...... 103
Table 4.3: Estimation of Young’s modulus ...... 104
Table 4.4: Typical compressive/tensile strength relations for coal measure rocks ...... 105
Table 4.5: Adopted mechanical properties for stratigraphic units used in the models ...... 106
Table 4.6: Typical discontinuity spacing in main stratigraphic units in the Southern Coalfield
...... 108
Table 4.7: Adopted mechanical properties for discontinuities used in the models ...... 111
Table 4.8: Summary of the parameters for model setup and the subsidence results ...... 123
Table 4.9: Summary of parameters selected in the UDEC validation together with the
comparison of the modelled and empirical predicted results ...... 136
Table 6.1: Valley closure and subsidence when mining parallel/perpendicular to valley ...... 206
Table 6.2: Valley closure and subsidence for varying S1 and mining directions ...... 221
XV Chapter 1. Introduction
1.1 Problem statement
In the coalfields of New South Wales, Australia, particularly in the Southern Coalfield, numerous underground longwall panels have been, or are proposed to be, extracted beneath natural features such as river valleys, water catchments and cliffs. This is the nature of the region’s geomorphology. Figure 1.1 illustrates the geographic conditions, major water catchments and mining leases in the Southern Coalfield.
Figure 1.1: Major water bodies, mining leases and upland swamps in the Southern Coalfield (NSW Dept. of Planning, 2008a).
When mining occurs beneath or in the vicinity of valleys and other forms of irregular surface topography, the valley sides are observed to move inwards towards the valley centreline, while the observed vertical subsidence at the base of the valley is less than
1 would be expected in flat terrain. The convergence of two sides of the valley is termed valley closure, and the reduction in subsidence or relative upward movement at the bottom of the valley is referred to as upsidence.
Both the past and potential future impacts of mining induced ground movements on significant natural features in the Southern Coalfield have been recognised to be issues of significant community interest and government concerns. The mine subsidence effects on the natural features are associated with valley closure effects, and risks to water flows and ecosystems include loss of surface flow to the subsurface, river bed failure (cracking of river beds and underlying strata), loss of standing pools, methane emission, cliff falls and rock falls, as illustrated in Figure 1.2.
(a) (b)
Figure 1.2: (a) Buckling of near surface strata associated with upsidence; (b) override of bedding slabs caused by shear failures in Waratah Rivulet (NSW Dept. of Planning, 2008b).
Valley closure subsidence effects have been observed for many years in Australia, and research coupled with experience over past decades has identified the phenomenon of valley closure and upsidence as being a significant component of non-conventional subsidence. Currently valley closure subsidence is mainly predicted using the state-of- the-art empirical methods based on an extensive database of field measurements. The major prediction method for the observed features was originally developed at the
2 beginning of the 2000s (Waddington and Kay, 2001, 2002). It was built on a series of upper-bound prediction lines that illustrated the interrelationships between the valley related movements and four contributory factors, being the lateral distance from the main gate edge of the mined incremental panel, the longitudinal distance from the nearest edge of the mined panel, the maximum subsidence from the mined panel and the valley depth. This method was based on valley closure subsidence movements primarily gathered in the Southern Coalfield. Due to the lack of ongoing monitoring data, the empirical method was deliberately conservative over-predicting both upsidence and closure for many cases. An updated project was conducted to deliver more appropriate and accurate predictions of valley closure movements, by assessing a new extensive raw survey database including more than 9000 observed valley related movement cases in a huge diversity of coalfields (MSEC, 2014). Multi-variant data were analysed to provide probabilistic predictions for varying factors rather than merely introducing an upper- bound prediction. Thirty additional potential factors were identified that could influence the observed valley closure movements (MSEC, 2014). Despite the notable results, it should be recognised that there are some key limitations to empirical prediction techniques. These methods are generally based on field measurements over similar mining conditions, which may limit application to other areas. For example, where the depth of cover is much shallower than that in the Southern Coalfield, the prediction of the valley closure movements needs to be modified, because the conventional mining induced horizontal movement can represent a larger proportion of the observed closure movements (MSEC, 2014). Another common problem among these empirical methods is the failure to address the influence of sub-surface geological features including the material properties of rock mass within the overburden, pre-mining stress conditions
3 and mining induced failure occurrences around the valley such as shearing and fracturing.
Compared to the empirical methods, numerical modelling has the potential to replicate the observed valley related movements and investigate the fracture development. A comprehensive literature review of the current numerical modelling methods for analysing valley closure subsidence is performed in this study, including a wide range of numerical models using both continuum and discontinuous modelling codes. The accuracy of prediction of valley closure subsidence is found to be sensitive to the input material properties and the quality of the model calibration process. Moreover, previous numerical modelling studies were usually limited to a general description of the influence of potential contributing factors on valley closure subsidence and the results were qualitative rather than quantitative.
To sum up, there is now an extensive database of measurements of valley closure subsidence, and a number of empirical and numerical prediction techniques are available in the literature. Valley closure subsidence development has been recognised as a result of interplay between several governing factors. However the failure mechanisms contributing to this phenomenon are still not fully understood, due to the complexity of the local geology and inter-relationships of the contributing factors. This is the current knowledge gap in the prediction of such behaviour.
1.2 Research objective and scope
The objective of this research is to conduct a fundamental investigation to improve the understanding of the underlying mechanisms accounting for the observed valley closure subsidence behaviour. A comprehensive program of numerical modelling investigations
4 is carried out in this study in order to evaluate and quantify the influence of a number of significant geological and geotechnical factors and their inter-relationships. This study is designed to develop a new, advanced numerical method (or tool) to predict the movement in valley areas associated with underground mining.
The primary focus of this study is to determine the failure mechanisms and the role of key geological variables influencing valley closure subsidence in the Southern
Coalfield. This study does not attempt to address all the elements that may contribute to valley closure subsidence. Other aspects, for example hydrogeological factors such as the rainfall over the valley, far field mining induced movements and time-dependent subsidence are beyond the scope of this study, all of which could be the recommendations for the future research.
1.3 Methodology
The methodology used in this study includes:
At the earliest stage of this study data is acquired from existing Southern
Coalfield databases. The data is mainly derived from the Metropolitan Mine in
the Southern Coalfield and Mine Subsidence Engineering Consultants (MSEC),
regarding mine subsidence and geological details, mining geometry, measured
ground movements and empirical prediction methods.
A series of hypothetical mining/surface topography/overburden geology
scenarios are developed to represent the typical conditions in the Southern
Coalfield, in order to study the mining induced valley closure subsidence effects.
5 A range of numerical models, both in two-dimensional and three-dimensional,
using continuum and discontinuous numerical modelling codes are reviewed and
evaluated for applicability to represent these scenarios. As a result, Universal
Distinct Element Code (UDEC) and Three Dimensional Distinct Element Code
(3DEC) are selected for this study, and the models are validated against the
extensive calibration database.
Comprehensive numerical modelling investigations are conducted using UDEC
and 3DEC including a series of parametric analyses for varying geological and
geotechnical factors. The outcomes of the modelling investigations are then
analysed against the calibration data.
Recommendations are developed with regard to the underlying mechanisms
contributing to the observed valley closure subsidence behaviour.
1.4 Thesis outline
This thesis has been structured to analyse the valley closure subsidence effects systematically through seven chapters. A general introduction to the study, including the research problem, objective and methodology, is presented in this introductory chapter.
A comprehensive literature review follows in Chapter 2. The review summarizes the knowledge of conventional subsidence, the stress state in the Sydney Basin and the mining induced horizontal movements. Then the relative numerical modelling investigations and the available hypothesised mechanisms for valley closure subsidence are critically evaluated, with a number of limitations being identified.
Chapter 3 of this thesis introduces the geology of the Sydney Basin. A range of laboratory tests on rock specimens obtained from the Metropolitan Mine are conducted
6 to obtain the material mechanical properties. After the assessment of the geomechanical conditions in the Southern Coalfield, Chapter 4 then focuses on the development of the numerical models and validation of their capacity to satisfactorily reflect the mining induced ground movements. The estimation of the strength and deformation parameters of the overburden is determined taking into account the laboratory test results of standard specimens as described in Chapter 3, along with the scale effects and empirical experiences in assessment of the rock engineering geologic features. Thereafter, the modelling results are validated qualitatively and quantitatively against the field measurements and a series of empirical prediction methods, which demonstrate high predictive accuracy for surface horizontal movement and subsidence.
Chapter 5 provides fundamental understanding of the mechanisms contributing to the observed valley closure subsidence behaviour. A series of hypothetical overburden geology/surface topography scenarios are developed using the two-dimensional UDEC program, representing the broad range of typical conditions in the Southern Coalfield.
Critical geological and geotechnical factors governing the valley closure subsidence behaviour have been identified in this chapter.
Chapter 6 further addresses the significant contributing factors of valley closure subsidence effects by extending the two-dimensional modelling to the three- dimensional simulation analyses. A series of analyses are done using the three- dimensional distinct element code 3DEC to investigate in particular the effects of the three-dimensional longwall positional factors and the three-dimensional major horizontal stress on valley closure subsidence behaviour.
7 Ultimately, Chapter 7 summarises the major finding of this research, and outlines the key contributions of this study. Future research avenues are discussed at the end of the chapter.
8 Chapter 2. Literature review
2.1 Introduction
This chapter presents a review of available literature covering the systematic and non- systematic mine subsidence, in situ stresses in the Sydney Basin, previous numerical modelling for mining induced subsidence and overall summaries of the hypothesized mechanisms contributing to valley closure subsidence. The objective of this review is to present the background knowledge for the investigation of the valley closure subsidence effects.
2.2 Conventional mine subsidence
Subsidence is typically defined as the downward motion of ground surface resulting from the underground extraction of an orebody. This motion occurs when the orebody is extracted from either a moderate or a greater depth of cover. As suggested by Brady and
Brown (2004), the mine subsidence can be divided into two categories: discontinuous subsidence and continuous subsidence. Discontinuous subsidence, which is also referred to as pit subsidence (Holla and Barclay, 2000), is characterised by large surface displacement over limited surface area, where the mining depth is shallow and the overburden is brittle and weak. There are several types of discontinuous subsidence, as presented in Figure 2.1. It should be noted that in the Southern Coalfield the underground coal mining depth is generally more than 300 m, and discontinuous subsidence is not a common phenomenon. It is beyond the scope of this study to discuss the discontinuous subsidence in detail.
9
Figure 2.1: Different types of discontinuous subsidence: (a) crown hole; (b) chimney caving; (c) plug subsidence; (d) solution cavities; (e) block caving; (f) hangingwall caving (Brady and Brown, 2004).
Continuous subsidence is also termed as trough subsidence, with the subsidence generally in the form of a trough and occurs in most longwall mining operations. In longwall mining, large rectangular blocks of coal (referred to as longwall panels typically several hundred metres wide and several kilometres long) are totally extracted by longwall shearer. A typical section through a coal longwall face is presented in
Figure 2.2.
10
Figure 2.2: Typical cross section of a longwall face.
As can be seen from the figure, when coal seam is extracted the immediate roof strata fail in tension with fractures developed across the unit, and then collapse into the goaf area. The rock units above the immediate roof gradually sag to the void beneath them and fractures in rock mass propagate upwards to the surface. The overlying strata bending and subsidence develop upward until reaching the surface and forming a subsidence basin. The mining induced strata displacement zones have been recognised in numerous studies using different terms. For example, Kratzsch (1983) introduced four zones to describe the overburden, those zones being the immediate roof, the main roof, the intermediate zone and the surface zone. McNalley et al. (1996) identified three zones, namely the caved zone, fractured zone and elastic zone. Forster (1995) presented a hydrogeological model for the Central Coast and the overburden strata were divided into four zones with different deformation characteristics, namely the caved zone, fractured zone, constrained zone and surface zone, as illustrated in Figure 2.3. This model was further calibrated against the hydrogeological data in the Central Coast
Region to be applicable to underground mining designs.
11
Figure 2.3: Schematic model of mining induced overburden deformation zones (Forster, 1995).
The key features of each zone are described as follows:
Caved zone – Comprising loose rock blocks detached from the roof. Large voids
can exist.
Fractured zone – Lying on top of the caved zone and rock masses experience
significant bending, fracturing, joint opening and bedding separation.
Constrained zone – Comprising confined rock strata above the disturbed zone
which have sagged slightly but have absorbed most of the strain energy without
suffering much fracturing. Some bedding separation and slippage can be presented
and plastic deformation occurs for weak and soft units.
Surface zone – Unconfined strata at the surface in which mining induced horizontal
tensile and compressive strains may result in surface cracking or ground heaving.
There are a range of parameters to describe and quantify mining induced subsidence, and these subsidence parameters mainly include vertical subsidence, horizontal displacement, angle of draw, tilt, curvature and strain, as illustrated in Figure 2.4.
12
Figure 2.4: Typical section of subsidence trough, illustrating various subsidence parameters (MSEC, 2007a).
It is well known that ground subsidence includes both vertical and horizontal displacements. Maximum horizontal displacement occurs at the point of maximum tilt and becomes negligible at the limit of subsidence and at the point of maximum subsidence.
The limit of subsidence influence is indicated by the angle of draw, which is referred to as the angle between the vertical and the line joining the edge of the mining void with the edge of the subsidence trough. A 20 mm cut-off limit for subsidence was introduced by Kratzsch (1983) and has been generally accepted in mining practice, which suggested that subsidence of less than 20 mm has negligible effect on surface and it is generally used as the cut-off point for determination of the angle of draw. Angle of draw of 26.5° has been used to define the limit of the proposed extraction area in mining leases in the NSW Coalfields. However the Southern Coalfield is excluded from the recommended 26.5° angle of draw for reasons related to geology, surface topography and depth of cover, and a value of 35° is recommended for the Southern Coalfield
(NSW Department of Mineral Resources, 2003). Both of these values have been well validated by extensive field measurement data (DGS, 2011, Holla and Barclay, 2000,
13 MSEC, 2013). The average angle of draw in the Southern Coalfield is 29° with nearly
70 per cent of the observed values below 35° (Holla and Barclay, 2000). The variation of the values is attributed to the difference in the strength of the overburden strata and the depth of cover to the coal seam.
Tilt is calculated as the change in subsidence between two points divided by their distance and it is essentially the first derivative of the subsidence profile. Curvature is the second derivative of subsidence. It is convex over the goaf edges and concave toward the bottom of the subsidence trough. Strain is caused by bending and differential horizontal movements in the strata. It is measured as the amount of horizontal extension or compression induced over a given distance on the surface. Positive strain represents that the ground is in tension while the negative value indicates the ground is being compressed.
The magnitude of maximum subsidence depends on the extracted seam thickness, depth of cover and panel width. The effect of panel width and depth of cover on subsidence is illustrated in Figure 2.5, which is based on longwall operations in a number of coalfields throughout the world. Mining induced subsidence is also influenced by a range of geological factors such as lithology, joints and fractures, beddings and topography (McNally et al., 1996), and it should be noted that it is outside the scope of this study to discuss those impacts on the normal subsidence behaviour in detail.
14
Vz=vertical subsidence, h=extracted seam thickness, W=panel width, H=depth of cover
Figure 2.5: Effects of panel width, depth of cover on subsidence (Whittaker and Reddish, 1989).
2.3 Horizontal stresses in the Sydney Basin
A large amount of stress measurements at various mining and civil engineering sites in
Australia and around the world have been carried out using techniques such as hydraulic fracturing and overcoring to improve the understanding of natural stress states. Studies such as the World Stress Map project (Zoback, 1992) and the subsequent Australian
Stress Map project (Hillis et al., 1998) have provided detailed descriptions of stress states and have been highly beneficial for mining operations.
In 1958, Hast (1958) measured horizontal stress to be several times that of vertical stress at a number of sites in Scandinavia. At the time this work was treated with extreme scepticism. However since then such a pattern has been recorded by a large number of researchers in a wide range of locations and it is now considered to be the rule rather than the exception. It is well accepted that the presence of high in situ horizontal stress plays an important role in the mining induced strata movements and
15 subsidence development, and the existence of near-surface horizontal stress greater than the vertical stress appears to be a worldwide phenomenon. A detailed summary of high horizontal stresses throughout the world is provided by Waddington and Kay (2002).
With regard to the natural stress field in the Sydney Basin, the existence of a significant horizontal stress field has been well recognised for many years, and a stress map of the
Sydney Basin is presented in Figure 2.6 (Schreibner, 1993).
16
Figure 2.6: Sydney Basin Stress Map (Schreibner, 1993).
Stress orientations in the Sydney Basin are found to be variable and there is no consistency at a large scale of greater than 100 km (Gale, 1986, Hillis et al., 1999).
Nemcik et al. (2005) also confirmed that even with a very large data base, consisting of approximately 170 stress measurements from mines all over NSW, no consistent trends
17 in stress direction were found. Moreover the stress is often oriented with respect to local and regional structural features, such as density contrasts, faults and other geological structures (Hillis et al., 1999).
With respect to stress magnitudes, the high ratio of horizontal to vertical in situ stress has been a feature of the geological environment in the Sydney Basin, resulting from extensive stress measurements in the sedimentary basin. As can be seen in Figure 2.6, the horizontal to vertical stress ratios are typically in the range of 1.5 to 2.0 across the
Sydney Basin. A higher horizontal to vertical stress ratio in excess of 3.0 has been measured in a number of locations at Tower Colliery in the Southern Coalfield
(Hebblewhite et al., 2000). Table 2.1 presents some measurement results of the principal stress magnitudes in the Southern Coalfield.
Table 2.1: Measured principal stresses in the Southern Coalfield (Waddington and Kay, 2002).
Colliery Depth (m) 흈푯 (MPa) 흈풉 (MPa) 흈푯/흈풗
West Cliff 480 18.1 11.3 1.4
Appin 500 25.1 14.1 1.75
Tower 500 44.0 31.4 3.26
It is well known that the vertical stress is generally proportional to the weight of overburden, and the overburden stress gradient increases with depth, particularly for the sedimentary basins. The gradients of horizontal stress with depth have been found to vary over a considerable range. Figure 2.7 illustrates how the vertical and horizontal stress magnitudes vary with depth in the Sydney Basin. It is clear from the figure that the horizontal stress gradient is consistently higher than the vertical stress gradient.
18
Figure 2.7: Relationships of stress magnitude and depth in the Sydney Basin and in its sub- regions. 흈푯, 흈풉 and 흈풗 is the major, intermediate and minor principal stress (Hillis et al., 1999).
19 When correlated with geological structures, stress magnitudes can be described in terms of the associated fault condition, i.e. normal fault (σv > σH > σh ), strike slip fault
(σH > σv > σh) and reverse fault (σH > σh > σv). The fault condition in the Sydney
Basin is dominated by reverse fault (90 per cent of the measurements), with 8 per cent data indicating a strike slip and only 2 per cent indicating a normal fault condition
(Hillis et al., 1999). The typical reverse thrust (σH > σh > σv) pattern is also presented in Figure 2.7. Based on stress measurements in coalfields in the Sydney Basin,
Shepherd and Huntington (1981) presented the relationship of ratio of horizontal to vertical stresses and depth, as illustrated in Figure 2.8. It was found that the ratio of average horizontal to vertical stress was outside usual limits for depth of 200 m to
500 m using Equation 1 as suggested by Brown and Hoek (1978). Therefore, the horizontal stresses at these collieries are more extreme, indicating an affiliation with thrust faulting stress conditions.
100 1500 + 0.30 < k < + 0.50 Z Z (1)
Where k is the horizontal to vertical stress ratio and Z is depth in metres.
20
Figure 2.8: Relationship of horizontal to vertical stress ratio and depth (Shepherd and Huntington, 1981).
There are a number of causes proposed for the observed high levels of horizontal stress in the literature. The major contributing factors are concluded below and are based on previous studies (Brady and Brown, 2004, Branagan, 1985, Brown and Windsor, 1990,
Su et al., 2000, Waddington and Kay, 2002, Zoback and Zoback, 2007).
The in situ stress state is affected by the loading conditions in the rock mass. The weight of overburden causes elastic horizontal stresses that are restrained by the surrounding and confining stress fields. It is suggested that erosion of the surface rocks could lead to an increase in the ratio of horizontal to vertical stress, because unloading of the strata during surface erosion reduces the vertical stress at a point and the horizontal stress by a lesser amount which leads to the ambient stress state characterized by high horizontal to vertical stress ratio.
Elevated horizontal stress is also considered to be associated with tectonic processes
(Nemcik et al., 2006, Nemcik et al., 2012). The theory of plate tectonics implies that
21 lateral forces are constantly being applied to the brittle upper crust. These forces would continue to build unless there was some mechanism for their release. That mechanism is failure of the crust itself through faulting (Mark and Gadde, 2010). It should be noted that the regional tectonic stresses are likely to be modified appreciably by the local effects, such as the topographic variations (Brown and Windsor, 1990, Ferguson and
Hamel, 1981, Hebblewhite et al., 2000, Molinda et al., 1992). Zoback and Zoback
(2002) stated that worldwide studies had found “no evidence” that “residual stresses” from past tectonic events played any role in today’s stress fields. They speculated that if such stresses existed at all, they could only be important in the near-surface area where the tectonic stresses were small.
The presence of geological fracture is another reason for the high horizontal stresses. In situ stress might be heterogeneous locally because of the differences in the stiffness of adjacent lithological units and the presence of fractures faults and joints in the rock mass (Brown and Windsor, 1990). Su et al. (2000) made a similar conclusion, indicating that the presence of geological fracture is one reason for high horizontal stresses. Site measurements in Sweden and China indicated that stresses reorient and their magnitudes varied in the vicinity of fractures. Additionally they found two distinct stress domains separated by a major fracture zone where the shaft intersected two major thrust faults as reported at the Underground Research Laboratory. They concluded that the influence of fracture patterns on in situ stresses is very complex and that the stress magnitude, in rock masses composed of different rock types, is determined by the relative volume of rock masses with different competence and elastic characteristics.
22 2.4 Horizontal ground movements
A large amount of literature has been published on the subsidence behaviour in New
South Wales. Recently, with developments in three-dimensional surveying techniques to measure ground displacements, the understanding of mining induced horizontal movements has progressed. Improvements in monitoring techniques for characterising subsidence include the application of GPS technology and satellite based remote sensing such as LiDAR and DInSAR. These techniques have provided better basis for understanding the extent and the mechanics of the mining induced horizontal displacements (Mills, 2011a).
Measurements of mine subsidence indicate that horizontal ground movements associated with longwall mining can be broadly divided into three components namely, systematic horizontal movements that occur toward the goaf area, far field horizontal movements and horizontal movements associated with topographic relief. The following sections provide discussions on these movements.
2.4.1 Systematic horizontal movements
In areas where the surface topography is generally flat, horizontal movements are observed directly towards the centreline of the subsidence area. For example Peng
(1992) indicated that horizontal displacement vectors generally pointed towards the centre of the subsidence basin where the terrain was flat. These mining induced systematic horizontal movements are observed in all types of surface topography but are most clear in flat terrain. Typically the systematic horizontal movement occurs toward the subsidence trough, which results from the bending curvature of the overburden strata over the extracted longwall panel. Figure 2.9 presents an example of the
23 horizontal movement vector as a result of the extraction of longwall 17 at Tower
Colliery (Hebblewhite et al., 2000).
Figure 2.9: Horizontal movements due to extraction of Longwall 17 at Tower Colliery (Hebblewhite et al., 2000).
Kratzach and Helmut (1983) discussed the mining induced horizontal movements in flat terrain and the indicated ground surface experienced horizontal movements towards the centre of the subsidence basin, when extractions occurred beneath, with the result that a tensile strain rose at its boundary and a compressive strain at its centre, as illustrated in
Figure 2.10.
24
Figure 2.10: Horizontal deformation of rock masses above longwall extraction (Kratzsch and Helmut, 1983).
In terms of magnitude, it is generally found that the maximum horizontal movement is smaller than the vertical movement. For example, both Kratzsch and Helmut (1983) and
Holla (1997) found that the maximum horizontal displacements were as large as 0.4 of the observed vertical movements. This ratio is even smaller in the investigation of Peng
(1992), varying from 0.12 to 0.3, which is dependent on the inclination of the coal seam. However there are some specific cases where the magnitude of horizontal movement is larger than that of vertical movement. For example, Milner and Seedsman
(2000) referred to a geographic region around the Cataract area and pointed out that if elastic techniques were employed to evaluate horizontal deformations, rather than the empirically-derived prediction methods based primarily on vertical subsidence, then the observed horizontal movements at angles of draw in excess of 45° would exceed the vertical movements.
Preston (1992) presented the hypothesis that the roof collapsed and subsided as extraction happened, and horizontal bedding planes in the area above the longwall panels opened in response to the changing stress regime. Preston (1992) added that once these planes had opened, friction was lost and any sub-horizontal forces would be 25 unrestrained, resulting in horizontal displacement. The movements were dependent upon the dip of strata relative to the topography of the surface. Holla and Barclay (2000) drew a similar conclusion and the authors believed that changes in horizontal stress field caused by mining had the potential to generate movement along the bedding planes in the overburden.
2.4.2 Regional horizontal movements
In the last couple of decades, empirical database of mining induced subsidence has indicated that far field horizontal movements occur at considerable distances (up to several kilometres) from the current limit of mining. Far field movements have been recognised as another important component of the mining induced horizontal movements, which is usually referred to as the measured horizontal displacements at pegs located beyond the longwall panel edges and outside the accepted angles of draw zone. Such movements are usually several times higher than the vertical displacements measured at the same locations. This section reviews issues of far field horizontal displacements, mainly focused on coalfields in New South Wales.
Far field mining induced horizontal movements were first noted at Cataract Dam in the
Southern Coalfield in the 1990s. Preston (1992) firstly postulated that the recorded movements around the shores of Cataract Reservoir were associated with mining activities, indicating that horizontal movements of 25 mm had occurred over a period of six years, due to mining activity at South Bulli Colliery, at a distance of more than
1.5 km to the survey mark. Reid (1998) referred to the study, commenced in 1972, by
Sydney Water Corporation regarding precise measurements near Cataract Dam. Results show that significant horizontal displacements occurred even when underground mining was one to two kilometres away from the survey points.
26 Since then extensive survey data around many longwall areas mainly located in the
Southern Coalfield have indicated clear horizontal movements up to two kilometres from active mining. On the Central Coast, Seedsman and Watson (2001) discussed the impact of longwall mining at Newstan Colliery on major pipelines and reported that horizontal movements of 20 mm to 50 mm had been measured well outside the angle of draw. Waddington and Kay (2002) reviewed the regional horizontal displacements in the Southern Coalfield in studying the impact of subsidence on natural features, and provided case histories where far field mining induced horizontal movements were clearly observed on roads, bridges and tunnels far away from the mining activities.
Reid (2001) also stated that horizontal movements had occurred as far away as 1700 to
1800 m from the longwall being mined in the area of the Cataract Reservoir. The author noted that the far field movements were generally aligned in a similar direction to the maximum in situ horizontal stress and postulated that far field movements were triggered by mining and were driven by the in situ horizontal stress. Hebblewhite et al.
(2000) conducted a case study in Tower Colliery southwest of Sydney regarding regional horizontal movements. The study found that bridges and freeway at a considerable distance outside the conventional angle of draw subsidence influence criteria demonstrated widespread evidence of regional horizontal deformation of the land surface.
An empirical database of the observed far field horizontal movements from the NSW
Coalfields, predominately the Southern Coalfield, was established by MSEC (2014).
Figure 2.11 presents the recorded horizontal displacements, at a range of sites in the
Southern Coalfield, against the distance from the goaf edge of the mined panel. The confidence level is also included in the figure to show the spread of the data.
27
Figure 2.11: Far field horizontal movement database in the Southern Coalfield (MSEC, 2014).
There are a number of postulated mechanisms involved in the observed far field horizontal movements. Many investigations suggest that the redistribution of horizontal stress regime plays an important role in driving the far field horizontal displacements, particularly in the high horizontal stress environment as described in Section 2.3.
MSEC (2007) stated that far field movements are caused by redistribution of the stresses in the strata between the seam and the surface due to regional mining activity. A similar conclusion was drawn by Hebblewhite (2009), indicating that stress relaxation towards mining excavation and horizontal movements aligned with the principal in situ compressive stress direction could be a cause of the far field horizontal movements.
Hebblewhite (2009) also included some contributing mechanisms such as pre-mining stress relaxation, valley notch effect, valley bulging, regional joint patterns, vertical gorge shearing and shear failure of horizontal bedding planes below the surface. Pells
(2011) demonstrated that far field movements were consistent with perturbation of the
28 natural horizontal stress field due to macro changes to the stiffness of the strata above areas of total extraction. It was suggested that the coal seam had a much lower shear stiffness than the overlying strata, and acted as a potential discontinuity in the horizontal stress field. Consequently the extraction of the coal seam formed a free face at the base of the overlying strata in respect to perturbations of the horizontal stress field.
It should be noted that far field horizontal movements are accompanied by very low levels of strain (Hebblewhite, 2009, MSEC, 2013). As such the differential horizontal movements over a particular length of surface are minimal, and these movements seem to have little impact on built environments and natural features, except for those structures that are very sensitive to differential horizontal movements.
2.4.3 Topography related movements
In addition to the systematic and far field horizontal movements, topography related movements have also been recorded in numerous cases. These movements predominantly occur in cases where non-conventional surface topography exists such as steep slopes, gorges, river valleys and other surface incisions. The presence of these features influences both the magnitude and the direction of ground movements, particularly the horizontal displacements.
In high relief areas, such as valleys and gorges, mining induced ground movement tends to be asymmetrical about the centre of the mining panel (Holla, 1997). In addition, most of the horizontal movements take place towards both the gorge and the active goaf area, although some movements have been observed towards old goaf (Hebblewhite et al.,
2000). The topography related movements and the associated mechanisms are discussed in Section 2.6.
29 2.5 Previous numerical modelling studies
A wide range of numerical methods such as the Boundary Element Method, Discrete
Element Method, Finite Element Method, Finite Difference Method and the hybrid method can be employed to conduct stress and deformation analysis in geomechanics.
The theories and the applications of these numerical methods in rock mechanics are well documented in the literature and will not be repeated herein (Brady and Brown,
2004, Jing and Hudson 2002, Bobet et al., 2009). This section reviews the scope and focus of various numerical modelling techniques for modelling of non-conventional subsidence behaviour.
2.5.1 Finite element method
The finite element method (FEM) was first introduced by Clough (1960) for the numerical investigation of stresses and displacements in continuous structures. Soon after that Franks and Geddes (1986) used 2D FEM to model longwall mining beneath idealized steep slopes based on a homogeneous linearly elastic continuum. The main attempt was to compare the mining induced subsidence and tilt with mining under flat terrain. As a result, ground strains and horizontal movements were vastly affected by surface slope while vertical displacements were not. Appropriate modelling and validation methods were not well developed in this study, which led to uncertain conclusions. Additionally the factors contributing to this difference at a point on the slope were identified as: (i) the direction of mining, (ii) the horizontal distance of the surface point from the mining edge, (iii) slope angle, and (iv) the proximity of the surface point to slope toe or slope crest.
30 During the early 1990’s, Shu and Bhattacharyya (1992) employed a two-dimensional
FEM code DEMON to simulate the subsidence on a sloping surface above a completely mined panel. The researchers compared the subsidence, horizontal displacements and strains from mining at ground surface for different inclinations.
The above numerical studies focus on subsidence characteristics of steep topography and generally indicate that the presence of steep topography affects horizontal movements and ground strains more than vertical displacements and tilts. However the failure mode of the valley slope had not been well identified.
A numerical procedure based on nonlinear finite element analysis was developed for predicting subsidence profile over mine panels located in hilly terrain (Siriwardane and
Amanat, 1984). The complex behaviour of the overburden strata was modelled by the
Drucker-Prager yield criterion (Drucker and Prager, 1952). The failure zones detected after excavation are shown in Figure 2.12. The plastic zone had propagated up to the surface while tensile zones had developed near the mine roof and on the slope near the ground surface. The development of tensile zones near the slope was identified as the contributing factor to the valley side failure as a result of subsidence.
31
Figure 2.12: Predicted plastic and tensile zone with FEM (Siriwardane and Amanat, 1984).
Recently the three-dimensional FEM code, COSFLOW has been employed in analysing mining induced subsidence. This code introduces a plasticity formulation to the elastic model to simulate rock fracturing. It is similar to conventional formulations (such as ubiquitous joints) in that the effect of joints between layers is smeared out over the whole element. It does however differ in that the bending stiffness of the layer can be represented and a realistic model of rock fracturing can be modelled (CSIRO, 2006).
Guo et al. (2004) conducted a research at Appin Colliery and West Cliff Colliery for mining induced subsidence using 3D COSFLOW. They demonstrated the capability of the code by simulating detailed subsidence profiles as compared with standard elastic- perfectly plastic models. More importantly, the computation of COSFLOW is faster and more stable than discontinuous code when doing large scale analysis. Few attempts have been made to employ COSFLOW to investigate the irregular valley related subsidence movements.
There are other FEM studies available in the literature such as the ABAQUS analysis regarding surface subsidence (Capasso and Mantica, 2006). Generally, FEM has good flexibility in treating material inhomogeneity, anisotropy and complex boundary
32 conditions. It should be noted that the FEM suffers from the treatment of block rotation, fracture growth and the mesh generation for complex three-dimensional geometry.
However development of discrete methods (Belytschko et al., 2001, Li et al., 2001,
Stolarska et al., 2001) has enhanced the applicability of FEM to the rock fracture representation.
2.5.2 Finite difference method
The finite difference method (FDM) is similar to the FEM, and is based on the discretization of governing partial differential equations in problem regions with simple boundaries (Jing and Hudson, 2002). The numerical modelling code FLAC, based on a spatial discretization derived from the finite difference perspective, has been widely used for studies of irregular subsidence behaviour. Molinda et al. (1992) investigated the coal mine roof failure underneath river valleys. FLAC 2D was used to model 13
West Virginia valleys overlying mines and it illustrated the presence of valleys with concentrated horizontal stresses beneath the valley apex. Later, FLAC 2D was employed to predict subsidence by Alejano et al. (1999) in British basins, using an elasto-plastic material model with the following features: (i) transversely isotropic elastic pre-failure behaviour, (ii) anisotropic yield surface, and (iii) isotropic elastic post-failure behaviour.
Although the subsidence modelled by FLAC fitted well with the empirical observations, the horizontal displacement showed divergence due to the surface tension cracks which turned a continuous material into a discontinuous one. They hypothesised that the post- failure deformability behaviour of rock mass certainly controlled the subsidence behaviour. However no further investigation was conducted.
33 Lee (2005) investigated the strata interaction when mining beneath river valleys using
FLAC 2D. The ubiquitous joint model was chosen which simulated strain softening by resetting cohesion and tension to zero in the event of element failure. A parametric analysis was conducted regarding the k value (horizontal stress to vertical stress ratio), reduction factors for material properties, and loading type. It was found that valley closure and upsidence occurred (Figure 2.13a) when the valley was extracted for all k values, i.e., k=1, 3. The sliding of bedding planes due to applied lateral displacement is presented in Figure 2.13b. Shear separation of bedding planes was observed when a weak bedding plane 10 m below the surface was added in the model. However multiple joints or bedding planes were not well represented in the study.
(a)
34
(b)
Figure 2.13: (a) closure and upsidence after equilibrium, (b) sliding of bedding plane (Lee, 2005).
It should be noted that the regular grid systems in the conventional FLAC code usually limit its application in representing the explicit fractures associated with mining subsidence This shortcoming has been overcome using the internal programming language FISH. Gale (2004, 2006, 2008, 2011) pioneered the trend to simulate the overburden subsidence characteristics of the strata and fracture mechanics as well as the geometry of the fracture systems using modified FLAC. For example a valley case was modelled in FLAC to investigate the effects of subsidence on water catchment and groundwater regimes in the Wallarah 2 Coal Project (W2CP, 2010). The fracture distribution in the overburden for the progressive stages of extraction is illustrated in
Figure 2.14. It was indicated that the strata above the caving-related fracturing were forming shear bounded zones that subside and fracture onto the caved area. In terms of discontinuity representation, FLAC is capable of including a small number of joints and interfaces. However, with the increase in the number of structures and the complexity of
35 the joint pattern, it is natural to employ the models based on the discontinuous methods that are discussed below. These FLAC model studies have led to an improved understanding of the influence of surface topography on subsidence troughs and the redistribution of stresses beneath the valley within the limitations of the models and the assumed material behaviour.
Figure 2.14: Failure modes of the strata in the weathered zone (W2CP, 2010).
2.5.3 Boundary element method
The boundary element method (BEM), which only requires discretization along the boundary of the problem, tends to be more efficient than the finite element and finite difference approaches mentioned above, due to less computing time and memory consumption. Ahola (1990) employed a two-dimensional BEM to assess the effect of extracting two longwall panels beneath massive sandstone escarpments in Cottonwood
Mine, Utah. The model represented the rock mass as an infinite, homogeneous and elastic continuum. A displacement discontinuity boundary was placed along the slope and escarpment regions, prescribing zero normal and shear stresses to model the free surface. The study found that a small outward horizontal displacement occurred near the
36 escarpment, and tensile cracking in the area could trigger toppling or slope failure. A third panel distanced from the escarpment was extracted as an additional study, which led to the high compressive stress concentration at the toe and to inward movements of the horizontal component in the vicinity of the escarpment. Localized yielding could influence escarpment instability according to the magnitude of the compression stress in the region.
It is important to note that two-dimensional analyses are always limited to panels parallel to the cliff lines. For the first time, full three-dimensional topographic effects were taken into account in model mining induced subsidence by Wardle and McNabb
(1992). The displacement discontinuity method (DDM) was developed, along with finite elements that were used to model the surface topography in Baal Bone Mine.
DDM was extended to incorporate non-homogeneity with the rock mass being represented by a number of parallel layers with distinct anisotropic elastic properties
(Wardle, 1986). The authors referred to the work by Kay et al. (1991) to present the unusual displacement pattern observed above the panel, as shown in Figure 2.15.
Horizontal displacements were towards the valley centreline rather than towards the goaf. Given the limited database and time, no attempt was made to model the detailed failure (such as collapse) behaviour of the cliffs in this study.
Figure 2.15: Mining induced displacement pattern of valley (Kay et al., 1991).
In the boundary element method, distribution of stress and displacement throughout the domain is continuous and far field boundary conditions are satisfied correctly. Although
37 this method has been extensively used in geomechanics its application in subsidence analysis is still restricted because of the limited flexibility in the simulation of non- homogeneous and non-linear material behaviour.
2.5.4 Distinct element method
Most of the methodologies described so far have treated the overburden rock mass as an equivalent continuum material. In the distinct element method (DEM) it is assumed that rock is represented as an assemblage of rigid or deformable components for the analysis of rock mechanical behaviours. Research has been reviewed of the development of distinct element method to gain a better understanding of the mechanisms of the irregular subsidence.
Jones et al. (1990) modelled the same sandstone escarpments as Ahola (1990), at the
Cottonwood Mine in Utah. Since the initial stability study using a two-dimensional
FEM could not model the large horizontal movement observed, a DEM approach was then employed to investigate escarpment stability. It was found that localized structures, such as joint sets, primarily controlled the stability of escarpment. In conjunction with the observations of escarpment spalling, the coupled modelling work showed that rotation accounted little for the escarpment failure.
O’Conner and Dowding (1992) addressed the impacts of rock discontinuities on mining induced subsidence using a rigid block computer code. The edge-to-edge contact between blocks was the main feature of the code, which could eliminate unrealistic interlocking occurring in the Rigid Block Model (RBM) developed by Cundall (1971).
It was found that all rigid blocks on the surface experienced a horizontal component of displacement towards the valley floor and small surface blocks experienced heave due
38 to rigid body rotation of adjacent blocks but were displaced towards the valley floor.
Although the rigid block model is capable of simulating the trend of horizontal and vertical displacements in the overburden and on the surface, it does not work on fracturing and shearing behaviours of the strata.
The two-dimensional blocky discrete element models using computer code UDEC have been used to meet major aspects of the study. In UDEC models the discontinuities are treated as boundary conditions between blocks, and large displacements along discontinuities and rotations of blocks are allowed. The work undertaken by Wold et al.
(1999) has resulted in the development of UDEC for undermining cliffs and gorges at
Tower Colliery southwest of Sydney. In this model, a relatively high intensity of fracture discretization was used at the base of the gorge and transecting the Hawkesbury and Bulgo Sandstones, together with fully deformable blocks. The predominating rock failure behaviour at the base of the gorge was shear and opening of bedding planes in the shoulders and near the gorge floor horizon, buckling of rock units in the floor and vertical tensile cracking to about 5 m depth from the surface. The models showed that the depth of the stress relieved surface zone was between 2 m and 8 m, extending along either side of the gorge for a distance of up to 300 m. The depth of yielding at the base of the gorge was about 3 m to 5.5 m. The major fracture zones extended up to 150 m laterally into the gorge walls, with a depth of up to 100 m. Another finding was that vertical fractures, which opened during mining, tended to close again as the mining face passed. The vertical displacement contours were close to the gorge, as can be seen in
Figure 2.16. These were difficult to identify due to the discontinuous nature of the buckling and blocky movement, but displacements of 500 mm to 150 mm had been known on the gorge floor.
39
(a)
(b)
Figure 2.16: (a) Total fractures during undermining of gorge, (b) fractures remaining open at completion of mining (Wold et al., 1999).
Further research utilizing UDEC was conducted to investigate the impact of mine subsidence beneath the Cataract River at Tower Colliery (Waddington and Kay, 2001,
2002). The gorge was firstly excavated to simulate the erosional process of gorge formation, followed by excavation of four longwall panels in a linear sequence. As observed, bedding planes sheared at a relatively shallow depth, less than 50 m from the gorge floor, and were predominantly towards the gorge rather than towards the mined area, as illustrated in Figure 2.17. These shear movements were mainly horizontal, with relatively large shearing displacements concentrated on bedding at the base horizon of the gorge, particular at the toes of the gorge walls. Only limited shear failure was observed at the base of the gorge, while yielding near the surface was spread over a large area.
40
Figure 2.17: Shearing displacements in gorge region after extraction (Waddington and Kay, 2002).
Another popular representation of the distinct element method is implemented with
Particle Flow Code (PFC), where rock is represented as a number of small rigid, spherical grains that are either bonded together at their contacts with shear and tensile strengths or that interact by a grain-to-grain friction angle after the contact bond has been broken (Potyondy and Cundall, 2004). Sainsbury (2008) used PFC to study river bed cracking above underground longwall mining in the Southern Coalfield of NSW.
Sliding joints were included within the PFC model to produce a synthetic rock mass
(SRM) to simulate Hawkesbury Sandstone. The SRM was developed in the Mass
Mining Technology Project (MMT) to quantify rock mass behaviour at large scales, 10 to 100 m (Pierce et al., 2007). This approach combined intact rock mechanisms, deformation and brittle fracture, with mechanisms involving discontinuities. The SRM method simulated the low angle, explicit shear fractures in the river bed (Figure 2.18)
41 extending to a depth of between 10 m and 20 m, which was in compatible with the
UDEC results mentioned above.
Figure 2.18: Explicit fracturing and displacement at the base of the gorge (Sainsbury, 2008).
It is necessary to note that the PFC modelling has some inherent limitations, as explained by Potyondy and Cundall (2004) and Damjanac et al. (2007). This limitation can be overcome either by using clumped-particle geometry (Cho et al., 2007) or by introducing a polygonal grain structure and thereby transforming the fundamental grain shape from circular to polygonal structure to more closely mimic the real microstructure in rock. However PFC is currently restricted to modelling the mining induced
42 subsidence at field scale with the near-surface explicit fracture propagation represented in detail, due to the high computational intensity.
An alternative representation of the particle model is implemented with UDEC using
Voronoi tessellation, which creates randomly sized rigid or deformable polygonal blocks that are similar to the clumps mentioned before. The Voronoi tessellation technique has been well accepted for constructing microstructures and simulating crack propagation at laboratory test scale (Kazerani and Zhao, 2010, Lan et al., 2010, Li et al.,
2006, Nygårds and Gudmundson, 2002, Zhang et al., 2005,). Currently the field scale applications of this Voronoi method for solving practical engineering problems can be found in the studies (Alzo’ubi, 2009, 2011, Damjanac et al., 2007, 2012). UDEC with the Voronoi discretization is a promising modelling approach in the study of fracture developments around valleys. However attention should be paid to the calibration of the micro-properties to match the macro-properties when using these particle models.
2.5.5 Hybrid method
In some cases one only specific simulation method cannot meet the needs of modelling a large scale mining area or a specific area of interest. Hence two different modelling methods need to be combined to solve a problem, thereby aiming to decrease the modelling work. In practice a combination of continuum and discontinuous methods or the multi-scale hybrid methods can model the large-scale mining area with sufficient detail to examine the explicit fracture development around valleys.
During the ACARP Project (Waddington and Kay, 2002), three-dimensional FDM/FEM code FLOMEC was employed to reflect strata movement in the Cataract and Nepean
River Gorges. FLOMEC can conduct analysis on coupled fluid flow-mechanical
43 processes in rock containing joints, faults and aquifers. The code is suitable for representing the three-dimensional geometry of the intersection of the river gorges
(Figure 2.19a). In this case the rock mass was considered as a deformable continuum with ubiquitous jointing throughout. As can be seen in Figure 2.19b incremental displacements induced by mining are presented as horizontal displacement vectors.
(a)
44
(b)
Figure 2.19: (a) FLOMEC Mesh and surface displacement contours, (b) vectors of horizontal displacement due to underground extraction (Waddington and Kay, 2002).
There were two key factors included in the research: (i) orientation of principal stress with respect to the major river gorges, and (ii) general topographical features with respect to mining layout.
It was found that horizontal convergence of valley sides was greater if the direction of the major surface movement was perpendicular to the gorge. Of particular note were the patterns of horizontal movements which were generally focussed towards the gorge junction, rather than on the mined excavation that was laterally offset from the junction.
The combined FEM/DEM method is also employed to investigate mechanisms governing subsidence development. This method is based on a hybrid 2D code ELFEN.
Vyazmensky et al. (2007) presented examples of conceptual ELFEN models simulating mining induced surface subsidence development using equivalent continuum, discrete
45 network and mixed approaches. In the application of fracturing modelling using
ELFEN, one important issue was the transformation of the continuous finite element mesh to one with discontinuous fractures. Physical fractures were inserted into a finite element mesh such that the initial continuum was gradually degraded into discrete blocks. The subsidence profiles are shown in Figure 2.20, where the maximum extent of the zone of influence is 110 m and 150 m respectively. The modelling based on equivalent continuum approach did not present the discontinuities satisfactorily since the directional strength weakening was not allowed. In the mixed approach the fracture network was developed using the proprietary code FRACMAN and exported into
ELFEN. In this case the initial combined finite/discrete modelling of surface subsidence showed encouraging results.
(a)
(b)
Figure 2.20: Simulation of subsidence development using (a) an equivalent continuum approach, (b) a mixed discrete/equivalent approach (Vyazmensky et al., 2007).
46 Chugh et al. (1994) conducted work on mining induced subsidence at Appin and Tower
Collieries using a 3D boundary element/displacement discontinuity program called
Longwall Ground Mechanics (LGM) as shown in Figure 2.21a. The gorge area was further studied using linear elastic 2D finite element code, ANSYS (Figure 2.21b). The authors concluded that mining induced compressive stresses at the gorge base increased pre-mining stress gradients below the gorge and could result in bedding plane failures, and compressive stresses would be increased over a very wide area (500 m) leading to gorge base failure far away from mining areas.
(a)
(b)
Figure 2.21: (a) LGM 3-D Model, (b) ANSYS 2-D valley model (Chugh et al., 1994).
Hybrid methods are widely employed to analyse the valley related movements because of flexibility and efficiency in representing the far-field to the near-field rock mass.
Popular explicit codes FLAC, UDEC and PFC are usually coupled to investigate the
47 mining induced fracture development because of the similarities in programming and methodologies (Katsaga and Potyondy, 2012, Sainsbury, 2008). Although hybrid models can take advantage of each modelling code, there are some issues to be considered in the process of modelling such as the continuity conditions at the different region boundaries, the mechanical coupling scheme and the communication between the codes.
2.6 Postulated mechanisms of valley closure and upsidence
Some of the previous literature on mining induced ground movements beneath or in the vicinity of valleys, gorges and cliffs proposed various hypothesised mechanisms. These have been regularly discussed as possible contributory factors in valley closure and upsidence. The following section provides a brief review of these mechanisms, which include, but are not limited to those below.
2.6.1 Valley stress relief
In the natural process of valley formation and development, valley stress relief has been detected as one of the possible causes for valley related deformation and movements.
The erosional process of valley formation removes the horizontal support from the valley walls (in the state of extension) and vertical support from the valley floor (in the state of compression). Hence valley walls tend to converge while upward movement occurs at the base of valley. Many historical studies have confirmed the impacts of valley stress relief on valley feature deformation, as discussed below.
The stress relief features and the relative valley related movements are affected by many factors, such as variation in rock strength and topographic factors. Ferguson (1981)
48 studied the stress relief features in flat-lying sedimentary rocks accompanying valley erosion (Figure 2.22). During the process of valley wall stress relief, inward movement of valley sides concentrated in weaker, more deformable strata. Vertical to sub-vertical tension joints were developed in stiffer beds which typically did not extend across weaker beds or bedding contacts. At the bottom of valley, buckling of beds occurred as a result of lateral compression and vertical load removal. Bedding planes opened in stronger strata while anticlines and thrust faults developed in weaker beds.
Figure 2.22: Valley cross-section (Ferguson, 1981).
In these situations, the formation of vertical and sub-vertical tension joints can be considered as the evidence of valley stress relief. Gray (1982) studied the surface relief of stress in the Sydney Basin, and found that because of valley erosion, concentration of stress and valley bulging took place beneath the river bed with the formation of horizontal cracking, and sub-vertical joints formed in the gorge walls (Figure 2.23). Fell et al. (1992) obtained a similar conclusion that the steeply dipping joints next to the cliff faces opened up due to the expansion of the rock layers under the influence of horizontal stresses both across and parallel to the valley.
49
Figure 2.23: Valley stress pattern: A-pre-erosion, B-post erosion (Gray, 1982).
In order to better understand the pre-mining valley deformation, the influence of the valley geometry on valley stress relief also needs to be considered. Myrvang (1976) postulated that the shape of a valley could have significant influence on the direction of stress around valleys, as shown in Figure 2.24. Molinda et al. (1992) studied the coal mine roof failure pattern beneath valleys in the Appalachian Coalfields, USA, and evidence of valley stress relief was found beneath several valleys by in situ horizontal stress measurements in the research. The survey showed that the broad valleys might be in the stress-relieved state because of valley rebound due to unloading as well as the transfer of stress laterally due to loading of the valley walls, while the V-shaped valleys tended to concentrate the horizontal stresses near their apex and produced active stress related failures such as cutter roof and snap top.
50
Figure 2.24: Measured stress directions by comparing with topography (Myrvang, 1976).
There have been cases where valley closure and horizontal deformation can be attributed to mining induced stress relaxation. Enever et al. (1990) conducted stress measurements in the Sydney Basin and gave an example of stress measurements in three surface holes at Warragamba Dam, as illustrated in Figure 2.25.
Figure 2.25: Stress measurement plan of Warragamba Dam showing the in situ stress direction (Enever et al., 1990).
51 During excavation of the Warragamba Dam gorge bottom rock units, there was evidence of a significant horizontal stress field existing at the bottom of the gorge.
There was noticeable lateral displacement of drill holes, opening of joints and measured inward movements of the vertical walls. The results clearly indicated the influence of topography on the stress field. The stress measurements showed stress relief in the near surface and stress concentration effect at the base of the gully, as presented in Figure
2.26.
Figure 2.26: Cross section of gorge showing in situ stresses in Warragamba Dam (Enever et al., 1990).
2.6.2 Effects of horizontal stress
Within the coalfields of New South Wales, the existence of near surface horizontal stresses greater than the vertical stresses has been well identified and this phenomenon has been reviewed in Section 2.3. Several studies have concluded that the redistribution of horizontal in situ stress resulting from longwall mining can be employed to explain the valley related closure and upsidence. Underground mining activities result in the redistribution of horizontal stresses around the void and the collapsed area above, as shown in Figure 2.27. This schematic graph was derived in a research study of mining-
52 induced valley related movements mainly in Cataract and Nepean River Gorges in the
Southern Coalfield (Waddington and Kay, 2001, 2002). As can be seen from Figure
2.27, incisions in the surface where the redistribution of horizontal in situ stress is concentrated act as a discontinuity in the transfer of the high horizontal stresses from one side of the valley to the other. These stresses then transfer to the strata beneath the base of the valley. At the base of the valley where the strata is not strong enough to support the increased mining-induced horizontal stresses an upward dilation of the strata occurs.
Figure 2.27: Redistribution of horizontal stresses due to underground mining beneath a valley (Waddington and Kay, 2002).
The effects of horizontal stress on valley related movements are also illustrated in other cases. Reid (1994) reported on the measured movements at the base of the Cataract
River Gorge as a result of mine subsidence over Longwall 8 at Tower Colliery. The author noted that the measured subsidence and strain had been greatly affected by the
53 gorge and pointed out that bulging up to 250 mm to 300 mm had occurred in the centre of the gorge. The bulge appeared progressively as mining extended underneath the gorge.
In 1997, Holla studied the mining induced ground movements in high relief areas in
New South Wales, and the asymmetrical ground movements about the centre of the extraction panel were observed as a result (Holla, 1997). Figure 2.28 depicts the postulated mechanism for subsidence hump. Gravitational stresses on the rock mass in the valley sides are induced by underground mining activities. The horizontal stresses would be resisted by the rock mass in the creek bed. Due to the lateral compression the compressed rock mass in the creek bed would be pushed up resulting in hump and upsidence. Moreover the increased horizontal compression of the rock mass would also produce large compressive strains in the creek bed.
Figure 2.28: Notch effect with concentration of stresses (Holla, 1997).
Fell and Robin (1992) studied the valley related movements, focusing on a valley cutting through an interbedded sequence of sandstones, siltstones and claystones in New
South Wales. The authors found that most of the features associated with valley bulging
54 resulted from buckling and shear failure under high horizontal compressive stresses, as shown in Figure 2.29.
Figure 2.29: Valley structures in flat-lying rocks (Fell and Robin, 1992).
Hebblewhite et al. (2000) studied the regional horizontal subsidence displacements associated with longwall mining in Tower Colliery. As an explanation for the mining- induced regional horizontal deformation, the authors postulated that within the prevailing high horizontal stress field the caving above the longwall panel would trigger significant directional, horizontal closure towards the sides of the caved void, and this would cause more widespread subsidence on the surface with major changes in topography in the direction of the principal horizontal stress field.
Orientation of the horizontal stress relative to the valley also influences the degree of valley related movements. Gray (1982) concluded that orientation of the horizontal stress relative to the valley affected the degree of valley bulging and opening of the foundation rocks, which was more pronounced if the stress was perpendicular to the direction of river flows than when the flow was parallel to it.
55 2.6.3 Lateral dilation mechanism
Mills (2001) reported the results of subsidence monitoring at Baal Bone Colliery in the
Western Coalfield of New South Wales. The horizontal stress measurements at Baal
Bone Colliery showed that the principal horizontal stress was of low magnitude, suggesting that horizontal stress played a minimal role in the mechanism driving horizontal displacements at the sites. Horizontal movements were observed in a downslope direction unrelated to in situ stress magnitude. In this situation the mechanism includes lateral dilation which occurs when rock strata moves differentially across vertical joints and rock block rotates relative to one another as the rock strata subside differentially with the incremental retreat of longwall panel.
The author found that the horizontal movements consisted of several components, including movement towards the goaf, movement towards the work face and downslope movements. The presence of bedding planes within the overburden provides additional freedom for the rock strata in the hillside to move laterally in the downslope direction.
The movement itself is horizontal, but the direction of movement is down the slope. It was observed that the magnitude of downslope movements was sometimes even greater than that of vertical subsidence and usually much greater than that of systematic horizontal subsidence movements in flat terrain. Hence downslope movement tends to dominate horizontal subsidence behaviour in steep terrain.
As underground mining occurs downward movement of the subsiding strata leads to lateral dilation of the overburden strata. In flat terrain the trend of lateral dilation is suppressed by the confining effects of the adjacent rock strata outside the longwall goaf.
However in steep terrain the presence of a valley provides a free surface in the direction
56 of the slope, resulting in less resistance to balance the dilation force on the downslope side, as shown in Figure 2.30.
Mills (2001) also reported that lateral dilation occurred much more easily when the ground was being stretched in the tensile phase of subsidence, compared to later stages of subsidence when the ground was being compressed, when dilation and lateral movements appeared to be suppressed.
When mining occurs from the plateau side towards a valley, the stretching phase is observed on the slope. When mining away from the valley the dilation occurs when the systematic ground movements are compressional so that dilation and lateral movements are curtailed.
57
Figure 2.30: Imbalance of dilation forces in sloping terrain (Mills, 2011b).
58 2.6.4 Movements towards goaf area
A two year research study was conducted by Kay (1991) to improve the understanding of strata mechanisms involved when cliffs and valleys were undermined. A monitoring program was carried out over nine cliffs at Baal Bone Colliery with reflectors attached at the valley floor, cliff face and plateau area.
It was found that where the base of the cliff was undermined first, cliff face moved towards the valley significantly. When cliffs were mined from their plateau side first, cliff face still moved towards the valley, which was away from the goaf. In other words, horizontal movements around the cliff face moved towards the valley, regardless of the direction of mining and the direction of cliff lines.
The following mechanism for the horizontal movements was proposed as shown in
Figure 2.31. Due to the extraction of the coal seam, the immediately overlying strata collapse into the goaf area. Above the collapsed area, the strata remain relatively intact and bend into the void. Some of the strata blocks break off and fall into the goaf while others remain overhanding into the void. The movements of these blocks result in a horizontal force on the overlying strata by horizontal shear transfer, causing localised failures within the strata. This horizontal “dragging in” of strata over the void propagates upwards through each formation to the surface.
In flat terrain, the horizontal forces would be in balance. However, in cases where cliffs are undermined, the presence of the cliff face removes the resistance to the horizontal stresses acting from the plateau to the cliff; hence inward movement of the cliff face is generated.
59
Figure 2.31: Assumed horizontal movements within strata layers near cliff line (Kay et al., 2007).
2.6.5 Rigid block model
Unlike the mechanisms introduced above, where the rock strata is treated as deformable materials the rigid block model is proposed, based on the presence of local structures, such as joints, bedding planes and faults in real rock mass which can define the rigid blocks of rock within the overburden.
As a support of the notion of the blocky movement, the horizontal movements of valley sides and outside the goaf area were observed to be constant with small deformations, indicating a rigid body type movement (Holla and Barclay, 2000). In the rigid model an assemblage of rock blocks that stand on the subsidence profile rotate against the adjacent one and translate sideways (Figure 2.32). Topographic reliefs such as river valleys which interrupted the rock blocks are represented by providing spaces between adjacent rigid blocks.
60
Figure 2.32: Block displacements on curved slope (Nemcik, 2003).
Seedsman and Dawkins (2006) presented an explanation of valley closure based on rigid block rotation and interaction with sagging and hogging type subsidence behaviour
(Figure 2.33), and based on the research of subsidence cracking associated with shallow longwalls using case studies from Beltana and Crinum Mines.
The model of rigid blocks provides the explanation for horizontal closure movements at valley sides at locations of both hogging and sagging curvature. In the sagging phase of subsidence, the sagging wave will induce lateral stresses on each block that are resisted by other blocks and this is in fact the basis of voussoir beam behaviour. In the presence of a valley which provides free face the geometry is such that that there may be no immediately adjacent block (Block B in Figure 2.33). In the absence of a reaction, the interaction of blocks A and B results in an unrecoverable translation of Block B into the valley. If there is no valley at the point of sagging, it is possible that the far field translation occurs, resulting in the valley closure in an area of no curvature or in an area of hogging curvature.
61
(a)
(b)
Figure 2.33: a) Valley closure in sagging phase of dynamic wave; b) valley closure in hogging phase (Seedsman and Dawkins, 2006).
This model also provides a possible explanation for upsidence at the base of a valley. If the translation plane is above the valley floor, there will be just valley closure movements. In the case where the translational plane is located beneath the valley floor, the thin strata at the base of the valley may fail to resist the induced translation of the adjacent blocks, resulting in upward deformation of the valley floor.
The kinematics of rock blocks sliding along a known path to produce valley closure was also studied by Keilich (2009), and the author suggested that the movement of each block was a function of the radius of curvature or tilt of the subsidence profile and the height of each individual block. By referring to the theory of kinematics of a particle
62 moving along a known path (Hibbeler, 2007), the author employed the theory of block rotation for simple trigonometric analysis. However, Keilich (2009) provided limited explanations regarding the possible strata mechanisms causing the observed valley closure movements.
Figure 2.34: Model for valley closure and upsidence (Keilich, 2009).
2.6.6 Other field observations
Apart from the mechanisms of lateral dilation, movements towards goaf, and rigid block model, there are some other field observations worthy to be included as potential causes of the valley closure subsidence behaviour. The nature of rock fracturing due to valley closure in river channels in the Southern Coalfield was addressed by Mills (2007), employing a range of monitoring techniques. Figure 2.35 depicts the observed fracturing system around river valleys, which include several major fracture zones such as surface fracture zone, main upsidence zone, dilated bedding planes, basal shear fracturing and hillside bedding plane shear. Most of the differential horizontal movements and resulting vertical dilation is observed concentrating within a 20 m to
30 m wide corridor where the main upsidence occurs. Subsidence monitoring shows that the dilation volume created in the main fracture zone is typically distributed in
63 triangular pattern with the greatest upsidence observed in the centre. Low angle shear fracture sets develop under the effect of horizontal compression in the horizontal strata, and the wedging of the fractured rock strata along the low angle shear failures caused the upward movement. There is also shear fracturing observed on the valley shoulders, which has the potential to result in additional valley closure movements.
Figure 2.35: Rock fracturing features associated with valley closure in the Southern Coalfield (Mills, 2007).
2.7 Conclusions
This chapter has reviewed the mining induced systematic and non-systematic mine subsidence with particular attention on the horizontal ground movements. Three components of the horizontal ground movements have been discussed, and it has been recognised that horizontal stress plays an important role in the development of these movements in terms of horizontal stress release and redistribution. What follows is a description of the horizontal stresses in the Sydney Basin. The high levels of in situ horizontal stress measured in the coalfields of New South Wales are not unique as can
64 be seen from the examples cited above, and there are several possible causes which account for this phenomenon, as has been discussed above.
There are a range of numerical modelling techniques that can be used to study mine subsidence, particular the valley closure subsidence behaviour. While the advantages and disadvantage of different method are usually problem-dependent, their essential differences and capabilities of modelling mining induced subsidence have been discussed. It should be noted that not all of the numerical modelling literature presented in this paper is specific to the mathematical modelling of cliffs and valleys. The numerical modelling studies included in this chapter indicate that, notwithstanding the assumptions of strata behaviour used in the models:
i. the presence of irregular topography affects horizontal movements and
ground strains more than vertical displacements and tilts;
ii. stress concentrations cause notch effects below valley features ;
iii. the major fracture zones extends laterally into the gorge walls;
iv. shear on bedding planes near the base of valleys is significant; and
v. horizontal movements towards valleys are of larger magnitude than the
systematic movements towards the goaf.
The numerical models reviewed in this paper are based on various assumptions of material behaviour. As with all numerical modelling, the assumptions of material behaviour tend to be significant in terms of the model outputs. Thus field monitoring and validation is essential to modelling valley closure and upsidence. It is also recognised that the near surface geology and surface topography play an important role in capturing the fractures within the overburden and getting an accurate prediction of the surface displacements, so more information on geology and topography would be
65 vastly helpful to establish a greater degree of accuracy on the prediction. Moreover, valley closure subsidence effects analyses are normally performed in 2D. There are clear trends to addressing this problem with the routine application of 3D models incorporating the three-dimensional mining, geometry and stress factors into the mining scenarios.
Various hypothesized mechanisms that are regularly discussed as contributing factors in valley closure subsidence effects have been introduced. Based on the review, a number of parameters are found to be responsible for the mining induced valley related movements, including valley geometry, geotechnical factors and mining factors. The mechanics of lateral stress relief and lateral dilation of subsiding strata are well understood and have been for almost a decade. Although there is now an extensive database of valley closure subsidence data available, the geotechnical mechanisms that drive valley closure and upsidence have yet to be included into numerical models that can be used to predict non-conventional subsidence behaviour. It is only when there is a better understanding of such mechanisms, that the ability to predict this kind of behaviour can be achieved.
66 Chapter 3. Laboratory investigation of material properties
3.1 Introduction
This chapter addresses the laboratory testing undertaken to determine the material properties of rock derived from the Southern Coalfield. The geology in the Southern
Coalfield is reviewed first, then standard laboratory testing procedures for indexing rock strength and deformation properties are introduced. This will support the evaluation of the geomechanical conditions for the numerical models described in the following chapters.
3.2 Overview of geology in the Southern Coalfield
In New South Wales numerous mining activities have already been completed and many more operations are planned in several basins which contain significant mineral deposits, as can be seen in Figure 3.1. The Sydney Basin is part of the Sydney-
Gunnedah Basin, a major broad structural basin which extends from southern coastal
New South Wales to Central Queensland. It is approximately 350 km long and an average of 100 km wide, and has a depth of about 3 km in the central area. The Sydney
Basin has a long history of coal exploration and mining, with several thousands of boreholes being drilled in the basin.
67
Figure 3.1: The Sydney-Gunnedah Basin and its coalfields (NSW Department of Primary Industries, 2007).
There are five major coalfields within the Sydney Basin, namely Hunter, Newcastle,
Southern, Western and Gunnedah, as illustrated in Figure 3.1. As one of the five main coalfields within the Sydney-Gunnedah Basin, the Southern Coalfield is bounded approximately by the towns of Campbelltown and Mittagong in the west, and
Wollongong and Helensburgh in the east. The basin within the Southern Coalfield is filled mainly with sedimentary rocks deposited in the Permian and Triassic Periods. The geological sequence of strata in the Southern Coalfield in descending order is shown in
Table 3.1. Figure 3.2 depicts a generalised stratigraphic column of those strata in the
Southern Coalfield.
68 Table 3.1: Subdivision of the sequence in the Southern Coalfield (Holla and Barkley, 2000). Group Formation Wianamatta Group Mittagong Formation Hawkesbury Sandstone Narrabeen Group Newport Formation Bald Hill Claystone Bulgo Sandstone Stanwell Park Claystone Scarborough Sandstone Wombarra Shale Coal Cliff Sandstone Illawarra Coal Measures Bulli seam Balgownie seam Wongawilli seam
Figure 3.2: Sydney Basin sedimentary stratigraphic sections (NSW Department of Primary Industries, 2008).
69 There have been several notable studies to generalise the geological information of the
Southern Coalfield up to approximately the end of 1967, for example Bowman (1974),
Bunny (1972), Ghobadi (1994), Holla and Barclay (2000), Hutton and Bamberry
(1999), Hutton (2009), Moffitt (2000), Packham (1969) and Pells (1985, 1993). Based on the outcomes of these comprehensive studies, a brief summary of the geology, stratigraphy, and sedimentary structures of the formations within the Southern Coalfield is represented below.
Wianamatta Group
The Wianamatta Group is the uppermost unit of the thick Permo-Triassic stratigraphic sequence, and consists of a conformable sequence of inter-bedded grey shales and lithic sandstones. The Wianamatta Group underlies the central portion of the Sydney Basin and, close to the coastline, has been removed by erosion. Outcrops cover large areas of the Cumberland Plain, and small exposures, often confined to ridges and hilltops, occur on the adjoining plateaux.
Hawkesbury Sandstone
The Hawkesbury Sandstone tends to dominate the Sydney area, outcropping over large areas. The Hawkesbury Sandstone ranges in thickness from around 330 m in the west and southwest to a maximum thickness of about 240 m north of Sydney (Packham,
1969). While the formation contains a certain amount of mudstone, inter-bedded with fine sandstone or occasionally in layers of up to 2 m thickness, it consists dominantly of quite massive sandstone beds, typically 2 m to 5 m but at times up to 15 m in thickness.
Cross-bedding is the most noticeable feature of the Hawkesbury Sandstone with about half the beds in any one section being cross-bedded. Apart from the strongly cross-
70 bedded layers, massive units are also very common. The Hawkesbury Sandstone also contains a significant amount of dark grey shale and siltstone, which lenses up to several metres thick.
Narrabeen Group
The entire Narrabeen Group is developed below the Hawkesbury Sandstone and above the Illawarra Coal Measures and includes the main sequence of rocks along the coastal cliffs between Stanwell Park and Scarborough. The lowest units of the Narrabeen Group are Late Permian and the upper unit is Middle to Late Triassic in age. The group covers the major part of the Sydney Basin from the Goulburn Valley in the north to the
Shoalhaven River in the south (Packham, 1969).
Newport Formation
The Newport Formation consists of inter-bedded shale and sandstone sequence and passes upwards from the Garie Formation (Bald Hill Claystone) with the upper boundary inter-bedded with the overlying Hawkesbury Sandstone.
Bald Hill Claystone
The Bald Hill Claystone is persistent throughout most of the Southern Coalfield. It can be traced from the southern extremity of the basin, northwards along the coastal escarpments to Garie where it passes below sea-level, to bores inland almost as far as
Mittagong. It has thickness ranging from 15 m to 80 m in the Sydney Basin, and the typical formation thickness in the Southern Coalfield is about 20 m to 30 m. The Bald
Hill Claystone is generally a chocolate brown claystone with some silty and sandy grey and mottled grey-brown zones. It also contains minor laminated and thinly bedded
71 siltstones and sandstones ranging in thickness from fractions of a metre to three metre
(Pells, 1993).
Bulgo Sandstone
As the thickest unit of the Narrabeen Group in the southern part of the Sydney Basin, the Bulgo Sandstone has a maximum thickness of around 260 m. It consists of thickly bedded and laminated sandstone beds with intercalated siltstone and claystone bands ranging from fractions of a metre to more than ten metre.
Stanwell Park Claystone
The Stanwell Park Claystone outcrops at the village of Stanwell Park after which it has been named. However the best natural sections are exposed at Bulgo, and between Coal
Cliff and Clifton. The Stanwell Park Claystone consists of green and grey mudstones and sandstones. The green shales are lithologically very weak and fret easily on exposure. The sandstones are lithic and range from fine to coarse grained and conglomeratic in places.
Scarborough Sandstone
The Scarborough Sandstone consists mainly of thickly bedded sandstone with shale and sandy shale lenses up to several metres thick. It is conglomeratic with coloured chert clasts especially in the basal half. It is approximately 24 m thick and forms prominent cliffs in the Clifton, Coal Cliff, Stanwell Park and Bulgo areas.
72 Wombarra Shale
The Wombarra Shale extends from the base of the Scarborough Sandstone (the presence of thick green mudstones) and extends downwards to the top of the Coal Cliff
Sandstone. The unit consists of shales, claystones and siltstones with intercalated sandstones, and is similar to the Stanwell Park Claystone described above. Thickness varies from 30 m to 37 m because the base of the Wombarra Shale is identified by the transition to strata of the Coal Cliff Sandstone.
Coal Cliff Sandstone
The Coal Cliff Sandstone extends from the base of the Wombarra Shale downwards to the top of the Illawarra Coal Measures. The unit consists essentially of medium to coarse, quartz-lithic sandstone with pebbly bands. Fine sandstone near the base of the formation commonly changes into shales. There is no discernible break or change of sedimentation between this unit and the underlying Illawarra Coal Measures except for the absence of coal.
Illawarra Coal Measures
The Illawarra Coal Measures consist of inter-bedded shales, mudstones, and coal seams.
This unit is bounded on the east and south by the outcrop of the coal measures which appear above sea level at Coal Cliff, around 20 km north of Wollongong, and traverses the escarpment of the Illawarra Coastal Range. The coal measures are of Permian age and lie conformably upon the Shoalhaven Group. Triassic rocks lie conformably upon the coal measures. The basal formation of the Triassic System is the Coal Cliff
Sandstone of the Narrabeen Group. The economic coal seams in the Southern Coalfield are all in the Illawarra Coal Measures.
73 Bulli Seam
The Bulli Coal is overlain by the Coal Cliff Sandstone of the Narrabeen Group and is the uppermost coal unit in the Illawarra Coal Measures. The Bulli Coal area has been worked extensively in the northern portion of the Southern Coalfield, from outcrop mines on the coastal margins to inland mines. Bulli seam is the major formation where hard coking coal is mostly found, and generally requires underground mining at depths of more than 400 m. Thickness varies from 30 cm in the far south to approximately 4 m in the northern part of the field. In areas north of Mt Kembla the Bulli Seam consists essentially of clean coal. The ash content is remarkably consistent, only rising above the general range of 9 per cent to 12 per cent at the northern end of the field. The coking properties of the seam vary generally from medium to strong with weakness in just one or two localities.
Balgownie Seam
The Balgownie Seam is 5 m to 30 m below the Bulli Seam, with the average thickness of the seam generally at 1.5 m to 2 m. This seam has limited potential for mining.
Wongawilli Seam
The Wongawilli Seam is about 30 m to 60 m below the Balgownie Seam and is the thickest seam in the Southern Coalfield. The full geological section varies from 6 m to
15 m with a maximum of 18 m in the Campbelltown area (Moffitt, 2000). There is a
1 m thick Farmborough Heights Claystone in the seam, which is also known as the
Sandstone Band, Nolan Band and Three-foot Band. The Wongawilli Seam contains 60 per cent to 80 per cent vitrinite, mineral matter free, and is a valuable addition to a coking coal blend.
74 Tongarra Seam
The Tongarra Seam consists of inter-bedded dull and bright coal usually with one or two claystone interbeds, and it sits around 70 m to 80 m below the Wongawilli Seam.
The seam is medium to high in ash yield with medium volatile matter and sulphur content.
3.3 Sampling of rock specimens
Rock cores samples used in this study were retrieved from the PM03 Borehole in
Metropolitan Colliery. The site is located in the Southern Coalfield, approximately
30 km north of Wollongong in New South Wales. It has a deep overburden of sedimentary succession, with depth of cover generally over 300 m. Coal is extracted from the Bulli Seam, the upper seam of the Illawarra Coal Measures in the Southern
Coalfield. A generalised stratigraphy column of the strata presented in the Southern
Coalfield is depicted in Figure 3.3. Stratigraphic thickness from the lithology log is listed in Table 3.2. The experimental study is performed on the specimens drilled from the logs as shown in Figure 3.4.
75
Figure 3.3: Generalised stratigraphy column of the Southern Coalfield (Geosensing Solutions, 2008).
Table 3.2: Stratigraphic horizons from PM03.
Stratigraphic horizons Picks from Lithology Log Formation Drillers From (m) Drillers To (m) Thickness Hawkesbury Sandstone 0 156.07 156.07 Newport Formation 156.07 175.64 19.57 Garie Formation 175.64 176.06 0.42 Bald Hill Claystone 176.06 202.24 26.18 Bulgo Sandstone 202.24 394.26 192.02 Stanwell Park Claystone 394.26 443.93 49.67 Scarborough Sandstone 443.93 472.37 28.44 Wombarra Claystone 472.37 507.4 35.03 Coal Cliff Sandstone 507.4 524.56 17.16 Bulli Seam 524.56 526.14 1.58 Loddon Sandstone 526.14 536 9.86
76
Hawkesbury Sandstone
Newport Formation
Bald Hill Claystone
Bulgo Sandstone
Stanwell Park Claystone
77
Scarborough Sandstone
Wombarra Shale
Coal Cliff Sandstone
Figure 3.4: Field logs for laboratory tests.
The test specimens were cored (with diameter of 42 mm) using a core drilling machine
(Figure 3.5a) in the laboratory. The cylinder specimens were cut in lengths of 105 mm, with a height to diameter ratio of 2.5, which is in agreement with suggestions outlined by the International Society of Rock Mechanics (ISRM, 1978). Flatness of both ends of the samples was 0.02 mm. During the process of coring, water was extensively used for cooling and removing the cuttings, which had increased the sample moisture content.
Therefore, the drilled specimens were left in the drying oven (Figure 3.5b) for at least
72 hours prior to testing to eliminate the influence of additional moisture on sample preparation. It should be noted that the in situ moisture conditions of the cores were not
78 well preserved in field before the collection for the test; and it is beyond the scope to re- drill the logs from underground to conduct the tests. The influence of moisture on the strength for sedimentary rock types has been reported in the literature (Hawkins and
McConnell, 1992, Zainab et al., 2007, Prakoso and Kulhawy, 2011). The reduction from dry to saturated strength for sandstones usually varies from 10% to 34%. In this study, the strength of rock specimens was scaled based on the scale effects for the model input; therefore the effect of moisture was also minimized. The scaled properties were back analysed against empirical data from similar areas, and the values were found in the reasonable range. Thus, the in situ moisture conditions did not have significant impacts on the model input strength properties.
(a) (b)
Figure 3.5: (a) Core drilling machine and (b) drying oven.
79 3.4 Laboratory testing of rock materials
This section addresses the mechanical properties of rock materials. Details of the laboratory testing procedure and the results of the testing are presented. Laboratory tests are carried out in accordance with the suggested methods by ISRM (1978), and compressive strength, elastic modulus and tensile strength of the rock materials are measured in this section.
3.4.1 Determination of the uniaxial compressive strength
Compressive strength is one of the most important mechanical properties of the rock material, indicating the capacity of a material to withstand axially directed compressive stress. The compressive strength is usually measured experimentally by means of a compressive test. The tests described in this section cover the determination of the uniaxial compressive strength and deformability of intact rock core specimens in uniaxial compression, and includes uniaxial strength, Young’s modulus and Poisson’s ratio.
MTS 815 Rock Mechanics Test System was used for all the laboratory experiments described in this Chapter, and the tests have been conducted at the Geomechanics
Research Laboratory at the School of Mining Engineering, UNSW. A basic system for uniaxial testing includes a load frame assembly, a hydraulic package and a controller, and the systems components are described in Figure 3.6. It offers high axial force capacity with maximum compression rating of 4600 kN. The high-capacity, high stiffness design of the load frames coupled with high-response servo-hydraulics and digital control technology meet the need for investigating the complete deformation of the rock samples in this study.
80
Figure 3.6: Schematic of MTS rock mechanics testing system (MTS Systems Corporation, 2004).
As shown in Figure 3.7, the specimen was placed on a spherical seat to ensure the load was applied evenly. During the compression tests, axial stress on the sample was recorded as reaction force at the compression platen. Uniaxial strain measurement kits and circumferential strain measurement kits were attached to the cylinder specimen to measure the axial and circumferential deformation using Linear Variable Differential
Transducer (LVDT), as illustrated in Figure 3.7. It should be noted that the uniaxial strain that was measured directly from the built-in sensors within the MTS machine was not used in this study, because during the testing procedure micro fractures developed at the ends of the sample, which affected the accuracy of the strain measurement based on the change in length of the sample. The uniaxial and circumferential LVDT were attached in the centre part of the specimen in order to eliminate the end friction effects
(Hawkes and Mellor, 1970).
There were five specimens tested for each rock type and the tests were conducted based on the ISRM suggested methods for determining the uniaxial compressive strength and
81 deformability of rock materials (ISRM, 1978). Loading rate must be set sufficiently slow to ensure the sample remains in quasi-static equilibrium and there is no strength increase or unexpected material responses during the process. The loading rate was set to maintain the displacement rate at 0.002 mm/s for the compression tests. The fracture mode of the failed specimens of different rock types are presented in Figure 3.8.
Figure 3.7: Uniaxial compression test setup for rock specimen.
82
HBSS: Hawkesbury Sandstone, BGSS: Bulgo Sandstone, SPCS: Stanwell Park Claystone, SBSS: Scarborough Sandstone, CCSS: Coal Cliff Sandstone.
Figure 3.8: Typical failure mode of specimens after uniaxial compression tests.
The Uniaxial Compressive Strength of the specimen is calculated by dividing the maximum load from the MTS by the original cross-sectional area. Table 3.3 summarises the results of uniaxial compressive strength tests for each rock type.
83 Table 3.3: Results of uniaxial compressive strength tests.
Uniaxial Compressive Strength (MPa) Rock Type Depth (m) Range Mean Std. Deviation
Hawkesbury Sandstone 137-140 57.5 – 61.5 59.5 1.6
Bulgo Sandstone 344-347 50.9 – 112.9 90.1 17.4
Stanwell Park Claystone 426-429 85.8 – 88.0 83.9 12.1
Scarborough Sandstone 457-460 108.6 – 125.9 119.6 6.7
Coal Cliff Sandstone 511-514 110.0 – 114.1 118.7 11.3
Young’s modulus is “elastic modulus” which is a measure of the stiffness of rock material. It can be experimentally determined from the slope of a stress-strain curve created during the compressive tests conducted on rock samples. Poisson’s ratio is a measure of the ratio of lateral strain to axial strain at the linearly-elastic region. It is the ratio of the fraction of expansion divided by that of compression. For determining
Young’s modulus and Poisson’s ratio, stress-axial strain and stress-lateral strain curves of rock specimens from compressive tests are plotted. Figure 3.9 shows a typical stress- strain curve of the Coal Cliff Sandstone specimen that has been obtained from the laboratory testing.
84 160
140
120 Stress (MPa) Stress 100
80
60
40
20
0 -0.004 -0.002 0 0.002 0.004 0.006 0.008 Lateral Strain Axial Strain
Figure 3.9: Stress-strain curve of one of the Coal Cliff Sandstone specimens.
The values of Young’s modulus and Poisson’s ratio are usually calculated using the engineering practice method suggested by ISRM (1978). Young’s modulus, E, is calculated by dividing the compressive stress by the axial strain in the elastic portion of the stress-strain curve. Poison’s ratio, ν, is calculated from the equation:
slope of axial stress − strain curve ν = − slope of lateral stress − strain curve
E = − slope of lateral curve (1)
Table 3.4 lists the values calculated for different rock types.
Table 3.4: Calculated Young’s modulus and Poisson’s ratio. Rock Type E (GPa) ν Hawkesbury Sandstone 10.08 0.24 Bulgo Sandstone 20.48 0.25 Stanwell Park Claystone 22.06 0.23 Scarborough Sandstone 20.43 0.20 Coal Cliff Sandstone 25.62 0.24
85 The relation between the shear and bulk modulus and Young’s modulus and Poisson’s ratio are given in Equation 2 and 3, and the modulus and Poisson’s ratio for different rock units are listed in Table 3.5.
E G = 2(1+ν ) (2)
E K = 3(1−2ν ) (3)
Where:
퐺 = shear modulus,
퐾 = bulk modulus,
퐸 = Young’s modulus, and
ν = Poisson’s ratio.
Table 3.5: Calculated Young’s modulus, Poisson’s ratio, bulk modulus and shear modulus.
E ν K G Rock Type (GPa) (GPa) (GPa)
Hawkesbury Sandstone 10.08 0.24 6.46 4.06
Bulgo Sandstone 20.48 0.25 13.65 8.19
Stanwell Park Claystone 22.06 0.23 13.62 8.97
Scarborough Sandstone 20.43 0.20 11.35 8.51
Coal Cliff Sandstone 25.62 0.24 16.42 10.33
3.4.2 Determination of the tensile strength
Tensile strength of rock material is normally defined by the ultimate strength in tension.
It can be determined from several tests including direct tensile test and indirect tensile test. Due to the difficulty of sample preparation in direct test, the indirect tensile test, also known as the Brazilian test is used to obtain the tensile strength of rock specimens,
86 producing tensile failure in the loaded diametral plane of cylindrical specimens. The justification of the indirect Brazilian test is that most rocks in biaxial stress fields fail in tension at their uniaxial tensile strength, when one principal stress is tensile and the other one is compressive (less than 3 times the tensile principal stress)(ISRM, 1978).
The Brazilian test setup is shown in Figure 3.10.
Figure 3.10: Typical testing setup for Brazilian test.
The test specimens were also prepared under the guideline of ISRM (1978). A constant load was applied on the specimen at a constant rate such that failure in the weakest rocks occurs within 15 to 30 seconds.
87 The suggested formula for calculating the tensile strength based on the Brazilian test is
(ISRM, 1978):
2P P σ = = 0.636 t πDt Dt (4)
Where 푃 is the load at failure (N), 퐷 is the diameter of the specimen (mm), and 푡 is the thickness of the test specimen measured at the centre (mm). Table 3.6 summarises the results of the tensile tests.
Table 3.6: Results of the Brazilian test.
Tensile Strength (MPa) Rock Type Mean Std. Deviation
Hawkesbury Sandstone 4.6 1.9
Bulgo Sandstone 4.5 1.5
Stanwell Park Claystone 5.0 2.0
Scarborough Sandstone 7.9 1.5
Coal Cliff Sandstone 8.3 1.1
3.4.3 Determination of the strength in triaxial compression
A series of triaxial compression tests were performed to measure the strength of specimens as a function of confining pressures. The Hoek Triaxial Cell is used to apply confining pressures, and it consists of a hollow steel cylinder with threaded removable ends. A urethane rubber sleeve incorporating a U-shaped seal to form a pressurization chamber for the hydraulic fluid is mounted within the cell. The confining pressure is applied to the cell using a hydraulic pump with pressure gauge which has a maximum capacity of 68.9 MPa (10000 PSI).
88 Figure 3.11 illustrates the test setup for the triaxial test, where the specimen is placed in the MTS and subjected to confining pressure. First, the ends of the rock specimens were grounded flat as suggested by ISRM (1978). After saturation of the pressurization chamber, the sample was inserted into the chamber. After applying a small confining pressure to hold the rock specimen in place, the cell with its spherical plates was then placed in MTS loading frame which then applied an axial load to the flattened ends of the sample inside.
Figure 3.11: Triaxial test setup in the laboratory.
89 The triaxial tests were carried out under different confining pressures ranging from
3 MPa to 16 MPa. The confining pressure was increased to the set value at a constant rate to ensure that the specimen was under uniform hydrostatic stresses. The loading procedure was the same as that of the UCS test, and which was also in accordance with the suggested methods by ISRM (1978).
There are two types of strength criterion namely Hoek-Brown criterion and Mohr-
Coulomb criterion which are widely used in rock mechanics. As the most common failure criterion for rock mechanics, the Mohr-Coulomb criterion describes a linear relationship between the normal and shear stresses or maximum and minimum principal stresses at failure. The stresses developed on the failure plane are on the strength envelope when failure occurs. The Mohr-Coulomb criterion for triaxial data is expressed as:
1+sin ∅ 2c cos ∅ σ = σ + 1 1−sin ∅ 3 1−sin ∅ (5)
Where 푐 is cohesion and ∅ represents the internal friction angle.
Due to the fact that the classic strength theories used for other engineering materials are not applied to rock materials, many empirical strength criterion have been proposed for mining and civil engineering use. The Hoek-Brown criterion is the most widely used empirical failure criterion. It relates the major principal stress to the minor principal stresses as proposed by Hoek and Brown (1980). The peak triaxial compressive strength is described as:
2 0.5 σ1 = σ3 + (mσcσ3 + sσ1 ) (6)
Where 푚 varies with rock type and 푠 = 1 for intact rock.
90 In addition to the triaxial tests, the data for uniaxial compression and Brazilian tensile tests derived above were used for the calculation of the Hoek-Brown parameters and the
Mohr-Coulomb parameters of the intact rock. Table 3.7 summarises the triaxial compression test results.
Based on the results from uniaxial, triaxial and Brazilian tests, the strength of the different types of intact rock is plotted as a function of the maximum and minimum principal stresses, as illustrated in Figure 3.12. The strength envelop is fitted with both the Hoek-Brown criterion and the Mohr-Coulomb criterion, as can be seen in Figure
3.12.
Table 3.7: Results of the Triaxial Compression tests for different rock types.
흈풄풊 흈풕 Cohesion Friction Rock Type 풎풊 (MPa) (MPa) (MPa) angle
Hawkesbury Sandstone 57.0 4.6 12.3 10.1 47.9
Bulgo Sandstone 82.5 4.5 18.3 12.8 53.0
Stanwell Park Claystone 93.5 5.1 18.5 14.4 53.2
Scarborough Sandstone 97.6 8.0 12.2 17.5 47.6
Coal Cliff Sandstone 118.5 8.3 14.3 19.6 50.6
91
Hawkesbury Sandstone 120
100
80
60 Hoek-Brown Mohr-Coulomb 40 Data points 20 Major principal stress (MPa) stress principalMajor
0 -10 -5 0 5 10 Minor principal stress (MPa)
Bulgo Sandstone 200
180 160 140 120 100 Hoek-Brown 80 Mohr-Coulomb 60 Data points 40 Major principal stress (MPa) stress principalMajor 20 0 -10 -5 0 5 10 15 Minor principal stress (MPa)
92
Stanwell Park Claystone 250
200
150 Hoek-Brown 100 Mohr-Coulomb Data points 50 Major principal stress (MPa) stress principalMajor
0 -10 -5 0 5 10 15 Minor principal stress (MPa)
Scarborough Sandstone 250
200
150 Hoek-Brown 100 Mohr-Coulomb Data points 50 Major principal stress (MPa) stress principalMajor
0 -10 -5 0 5 10 15 20 Minor principal stress (MPa)
93 Coal Cliff Sandstone
250
200
150 Hoek-Brown 100 Mohr-Coulomb Data points 50 Major principal stress (MPa) stress principalMajor 0 -10 -5 0 5 10 15 20 Minor principal stress (MPa)
Figure 3.12: Curve fitting of Hoek-Brown and Mohr-Coulomb envelope to triaxial test data.
3.5 Conclusions
A review of the Southern Coalfield geology has been initially presented in this chapter.
The Southern Coalfield area is mainly filled with sedimentary rocks deposited in the
Permian and Triassic ages. The principal coal-bearing sequence in the Southern
Coalfield is the Illawarra Coal Measures, and the Bulli seam represents the majority of the coal reserves in the Southern Coalfield.
Rock cores specimens used in this study were retrieved from Metropolitan Colliery. The numerical modelling analyses are based on the geology of Metropolitan Colliery which is presented in the following chapters. The specimens were made in accordance with guidelines suggested by ISRM (1978) for laboratory testing. It should be noted that the lack of tests on some claystones and shale could be explained by the difficulty in drilling the cores and obtaining suitably sized samples.
94 Uniaxial compression, triaxial compression and tension test were performed to investigate the strength, deformation behaviour and failure characteristics of sandstones and claystone from the Southern Coalfield. The uniaxial compression strength of the rock materials ranges from 59.5 MPa to 119.6 MPa, with the Young’s modulus ranging from 10.1 GPa to 25.6 GPa. These sandstones and claystone have Poisson’s ratio values between 0.20 and 0.25. The tensile strength of the rock samples increases from 4.6 MPa to 8.3 MPa as the retrieve depth of the core specimen becomes deeper. For the strength data in triaxial compression testing, the nonlinear Hoek-Brown criterion better reflects the peak strength properties than does the linear Mohr-Coulomb criterion. It can also be concluded that the Mohr-Coulomb criterion is generally suitable for the low range of minor principal stress, and at high minor principal stress it tends to overestimate the strength. At high stress level, the Hoek-Brown criterion gives a lower strength estimate than the Mohr-Coulomb envelope.
This chapter describes the determination of the strength and deformation properties of intact rock using laboratory testing. These material properties will be used for the numerical models to conduct a detailed assessment of geomechanical conditions in the following chapters.
95 Chapter 4. Model setup and validation
4.1 Introduction
This chapter outlines the development of numerical models simulating mining induced subsidence, followed by the model validation for subsidence movements against pre- existing field results. Geomechanical conditions of the overburden rock units in the
Southern Coalfield providing model input are firstly assessed, and the estimation of strength and deformation parameters of the overburden is discussed taking into account the laboratory test results of standard specimens, the relative scale effects and experiences in assessment of the engineering geologic features of rocks from the literature. The modelling technique is then evaluated to justify its capability to predict the actual field movements in a realistic manner.
4.2 Modelling technique
A wide range of numerical modelling approaches for simulating subsidence issues has been reviewed and discussed in Chapter 2. In this study, the Universal Distinct Element
Code (UDEC) (Itasca, 2011), which is a two-dimensional numerical package based on the distinct element method for discontinuous modelling, is selected as the most suitable software to establish field scale subsidence models. Within this package the jointed rock mass is represented as an assemblage of discrete blocks. The individual blocks used in this study are deformable, representing the intact rock, and the discontinuities are treated as boundary conditions between blocks. The Mohr-Coulomb plasticity block material model is applied to the modelling in this research.
96 4.3 Estimation of model input data
This section describes the selection of the geomechanical input data required for modelling. A considerable amount of literature has been published on the material properties of the stratigraphic formations in the Sydney Basin providing informative background for the collection of the model input data. In this study, the estimation of geomechnical properties of rock units for the numerical modelling input is based on a combination of laboratory tests on standard size samples, as described in Chapter 3, taking into account the scale effects and empirical field observation results.
4.3.1 Mechanical properties of intact rock
The detailed stratigraphic information in the Southern Coalfield has been presented in
Chapter 3, and a generalised stratigraphic column across the Southern Coalfield area was used for numerical modelling in this research, as can be seen in Table 4.1.
Table 4.1: Stratigraphic columns for numerical modelling.
Stratigraphic unit Thickness (m)
Hawkesbury Sandstone 129
Newport Formation 24
Bald Hill Claystone 11
Bulgo Sandstone 172
Stanwell Park Claystone 44
Scarborough Sandstone 34
Wombarra Shale 36
Coal Cliff Sandstone 22
Bulli coal seam 3
Sub-Bulli Sediments 75
97 Mechanical properties of 42 mm diameter specimens drilled from these formations have been measured, as described in Chapter 3. The mechanical properties of the different types of rock can be obtained from laboratory tests; however, in order to use the standard specimens to properly estimate the strength and deformation characteristics of field scale rock, changes in the mechanical properties of rock with size should be taken into consideration. Scale effects of rock strength and deformation properties have been well recognised in the literature (Gupta and Seshagiri Rao, 2000, Ozkan et al., 2009,
Poulsen and Adhikary, 2013, Pratt and Black, 1972). There are typically two parts of the rock mass strength: the strength of the intact rock and the strength of the fractures, with each of them exhibiting different scale effects. The scale effects of discontinuities on the mechanical behaviour of rock mass have been discussed in the literature with the help of highly developed numerical methods (Elmo et al., 2011, Martin et al., 2012,
Natau, 1990, Yoshinaka et al., 2008), which are outside the scope of this study. This section focuses on the determination of the mechanical properties of intact rock within a rock mass to be considered in the numerical simulation. The following material properties are required for modelling, where the Mohr-Coulomb failure criterion is applied:
Density (kg/m3) Young’s modulus (GPa) Poisson’s ratio Cohesion (MPa) Friction angle (Degree), and Tensile strength (MPa)
98 Uniaxial compressive strength
It is well informed that the observed peak strength of rock decreases as sample size increases. Many empirical relationships have been proposed to estimate the strength of rock at field scale, and the most common one is presented in Figure 4.1 (Hoek and
Brown, 1980). The figure illustrates the influence of size on the strength of different rock types based on laboratory testing conducted by a number of different researchers.
Figure 4.1: Scale effect on uniaxial compressive strength of intact rock (Hoek and Brown, 1980).
The decreasing trend in Figure 4.1 is expressed as:
−0.18 휎푐 = 휎푐.50(푑/50) (1)
99 Where σc.50 is the uniaxial compressive strength of the cylindrical specimen with diameter of 50 mm and σc is the uniaxial compressive strength with a given diameter.
Yoshinaka et al. (2008) have developed a similar relationship in size effects that was derived from Weibull’s theory:
1 ⁄m σc1 = σc2(V2/V1) (2)
Where 휎푐1 and 휎푐2 are the strength of samples with volumes 푉1 and 푉2 respectively, and
푚 is a constant of the material. The equation was modified using equivalent length:
1⁄ 3 푑푒 = 푉 3, and an exponent, 푘 = ⁄푚, as follows:
−k σc = σc0(de/de0) (3)
This equation derived from Weibull’s theory can be applied to intact hard rock in a wide range of size. Based on the results of numerous laboratory tests and in situ tests, exponent k in Equation 3 varies for different types of rock: k ranges from about 0.1 to
0.3 for homogeneous hard rock, and from about 0.3 to 0.9 for weathered and/or extensively microflawed rock (Yoshinaka et al., 2008). These relations can be plotted together with the Hoek and Brown curve in Figure 4.2.
100
Figure 4.2: Scale effects relations developed by Yoshinaka and Hoek & Brown (Mas Ivars, 2010).
There are also some notable studies focusing on the selection of mechanical properties of the local sedimentary rock units in the Southern Coalfield. Pells (2002) studied the engineering geologic features of the Sydney sandstones, mainly Hawkesbury Sandstone, and a reduction factor of 0.6 was adopted with respect to 50 mm diameter core strength to be used at the tunnel scale. Similarly, Tarrant (2006) conducted numerical modelling studies to investigate the strata behaviour in Metropolitan Colliery, and the laboratory
UCS was reduced by 0.58 for the model input. In this study, the geological information was mainly derived from the Southern Coalfield, especially the Metropolitan Colliery.
Therefore, based on the discussions of the scale effects above, a reduction factor of 0.58 is used for the uniaxial compressive strength in the current study.
101 Elastic modulus
In sedimentary rocks there is an approximate relationship between deformation modulus and compressive strength. Deere and Miller (1966) proposed the distribution of the data in the modulus ratio (E/UCS) graph for sedimentary rocks, as shown in Figure 4.3.
Figure 4.3: Relationship of UCS and Young’s Modulus at 50% of UCS, showing engineering classification for intact sedimentary rocks (Bell and Lindsay, 1999) after (Deere and Miller, 1966).
Gupta and Seshagiri Rao (2000) further studied the strength and deformational behaviour of rock taking into account weathering effects, and developed the relationships for different rock types including granite, basalt, quartzite and other rocks, as illustrated in Figure 4.4.
102
Figure 4.4: Relationship between E and UCS for different rock types (Gupta and Seshagiri Rao, 2000).
As can be seen from Figure 4.4, a close variation of the points fall between modulus ratio of 100 to 500, and the functions of these relationships is summarised in Table 4.2:
Table 4.2 Expression of relationship between E and UCS.
Rock type E
2 0.98 Igneous rocks (푟 = 0.87) 휎푐 × 286
2 1.91 Sedimentary rocks (푟 = 0.60) 휎푐 × 80
2 1.11 All rocks (푟 = 0.83) 휎푐 × 150
103 Based on the study of the geomechanical properties of coal measure rocks, McNally et al. (1996) proposed an empirical estimation of the Young’s modulus using the modulus ratio listed in Table 4.3.
Table 4.3: Estimation of Young’s modulus (McNally et al., 1996).
Modulus ratio, E/UCS
500+ Exceptionally brittle cherty claystone
300 Strong, massive sandstone and conglomerate
200 Most coal measures rock types, especially sandstone
200 Strong, uncleated coal, UCS > 30 MPa
150 Medium to low strength coal
100 Weak mudstone, shale, non-silicified claystone
Taken together these results suggest that a modulus ratio of 200 can be assumed to estimate the Young’s modulus of the sandstones using the UCS values in this study.
Tensile strength
The tensile strength of rock mass can be estimated using the relationships given in
Table 4.4 (McNally et al., 1996). It suggests that for the sedimentary formations in the
Southern Coalfield, tensile strength for rock blocks can be estimated as one tenth of their uniaxial compressive strength.
104 Table 4.4: Typical compressive/tensile strength relations for coal measure rocks (McNally et al., 1996).
UCS/ITS UCS/UTS (indirect tensile strength) (uniaxial tensile strength)
20 14 Strong sandstone and conglomerate
20 14 Strong coal
15 10 Sedimentary rock generally
15 10 Medium to low strength coal
12 8 Shale, siltstone, mudstone
10 7 Weak shale, siltstone, mudstone
In general, it can be concluded that the mechanical properties of field scale rock blocks are based on the laboratory testing taking into account the scale effect for rock strength as well as engineering assessments of the geomechanical conditions of the Southern
Coalfield (Keilich, 2009, Pells, 2002, Waddington and Kay, 2002). The adopted material properties of the stratigraphic units for the UDEC model are listed in Table 4.5.
105 Table 4.5: Adopted mechanical properties for stratigraphic units used in the models.
Stratigraphic unit Bulk Shear Friction Tensile Density Cohesion modulus* modulus* angle strength (kg/m3) (MPa) (GPa) (GPa) (degree) (MPa)
Hawkesbury 2397 2.6 1.2 4 47 0.5 Sandstone
Newport Formation 2290 3.5 2.5 4 35 0.5
Bald Hill Claystone 2719 3.5 2.5 6 46 0.5
Bulgo Sandstone 2527 7.2 4.3 9 46 0.5
Stanwell Park 2693 6.5 4.3 11 32 0.5 Claystone
Scarborough 2514 8.2 6.1 12 46 0.5 Sandstone
Wombarra Shale 2643 6.9 5.0 12 40 0.5
Coal Cliff Sandstone 2600 9.1 5.7 11 46 0.5
Bulli Seam 1500 1.6 1.0 1.9 25 0.5
Sub-Bulli units 2500 9.1 5.7 18.7 40 0.5
(*Calculated based on Equation 2 and 3 in Section 3.4.1)
4.3.2 Properties of rock structures
Discontinuities in geotechnical engineering include bedding, joints, fault and other structures, which play an important role in the rock mass behaviour by providing planes of weakness. It should be noted that data on these structural features is sparse. Therefore it is unrealistic to generalize a detailed description about the geometrical and mechanical characteristics of rock discontinuities in the Sydney area. However recommendations from numerous onsite projects could give some clues about the range of values that should be used as the input data for numerical modelling.
Based on the observations on cliff faces in the Sydney Basin, McNally et al. (1996) reported that:
106 Joint spacing is roughly proportional to bedding thickness. Joints commonly
occur in clusters from the same set, with wide gaps between clusters. In these
situations the average joint spacing tends to understate the most common block
sizes.
The great majority of joints are vertical or sub-vertical, and occur in well-
defined sets with one dominant orientation. The secondary set tends to be more
or less perpendicular to this and less persistent.
Joints tend to have a vertical persistence roughly equal to bedding thickness, but
tend to be laterally offset between beds. This creates a large scale “brick wall”
effect, but with the dimensions of the “bricks” being proportional to bedding
thickness.
In general it can be suggested that the horizontal and sub-horizontal bedding planes have been recognised as the most important form of discontinuity in the Sydney area from a large amount of field investigations. Vertical joints are typically in orthogonal pattern, thus large scale brick-shaped blocks can form in the sedimentary rocks, especially in Hawkesbury Sandstone. Based on numerous studies on the geometry of cross beddings and joints in Sydney rocks by several methods such as field mapping and aerial photo interpretation (Holla and Barclay, 2000, Waddington and Kay, 2001,
2002), the pattern of beddings and joints in the major stratigraphic units within the
Southern Coalfield is listed in Table 4.6.
107 Table 4.6: Typical discontinuity spacing in main stratigraphic units in the Southern Coalfield.
Rock unit Bedding spacing (m) Joint spacing (m)
Hawkesbury Sandstone 0.3-7 7-15
Newport Formation 0.1-2 1-5
Bald Hill Claystone 0.1-1 < 2
Bulgo Sandstone 0.5-5 2-13
The mechanical properties of beddings and joints, normal and shear deformation characteristics are important parameters which are usually difficult to measure in the laboratory environment and therefore are generally limited. Using the model of Duncan and Goodman (1968), normal and shear stiffness can be calculated as:
EE′ k = n (E−E′)S (4)
GG′ k = s (G−G′)S (5)
Where:
푘푛, 푘푠 = joint normal and shear stiffness
퐸, 퐺 = Young’s modulus and shear modulus
퐸′, 퐺′ = the effective modulus
S = joint spacing
There are also many mathematical models describing the normal and shear stiffness relationship. For example, the relationship between normal and shear stiffness can be written as (Bertuzzi and Pells, 2002):
k k = n s 2(1+ν) (6)
Which suggests that 푘푠 should be 0.33 to 0.5 times 푘푛.
108 Bandis et al. (1983) investigated the joint deformation and the relationship between kn⁄ks ratio and normal stress in the laboratory test, as shown in Figure 4.5.
Figure 4.5: Joint behaviour under normal and shear loading (Bandis et al., 1983).
It is clear that the change on normal stress has a pronounced influence on the kn⁄ks ratio, and it is recommended that in cases where normal stress is greater than 1 MPa, a shear/normal stiffness ratio of around 0.1 could be used in the absence of specific data, as also noted by Bertuzzi and Pells (2002). This is therefore the ratio used in the current research.
The scale effect of joint shear stiffness is presented in Figure 4.6, and represents a wide range of discontinuities from the literature.
109
Figure 4.6: Scale effect on peak shear stiffness (Bandis et al., 1983).
The shearing resistance of joints is expressed in terms of effective angle of friction, which is derived by small scale roughness and large scale waviness and also by cohesion. The value of friction angle depends on factors such as rock type, scales, and moisture conditions (Coulson, 1972, Deere and Miller, 1966), and indicative ranges for the angle of friction is around 25° to 35° for the majority of rocks.
Thus together with the information on the rock discontinuities from numerous site investigations for specific projects in the Sydney region (Bertuzzi and Pells, 2002,
Keilich, 2009, Waddington and Kay, 2001, 2002), the adopted discontinuity properties for the numerical modelling in this study have been decided as shown in Table 4.7.
110 Table 4.7: Adopted mechanical properties for discontinuities used in the models.
Normal Shear Friction Tensile Cohesion stiffness stiffness angle strength (MPa) (GPa/m) (GPa/m) (degree) (MPa)
26 2.6 0 25 0
4.3.3 In situ stress
Apart from the determination of material properties of rock mass, the selection of in situ stress is also fundamental to the numerical modelling. There is a large volume of published literature describing the in situ stress in the Sydney Basin. Measurements of principal stresses within coalfields in New South Wales indicate that the horizontal stresses are typically greater than the vertical stresses near the surface.
In the 2D UDEC model, where the plane strain condition is applied, the major principal stress is horizontal and the minor principal stress is in the vertical direction. The vertical stress is usually proportional to the weight of overburden, especially in the sedimentary rock units with flat or low dipping angle as simulated in this study. However, the gradient of horizontal stress as cover depth increases is different. Waddington and Kay
(2002) summarized that the ratio of horizontal stresses to vertical stress was found to be in the range of 1.5 to 2.0 within the Sydney Basin, based on the literature review of the horizontal stress within the coalfields in New South Wales. The ratio was even found to be over 3.0 in a number of locations at Tower Colliery (Hebblewhite et al., 2000). Thus the pre-mining stress regime in the numerical models has a typical horizontal/vertical stress ratio of 2.0.
111 4.4 Model validation
The study in this section is designed to evaluate the capability of the numerical models to predict the actual monitored results from existing subsidence database in order to build confidence in the numerical modelling. The UDEC models built using the parameters described in Section 4.3 were validated against both the field monitoring data and the empirical database from a number of specific mining sites in this section.
Each of them included two scenarios: single longwall extraction and multiple longwall extractions. The observed subsidence and horizontal displacements from Metropolitan
Colliery and the empirical subsidence prediction curves using Incremental Profile
Method (Waddington and Kay, 1998) were used for the validation.
4.4.1 Overview of mining setting in Metropolitan Colliery
Metropolitan Colliery is located at Helensburgh between Wollongong and Sydney within the Southern Coalfield as shown in Figure 4.7. Metropolitan Colliery has been mining Bulli Seam coal since 1886. The Bulli Seam varies from approximately 2.6 m to
3.5 m in thickness and it is expected that its full thickness would be extracted during the underground mining operations. The typical overburden depth in Metropolitan Colliery is around 400 m to 470 m and most of the variation in overburden depth is a result of surface topography, which typically ranges up to 70 m above the river level and occasionally 120 m at the high points.
112
Figure 4.7: Location of Metropolitan Colliery (Metropolitan Coal, 2011).
The conventional longwall coal mining method is employed at the Metropolitan
Colliery. The longwall extraction panel widths have increased from nominally 123 m for Longwall 1 to 163 m for Longwall 18 over a series of panels. The length of the longwall panels is approximately 1500 m. The chain pillar lengths have varied from
50 m over Longwall 7, 75 m over Longwall 8 and then 90 m over the remaining
113 longwalls, and the pillar widths have remained constant at 35 m (centres). A main subsidence monitoring line, D-Line, was established perpendicular to the Longwall 1 to
18, and it was extended to beyond the 35° angle of draw prior to extraction of the relevant panels. Figure 4.8 illustrates the layout of Longwall 1 to 18 and the D-Line.
Figure 4.8: Layout of Longwall 1 to 18 and D-Line in Metropolitan Colliery (DeBono and Tarrant, 2011).
Longwall 20 was the first longwall developed within the Central Area, and it has panel width of 163 m and the overall longwall length is 3603 m. Its extraction started in May
2010 and completed in September 2011. Survey lines, Line 9C and Line 9C West, run oblique to Longwall 20 panel with survey marks established at a spacing of around
20 m (typically being less than 1/20th the depth of cover) to monitor the subsidence movements induced by the extraction of Longwall 20. The location of Longwall 20, the monitoring line Line 9C (marked as C Line in Figure 4.9) and Line 9C West are shown in Figure 4.9.
114
Figure 4.9: Location of Longwall 20, monitoring line C Line and 9CW Line (Metropolitan Coal, 2011).
115 As information for the reader the following support method is introduced. In
Metropolitan Colliery, the roadways are nominally 5.2 m wide and 3 m to 3.2 m high.
The mine used 6 × 2.1 m long “X” grade bolts (30 tonne capacity) per row at a row spacing of 1.2 m as illustrated in Figure 4.10. Additional bolts were installed through intersections, and all bolts were fully encapsulated.
Figure 4.10: Primary bolt pattern in Metropolitan Colliery (Tarrant, 2006).
4.4.2 Overview of the Incremental Profile Method
Subsidence movements can be predicted using the Incremental Profile Method (IPM) which was developed by Waddington and Kay (1998). The Incremental Profile Method is an empirical approach that is based on extensive measurements of mining induced subsidence that has taken place in areas mined for more than fifty years. This method has been continually refined to suit a wide variety of mine layouts with differing geological conditions in New South Wales and Queensland coalfields.
116 Incremental subsidence is obtained by subtracting the subsidence observed at a point before the extraction of a panel from the subsidence monitored at that point after the longwall is extracted. It thus indicates the influence of a particular longwall panel extraction on the change of subsidence. In the Incremental Profile Method a series of curves have been generated combing aspects of the empirical, graphical, influence line and profile function method. Figure 4.11 shows the trend of values of maximum incremental subsidence/seam thickness increasing as the values of panel width/depth of cover increase; and decreasing as the values of chain pillar width/depth of cover increase.
Figure 4.11: Graph for the prediction of maximum subsidence for various extraction conditions using Incremental Profile Method (Waddington and Kay, 1998).
117 4.4.3 Single longwall panel excavation
For the prediction of the subsidence resulting from single panel extraction it is assumed that there is negligible influence from mining in the adjacent panels on the observed maximum subsidence where only one flat coal seam is extracted. The modelled subsidence troughs were compared with the field observed subsidence and the
Incremental Profile Method curves, as provided below.
4.4.3.1 Validation against the Extraction of Longwall 20
As can be seen in Figure 4.9, Longwall 20 is the first longwall developed within the
Central Area, and it is single and isolated panel that is unaffected by adjacent and previously mined Longwall 1 to 19, which can be simulated as a single longwall panel.
The UDEC model was built based on a vertical section as shown in Figure 4.12. The model was 1272 m wide and 550 m deep, with a 3 m thick coal seam sitting at a depth of 472 m below the surface. A monitoring line was placed on the surface to record the displacements of the “pegs”.
118
Figure 4.12: Layout of Longwall 20 and Monitoring Line 9C (Metropolitan Coal, 2011).
The grid system within the UDEC model is illustrated in Figure 4.13. The rock mass properties presented in Section 4.3 were used for model input. In the model the minor principal stress is assumed to be vertical and equivalent to the overburden stress. The major principal stress is in the horizontal direction and the magnitude of the horizontal stress is twice that of the vertical stress where a high horizontal stress condition is applied, as described in Section 4.3.3. The model was first cycled to allow the system to consolidate and reach equilibrium; then the panel was excavated, and the model was cycled sufficiently to reach equilibrium for the analysis.
119
Figure 4.13: UDEC model geometry for simulating extraction of Longwall 20.
0 0 100 300 500 700 900 1100 -10 -10 -20 -20
-30
-40 -30
-50 -40 Surface level (m) level Surface Subsidence (mm) Subsidence -60 -50 -70 -60 -80
-90 -70 Distance along monitoring line (m)
Field Modelled LW20 Surface level
Figure 4.14: Comparison of modelled and monitored subsidence troughs induced by extraction of Longwall 20.
120 Figure 4.14 presents results obtained from UDEC modelling compared with the previous field monitoring data. As can be seen from the figure, a good correlation is found between the modelled and observed subsidence troughs. The shape of the subsidence trough from the numerical modelling is slightly less than the observed subsidence curve on field. The maximum modelled and observed subsidence during the extraction of Longwall 20 is 74 mm and 83 mm, with a difference being 17 mm. This difference of movements are within the level of accuracy, based on the recommendation of Mining Subsidence Engineering Consultants (MSEC) (2008), which suggests that the calculations should be accurate to within 50 mm difference of subsidence. It should be noted that at the time the base survey for Line 9C was undertaken there was still 85 m of extraction remaining for the previously extracted Longwall 18 and therefore the survey results along Line 9C included a small component from the extraction of the previously mined Longwall 18, which is likely to have contributed to the difference between the modelled and observed subsidence. It should also be noted that there is a difference in the subsidence value when the monitoring line (X axis) exceeds 900 m, because the UDEC model could not simulate the real infinite far field boundary due to the limitation of computation.
4.4.3.2 Validation against the Empirical Incremental Profiles
For the prediction of typical subsidence using the Empirical Incremental Profiles it is generally accepted that the magnitude of maximum subsidence over a mined panel depends on the panel width to depth ratio and the extracted coal seam thickness. These methods only allow for the prediction of maximum values of mining induced systematic subsidence and do not precisely present the whole subsidence trough in this section.
121 The UDEC model developed for single panel extraction was based on a vertical section through an idealised flat surface (Figure 4.15). The dimension of the model was 1500 m by 550 m, with a 3 m thick coal seam sitting at a depth of 472 m below the surface. A monitoring line was also established on the surface for displacement measurement.
Taking into consideration the typical mining conditions in the Southern Coalfield, a series of panel width on depth ratios were used for the calibration study, as listed in
Table 4.7.
Figure 4.15: Generic UDEC model for single longwall extraction.
122 Table 4.8: Summary of the parameters for model setup and the subsidence results.
Model No. 1 2 3 4
Cover of Depth (m) 472 472 472 472
Panel width (m) 141.6 236.0 330.4 424.8
W/H 0.3 0.5 0.7 0.9
Predicted Smax (m) 0.06 0.33 0.97 1.54
Modelled Smax (m) 0.07 0.24 1.08 1.64
A summary of the maximum modelled and predicted incremental systematic subsidence due to the extraction of each of the single longwall with different panel width to depth ratios is also provided in Table 4.7.
Figure 4.16 represents maximum subsidence as a ratio of the extracted coal seam thickness for both the UDEC and Empirical Incremental Profiles method, and positive correlations between modelled and predicted maximum subsidence are found in the figure. It is clear that the value of modelled maximum subsidence increases as the panel width to depth ratio increases from 0.3 to 0.9 and thus follows the trend of the Empirical
Incremental Profiles. The scatter in the modelled and predicted maximum subsidence is in the order of ±0.01 m to ±0.11 m. It should be noted that the Empirical Incremental
Profiles Method is based on observed subsidence monitoring data from coalfields with various types of topography. However the maximum subsidence obtained from the numerical modelling is based on the generic idealised flat terrain.
123 0.60
0.50
0.40
0.30 Smax, inc/ T inc/Smax, 0.20
0.10
0.00 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 Panel width on depth
IPM Modelled
Figure 4.16: Modelled maximum subsidence, compared with the prediction profile developed by the Empirical Incremental Profiles method.
4.4.4 Multiple panel excavations
In practice, longwall panels are usually extracted side by side and are separated by chain pillars. When analysing subsidence over a series of longwall panels, it has been suggested that chain pillar plays an important role in the development of subsidence
(Holla and Barclay, 2000, Mills, 1998, Mills and Edwards, 1997). This section examines the influence of chain pillar on subsidence movements.
4.4.4.1 Validation against the Extraction of Longwall 11 and 12
Validation of multiple longwall panel extractions was conducted by comparing the modelled and observed subsidence profiles for Longwall 11 and 12 in Metropolitan
Colliery. The panels were extracted beneath Waratah Rivulet, a major tributary of the
Woronora Water Supply. Figure 4.17 shows the layout of Longwall 11 and 12 as well as
124 the surface level (the horizontal scale is exaggerated compared with the vertical scale in the figure). A UDEC model of these two longwall panels and the same surface topography as exists in the field was established, with the same “pegs” placed on the surface to record the mining induced vertical and horizontal displacements.
Figure 4.17: Location of Longwall 11, 12 and the ground surface level (Metropolitan Coal, 2010).
The model geometry is presented in Figure 4.18. Longwalls 11 and 12 have panel widths of 163 m and an extraction height of 3 m, and the chain pillar is 35 m wide.
Longwall 11 and 12 were excavated in a linear sequence, and the model was cycled to reach equilibrium after each stage of extraction.
125
Figure 4.18: UDEC model geometry for multiple longwall extractions.
Initially the observed incremental subsidence for Longwall 12 was around ten times larger than the modelled one. In order to find the reasons behind this, the pillar performance during the mining process was examined. It was observed that the boundaries of the pillar had yielded. The plasticity failure zone within the pillar in the
UDEC model is illustrated in Figure 4.19.
126
Figure 4.19: Chain pillar failure in the UDEC model.
In order to eliminate the potential numerical effects due to the use of DEM software, the
2D elasto-plastic finite element stress analysis program Phase2 was also employed as a comparative study. The continuous model was established using the same parameters as the UDEC model, and an overview of the model setup is shown in Figure 4.20. As can be seen in Figure 4.21, the boundary of the pillar yielded and could not carry any further stress, and is similar to the UDEC model.
127
Figure 4.20: Phase2 model setup for extraction of Longwall 11 and 12.
Figure 4.21: Chain pillar failure and stress contour from Phase2 modelling.
Overall, the UDEC and Phase2 modelling results indicate that the chain pillar has yielded during the development of subsidence. Before any yielding has occurred, there is a lower applied vertical stress in the centre of the pillar than there is on the pillar edges. The outer section of the pillar is unconfined and at the points of highest stress, therefore, yielding would be expected to occur there first. During longwall retreat, there is evidence of yielding of the outer section of the pillars. The load carried by the
128 boundary part of the pillar dramatically decreases, and it is obvious from the field stress measurements that the centre section of the chain pillar is accepting the major portion of the vertical load. Monitoring data therefore is indicative of the coal pillar retaining an intact core with the boundary part of the pillars yielded under the action of the vertical abutment loading, losing its load carrying capacity. It is also worth noting that the ratio of pillar width (w) to height (h) of 10 or above are generally regarded as indestructible pillars. Fractured coal at the pillar circumference is always observed for large width/height ratios pillars.
Subsidence components relative to chain pillars have been identified as failure of the chain pillar system and elastic compression of the chain pillar (Mills, 1998). In cases where chain pillars are overloaded, the pillar can fail thus resulting in more surface subsidence. The relationship of pillar size and the maximum subsidence has been illustrated in Figure 4.22, which was based on subsidence back analysis studies in the
Newcastle Coalfield (Mills and Edwards, 1997). As can be seen from the figure, the data set falls into two major groups, with the lower group representing intact pillar where the subsidence is low, while the upper group representing yielded pillars where subsidence is greater. The modelled scenario for Longwall 12 with the pillar width to height ratio of 11.7 and maximum subsidence of 0.83 m, falls into the pillar failed zone represented by the red dot in Figure 4.22. It should be noted that the Lake Macquarie data is significantly different to the Southern Coalfield, as the performances of the pillars are influenced by the Awaba Tuff which provides soft floor conditions. The use of Figure 4.22 would likely provide conservative or an upper-bound result, as the
Loddon Sandstone in the Southern Coalfield provides a more competent floor compared to the Awaba Tuff. However, this figure has been used as a guide only, as the failure of
129 the outer section of chain pillars had already been identified through numerical modelling (UDEC and Phase2) and field observations.
Apart from numerical modelling and empirical prediction methods, the chain pillar failure has also been confirmed by observation from Metropolitan Colliery which indicated that some of the chain pillars had softened for 30 m of its total 40 m width for
Longwalls 1 to 18 (Tarrant, 2006).
Figure 4.22: Pillar failure subsidence (Mills and Edwards, 1997).
Having identified the mining induced pillar failure, the subsidence relative to pillar elastic compression was addressed. Figure 4.23 illustrates the vertical stress distribution around a loaded chain pillar. It is clear that the vertical stress concentrates in the chain pillar and the strata above and below it, and gradually diminishes with the vertical distance. A similar vertical stress distribution pattern is found in the UDEC model, as shown in Figure 4.24.
130
Figure 4.23: Pillar compression concept for subsidence (Mills, 1998).
Figure 4.24: Contour of vertical stress around chain pillar in the UDEC model.
131 In the Southern Coalfield the subsidence mechanism over multiple panel layouts is affected primarily by the compression of pillars (or the pillar system including roof and floor strata), and pillar deformation is the significant contributor to the maximum observed subsidence (Holla and Barclay, 2000). For example, Mills and Huuskes (2004) carried out subsidence measurements at Metropolitan Colliery, and noted indications that the overall magnitude of subsidence was controlled by elastic compression of the chain pillars and the surrounding strata.
Therefore, based on the understanding of the influence of pillar failure on subsidence, two methods were thus tested to calibrate the modelled results to the field observed data. One of the methods was to simulate a softer and weaker pillar while the other method was to model a narrower pillar. It should be noted that due to the locking effects of lower order elements in the blocks, it is difficult in UDEC to accurately represent the plasticity and post-failure behaviour (Itasca, 2011). Thus for the first approach, although the pillar was assigned lower strength, the yielded pillar blocks still provided support to the roof, reducing the subsiding effects of the overburden strata. Sensitivity studies were also conducted to study the effect of pillar deformation properties, strength and width on subsidence movements. The results indicated that reducing the width of the chain pillar provided more consistent results for the incremental subsidence curves than varying strengths of coal in the chain pillar. As a consequence a pillar width reduction factor of 0.25 was applied to the pillar in the model. Of note is that numerical modelling gives approximate solution not exact solution, and there are always assumptions made to solve the problem, therefore, it seems impossible to represent the problem as exactly the same as that in the real world, as it do has certain limitations. More detailed modelling of the pillars could be undertaken, however, this would not greatly improve the outcome of the modelled surface movements, which was the intent of the model.
132 Reduction of the pillar width increased the pillar stress and pillar compression, dlea ing to more pillar compression subsidence. A comparison of the observed and modelled incremental subsidence and horizontal displacement profiles using the pillar width reduction factor of 0.25 are presented in Figure 4.25 and Figure 4.26 respectively. It is apparent that a positive correlation was found between the modelled and the field observed vertical and horizontal displacements. Moreover, a good agreement was found between the maximum modelled subsidence (790 mm) and the empirical predicted subsidence, which suggests that for the 35 m wide chain pillar and 163 m wide longwall, the magnitude of subsidence that affected by the pillar is typically of the order of 750 mm to 1400 mm (Mills, 1998).
A similar approach has been successfully applied in the calibration of the subsidence prediction method. As an example, MSEC (2008) simulated narrower (weaker) chain pillars, introducing a pillar proportion factor of 0.65, in calibrating the Incremental
Profile Method. A good correlation was found between the observed and predicted subsidence profiles.
133 100 350
0 1100 1300 1500 1700 1900 2100 2300 2500 300
-100 250
-200
-300 200 -400 150 -500 Surface level (m) level Surface -600 100 Incremental Susidence (mm) Susidence Incremental -700 50 -800
-900 0 Distance along monitoring line (m)
Field Modelled Surface level LW 11,12(right to left)
Figure 4.25: Comparison of the modelled subsidence troughs to those measured in field.
200 350
100 300
250 0 1100 1300 1500 1700 1900 2100 2300 2500 200 -100 150
-200 (m) level Surface 100 Incremental horizonal disp. (mm) disp. horizonal Incremental -300 50
-400 0 Distance along monitoring line (m)
Field Modelled Surface level LW 11,12(right to left)
Figure 4.26: Comparison of horizontal displacement predicted by UDEC to those measured in field.
134 It should be noted that as Figure 4.25 presents, the modelled incremental subsidence profile along the monitoring line is slightly narrower than for the field measured results, indicating the curvature of the subsidence profile for the numerical modelling is a little bit higher than the real situation. One possible solution could be increasing the overburden strata shearing and chain pillar compression and/or reducing the overburden bending by decreasing the elastic modulus of the chain pillar blocks. Since adding another variables would make the validation much more complex and the current results is well within the level of accuracy, this refinement was not made in the model and could be used as recommendation of future investigations regarding the chain pillar design.
4.4.4.2 Validation against the Empirical Incremental Profiles
In the Incremental Profile Method of predicting subsidence the chain pillar width to depth of cover ratio is introduced to examine the influence of pillar width on the maximum subsidence. Based on the typical mining conditions in the Southern Coalfield where the widest mining void is around 220 m and chain pillars are around 40 m in width, UDEC models have been established with various pillar widths, as listed in Table
4.8. As the only variable in the model, the pillar width increases from 19 m to 47 m.
The adjustment factor of 0.25 was applied to these four pillars for the models.
135 Table 4.9: Summary of parameters selected in the UDEC validation together with the comparison of the modelled and empirical predicted results.
Model No. 1 2 3 4
Cover of Depth (m) 472 472 472 472
Panel width (m) 163 163 163 163
Pillar width (m) 19 28 38 47
Wpi/H 0.04 0.06 0.08 0.10
Predicted Smax (m) 0.90 0.68 0.48 0.33
Modelled Smax (m) 0.89 0.72 0.50 0.42
In Figure 4.27 a comparison is presented between the modelled maximum subsidence with different pillar width/cover depth and values obtained from the Empirical
Incremental Profiles method, and it is found that the values closely matched each other.
The scatter in the modelled and predicted maximum subsidence is in the order of ±0.01 m to ±0.09 m, which is well within the level of accuracy. This indicates that the UDEC model is accurately simulating the subsidence induced by multiple panel extractions when the pillars are involved.
136 0.600
0.500
0.400
0.300
Smax, inc/ T inc/Smax, 0.200
0.100
0.000 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 Panel width on depth
Smax, Wpi/H 0.04 Smax, Wpi/H 0.06 Smax, Wpi/H 0.08 Smax, Wpi/H 0.1 Modelled
Figure 4.27: Comparison of modelled incremental maximum subsidence to the prediction profiles developed by the Empirical Incremental Profiles with different modelled width to depth ratios.
4.5 Conclusions
This chapter provides an overview of development of the numerical models and of how the model is validated for mining induced subsidence movements against field data. The
UDEC models are built based on estimations of the geomechanical characteristics of the overburden rock units, which are based on a combination of laboratory tests on standard size samples, scale effects and a large number of previous field study results. It should be noted that not all the properties used in the modelling could be obtained from the direct laboratory or field test, and previous empirical estimations and sensitivity analyses of the influence of variation in specific properties are necessary to deliver a reasonable numerical model input data selection.
137 There were four scenarios established for the comprehensive UDEC model validation.
For the single longwall panel extraction a good match was found between the modelled subsidence movements, those derived from field monitoring, and the empirical prediction method. For the multiple longwall panel extractions, where chain pillars failure was involved, a chain pillar width reduction factor of 0.25 was adopted for the modelling in order to represent the yielding at outer portions of the pillar as observed in the field. This approach generated a close match between the modelled and target results.
The results of model validation in this chapter provide confidence in the modelled subsidence and horizontal displacement being consistent with the field monitored database as well as the empirical prediction profiles. Thus the validation process has established the credibility of the calibrated numerical models in predicting the mining induced displacements. As a result a guide line has been built for the comprehensive modelling investigation that will be conducted in the following chapters both in 2D and
3D to study the impact factors contributing to the unconventional valley closure subsidence effects.
138 Chapter 5. Numerical investigation of the mechanisms contributing to valley closure subsidence
5.1 Introduction
This chapter focuses on the development of numerical models to address the fundamental understanding of mechanisms contributing to valley closure subsidence. A series of hypothetical overburden geology/surface topography scenarios are developed using the two-dimensional UDEC program, representing the broad range of typical conditions in the Southern Coalfield. Critical geological and geotechnical factors governing the valley closure subsidence behaviour have been identified in the following sections.
5.2 Development of UDEC models
This chapter uses UDEC for valley closure subsidence analysis. The modelling technique and assessments of geomechanical conditions for the UDEC models have been described in Chapter 4 and the parameters needed for model setup include model dimensions and meshing, mechanical properties of rock masses and in situ stress.
Results of the model validation analysis carried out in Chapter 4 illustrate material properties used for the modelling produced realistic mining induced subsidence and horizontal displacements.
The two-dimensional UDEC model represents a vertical section normal to the valley axis and the direction of longwall panel advance. A typical river valley in the Southern
Coalfield (Figure 5.1) with depth of 60 m was modelled, and the valley side slope is 45° for the benchmark model. Bulli seam with an extraction of 3 m was modelled, and the
139 depth of cover above the proposed longwalls was 472 m. The benchmark model ranged up to 2200 m wide and 550 m deep, placing boundaries sufficiently far from the valley and longwall panels to avoid adverse boundary effects. The model was cycled to consolidate and reach equilibrium prior to the longwall panel extraction. Variations in model setup for different geological and topographical scenarios are presented in the following sections.
Figure 5.1: Vertical section of a river valley in the Southern Coalfield (MSEC, 2012).
5.3 Valley bulging movements
Valley bulging movements were modelled in this section to provide basic understanding of valley formation and the relative failure development. An example of valley bulging is presented in Figure 5.2. As a natural phenomenon valley bulging is developed in the process of valley formation and has been identified and recognised by many engineers and geologists (Fell et al. 1992, 2000, McNally, 1981, Zaruba and Mencl, 1976).
Typically weathering and erosion processes occurring during valley development result in the stress relief in the valley sides and concentration of horizontal stress compression at the bottom of the valley. Due to the increase of horizontal stress the valley base strata fail in compression and tend to buckle upward, as illustrated in Figure 5.3.
140
Figure 5.2: Bulging effect at the base of a valley (NSW Dept. of Planning, 2008b).
Figure 5.3: Valley base failure mechanisms (Waddington and Kay, 2002).
141 The valley bulging effect was modelled using UDEC. Figure 5.4, 5.5 and 5.6 presents the displacement vectors, horizontal stress distribution and failure zones around the valley respectively. Bulging movements at the valley base and valley closure movements can be clearly observed in Figure 5.4. It can be seen from Figure 5.5 that there is a high horizontal stress concentration at the valley base and this increase in horizontal stress causes shearing and bending in the surface strata, which then leads to the strata separation and buckling. The buckling releases the horizontal confining stress in the surface strata and the stress relief transfers additional stress into the strata below.
Due to the redistribution of horizontal stress, the surface strata at the valley bottom fail in bending tension, and the strata below mainly fail in horizontal compression and low- angle shearing, as illustrated in Figure 5.6.
Figure 5.4: Displacement vectors showing valley bulging.
142
Figure 5.5: Horizontal stress contours in the valley.
Figure 5.6: Yield, tensile and shearing failure for the valley.
143 There are several factors contributing to the valley bulging movement. Numerical modelling analysis in this section indicates that the local geology of the valley base rock mass, which includes cohesion of vertical joints, discontinuity pattern and rock block dimensions, and the in situ horizontal stresses play an important role in the development of this behaviour.
It should be noted that the material properties for numerical modelling in this study are based on the geological conditions in Metropolitan Colliery, where rock bar fracturing and shearing dominate the valley base failure rather than the valley bulging. These calibrated geological and geotechnical parameters do not necessarily have to experience the pronounced bulging effects at the valley base. Therefore in this section for the modelling of valley bulging, as shown in Figures 5.4 to 5.6, weaker joint normal and shear stiffness were applied in comparison to the validated properties as described in
Chapter 4.
5.4 Effects of geological and geotechnical factors on valley closure subsidence
A series of hypothetical overburden geology/surface topography scenarios have been developed here to represent the typical conditions in the Southern Coalfield. Parametric analyses were conducted with varying geological and topographical factors affecting the extent of valley closure subsidence.
Valley closure is measured as “the reduced distance between any two points across a valley” and upsidence is seen as the “difference between the observed subsidence profile within valleys and the conventional subsidence profile that would have otherwise been expected in flat terrain” (MSEC, 2014). The challenge in determining upsidence is
144 caused by not always correctly estimating what levels of subsidence would have been expected “in flat terrain”. A smooth line is usually drawn using engineering judgment to estimate the “flat terrain” subsidence assuming the valley is not there, and this subjective estimation could result in huge deviation and inaccuracy. Moreover the measured upsidence values also depend on the placement of survey pegs which can miss the points of maximum upsidence within the cross-section (MSEC, 2014). On the other hand, valley closure is easier to determine and is measured as the maximum change in length between any two monitoring points. Thus when it comes to the interpretation of the mining-induced movements relative to valley, valley closure is a more indicative, accurate and reliable parameter in comparison to upsidence to show the behaviour of the valley. Valley closure is therefore used as the major criteria for the parametric analysis as shown below.
5.4.1 Influence of longwall location relative to valley
In this section the influence of the location of the longwall relative to the valley was simulated. The UDEC model was built based on a typical vertical section of the stratigraphy in the Southern Coalfield. The model was 2200 m wide and 550 m deep, with a 3 m thick coal seam lying at a depth of 472 m below the surface. A monitoring line was set on the surface, spaced at intervals of cover depth divided by 20, to record the horizontal and vertical displacements.
5.4.1.1 Single longwall extraction
In this scenario single and isolated longwall panel extraction was simulated with varying offset distance to the valley. The offset of the longwall from the valley is defined as the lateral distance between the valley centreline and the advancing goaf
145 edge (on the valley side), as illustrated in Figure 5.7. Figure 5.8 shows the layout of the longwall panels with different offsets to valley in the modelling. The observed valley closure and the maximum subsidence in the valley (Smax) against the longwall offset distances are presented in Figures 5.9 and 5.10 respectively.
Figure 5.7: Sketch showing the offset distance from longwall to valley.
146
Figure 5.8: Locations of single longwall panel with varying offset distance to valley, showing mining induced failure zones.
147 100 90 80
70 60 50 y = 73.044e-0.003x 40
Valley closure (mm) closure Valley 30 20 10 0 0 200 400 600 800 1000 1200 Distance between valley centreline and panel edge (m)
valley closure Expon. (valley closure)
Figure 5.9: Valley closure for single longwall extraction.
100
80
60 y = 0.0001x2 - 0.2312x + 92.639 40
20 Max subsidnece in valley (mm) (mm) valley in subsidneceMax 0 0 200 400 600 800 1000 1200
-20 Distance between valley centreline and panel edge (m) Smax Poly. (Smax)
Figure 5.10: Smax in the valley for single longwall extraction.
148 It is clear that both the valley closure and the maximum subsidence in the valley decreases as the longwall panel moves away from the valley (or increases when the longwall goes towards the valley), and mining beneath valley induces the largest valley closure movement. The effect of offset distance on valley closure appears to be that of an exponential relationship, and a polynomial relationship is identified for the Smax.
A critical zone is then identified for valley closure, where the offset is less than 310 m
(angle of draw being 33°), i.e. the longwall panel is located less than two longwall widths away from the valley. It is recognised that the most obvious changes of the valley closure movement occur within the critical zone, with the value decreasing by
80%. The magnitude of closure at specific locations outside of mining has been quantified: 54% of the maximum closure occurs at a distance of one panel width, 19% at two panel widths, 18% at three panel widths, 10% at four panel widths, where the maximum value is that when directly mined beneath. The findings are in good agreement with field observations from a range of operations, which states that when the panel is mined beneath the valley, great valley closure value is observed; as the longwall is located more than one panel width from the valley, the closure movement is halved and if the offset is more than two panel widths, the value of valley closure is decreased to less than a quarter of the high closure movement induced by mining beneath the valley (MSEC, 2014).
Measurable closure is predicted to occur beyond a 35° angle of draw which is consistent with field observations. The furthest distance that valley closure movements have been measured from longwall mining at Metropolitan Colliery is around 960 m (angle of draw being 64°), and the modelling shows that closure occurring at distances greater than 940 m (around six panel widths) was less than 5 mm (typical survey tolerance), which is negligible. 149 5.4.1.2 Multiple longwall extractions
In practice several longwall panels are mined in a series and chain pillars are left between the panels. There were a series of six longwall panels each having a range of offsets from the valley mined in linear sequence in the model (Figure 5.11). Extraction was carried out in two directions, mining towards the valley, and mining away from the valley. The model was cycled sufficiently after each extraction to reach equilibrium.
Figure 5.11: Failure zones for sequential extraction of 6 panels towards the valley, showing the longwall layouts in regards to the valley.
When mining towards the valley the modelled valley closure and maximum subsidence in the valley is presented in Figures 5.12 and 5.13 respectively. As can be seen in the figures, there is a clear trend of increasing valley closure and Smax as the longwall panel moves close to the valley, and is identical to the pattern for single longwall panel extraction.
150 In the scenario where mining is done away from the valley, the first longwall panel was extracted beneath the valley and progressed away from it. The modelled results are shown in Figures 5.14 and 5.15. It is clear that the major changes occur within the critical zone, and the numbers tend to stabilize afterwards.
600
500
400
300
200 Valley closure (mm) closure Valley 100 y = 421.03e-0.005x 0 0 200 400 600 800 1000 Distance between valley centreline and panel edge (m)
Multiple LW-Mining towards valley
Figure 5.12: Valley closure for multiple longwall mining towards valley.
151 900 800
700 600 500 400 300 200 Max subsidence in valley (mm) (mm) valley in subsidence Max
100 0 0 200 400 600 800 1000 -100 Distance between valley centreline and panel edge (m)
Multiple LW-Mining towards valley
Figure 5.13: Smax in valley for multiple longwall mining towards valley.
600
500
400 y = 3E-06x3 - 0.0044x2 + 2.3029x + 119.2
300
200 Valley closure (mm) closure Valley
100
0 0 200 400 600 800 1000 Distance between valley centreline and panel edge (m) Multiple LW-Mining away from valley
Figure 5.14: Valley closure for multiple longwall mining away from valley.
152 1400
1200 y = 5E-06x3 - 0.0093x2 + 5.5615x + 125.04 1000
800
600
400 Max subsidence in valley (mm) (mm) valley in subsidenceMax 200
0 0 200 400 600 800 1000 Distance between valley centreline and panel edge (m)
Multiple LW-Mining away from valley
Figure 5.15: Smax in valley for multiple longwall mining away from valley.
In order to validate the assumption regarding the influence of the longwall location, the modelling results were compared with field measurements. In Metropolitan Colliery,
Longwalls 8 to 12 were extracted towards Waratah Rivulet, and Longwalls 12 to 18 were mined away from the river, as illustrated in Figure 5.16. The field monitoring valley closure data is plotted against the UDEC model as shown in Figures 5.17 and
5.18. It is clear that the modelled valley closure is consistent with the field measured data. It should be noted that in the UDEC model there were no previous longwall panels extracted when mining away from valley, whereas in reality longwalls had already been mined prior to the extraction of Longwall 12. Therefore the accumulation effect for the previous extractions accounts for the difference in the valley closure when mining away from valley, as shown in Figure 5.18.
153
Figure 5.16: Longwall layouts in the Metropolitan Colliery (Metropolitan Colliery, 2010).
600
500
400
300
Valley closure (mm) closure Valley 200
100 y = 421.03e-0.005x y = 246.23e-0.004x 0 0 200 400 600 800 1000 Distance between valley centreline and panel edge (m)
Multiple LW-Mining towards valley LW8-12
Figure 5.17: Comparison of modelled (blue) and field monitored (red) valley closure when mining towards valley.
154 700
600 y = 1E-06x3 - 0.0021x2 + 1.1594x + 365.95
500
400 y = 3E-06x3 - 0.0044x2 + 2.3029x + 119.2
300
Valley closure (mm) closure Valley 200
100
0 0 200 400 600 800 1000 1200 1400 Distance between valley centreline and panel edge(m)
Multiple LW-Mining away from valley LW12-18
Figure 5.18: Comparison of modelled (blue) and field monitored (red) valley closure when mining away from valley.
Field observed valley closure movements also show a similar trend over different offset distances toward the valley (MSEC, 2014). Based on analyses of the valley closure database from a large number of sites, the position of extracted longwall panels relative to the valley has been recognised as one of the principal factors affecting the observed valley closures. The outcome of multivariate analyses of the measured valley closure movements indicates that when panels are extracted beneath or near a valley, high valley closure movements are observed. When the panel edges are located more than one longwall width away from the valley, the observed valley closure will be halved, and in cases where the longwall panel is extracted more than two panel widths away from the valley centreline, the observed valley closure is reduced to less than a quarter of the magnitude of the high valley closure observed directly over the mined panels
(MSEC, 2014).
155 The effect of mining sequence on valley closure movements has also been clearly seen from the modelling results. It is apparent that a marked increase of valley closure value is observed when mining towards the valley; when mining occurs away from the valley, the modelled closure only increases slightly for the first two longwalls mined beyond the valley, which are located in the critical zone, and tends to become stable as the longwall panels are located further away from the valley.
5.4.2 Influence of horizontal stress
Recent studies have examined possible contributing factors to valley closure movements, and recognised that the presence of in situ horizontal stresses is an important parameter affecting the overburden strata movements and valley closure subsidence. A detailed review of the in situ horizontal stresses in the Southern Coalfield can be found in Chapter 2.
This section discusses the influence of horizontal stress on the valley related movements by analysing the effect of change in the ratio between horizontal and vertical stresses.
Based on the literature review of in situ stress within coalfields in NSW, the existence of near surface horizontal stresses is always found to be greater than the vertical stresses. In the UDEC model the ratio of horizontal to vertical stress is set to 1.0, 2.0 and 3.0 for the sensitivity study and a comparative analysis is performed on the benchmark model by keeping the mining layout constant for the single longwall extraction.
Figure 5.19 describes the change of valley closure and Smax in the valley as the horizontal/vertical stress ratio increases for various panel locations. Generally the variation in horizontal stress has little significant influence on the results. However it is
156 interesting to note that for an offset distance of 110 m (within the critical zone) the valley closure steeply increases as the horizontal/vertical stress ratio changes from 1.0 to 3.0.
100 90 80
70 60 50 40 30 Valley closure (mm) closure Valley 20 10 0 0 1 2 3 4 Sigma H/ Sigma V ratio
D=1.5m D=112m D=310m D=510m D=1128.5m
100 90 80 70 60 50 40 30 20
Max subsidence in valley (mm) valley in subsidence Max 10 0 0 1 2 3 4 -10 Sigma H/ Sigma V ratio
D=1.5m D=112m D=310m D=510m D=1128.5m
Figure 5.19: Valley closure and Smax in valley for different horizontal stresses.
157 Figure 5.20 presents the contour plot of horizontal stress around the goaf and valley area for longwall panel extractions with different offset distances towards valley, where a horizontal/vertical stress ratio of 3 was used as the worst case scenario. The plot indicates there is an asymmetrical horizontal distribution on the two valley sides for the critical longwall extraction (Figure 5.20b). However for the other panels the horizontal stresses are all distributed in a symmetrical pattern on the valley. The extraction of the critical longwall panel results in redistribution of the stress, particularly the horizontal stress. The horizontal stresses on the longwall side transfer into the strata above and beneath it. Therefore a high horizontal stress concentration develops between the caved zone and the surface, and the valley acts as a discontinuity in the transfer of horizontal stresses from one side of the valley to the other.
(a)
158
(b)
(c)
Figure 5.20: Horizontal stress contour for different longwall extractions: (a) mining with the angle of draw of 33°, (b) mining of the critical longwall and (c) mining directly beneath the valley.
It is proposed that a stress arch concept explains this behaviour. It is apparent from
Figure 5.21 that the left valley wall is sitting on the shoulder of the stress arch. In the stress arch concept, horizontal stress concentrations occur within and beneath the left side valley wall, generating a strong pushing effect and thus inducing massive bedding
159 plane shearing at the base of the left valley wall. The presence of bedding plane shearing can also be used as an illustration of horizontal stress transfer. As the horizontal stress increases, larger stress concentrations are presented adjacent to the left valley side. The pushing effect induced by the horizontal stress arching becomes more pronounced, thus resulting in a sharp increase of the valley closure.
Figure 5.21: Shear displacements and horizontal stress contours in the model.
Apart from valley closure and upsidence, the failure state within the valley is also affected by the change in horizontal stress. Figure 5.22 compares the failure zones around the valley induced by the extraction of the critical panel for different horizontal/vertical stress ratios. It is apparent that higher horizontal compressive stress results in extensive shearing at the base of valley, especially concentrating on the valley side that is closest to the goaf, and more tensile failures develop as the horizontal stress increases.
160
(a)
(b)
161
(c)
Figure 5.22: Yield, tensile and shearing failure for different Sigma H/ Sigma V ratios: (a) 1, (b) 2 and (c) 3.
5.4.3 Influence of valley sloping angle
A series of analysis was done to investigate the influence of variable valley sloping angles. The angle of valley side varies from 30° to 75° with an increment of 15°. Valley closure and maximum subsidence in the valley for different angles are plotted in Figure
5.23.
162 90 80 70
60 50 40 y = -0.1847x + 53.62 30 Valley closure (mm) closure Valley 20 10 0 20 30 40 50 60 70 80
Valley sloping angle (degree)
90 80
70 60 y = -0.458x + 97.22 50 40 30 20 Max subsidence in valley in subsidenceMax 10 0 20 30 40 50 60 70 80
Valley sloping angle (degree)
Figure 5.23: Valley closure and Smax in valley for different valley sloping angles.
It can be seen that both closure and subsidence decrease as the sloping angle increases.
It is proposed that bedding planes play an important role in the change of the valley closure, and the properties of the bedding planes used in the model were illustrated in
Table 4.7. Figure 5.24 presents the shear displacement of the valley sides with different valley sloping angles. When the valley sloping angle is 30°, a major shear plane
163 develops near the valley base. As the valley sloping angle increases there is less shear displacement observed for the bedding planes close to the valley bottom. The increase in the valley sloping angle adds an extra weight on the valley wall which could result in more resistance, therefore less shearing along bedding planes and less valley closure is induced.
Figure 5.24: Shearing on the valley walls for different valley sloping angles.
In order to further investigate the effect of bedding plane shearing on valley closure, a weak bedding plane (stiffness was assumed as 1/10 of the original value) was introduced into the valley side, keeping all other parameters constant. The analysis results are shown in Figure 5.25. It is apparent from the contour plot that a large amount of shear failure is developed along the weak bedding plane, and the presence of the weak bedding plane has a pronounced influence on valley closure. It is also interesting to find that the lower part of the left valley side is drawn into the caved zone due to the
164 low friction on the weak bedding plane. It should be noted that the major issue addressed here is shearing along the weak bedding plane causes a dramatic increase of the valley closure.
Figure 5.25: Contour of the shear displacement around the weak bedding plane.
The effect of valley sloping angle on horizontal stress redistribution was also examined.
Figure 5.26 presents the horizontal stress contour after mining the longwall in each scenario. It can be seen that an increase in valley sloping angle results in a relatively minor increase of the horizontal stress concentration in the valley base, and the high horizontal stress concentration zones are moving towards the valley centreline as the valley become s steeper. In order to quantitatively analyse the horizontal stress change beneath the valley, a monitoring line was established below the valley floor to record the stress change. Figure 5.27 illustrates the magnitude of the horizontal compressive
165 stresses beneath the valley for the four valley sloping angles. Although the difference of stress magnitude in the figure is small, it aims to quantify the stress change, as a supplementary description to the stress contour as shown in Figure 5.26. It is apparent that the presence of the valley concentrates horizontal stress beneath the valley regardless of the valley sloping angle. Comparing these curves also demonstrates:
The increase of valley sloping angle results in a decrease of horizontal stress
beneath the left valley side, which means there is a stronger horizontal push
developed beneath valley wall on the goaf side when the valley is widely opened;
Increased horizontal compression is noted in the valley base when the valley side
becomes steeper, which could lead to more compressive and shearing failures at the
valley bottom. In other words, the steeper valley concentrates more horizontal stress
in the valley base.
Compared to the longwall positional factors and stresses, the valley sloping angle has less influence on valley closure. As indicated in Figure 5.23, the change of valley closure is less than 10 mm when the angle of valley sides varies from 30° to 75°. Figure
5.26 and Figure 5.27 illustrate how the stress is redistributed affected by different valley geometry, even though its influence is minor compared to other factors mentioned in the previous sections.
166 30°
45°
60°
75°
Figure 5.26: Horizontal stress contours for different valley sloping angles.
167 6.00 -50 5.50
-150 5.00 -250 4.50
4.00 -350 Surface level (m) level Surface 3.50 -450
3.00 -550
Horizontal compressive stress (MPa) stress compressive Horizontal 1000 1200 1400 1600 1800 2000 2200 Distance along monitoring line (m)
30 45 60 75 Surface level LW
Figure 5.27: Horizontal stress at a -70m monitoring line for different valley sloping angles.
5.4.4 Influence of cover depth above longwall
The depth of cover above the longwall working was discussed as a potential factor which could influence valley closure subsidence in this section. The effects of cover depth were studied by modelling various panel width/depth ratios. It should be noted that the panel width/depth ratios in the Southern Coalfield are usually limited to 0.9, and even the widest panel is still sub-critical, with a panel width/depth ratio less than 1.4
(Holla and Barclay, 2000, Whittaker and Reddish, 1989). The benchmark models in this study used a panel width/depth ratio of 0.35, and a further three scenarios were built with the ratios being 0.4, 0.5 and 0.6, thereby representing the typical mining conditions in the Southern Coalfield. The longwall panel width was fixed at 163 m, and the depth of cover above the coal seam was 472 m, 408 m, 326 m and 272 m respectively. Figure
5.28 illustrates the UDEC model setup.
168
(a)
(b)
(c)
(d)
Figure 5.28: Model grid for different cover depths: (a) 472 m, (b) 408 m, (c) 326 m and (d) 272 m, with location of the longwall marked in red.
169 The values of valley closure for these panel width/depth ratios are plotted in Figure
5.29. It is clear from the figure that the amount of valley closure increases as the panel width/depth ratio increases, indicating that mining at shallower cover depth induces more valley closure than deeper extractions in the modelled geometry.
80
70
60 y = 104.44x + 9.5712 50
40
30 Valley closure (mm) closure Valley 20
10
0 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65 Panel width to depth ratio
Figure 5.29: Relationship between valley closure and panel width/depth ratio.
Figure 5.30 compares the horizontal stresses around the valley for the different cover depth scenarios. The graph shows that the stress arching effect is more pronounced beneath the valley wall that is closest to the goaf. As the depth of cover becomes shallower, i.e. when the panel width/depth ratio is greater, redistribution of in situ stresses increases, and the magnitude of the stresses in the base of the valley also increases. In order to quantify the stress change in each scenario a monitoring line was placed at the level of valley base to record the horizontal stress magnitudes. Figure 5.31 presents the magnitudes of horizontal stresses along the monitoring line. As can be seen
170 from the figure, the horizontal stress beneath the left valley side increases by around
2 MPa as the panel width/depth ratio decreases from 0.6 to 0.35, which is positively correlated with the stress redistribution pattern illustrated in Figure 5.30. The other reason for the depth of cover influencing valley closure movements could be that the depth of cover is a mining geometry parameters affecting the value of maximum subsidence and lower width/depth ratio would result in little subsidence. Of note is that the influence of depth of cover on valley closure movements is smaller than the major factors such as longwall positions and horizontal stresses.
171
(a)
(b)
172
(c)
(d)
Figure 5.30: Mining induced horizontal stress contours for varying panel width/depth ratios: (a) 0.35, (b) 0.4, (c) 0.5 and (d) 0.6.
173 6.00
5.50 limit of valley 5.00
4.50
4.00
3.50
3.00
Horizontal compressive stress (MPa) stress compressive Horizontal 2.50
2.00 1400 1500 1600 1700 1800 1900 Distance along monitoring line (m)
0.6 0.5 0.4 0.35
Figure 5.31: Recorded horizontal compressive stresses at the valley base.
It should be noted that in the modelling of varying cover depths as described above the thickness of each stratigraphic formation within the overburden was proportionally changed together with the depth of cover. A supplementary modelling scenario was built using only one type of material in the overburden with uniform block size, so the effect of cover depth was further studied as the only variable. The influence of the variation in geological features (stratigraphy and rock mass properties) and numerical modelling setup (block size) on the valley closure subsidence were therefore eliminated in this analysis.
Figure 5.32 presents an example of the model grid system for one depth of cover value.
The result is showed in Figure 5.33 and can be compared with the numbers in Figure
5.29. This finding confirms the association between the valley closure and cover depths.
It is apparent that the magnitude of valley closure increases as the depth of cover
174 becomes shallower, regardless of the geological conditions and numerical modelling settings.
Figure 5.32: Model setup with uniform blocks in the overburden.
120
y = 97.627x + 46.147 100
80 y = 104.44x + 9.5712 60
40 Valley closure (mm) closure Valley
20
0 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65 Panel width to depth ratio
Multiple formations Uniform blocks
Figure 5.33: Comparison of valley closure for uniform block and multiple formations model.
175 The depth of cover has been examined as a potential factor influencing valley closure movements based on available field data (MSEC, 2014). Figure 5.34 shows the observed incremental valley closure plotted against the depth of cover over the longwall panel from a range of sites. It should be noted that the incremental valley closure is the additional valley closure resulting solely from the extraction of one panel. Hence it can be used to validate the valley closure movements induced by single longwall panel extraction in the UDEC models. The numerical modelling results were then added in the figure and compared with the empirical data set, and it is apparent that the models predict valley closure within the range of measured movements, which has a large scatter due to the measurements being taken from sites with varying valley heights and locations relative to mining.
Figure 5.34: Modelling results in comparison with filed observations (based on MSEC, 2014).
It should be noted that few panels have been extracted beneath the river valley since
2005, resulting in the reduction of the measured valley closure data in these sites.
Therefore the empirical database was limited by the lack of information on these
176 ongoing valley closure results. Although inclusion of the depth of cover as a potential factor influencing valley closure, according to the empirical analyses (MSEC, 2014), was not recommended, based on the modelling results valley closure does exhibit a minor increase as depth of cover becomes shallower.
5.4.5 Influence of valley shape
There are typically two types of valley shape profiles observed in the field: shallow valleys in “V” shape and deep “U” shaped valleys. The shape of the valley seems to be strongly related to its erosion pattern and the local geology. The V-shaped valley is usually formed in weak soil or rock formations such as shale and the less weathered U- shaped valley is found in the stiffer sandstone strata, and the shale valley is noticeably wider than the sandstone valley. Figure 5.35 shows the profiles of these two types of valley.
177
Figure 5.35: U-shaped and V-shaped valley (Waddington and Kay, 2001).
The effect of valley shape was examined by simulating a steep sided sandstone valley and a shallow flat shale valley. Figure 5.36 presents the geometry of the two types of valley. It should be noted that the strength of the shale valley in the model was lower in comparison to the sandstone valley.
178
(a)
(b)
Figure 5.36: (a) Sandstone valley model and (b) shale valley model.
179 The horizontal displacement and subsidence profiles for the two scenarios are presented in Figure 5.37 and 5.38. It is clear that there is no significant variation of the profiles in terms of horizontal displacement curve shape and magnitude, and no noticeable difference is observed in the subsidence profile for the two scenarios.
0.04
0.03
0.02
0.01
0.00 1,000 1,200 1,400 1,600 1,800 2,000 2,200 -0.01
-0.02 Horizontal disp. (m) disp. Horizontal
-0.03
-0.04
-0.05 Distance along monitoring line (m)
Sandstone Shale
Figure 5.37: Comparison of the horizontal displacement for the sandstone (U-shaped) and shale (V-shaped) scenario.
180 0.00 1,000 1,200 1,400 1,600 1,800 2,000 2,200 -0.01
-0.02
-0.03
-0.04
-0.05
-0.06 Subsidence (m) Subsidence -0.07
-0.08
-0.09
-0.10 Distance along monitoring line (m)
Sandstone Shale
Figure 5.38: Comparison of the subsidence for the sandstone (U-shaped) and shale (V-shaped) scenario.
Similar conclusions were drawn regarding the influence of the cross sectional shape of valley on valley closure subsidence in the literature. MSEC (2014) reviewed and compared numerous monitoring results of valley closure movements from the U-shaped and V-shaped valleys and gorges on a big diversity of mine sites, and found that there is no clear influence that the shape of the valley causes increased valley closure movements.
Following discussion with MSEC and based on reviews of valley movements since
2002, it is suggested that rather than the shale/sandstone valley shape being the issue, the accurate determination of the valley depth, which is the main aspect of valley shapes, is necessary. Currently, for the prediction of closure movements in the valleys, the equivalent valley depth is introduced to represent the valley shape using the One
Depth of Cover Method (MSEC, 2014), which is calculated by multiplying the
181 measured overall valley depth by a factor which reflects the shape of the valley and surrounding terrain within a radius equal to the full depth of cover.
5.4.6 Influence of cross bedding plane degree
Variation of cross bedding degree in the valley bed or around the valley is usually observed in field. Figure 5.39 shows an example of cross bedded sandstone units from a site.
Figure 5.39: Cross bedded sandstone units (MSEC, 2014).
The effect of cross bedding plane sloping angle on valley closure subsidence was evaluated based on two scenarios, where cross bedding plane angle was 5° and 10° respectively, as presented in Figure 5.40.
182
(a)
(b)
Figure 5.40: UDEC model with cross bedding plane angle of (a) 5° and (b)10°.
183 Comparison of the valley closure and maximum subsidence in valley for the models with degree of bedding planes 0°, 5° and 10° is presented in Figures 5.41 and 5.42. It can be seen from these figures that the magnitude of valley closure gradually increases as the degree of bedding plane increases from 0° to 10°, while the maximum subsidence in the valley slightly decreases. Figure 5.43 and 5.44 illustrate the horizontal stress distribution and the yield zones around the valley for the two scenarios respectively. It can be inferred from the figures that the increased shearing along bedding planes for 10° beddings is the reason for the increased valley closure and decreased maximum subsidence in the valley, and the change of the angle of the bedding planes negligibly affects the redistribution of horizontal stress in the valley.
To conclude, the variation of cross bedding degree around the valley has negligible effect on the valley closure subsidence movements. This finding is in agreement with the analysis results which are based on field monitoring results from more than 300 sites
(MSEC, 2014). The MSEC 2014 Report states “there is no clear influence in the available data to indicate that the degree of cross bedding causes increased valley closure movements”.
184
90
80
70
60 y = 0.8x + 44.767 50
40
30 Valley closure (mm) closure Valley 20
10
0 0 2 4 6 8 10 Angle of bedding plane (degree)
Figure 5.41: Valley closure for different degree of bedding plane.
90 80 70 y = -0.7x + 75.667 60 50 40 30 20
Max subsidence in valley (mm) valley in subsidenceMax 10 0 0 2 4 6 8 10 Angle of bedding plane (degree)
Figure 5.42: Smax in valley for different degree of bedding plane.
185 5°
10°
Figure 5.43: Horizontal stress contour around the valley for different bedding plane dipping angle.
186 5°
10°
Figure 5.44: Plastic zones and shearing within the valley for different bedding plane dipping angle.
187 5.5 Conclusions
The UDEC modelling studies conducted in this chapter improve the understanding of the mechanisms contributing to the valley closure subsidence behaviour. A series of hypothetical mining scenarios were established to investigate the influence of various geological and geotechnical factors. The UDEC models provide quantitative analyses on the effects of these factors which contribute to valley closure subsidence within the limitations of the model.
The study has shown that the location of the panel with respect to the valley has a pronounced effect on valley closure subsidence. This effect is also consistent with the field monitoring data. A critical zone has been identified for an angle of draw smaller than 33°, where the valley closure and maximum subsidence in the valley are highly sensitive. The most obvious changes of the valley closure movement occur within the critical zone, with the value decreasing by 80%. On the other hand, the valley-related movements can be ignored in cases where the offset distance between the valley and the longwall edge is greater than 940 m (around six panel widths).
The effect of horizontal stress has been studied and it is found that horizontal stress has more influence on valley closure than on subsidence. For extractions that occur within the critical zone a stress arch concept is proposed. The pronounced concentration of horizontal stress results in a pushing effect on the valley wall (on the longwall side) leading to a dramatic increase in valley closure compared to other factors. The model indicates that redistribution of horizontal stress increases as the depth of cover becomes shallower, leading to greater magnitude of valley closure movements, based on the geological conditions used in the modelling.
188 The role of bedding planes has also been identified as another important parameter that may affect valley closure. Shearing along bedding planes is a major cause of closure movement in a valley, and the mechanical properties of the beddings can greatly affect valley closure. The modelling indicates that other factors, such as variation in valley sloping angle and valley shape and degree of bedding plane around the valley have a limited capacity to induce valley closure subsidence.
189 Chapter 6. Three-dimensional numerical modelling of the potential factors contributing to valley closure subsidence
6.1 Introduction
The two-dimensional UDEC modelling analysis presented in Chapter 5 has provided improved understanding of the mechanisms of valley closure subsidence behaviour, taking into account a range of geological and geotechnical factors. It should be noted that UDEC is a two-dimensional plane strain DEM code which assumes that the deformation of the mass perpendicular to the model is zero. The UDEC approach therefore appears to be limited to representing the three-dimensional extraction layouts and the three-dimensional stress conditions when studying the valley closure subsidence effects.
In this chapter the two-dimensional modelling geometry is extended to simulate the three-dimensional mining conditions. A series of analyses were done using the Three- dimensional Distinct Element Code (3DEC) (Itasca, 2013) to investigate in particular the effects of the three-dimensional longwall positional factors and the orientation of the major horizontal stresses on valley closure subsidence behaviour, as described in the following sections.
6.2 Model setup
The 3DEC benchmark models had dimensions of 1632 m × 1000 m × 550 m (in the order of X, Y, Z). Model input parameters including model stratigraphy, valley geometry, mining geometry and mechanical properties were identical to those used in
190 the UDEC models as described in the earlier chapters. It should be noted that due to limitations of computational processing time, it was unrealistic to generate the grid system with high density for each stratigraphic formation as used in the UDEC models.
In the 3DEC model, the blocks were subdivided to increase the density of blocking over the regions of particular interest around the valley and above the panel in order to allow subsidence, valley deformation and cave propagation above the goaf to be developed sufficiently. The whole model was discretized using tetrahedral-shaped zones to become deformable. A general three-dimensional view of the model setup showing different mesh densities in varying formations is presented in Figure 6.1.
Figure 6.1: 3DEC full model mesh in different formations.
191 The stress conditions applying to the 3DEC benchmark models were the same as the
UDEC models, with the horizontal-to-vertical stress ratio being 2.0 to represent the high horizontal stress situation in the Southern Coalfield. The major horizontal stress was assigned perpendicular to the valley centreline for the benchmark models. The Mohr-
Coulomb constitutive model was used for all the 3DEC models. For boundary conditions, the roller boundary was applied on the vertical sides of the model and the bottom of the model was fixed by velocities equal to zero in the three directions.
To avoid an unstable response from the model and to realistically simulate the advance of the longwall excavation, the longwall panel was extracted in increments of 200 m, with a total uniform panel length being 1000 m, and the model was cycled with a sufficient numbers of steps to reach equilibrium after each stage of excavation. Figure
6.2 shows a typical consecutive stage run to equilibrium.
192
(a)
(b)
Figure 6.2: An example of the model equilibrium process, showing the displacement contour with (a) the unbalanced force, and (b) vertical displacements of three monitoring points.
193 6.3 Comparison of 3DEC and UDEC results
In this section, the applicability of 3DEC for modelling valley closure subsidence was evaluated. There were two modelling scenarios established for comparison with the
3DEC benchmark model which was described in Section 6.2, in terms of the horizontal movements and subsidence profiles.
The first comparative model was built using UDEC, employing the input data as described in Chapter 5. Since the typical 3DEC model used a lower-resolution grid system compared to the UDEC model, a 3DEC slice model was established with identical high-densified grids over each formation of the UDEC model to examine the influence of grid densities. Due to the limitation of the computation, the thickness of the
3DEC slice model (Y value) was limited to 100 m with the dimensions being 1632 m ×
100 m × 550 m.
For each modelling scenario the extraction of the longwall defined in Section 5.4.1 as critical was modelled, with a monitoring line placed on the surface for ground displacement measurements. It should be noted that all the geological parameters for the three models were the same, and the only difference was the model dimension and grid system. Figure 6.3 compares the model layouts for the three scenarios.
194
(a)
(b)
(c)
Figure 6.3: Model grid system for: (a) UDEC model, (b) 3DEC slice model and (c) 3DEC full model.
Figure 6.4 contains the summary of horizontal movements derived from the UDEC model, the 3DEC slice model and the 3DEC full model. Figure 6.5 compares the subsidence profiles at ground level for the three cases.
195 0.050 0.040 0.030
0.020 0.010 0.000 500 700 900 1,100 1,300 1,500 1,700 1,900 2,100 2,300 -0.010 -0.020 Horizontal disp. (m) disp. Horizontal -0.030 -0.040 -0.050 -0.060 Distance along monitoring line (m)
3DEC Slice 3DEC Full UDEC
Figure 6.4: Comparison of horizontal movements for the three scenarios at ground level.
0.020
0.000 500 700 900 1,100 1,300 1,500 1,700 1,900 2,100 2,300
-0.020
-0.040
Subsidence (m) Subsidence -0.060
-0.080
-0.100 Distance along monitoring line (m)
3DEC Slice 3DEC Full UDEC
Figure 6.5: Comparison of subsidence profiles for the three scenarios.
196 In comparing the 3DEC full model and the 3DEC slice model, the horizontal movements and subsidence profiles are close enough to be considered similar. In interpreting the curves from the three-dimensional modelling and two-dimensional modelling results, the 3DEC modelling results are in overall agreement with that of
UDEC although it appears that simulations based on 3DEC tend to have more horizontal movements in the valley area. On the other hand, the UDEC model is likely to have more subsidence compared to both the 3DEC full model and the 3DEC slice model. In summing up, the three-dimensional models provide reasonable correlations with the outcomes from the two-dimensional analyses in terms of the horizontal displacements and subsidence. It is therefore concluded that the 3DEC full model can be relied upon for the modelling of mining induced surface ground displacements.
6.4 Three-dimensional parametric analysis
Apart from the factors that have been evaluated in Chapter 5, there are other important factors which could affect the valley closure subsidence behaviour, including the three- dimensional longwall position relative to valley and the three-dimensional horizontal stresses. A range of 3DEC modelling scenarios were built in this section, considering two important factors:
Orientation of longwall relative to valley
Orientation of the major horizontal stress with respect to valley as well as
longwall advance
It should be noted that in the 3DEC modelling there are two components of horizontal movements, one in the direction of the monitoring line (along X axis) and the other perpendicular to the direction of the monitoring line (along Y axis). The monitoring
197 results show that the horizontal movement along the X axis direction is much larger than that in the Y axis direction, suggesting that the horizontal movement is predominately in the direction along the monitoring line (perpendicular to the valley) regardless of the mining direction. Therefore, the component of horizontal displacement along the monitoring line (X–displacement in the 3DEC model) is used in the following analysis.
6.4.1 Orientation of mining relative to valley
The influence of the direction of longwall with respect to the valley alignment on the valley closure subsidence was evaluated in this section. Two scenarios were modelled, where the longwall lay parallel, and where it was perpendicular to the valley centreline respectively. A schematic model layout is presented in Figure 6.6, and the position of the longwall relative to the valley is illustrated in Figure 6.7. The mining induced ground movements were monitored along a monitoring line that was established across the valley at the surface.
In these two cases, the direction of maximum principal stress S1 was normal to the valley centreline, that is, along the X axis, and the ratio of the horizontal stress to vertical stress applied in the models was 2.0.
198
Figure 6.6: Schematic view of longwall position relative to valley.
(a)
199
(b)
Figure 6.7: The position of longwall with respect to the valley: (a) mining parallel to valley and (b) mining perpendicular to valley.
6.4.1.1 Valley closure subsidence
The horizontal movement contour induced by mining parallel and perpendicular to the valley is presented in Figure 6.8. In both cases it is clearly observed that the valley wall that is closest to the longwall moves towards the valley centreline rather than the goaf area, indicating obvious closure movements of the valley walls. For comparison, identical 3DEC models were run with flat terrain to identify the effect of the presence of valley on the surface movements. The horizontal displacement contour derived from the flat terrain cases is illustrated in Figure 6.9. The presence of the valley clearly influences the surface ground movements, as can be observed by comparing Figure 6.8a with Figure 6.9a, and Figure 6.8b with Figure 6.9b.
200 Extracted panel
(a)
Extracted panel
(b)
Figure 6.8: Plot of the horizontal displacement contour for mining (a) parallel to valley and (b) perpendicular to valley.
201 Extracted panel
(a)
Extracted panel
(b)
Figure 6.9: Plot of the horizontal displacement contour in the flat terrain for mining (a) parallel to valley and (b) perpendicular to valley.
202 Horizontal displacements and subsidence profiles along the monitoring line are represented in Figures 6.10 to 6.13, and are the result of mining in the direction of valley and perpendicular to the valley. The displacements in flat terrain are also plotted in the figures to indicate the influence of the valley. For the horizontal movements the presence of valley apparently causes closure movements on the valley sides. The subsidence profile at the valley bottom is modified by a hump for both cases, which is resulted from the valley base not subsiding as much as in a flat terrain.
203 0.040
0.030
0.020
0.010
0.000 500 700 900 1,100 1,300 1,500 1,700 1,900 2,100 2,300 -0.010
-0.020
Horizontal disp. (m) disp. Horizontal -0.030
-0.040
-0.050
-0.060 Distance along monitoring line (m)
LW parallel to valley Flat terrain
Figure 6.10: Mining induced horizontal movements when the longwall is directed along the valley.
0.010
0.000 500 700 900 1,100 1,300 1,500 1,700 1,900 2,100 2,300 -0.010
-0.020
-0.030
-0.040 Subsidence (m) Subsidence -0.050
-0.060
-0.070
-0.080 Distance along monitoring line (m)
LW parallel to valley Flat terrain
Figure 6.11: Mining induced subsidence profiles when the longwall is directed along the valley.
204 0.020
0.010
0.000 500 700 900 1,100 1,300 1,500 1,700 1,900 2,100 2,300 -0.010
-0.020
Horizontal disp. (m) disp. Horizontal -0.030
-0.040
-0.050 Distance along monitoring line (m)
LW perpendicular to valley Flat terrain
Figure 6.12: Mining induced horizontal movements when the longwall is directed perpendicular to the valley.
0.010
0.000 500 700 900 1,100 1,300 1,500 1,700 1,900 2,100 2,300 -0.010
-0.020
-0.030
Subsidence (m) Subsidence -0.040
-0.050
-0.060
-0.070 Distance along monitoring line (m)
LW perpendicular to valley Flat terrain
Figure 6.13: Mining induced subsidence profiles when the longwall is directed perpendicular to the valley.
205 The magnitudes of valley closure and subsidence for the two mining orientations are listed in Table 6.1. It is apparent from the table that the direction of mining has a pronounced influence on the magnitude of valley closure and on maximum subsidence in the valley. The valley closure is much larger when mining in the direction of valley than when mining perpendicular to the valley direction, 76 mm and 48 mm respectively.
The maximum subsidence measured in the valley when mining parallel to the valley is
59 mm, which is more than twice the 26 mm measurement when mining perpendicular to valley.
Table 6.1: Valley closure and subsidence when mining parallel/perpendicular to valley.
Valley Maximum subsidence in the Case closure (mm) valley (mm)
Mining Parallel to valley 76 59
Mining Perpendicular to valley 48 26
The results of the 3DEC analyses were also compared to the findings of previous empirical work. MSEC (2014) reviewed the influence of the angle between the alignment of valley and the alignment of longwall on the observed valley closure movements, as illustrated in Figure 6.14. Based on extensive measurements of valley closure and the angle between the valley alignment and the longwall direction from more than 300 sites, it was concluded that the observed valley closure movements for those cases where the angles between the valley and longwall alignment were less than
30° was generally higher than those cases when the angles were greater than 60°. Figure
6.15 illustrates the observed and predicted valley closure for cases where the angles between the valley and longwall alignment were larger than 60° and less than 30°, and it was clear that “the mean values for the red cases are almost double the mean values of the blue cases” (MSEC, 2014). The 3DEC results regarding the relationships of the
206 orientation of mining relative to valley and the valley closure movements are consistent with the field observations.
Figure 6.14: Definition of angle between valley and longwall (MSEC, 2014).
Figure 6.15: Field observed and predicted valley closure for different alignment angles between valley and longwall (MSEC, 2014).
207 Figures 6.16 and 6.17 present the major horizontal stress around the valley in the
Hawkesbury Sandstone. As can be seen from the figures, mining parallel to valley concentrates more horizontal stress on the valley side which is close to the goaf (left valley side) than mining perpendicular to valley; the redistribution of horizontal stress is mainly perpendicular to the mining direction. Figures 6.18 and 6.19 show the shear displacements in the valley base. It is clear that there is more shearing developed beneath the left valley side when mining parallel to valley, around 1.3 times the value when mining perpendicular to valley. Furthermore, in both cases the majority of the shearing focuses beneath the valley side that is closest to the goaf.
208 Mining direction
Figure 6.16: Major horizontal stress distribution for mining parallel to valley.
Mining direction
Figure 6.17: Major horizontal stress distribution for mining perpendicular to valley.
209 Mining direction
Figure 6.18: Shearing displacement contour in the valley base for mining parallel to valley.
Mining direction
Figure 6.19: Shearing displacement contour in the valley base for mining normal to valley.
210 These results suggest that the redistribution of horizontal stress that dominates in the direction perpendicular to the mining advance could be a reason for the difference of the valley closure under different mining directions. The excavation of the underground longwall panel redirects horizontal stress into the roof and floor of the panel, causing the stress arching effect as described in Chapter 5. In cases where the orientation of the longwall is parallel to the valley, this increase of horizontal stress acts underneath the valley side which is closest to the longwall, developing a pronounced shearing plane for the valley side to move towards the valley centreline. However if the mining direction is perpendicular to the valley, the redistribution of horizontal stress mainly occurs along the valley axis, rather than across the valley. Therefore the valley wall on the goaf side seems not to be appreciably affected by the mining induced stress arching effect, and the horizontal stress is distributed in a more symmetrical pattern. The other reason may be that mining parallel to the valley creates a larger void to redistribute the stresses which are perpendicular to both the valley and longwall (i.e. the full length of longwall, rather than just the end). Apart from this, valley closure could also be affected by the conventional mining induced horizontal movements. When mining parallel to the valley the conventional horizontal movements are in the direction of valley closure, and this may add up to the valley closure values. On the other hand, in the case where the direction of mining is perpendicular to the valley, the valley closure seems not to be influenced by the conventional horizontal movements.
6.4.1.2 Stress change in the valley
From the previous discussion, it can be seen that the redistribution of horizontal stress plays an important role in the development of valley closure movements. This section
211 discusses the changing state of horizontal stress at the base of the valley as panels are extracted in different orientations.
The stress was measured in the base of the valley at the depth of 15 m as the longwall face approached and passed the monitoring point, which is presented in Figure 6.20.
The monitored horizontal stress results are illustrated in Figure 6.21 for mining along the valley.
Figure 6.20: Location of the stress monitoring point.
212
Sigma 1 4.40
4.35
4.30
4.25 Stress (MPa) Stress
4.20
4.15 -400 -200 0 200 400 600 Distance from monitoring point to longwall face (m)
Sigma 2 1.82 1.80 1.78 1.76
1.74 1.72 1.70 Stress (MPa) Stress 1.68 1.66 1.64 1.62 -400 -200 0 200 400 600 Distance from monitoring point to longwall face (m)
Figure 6.21: Monitored horizontal stress in the base of the valley when mining parallel to the valley.
213
As can be seen in Figure 6.21, in a case where the direction of mining was parallel to the valley, the major horizontal compressive stress (S1) (along X axis) at the valley bottom started to increase when the longwall face was 300 m in advance of the monitoring point. When the longwall passed under the monitoring point, the horizontal stress continued to increase in compression. When the longwall face was around 300 m past the monitoring point, the horizontal stress slightly increased, and seemed to remain relatively constant until the end of the panel. The pattern of the intermediate principal stress (S2) (along Y axis) change was generally similar to the major principal stress.
Recent studies in the measurement of the horizontal stress in valleys confirmed that the horizontal stress in the base of valley increased as mining occurred nearby. Shen et al.
(2010) conducted near-surface stress monitoring in a river valley in a study of subsidence control at West Cliff Colliery. It was found that the compressive stress in the river bed increased as mining approached the river, as shown in Figure 6.22.
Figure 6.22: Monitored horizontal stress changes in the near surface sandstone (Shen et al., 2010).
214 Walsh et al. (2014) measured stress changes when studying the valley closure movements at Sandy Creek Waterfall. The location of one of the monitoring points and the longwalls is presented in Figure 6.23. Monitoring results showed the stress change in the valley floor exhibited an increase pattern during the passage of the adjacent longwall panels (particularly Longwall 6 which was parallel to the alignment of the valley at the monitoring point), as presented in Figure 6.24.
Figure 6.23: Location of the stress cell and longwalls (Walsh et al., 2014).
Figure 6.24: Summary of horizontal compressive stress changes at the valley floor (Walsh et al., 2014).
Figure 6.25 illustrates the stress change when the longwall was extracted perpendicular to the valley. As can be seen from the figure, the major principal stress decreased as the longwall face moved towards the valley. The magnitude of S1 (along Y axis) reduced sharply when the longwall was approximately 500 m in front of the monitoring point. In contrast, the intermediate principal stress (along X axis) remained constant until the
215 longwall face was 310 m from the monitoring point then producing a steep increase in stress magnitude. The results suggest that as the longwall face approaches the valley the horizontal stresses across the valley increase, which could contribute to more closure movements.
Sigma 1 4.22
4.20
4.18
4.16
Stress (MPa) Stress 4.14
4.12
4.10 -1,000 -800 -600 -400 -200 0 Distance from monitoring point to longwall face (m)
Sigma 2 1.77 1.76 1.75
1.74 1.73 1.72
Stress (MPa) Stress 1.71 1.70 1.69 1.68 -1,000 -800 -600 -400 -200 0 Distance from monitoring point to longwall face (m)
Figure 6.25: Monitored principal horizontal stress in the base of the valley when mining perpendicular to the valley.
216 6.4.2 Orientation of the major horizontal stress relative to valley and mining
The influence of the major horizontal stress orientation with respect to valley and mining, on valley closure subsidence was studied. The direction of the major horizontal stress was assumed to be along the valley direction in the models in this section. The results were compared with the outcomes in Section 6.4.1, where the major horizontal stress was perpendicular to the valley.
Figure 6.26 illustrates the horizontal displacement distribution when the major horizontal stress is parallel to the valley. The monitored horizontal displacements are plotted in Figure 6.27 for different longwall orientations, in comparison to the profiles where the major principal stress is directed perpendicular to the valley. The summary of subsidence profiles are illustrated in Figure 6.28.
217 Extracted panel
(a)
Extracted panel
(b)
Figure 6.26: Plot of the horizontal displacement contour for mining (a) parallel to valley and (b) perpendicular to valley, when the major horizontal stress directed along the valley.
218 Mining parellal to valley 0.040 0.030 0.020
0.010 0.000 500 700 900 1,100 1,300 1,500 1,700 1,900 2,100 2,300 -0.010 -0.020
Horizontal disp. (m) disp. Horizontal -0.030 -0.040 -0.050 -0.060 Distance along monitoring line (m)
S1 perpendicular to valley S1 parallel to valley
Mining perpendicular to valley 0.020
0.010
0.000 500 700 900 1,100 1,300 1,500 1,700 1,900 2,100 2,300 -0.010
-0.020
Horizontal disp. (m) disp. Horizontal -0.030
-0.040
-0.050 Distance along monitoring line (m)
S1 perpendicular to valley S1 parallel to valley
Figure 6.27: Comparison of horizontal displacements for different S1 orientations.
219 Mining parallel to valley 0.010
0.000 500 700 900 1,100 1,300 1,500 1,700 1,900 2,100 2,300 -0.010
-0.020
-0.030
-0.040
Subsidence (m) Subsidence -0.050
-0.060
-0.070
-0.080 Distance along monitoring line (m)
S1 perpendicular to valley S1 parallel to valley
Mining perpendicular to valley 0.010
0.000 500 700 900 1,100 1,300 1,500 1,700 1,900 2,100 2,300 -0.010
-0.020
-0.030
-0.040 Subsidence (m) Subsidence
-0.050
-0.060
-0.070 Distance along monitoring line (m)
S1 perpendicularl to valley S1 parallel to valley
Figure 6.28: Comparison of subsidence profiles for different S1 orientations.
220 Comparing the results demonstrated in Figures 6.26, 6.27 and 6.28, it is clearly seen that for a single panel when the major principal stress is in the same direction as the valley, there is limited influence on the horizontal movements and subsidence. The change of major principal stress direction has a pronounced influence on valley closure but only minimal effect on the subsidence. Table 6.2 presents the values of valley closure and subsidence for the four scenarios. It can be seen that there is a significant drop (more than 50%) in the level of valley closure when S1 changes from perpendicular to valley to parallel to the valley, regardless of the mining direction. In contrast, no noticeable difference is observed in the magnitude of subsidence. In summing up, the discussion above indicates that when the major horizontal stress is directed across the valley, in the same alignment as valley closure, the valley sides will have much more lateral movements towards the valley centreline driven by the high horizontal compressive stress regardless of the mining direction relative to the valley.
Table 6.2: Valley closure and subsidence for varying S1 and mining directions.
Valley closure Maximum subsidence Maximum subsidence in valley (mm) Case (mm) (mm)
S1 S1 S1 S1 S1 S1 normal to normal to parallel to parallel to normal to parallel to valley valley valley valley valley valley
Mining Parallel 30 76 58 59 64 67 to valley
Mining normal 9 48 21 26 60 63 to valley
6.4.3 Longwall extractions with different offset distance to valley
As discussed in Chapter 5, the location of longwall relative to valley has been identified as a major impact factor contributing to valley closure subsidence effects. Single
221 longwall panel extractions were modelled using 3DEC to illustrate how the orientation of the major horizontal stress affects valley closure subsidence development for different longwall locations relative to valley.
The model setup is as described in Section 6.2. It should be noted that due to the limitation of computation for the 3DEC modelling, only four single longwall extractions were simulated with varying offset distances to the valley centreline. The single panel position and mining geometry were identical to the UDEC models as described in
Section 5.4.1.
In the first scenario, a series of models were built with varying longwall panel proximities to valley, and the major horizontal stress was assigned in the X direction, perpendicular to the valley centreline. Figure 6.29 presents the trend of valley closure and subsidence, plotted against the offset distance from the valley centreline to the panel edge on the valley side. It is apparent that high valley closure is observed when the longwall is located within the critical zone (angle of draw being not more than 33°) as identified in Section 5.4.1. Then the closure movements decrease to lower values for the increasing offset distance over the extracted panels, exhibiting realistic valley closure changes. The magnitude of subsidence decreases as the longwall panel is located further away from the valley. The results are in agreement with the UDEC modelling outcomes as presented in Section 5.4.1.
222 140
120
100
80
60 y = 0.0005x2 - 0.3837x + 118.34 Valley closure (mm) closure Valley 40
20
0 0 100 200 300 400 500 600 Distance between valley centreline and panel edge (m)
100
90
80
70
60
50
40
30 y = 0.0002x2 - 0.21x + 89.143 Max subsidence in valley (mm) (mm) valley in subsidence Max 20
10
0 0 100 200 300 400 500 600 Distance between valley centreline and panel edge (m)
Figure 6.29: Summary of the trend of valley closure and subsidence with different longwall locations, when S1 was perpendicular to valley.
223 The major horizontal stress was changed to be parallel to the valley centreline for the other scenario. Figure 6.30 illustrates the measured valley closure and subsidence in the valley in comparison with the values derived when the major horizontal stress was directed across the valley. For the longwall which is located directly beneath the valley, the valley closure is not much affected by the change of major horizontal stress direction. It can be explained that when the longwall lies directly underneath the valley, the mining induced valley closure movements are much greater than the values derived from other longwall extractions. Due to the unfavourable/critical location of the panel relative to the valley, the intermediate principal stress is high enough to cause high level valley closure movements. Moreover, the conventional mining induced horizontal movement above the panel is in the same alignment as the valley closure, which can be an additive to the closure values. Therefore, in this case the specific location of longwall is the dominant factor contributing to valley closure, rather than the orientation of the major horizontal stress. For other panel extractions, a notable difference of valley closure values is observed, indicating that when the major horizontal stress is in the same alignment as the valley, less valley closure movements are expected, and vice versa. This finding supports the discussion in Section 6.4.2.
Comparing the subsidence measured in the valley for the two scenarios, it can clearly be seen that the influence of major horizontal stress orientation relative to valley is minimal on the subsidence in the valley, regardless of the longwall location with respect to the valley.
224 140
120
100
80
60
Valley closure (mm) closure Valley 40
20
0 0 100 200 300 400 500 600 Distance between valley centreline and panel edge (m)
S1 parallel to valley S1 perpendicular to valley
120
100
80
60
40 Max subsidence in valley (mm) (mm) valley in subsidence Max 20
0 0 100 200 300 400 500 600 Distance between valley centreline and panel edge (m)
S1 parallel to valley S1 perpendicular to valley
Figure 6.30: Comparison of the trend of valley closure and subsidence with varying S1 orientations, for different longwall locations.
225 6.5 Conclusions
In this chapter, the potential factors that could influence valley closure subsidence were examined in the three-dimensional environment. The 3DEC modelling results are in positive correlation with the two-dimensional UDEC model, with respect to the surface displacements, indicating that the 3DEC model is suitable and capable of representing the three-dimensional mining conditions.
Two important factors were evaluated: (i) orientation of longwall relative to valley, and
(ii) orientation of the major horizontal stress with respect to valley as well as longwall advance. The results of 3DEC modelling studies of the orientation of longwall with respect to the valley indicate that mining parallel to valley induces more valley closure and subsidence in the valley compared to mining perpendicular to valley. It can be explained by the fact that when the orientation of mining is the same as the valley centreline, the redistribution of horizontal stress concentrates higher lateral compressive stress beneath the valley side that is closest to the longwall, leading to more shearing failures in the valley base. The other reason may be that mining parallel to the valley creates a larger void to redistribute the stresses which are perpendicular to both the valley and longwall (i.e. the full length of longwall, rather than just the end). Moreover, when mining parallel to the valley the conventional horizontal movements are in the same direction as valley closure movements, and it adds to the valley closure values.
Furthermore, the orientation of the major horizontal stress affects valley closure to a large extent. When the major horizontal stress is directed across the valley, in the same alignment as the valley closure, the valley sides will have significant lateral movements towards the valley centreline driven by the horizontal compressive stress. This is
226 regardless of the mining direction and the offset distance of longwall with respect to valley.
To conclude, trends of valley closure subsidence are exhibited during parametric analyses in the 3DEC modelling studies. The three-dimensional analysis provides insight into valley closure subsidence behaviour in a more realistic environment compared to the two-dimensional analysis, and takes into consideration the three- dimensional longwall positional factor, as well as the three-dimensional major horizontal stress conditions.
227 Chapter 7. Conclusions and recommendations
7.1 Major conclusions
This thesis is committed to improving understanding of the mechanisms which cause the observed valley closure subsidence behaviour. The objectives of this research, as mentioned in Chapter 1, have been fulfilled. The numerical modelling investigations have resulted in the development of both two-dimensional and three-dimensional models, which are capable of replicating valley closure subsidence movements. A number of factors influencing valley closure subsidence behaviour have been assessed through the numerical modelling:
Longwall positional factors including the offset distance from longwall to the
valley and the orientation of the longwall to the valley;
The magnitude and orientation of in situ horizontal stress;
Valley geometric factors including valley sloping angle and valley shape;
Number of longwall panels;
Depth of cover above the longwall, and
Geological structures around valley.
Several mechanisms that contribute to valley closure subsidence effects have been recognised in this study. The major influences on valley closure subsidence are the longwall positional factors. The location of the panel with respect to the valley is expressed as the offset distance between the valley centreline and the advancing goaf edge (on the valley side). A critical zone has been identified for an angle of draw being smaller than 33°, where the valley closure and maximum subsidence in the valley are highly sensitive. The valley-related movements can be ignored in cases where the offset
228 distance between the valley and the longwall edge is greater than 940 m (around six panel widths). When considering the orientation of the longwall with respect to the valley, mining parallel to valley induces more valley closure than mining perpendicular to valley.
The magnitude of the horizontal stress has a significant impact on valley closure movements by inducing more failure and shearing at the valley base. The way in which the redistribution of horizontal stress influences the valley closure subsidence effects has been identified. A stress arching concept is proposed. When the panel is in the critical zone, due to the unfavourable location of the longwall, intensive horizontal stress is concentrated within and beneath the valley side closest to the goaf, leading to potential for shear failures at the valley base. The three-dimensional numerical modelling indicates that the mining induced horizontal stress is predominantly redistributed perpendicular to the mining advance (across the panel). In addition, the redistribution of horizontal stress increases as the depth of cover becomes shallower leading to greater magnitude of valley closure movements, based on the geological conditions used in the modelling. When the major horizontal stress is directed across the valley, in the same alignment as the valley closure, the valley sides will have greater lateral movements towards the valley centreline. These movements are driven by the horizontal compressive stress.
The role of bedding planes has also been identified as another important parameter affecting valley closure movement. Shearing along the bedding plane is a major cause of closure movement in a valley, and the mechanical property of the beddings can greatly affect valley closure. The extent of valley closure subsidence is also influenced by valley geometry factors such as valley sloping angle and valley shape, as well as the
229 geological structures around valley area in the near surface; however, their influence is minimal compared to the factors mentioned above.
The conventional mining induced movements also have the potential to contribute to the valley closure movements. In cases where the conventional mining induced horizontal movements are in the same orientation as the valley closure, the conventional horizontal movements may add to the valley closure values.
7.2 Contributions of this study
The findings from this study make several noteworthy contributions to the research methodology of valley closure subsidence and to improving the understanding of this behaviour when mining in close proximity to valleys and other forms of irregular topographies.
A systematic validation process
The numerical modelling of mine subsidence is usually criticized because the results are sensitive to the model input properties and the lack of a systematic and state-of-the-art validation and calibration process. In this study the numerical models are capable of replicating the essential features of mining induced subsidence and horizontal movements, not only in flat terrain but also with valley surface features. Four scenarios are established for model validation against field measurements and empirical prediction curves that are built on an extensive database of measurements and observations. It should be noted that previous calibration of the numerical models in the literature is generally limited to surface subsidence only. Both subsidence and horizontal displacement are validated in this study. The models are tested initially using
230 fundamental methods, following generally established trends such as increasing maximum subsidence on seam thickness with increasing panel width on depth ratios using the Incremental Profile Method (Waddington and Kay, 1998). All the results from the four scenarios exhibit positive correlations with the empirical data. In addition the outcomes of the modelling parametric investigations are analysed against the calibration data from field measurements in order to gauge the reliability of the predictive models.
A quantitative approach
The numerical modelling developed in this study provides a quantitative approach to looking in detail at the valley closure subsidence behaviour. A number of recognised geological and geotechnical factors influencing the magnitude of valley closure movements have been isolated and assessed individually. The trend of the change of valley closure subsidence values is plotted against those factors. Consequently, the significance of each of these identified potential factors can be examined. With the calibrated values of valley closure movements, the parametric models are able to obtain quantitative predictions of valley closure subsidence.
A numerical predictive tool
The numerical analysis in this thesis has also resulted in the development of a useful tool for predicting valley closure subsidence. Compared with the empirical prediction methods, the numerical modelling has the potential to replicate the observed valley closure subsidence, investigate the failure development and provide the mechanics of the processes in the models.
The numerical predictive technique of valley closure subsidence in this research may be a useful source of reference for investigating valley related movements as well as
231 underground mine planning. For example, the longwall positional factors would be considered in the planning of future longwall. It should be noted however that the numerical models developed in this thesis are based mainly on the data from the
Southern Coalfield, where the depth of cover is generally around 350 m to 500 m and the horizontal stress is high. Thus for use in other cases these models need to be calibrated based on local data.
7.3 Recommendations for future research
Although this study has successfully delivered an improvement in the understanding of the mechanisms which cause observed valley closure subsidence, it has certain limitations. The numerical models are subject to the limited computation in this study, particularly the three-dimensional models, which had to be simplified. This meant that some features of rock mass such as the spacing of joint sets and rock block size could not be represented as accurate or detailed as that in the real situation. Moreover, the material properties used in the models are generally gathered from the Southern
Coalfield, and it seems that the numerical predictive tools may not be applicable to all other cases. It also should be noted that information regarding the properties of discontinuities within rock is always sparse. Although these model input parameters are reasonably and realistically estimated, the approximations of the rock mass material properties could still be a limitation.
The recommendations and ideas that have arisen during this study are also highlighted in this section. Future improvements for the understanding of valley closure subsidence behaviour could be made by considering the following aspects:
232 Time dependency. It is recommended that the role of time should be evaluated
in the development of valley closure subsidence using numerical modelling,
based on the available data of long-term subsidence and valley closure
movement. The assessments would include: time-dependent failure of rock mass
within the strata around the panels and within the valley; time-dependent chain
pillar stability since the chain pillar failure contributes to ground movements;
and residual subsidence effects on irregular topographic features.
Fracture development. It would also be interesting to investigate failure
mechanisms and fracture initiation and propagation from the goaf to the surface,
with particular interest in the fracture development around the valley. This is
because the key hydrogeological issue in the shallow aquifers and near surface
environment in the Southern Coalfield is the loss of stream flow to shallow
fractured sandstone formation, and mining-induced surface cracking on upland
swamps. Simulation of propagation of cracks and movement of water
particularly focusing on the valley base region could enhance the understanding
of mining induced valley failure patterns.
Pore pressure in the near surface strata. Future research might investigate the
impacts of pore pressure on the valley closure subsidence. It would be beneficial
to model the surface and groundwater; the hydrogeological changes in the river
valley, the losses and reappearance of water down-gradient, and the associated
response of the rock mass to underground mining. The findings would illustrate
how pore pressure would influence the stress redistribution within the
overburden, and then affect the failure mechanisms within the valley.
233 References
Ahola, M. P. 1990. Geomechanical evaluation of escarpments subjected to mining induced subsidence Rock Mechanics Contributions and Challenges, in Proceedings 31st US Symposium (eds: Hustrulid, W. A. and Johnson, G. A.), 129-136, Rotterdam: A. A. Balkema.
Alejano, L. R., Ramirez-oyanguren, P. and Taboada, J. 1999. FDM predictive methodology for subsidence due to flat and inclined coal seam mining, International Journal of Rock Mechanics and Mining Sciences, 36, 475-491.
Alzo’ubi, A. K. 2009. The effect of tensile strength on the stability of rock slopes, PhD thesis, Department of Civil and Environmental Engineering, University of Alberta, Edmonton.
Alzo’ubi, A. K., Martin, C. D. and Mughieda, O. S. 2011. Numerical modelling of buckling rock movement, in Proceedings Continuum and Distinct Element Numerical Modelling in Geo-Engineering (eds: Sainsbury, Hart, Detournay and Nelson), 205-212.
Bandis, S., Lumsden, A. C. and Barton, N. R. 1983. Fundamentals of rock joint deformation, International Journal of Rock Mechanics and Mining Sciences & Geomechanics Abstracts, 20, 249-268.
Bell, F. G. and Lindsay, P. 1999. The petrographic and geomechanical properties of some sandstones from the Newspaper Member of the Natal Group near Durban, South Africa, Engineering Geology, 53, 57-81.
Belytschko, T., Moës, N., Usui, S. and Parimi, C. 2001. Arbitrary discontinuities in finite elements, International Journal for Numerical Methods in Engineering, 50, 993-1013.
Bertuzzi, R. and Pells, P. J. N. 2002. Geotechnical parameters of Sydney sandstone and shale, Australian Geomechanics: Journal and News of the Australian Geomechanics Society, 37, 5, 41-54.
Bobet, A., Fakhimi, A., Johnson, S., Morris, J., Tonon, F. and Yeung, M. R. 2009. Numerical models in discontinuous media: review of advances for rock mechanics applications, Journal of Geotechnical and Geoenvironmental Engineering, 135, 1547-1561.
Bowman, H. N. 1974. Geology of the Wollongong, Kiama and Robertson, 1:50,000, Sydney Dept. of Mines, Geological Survey of New South Wales.
Brady, B. H. G. and Brown, E. T. 2004. Rock mechanics for underground mining. Dordrecht: Kluwer Academic Publishers.
Branagan, D. F. 1985. An overview of the geology of the Sydney region. In: Pells, P. J. N., ed. Engineering Geology of the Sydney Region. Rotterdam: A. A. Balkema on behalf of the Australian Geomechanics, 3-46.
234 Brown, E. T. and Hoek, E. 1978. Trends in relationships between measured in-situ stresses and depth, International Journal of Rock Mechanics and Mining Sciences & Geomechanics Abstracts, 15, 211-215.
Brown, E. T. and Windsor, C. R. 1990. Near surface in situ stresses in Australia and their influence on underground construction, in Proceedings 7th Australian Tunnelling Conference: "The underground domain", ACT: Institution of Engineers, Australia.
Bunny, M. R., 1972. Geology and coal resources of the Southern Catchment Coal Reserve, Southern Sydney basin, New South Wales, Geological Survey of New South Wales – Bulletin 22, 146.
Capasso, G. and Mantica, S. 2006. Numerical simulation of compaction and subsidence using ABAQUS, in Proceedings ABAQUS Users’ Conference, Boston, USA.
Cho, N., Martin, C. D. and Sego, D. C. 2007. A clumped particle model for rock, International Journal of Rock Mechanics and Mining Sciences, 44, 997-1010.
Chugh, Y. P., Yang, G., and Mehta, K. 1994. Subsidence modelling of large proposed mining areas and finite element analysis of subsidence impacts on gorge area, Australian Water Technologies, BHP Collieries and the Natural Gas Company Report.
Clough, R. W. 1960. The finite element method in plane stress analysis, in Proceedings 2nd ASCE Conference on Electronic Computation, Pittsburg, Pa.
Coulson, J. H. 1972. Shear strength of flat surfaces in rock, in Proceedings Stabilithy of Rock Slopes, 13, 77-105, American Society of Civil Engineering.
CSIRO, 2006. Numerical modelling leads the way, Earthmatters, 8-9.
Cundall, P. A. 1971. A computer model for simulating progressive large scale movements in blocky rock systems, in Proceedings Int. Symposium on Rock Fracture, 1, 129-136, International Society for Rock Mechanics, Nancy.
Damjanac, B., Board, M., Lin, M., Kicker, D. and Leem, J. 2007. Mechanical degradation of emplacement drifts at Yucca Mountain – A modelling case study: Part II: Lithophysal rock, International Journal of Rock Mechanics and Mining Sciences, 44, 368-399.
Damjanac, B., Martin, D., Diederichs, M., McCreath, D. and Lam, T. 2012. Long-term stability for a proposed nuclear waste deep geological repository: Bruce Nuclear Site, Ontario, Canada, in Proceedings Rock Engineering and Technology for Sustainable Underground Construction–EUROCK, Rock Engineering Research Foundation and Swedish National Group of ISRM.
Darlington, W. J., Ranjith, P. G. and Choi, S. K. 2011. The effect of specimen size on strength and other properties in laboratory testing of rock and rock-like cementitious brittle materials, Rock Mechanics and Rock Engineering, 44, 513- 529.
235 DeBono, P. and Tarrant, G. 2011. An analysis of long term subsidence at Metropolitan Colliery, in Proceedings 8th Triennial Conference on Management of Subsidence: State of the Art, 81-88, Mine Subsidence Technological Society, Newcastle.
Deere, D. U. and Miller, R. P. 1966. Engineering Classification and Index Properties of Intact Rock, Technical Report No. AFWL-TR-65-116, Air Force Weapons Laboratory, Kirkland Air Force Base, New Mexico.
DGS 2011. Subsidence prediction and impact assessment for the proposed Longwalls 415 to 417 at Springvale Colliery, Wallerawang, DGS Report No. SPV-003/1.
Drucker, D. C. and Prager, W. 1952. Soil mechanics and plastic analysis of limit design, Quarterly of Applied Mathematics, 10.
Duncan, J. M. and Goodman, R. E. 1968. Finite element analyses of slopes in jointed rock, Report S-68-3, Final Report to U.S. Army Corps of Engineers, Vicksburg, Mississippi.
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, University of Exeter, Exeter.
Elmo, D., Schlotfeldt, P., Beddoes, R. and Roberts, D. 2011. Numerical simulations of scale effects under varying loading conditions for naturally fractured rock masses and implications for rock mass strength characterization and the design of overhanging rock slopes, in Proceedings 45th US Rock Mechanics/Geomechanics Symposium, American Rock Mechanics Association.
Enever, J. R., Walton, R. J. and Windsor, C. R. 1990. Stress regime in the Sydney basin and its implications for excavation design and construction, in Proceedings 7th Australian Tunnelling Conference, 49-59.
Fell, R. and Robin, F. 1992. Geotechnical engineering of embankment dams. Rotterdam: A. A. Balkema.
Fell, R., Macgregor, P. and Stapledon, D. 1992. Weathering Processes and Profiles in Valleys. Rotterdam: A. A. Balkema.
Fell, R., Hungr, O., Leroueil, S. and Riemer, W. 2000. Geomechanical engineering of the stability of natural slopes, and cuts and fills in soil, in Proceedings International Conference on Geotechnical & Geological Engineering, International Society for Rock Mechanics.
Ferguson, H. F. and Hamel, J. V. 1981. Valley stress relief in flat-lying sedimentary rocks, in Proceedings Weak rock: soft, fractured and weathered rock, 2, 1235- 1240, Rotterdam: A. A. Balkema.
236 Forster, I. R. 1995. Impact of underground mining on the hydrogeological regime, Central Coast NSW, in Proceedings of the Conference on Engineering Geology of the Newcastle-Gosford Region (eds: Sloan, S. and Allman, M. A.), 156-168, Australian Geomechanics Society.
Franks, C. A. M. and Geddes, J. D. 1986. Subsidence on steep slopes due to longwall mining, Geotechnical and Geological Engineering, 4, 291-301.
Gale, W. J. 1986. The application of stress measurements to the optimization coal mine roadway drive age in the Illawarra Coal Measures, in Proceedings International Society for Rock Mechanics International Symposium, Stockholm, Sweden.
Gale, W. 2004. Rock fracture, caving and interaction of face supports under different geological environments. Experience from Australian coal mine, in Proceedings 23rd International Conference on Ground Control in Mining, 11-19.
Gale, W. 2006. Water inflow issues above longwall panels, in Proceedings Coal Operators' Conference, 175-179, University of Wollongong & the Australasian Institute of Mining and Metallurgy.
Gale, W. 2008. Aquifer inflow prediction above longwall panels , Final Report C13013, Australian Coal Association Research Program.
Gale, W. and Sheppard, I. 2011. Investigation into abnormal increased subsidence above longwall panels at Tahmoor Colliery NSW, in Proceedings 8th Triennial Conference Proceedings on Management of Subsidence: State of the Art, 63-80, Mine Subsidence Technological Society, Newcastle.
Geosensing Solutions, 2008. A compilation of surface geological features in the western portion of Metropolitan Colliery.
Ghobadi, M. H., 1994. Engineering geologic factors influencing the stability of slopes in the Northern Illawarra region, PhD thesis, University of Wollongong, Wollongong.
Guo, H., Chen, S. G., Craig, M. S. and Adhikary, D. P. 2004. Coal mine subsidence simulation using COSFLOW, Journal of the Australasian Institute of Mining and Metallurgy, 47-52.
Gupta, A. S. and Seshagiri Rao, K. 2000. Weathering effects on the strength and deformational behaviour of crystalline rocks under uniaxial compression state, Engineering Geology, 56, 257-274.
Gray, N. M. 1982. Direction of stress, southern Sydney Basin, Journal of the Geological Society of Australia, 29, 277-284.
Hamel, J. V. and Ferguson, H. F. 1983. J. V. Hamel & H. F. Ferguson write, Quarterly Journal of Engineering Geology and Hydrogeology, 16, 3, 244-245.
Hast, N. 1958. The measurement of rock pressure in mines, Sveriges Geologiska Undersoekning, Serie C: Avhandlingar och Uppsatser, 3, 560, 183.
237 Hawkes, I. and Mellor, M. 1970. Uniaxial testing in rock mechanics laboratories, Engineering Geology, 4, 177–285.
Hawkins, A. B. and McConnell, B. J. 1992. Sensitivity of sandstone strength and deformability to changes in moisture content, Quarterly Journal of Engineering Geology, 25, 115-130.
Hebblewhite, B. K., Waddington, A. and Wood, J. 2000. Regional horizontal surface displacements due to mining beneath severe surface topography, in Proceedings International Conference on Ground Control in Mining, 19, 149-157.
Hebblewhite, B. K. 2009. Outcomes of the independent inquiry into impacts of underground coal mining on natural features in the Southern Coalfield – an overview, in Proceedings Coal Operators' Conference (ed: Aziz, N.), 113-124, University of Wollongong & the Australasian Institute of Mining and Metallurgy.
Hibbeler, R. C. 2007. Engineering mechanics: Dynamics. Singapore: Pearson Prentice Hall.
Hillis, R. R., Meyer, J. J. and Reynolds, S. D. 1998. The Australian stress map, Exploration Geophysics, 29, 420-427.
Hillis, R. R., Enever, J. R. and Reynolds, S. D. 1999. In situ stress field of eastern Australia, Australian Journal of Earth Sciences, 46, 813-825.
Hoek, E. and Brown, E. T. 1980. Underground excavations in rock. London: The Institution of Mining and Metallurgy.
Holla, L. 1997. Ground movement due to longwall mining in high relief areas in New South Wales, Australia, International Journal of Rock Mechanics and Mining Sciences, 34, 775-787.
Holla, L. and Barclay, E. 2000. Mine subsidence in the southern coalfield, NSW, Australia. New South Wales Dept. of Mineral Resources.
Hutton, A. C. and Bamberry, W. J. 1999. Stratigraphy and terminology for the Southern Coalfield, Bulletin Coalfield Geology Council of New South Wales, 1, 120-152.
Hutton, A. C. 2009. Geological setting of Australasian coal deposits, in Proceedings Australasian Coal Mining Practice (eds: Kininmonth, R. and Baafi, E.), 40-84, the Australasian Institute of Mining and Metallurgy.
ISRM 1978. Suggested methods for determining tensile strength of rock materials, International Journal of Rock Mechanics and Mining Sciences & Geomechanics Abstracts, 15, 3, 99–103.
Itasca Consulting Group, Inc. 2011. UDEC – Universal Distinct Element Code, Version 5.0, Theory and Background, Minneapolis: Itasca.
Itasca Consulting Group, Inc. 2013. 3DEC – 3 Dimensional Distinct Element Code, Version 5.0. Minneapolis: Itasca.
238 Jamison, D. B. and Cook, N. G. W. 1980. Note on measured values for the state of stress in the Earth's crust, Journal of Geophysical Research: Solid Earth, 85, 1833-1838.
Jing, L. and Hudson, J. A. 2002. Numerical methods in rock mechanics, International Journal of Rock Mechanics and Mining Sciences, 39, 409-427.
Jones, R. E., Pariseau, W. G., Payne, V. and Takenaka, G. 1990. Sandstone escarpment stability in the vicinity of longwall mining operations, in Proceedings 31st U.S. Symposium on Rock Mechanics Contributions and Challenges (eds: Hustrulid, W. A. and Johnson, G. A.), 555–562, Colorado School of Mines, Rotterdam: A. A. Balkema.
Katsaga, T. and Potyondy, D. O. 2012. A generic stope model for investigation of fracturing mechanisms in deep gold mines, in Proceedings 46th U.S. Rock Mechanics Symp, Chicago.
Kay, D. R. 1991. Effects of subsidence on steep topography and cliff lines. Canberra: Energy Research and Development Corporation.
Kay, D. R., McNabb, K. E. and Carter, J. P. 1991. Numerical modelling of mine subsidence at Angus Place Colliery, Computer Methods and Advances in Geomechanics, 999-1004.
Kay, D. R., Barbato, J. P. and Mills, K. W. 2007. Review of mechanisms resulting in observed upsidence and closure movements, in Proceedings 7th Triennial Conference on Mine Subsidence: A Community Issue, Mine Subsidence Technological Society, Newcastle.
Kazerani, T. and Zhao, J. 2010. Micromechanical parameters in bonded particle method for modelling of brittle material failure, International Journal for Numerical and Analytical Methods in Geomechanics, 34, 1877-1895.
Keilich, W. 2009. Numerical modeling of mining subsidence, upsidence and valley closure using UDEC, PhD thesis, University of Wollongong, Wollongong.
Kratzsch, H. and Helmut, K. 1983. Mining subsidence engineering. Berlin: Springer- Verlag.
Lan, H., Martin, C. D. and Hu, B. 2010. Effect of heterogeneity of brittle rock on micromechanical extensile behaviour during compression loading, J. Geophys. Res., 115.
Lee, J. W. 2005. Development of a numerical model representative of rock failure mechanisms associated with the impact of lateral displacement, ME thesis, University of Wollongong, Wollongong.
Li, S., Qian, D., Kam Liu, W. and Belytschko, T. 2001. A meshfree contact-detection algorithm, Computer Methods in Applied Mechanics and Engineering, 190, 3271-3292.
239 Li, H., Li, K., Subhash, G., Kecskes, L. J. and Dowding, R. J. 2006. Micromechanical modelling of tungsten-based bulk metallic glass matrix composites, Materials Science and Engineering: A, 429, 115-123.
Mark, C. and Gadde, M. 2010. Global trends in coal mine horizontal stress measurements, in Proceedings 10th Underground Coal Operators' Conference (ed: Aziz, N.), 21-39, University of Wollongong & the Australasian Institute of Mining and Metallurgy.
Martin, C. D., Lu, Y. and Lan, H. 2012. Scale effects in a synthetic rock mass, in Proceedings 12th ISRM International Congress on Rock Mechanics – Harmonising Rock Engineering and the Environment, 473-478, Taylor and Francis Inc..
Mas Ivars, D. 2010. Bonded particle model for jointed rock mass, PhD thesis, KTH, Stockholm.
McNally, G. H. 1981. Valley bulging at Mangrove Creek Dam, Geological Survey of New South Wales, 42.
McNally, G. H., Willey, P. L. and Creech, M. 1996. Geological factors influencing longwall-induced subsidence, in Proceedings Geology in longwall mining (eds: McNally, G. H. and Ward, C. R.), 257-267, Coalfield Geology Council of New South Wales, St. Leonards, N.S.W., Australia.
Metropolitan Coal 2010. Observed total subsidence tilt and strain profiles along Monitoring Line D.
Metropolitan Coal 2011. 2011 annual review, Project No. MET-08-08/8.1 .
Mills, K. W. and Edwards, J. L. 1997. Review of pillar stability in claystone floor strata, in Proceedings Safety in mines; the role of geology (eds: Doyle, R., Moloney, J., Rogis, J. and Sheldon, M.), 161-168, Coalfield Geology Council of New South Wales, St. Leonards, N.S.W., Australia.
Mills, K. W. 1998. Subsidence mechanisms about longwall panels, in Proceedings International conference on Geomechanics/ground control in mining and underground construction (eds: Aziz, N. I. and Indraratna, B.), 745-756, University of Wollongong, Wollongong, N.S.W., Australia.
Mills, K. W. 2001. Observations of horizontal subsidence movement at Baal Bone Colliery, in Proceedings 5th Triennial Conference on Coal Mine Subsidence: Current Practice and Issues, 99-111, Mine Subsidence Technological Society, Newcastle.
Mills, K. W. and Huuskes, W. 2004. The effects of mining subsidence on rockbars in the Waratah Rivulet at Metropolitan Colliery, in Proceedings 6th Triennial Conference on Subsidence Management Issues, Mine Subsidence Technological Society, Maitland.
240 Mills, K. W. 2007. Subsidence impacts on river channels and opportunities for control, in Proceedings 7th Triennial Conference on Mine Subsidence: A Community Issue, 207-217, Mine Subsidence Technological Society, Newcastle.
Mills, K. W. 2011a. Effects of surface topography on mining subsidence damage to river channels, ACARP Research Project C15025.
Mills, K. W. 2011b. Developments in understanding subsidence with improved monitoring, in Proceedings 8th Triennial Conference on Management of Subsidence: State of the Art, 25-43, Mine Subsidence Technological Society, Newcastle.
Milner, J. and Seedsman, R. 2000. Geotechnical support for the turnaround at Bellambi West, in Proceedings 3rd Annual Longwall Mining Summit, Aust. J. Mining, Yeppoon, Qld, Australia.
Moffitt, R. S., 2000. A compilation of the geology of the Southern Coalfield: notes to accompany the 1:100 000 Southern Coalfield geological map, Dept. of Mineral Resources.
Molinda, G. M., Heasley, K. A., Oyler, D. C. and Jones, J. R. 1992. Effects of horizontal stress related to stream valleys on the stability of coal mine openings, U.S. Dept. of the Interior, Bureau of Mines, Washington, D.C..
MSEC 2007a. Introduction to longwall mining and subsidence, Revision A.
MSEC 2007b. General discussion on systematic and non systematic mine subsidence ground movements, Revision A.
MSEC 2008. The prediction of subsidence parameters and the assessment of mine subsidence impacts on natural features and surface infrastructure resulting from the proposed extraction of Longwalls 20 to 44 at Metropolitan Colliery in support of a Part 3a Application, Report No. MSEC285.
MSEC 2012. Subsidence predictions and impact assessments for natural features and surface infrastructure in support of the extraction plan, Revision B.
MSEC 2013. Subsidence predictions for Carmichael Project in support of EIS, Revision A.
MSEC 2014. Effects of geology on observed valley closure movements with updated valley closure prediction method, ACARP Project No. C18015.
MTS Systems Corporation, 2004. MTS rock and concrete mechanics testing systems, Technical description.
Myrvang, A. 1976. Practical use of rock stress measurements in Norway, in Proceedings Investigation of Stress in Rock: Advances in Stress Measurement, 92-99. ACT: Institution of Engineers, Australia.
241 Natau, O. 1990. Scale effects in the determination of the deformability and strength of rock masses, in Proceedings 1st International Workshop on Scale effects in Rock Masses (ed: Pinto Da Cunha, A.), A. A. Balkema, Rotterdam, Netherlands.
Nemcik, J. 2003. Floor failure mechanisms at underground longwall face, PhD thesis, University of Wollongong, Wollongong.
Nemcik, J., Gale, W. and Mills, K. 2005. Statistical analysis of underground stress measurements in Australian coal mines, in Proceedings Bowen Basin Symposium – The future for coal, fuel for thought (ed: Beeston, J. W.), 117-122, Geological Society of Australia, Coal Geology Group.
Nemcik, J., Gale, W. and Fabjanczyk, M. 2006. Methods of interpreting ground stress based on underground stress measurements and numerical modelling, in Proceedings 7th Underground Coal Operators' Conference (ed: Aziz, N.), 104- 112, University of Wollongong & the Australasian Institute of Mining and Metallurgy.
Nemcik, J., Aziz, N. and Ren, T. 2012. Large excavations and their effect on displacement of land boundaries, in Proceedings 12th Underground Coal Operators' Conference (ed: Aziz, N.), 8-14, University of Wollongong & the Australasian Institute of Mining and Metallurgy.
NSW Department of Mineral Resources 2003. Guideline for applications for subsidence management approvals, Revision 2.
NSW Department of Primary Industries, 2007. Map of NSW coalfields. Available through:
NSW Department of Primary Industries, 2008. Sydney Basin Stratigraphic Table. Available through:
NSW Department of Planning 2008a. Southern Coalfield – Colliery Holdings, Mine Workings and Upland Swamps.
NSW Department of Planning 2008b. Impacts of Underground Coal Mining on Natural Features in the Southern Coalfield – Strategic Review.
Nygårds, M. and Gudmundson, P. 2002. Micromechanical modelling of ferritic/pearlitic steels, Materials Science and Engineering: A, 325, 435-443.
O'Connor, K. M. and Dowding, C. H. 1992. Distinct element modelling and analysis of mining-induced subsidence, Rock Mechanics and Rock Engineering, 25, 1-24.
Ozkan, I., Ozarslan, A., Genis, M. and Ozsen, H. 2009. Assessment of scale effects on uniaxial compressive strength in rock salt, Environmental & Engineering Geoscience, 15, 91-100.
Packham, G. H., 1969. The geology of New South Wales. Geological Society of Australia.
242 Pells, P. J. N. 1985. Engineering geology of the Sydney region, Rotterdam: A.A. Balkema on behalf of the Australian Geomechanics Society.
Pells, P. J. N. 1993, The 1993 E.H Davies Memorial lecture, rock mechanics and engineering geology in the design of underground works, Australian Geomechanics Society, 3.1-3.33.
Pells, P. J. N. 2002. Developments in the design of tunnels and caverns in the Triassic rocks of the Sydney region, International Journal of Rock Mechanics and Mining Sciences, 39, 569-587.
Pells, P. J. N. 2011. A simple method of estimating far field movements associated with longwall mining, Australian Geomechanics Journal, 46, 1-8.
Peng, S. S. 1992. Surface subsidence engineering. Littleton, Colo.: Society for Mining, Metallurgy and Exploration.
Pierce, M., Mas Ivars, D., Cundall, P. and Potyondy, D. 2007. A synthetic rock mass model for jointed rock, in Proceedings 1st Canada-U.S. Rock Mechanics Symposium Rock Mechanics: Meeting Society's Challenges and Demands (eds: Eberhardt, E., Stead, D. and Morrison, T.), 342-349, Taylor and Francis Group, Vancouver.
Potyondy, D. O. and Cundall, P. A. 2004. A bonded-particle model for rock, International Journal of Rock Mechanics and Mining Sciences, 41, 1329-1364.
Poulsen, B. A. and Adhikary, D. P. 2013. A numerical study of the scale effect in coal strength, International Journal of Rock Mechanics and Mining Sciences, 63, 62- 71.
Prakoso, W. A. and Kulhawy, F. H. 2011. Effects of testing conditions on intact rock strength and variability, Geotechnical and Geological Engineering, 29, 101-111.
Pratt, H. R. and Black, A. D. 1972. The effect of specimen size on the mechanical properties of unjointed diorite, International Journal of Rock Mechanics and Mining Sciences & Geomechanics Abstracts, 9, 513-29.
Preston, A. 1992. Cataract Dam tectonic survey movements of Stations K03 and Y possible explanation of behaviour, unpublished report, Cataract Mining Investigation, Development Div., Geological Section.
Reid, P. 1998. Horizontal movements around Cataract Dam, Southern Coalfield, in Proceedings 4th Triennial Conference on Building and Structures subject to Ground Movement, 157-170, Mine Subsidence Technological Society, Newcastle.
Reid, P. 1994. Subsidence investigation, Tower Colliery, Dams Safety Committee.
Reid, P., 2001. Further analysis of horizontal movements around Cataract Dam, 1980 to 1997, in Proceedings 5th Triennial Conference on Coal Mine Subsidence: Current Practice and Issues, 211-218, Mine Subsidence Technological Society, Newcastle.
243 Sainsbury, D. P. 2008. Analysis of river bed cracking above longwall extraction panels in the Southern Coalfield of New South Wales, Australia, in Proceedings 1st Southern Hemisphere International Rock Mechanics Symposium, 325-338.
Seedsman, R. and Watson, G. 2001. Sensitive infrastructure and horizontal ground movement at Newstan Colliery, in Proceedings 5th Triennial Conference on Coal Mine Subsidence: Current Practice and Issues, Mine Subsidence Technological Society, Newcastle.
Seedsman, R. and Dawkins, A. 2006. Techniques to predict and measure subsidence and its impacts on the groundwater regime above shallow longwalls, ACARP Research Report No. C13009.
Schreibner, E. 1993. Structural framework of NSW, Quartley Notes, Geological Survey of NSW.
Shen, B., Alehossein, H., Pousen, B. and Waddington, A. 2010. Subsidence control using coal washery waste. ACARP Research Project No. C16023.
Shepherd, J. and Huntington, J. F. 1981. Geological fracture mapping in coalfields and the stress fields of the Sydney Basin, Journal of the Geological Society of Australia, 28, 299-309.
Shu, D. M. and Bhattacharyya, A. K. 1992. Influence of the sloping ground surfaces on the mine subsidence, in Proceedings 11th Int. Congress on Ground Control in Mining, 475-482, Wollongong.
Siriwardane, H. J. and Amanat, J. 1984. Prediction of subsidence in hilly ground terrain using finite element method, in Proceedings 2nd International Conference on Stability in Underground Mining, AIME, Lexington, Kentucky.
Stolarska, M., Chopp, D.L., Moës, N. and Belytschko, T. 2001. Modelling crack growth by level sets in the extended finite element method, International Journal for Numerical Methods in Engineering, 51, 943–60.
Su, S., Stephansson, O. and Peng, J. 2000. The effect of fracture pattern on in situ rock stresses studied by the distinct element method, in Proceedings ISRM International Symposium, International Society for Rock Mechanics, Melbourne, Australia.
Tarrant, G. 2006. New concepts in tailgate strata behaviour and implications for support design, PhD thesis, University of New South Wales, Sydney.
Vyazmensky, A., Elmo, D. and Rance, J. 2007. Combined finite-discrete element modelling of surface subsidence associated with block caving mining, in Proceedings 1st Canada-U.S. Rock Mechanics Symposium, 467-475, Vancouver.
Waddington, A. A. and Kay, D. R. 1998. Development of the Incremental Profile Method of predicting subsidence and its application in the Newcastle Coalfield, in Proceedings 4th Triennial Conference on Buildings and Structures Subject to Ground Movement, Mine Subsidence Technological Society, Newcastle.
244 Waddington, A. A. and Kay, D. R. 2001. Research into the impacts of mine subsidence on the strata and hydrology of river valleys and development of management guidelines for undermining cliffs, gorges and river systems, ACARP Research Project No. C8005.
Waddington, A. A. and Kay, D. R. 2002. Research into the impacts of mine subsidence on the strata and hydrology of river valleys and development of management guidelines for undermining cliffs, gorges and river systems, ACARP Research Project No. C9067.
Wallarah 2 Coal Project 2010. Subsidence impact assessment.
Walsh, R. V., Hebblewhite, B., Nicholson, M. A., Barbato, J., Mills, K., Li, G. and Brannon, P. J. 2014. Monitoring of ground movements at Sandy Creek Waterfall and implications for understanding the mechanics of valley closure movements, in Proceedings 9th Triennial Conference on Mine Subsidence: Risk Management in Action, 227-244, Mine Subsidence Technological Society, Newcastle.
Wardle, L. J. 1986. Boundary element methods for stress analysis of tabular excavations, PhD thesis, University of Queensland, Brisbane.
Wardle, L. J. and McNabb, K. E. 1992. Three-dimensional numerical modelling of effects of subsidence on escarpments, in Proceedings 11th Int. Congress on Ground Control in Mining, 506-510, Wollongong.
Whittaker, B. N. and Reddish, D. J. 1989. Subsidence: occurrence, prediction and control. Amsterdam: Elsevier.
Wold, M. B., Choi, S. K., George, S. C., Wood, J. H. and Williams, D. J. 1999. Coal mining beneath a gorge; induced fracturing and the release of reservoired gases, in Proceedings 9th Int. Congress on Rock Mechanics, 339-342, International Society for Rock Mechanics, Paris.
Yoshinaka, R., Osada, M., Park, H., Sasaki, T. and Sasaki, K. 2008. Practical determination of mechanical design parameters of intact rock considering scale effect, Engineering Geology, 96, 173-186.
Zainab, M., Kamaruzaman, M. and Cho Gye, C. 2007. Uniaxial compressive strength of composite rock material with respect to shale thickness ratio and moisture content, Electronic Journal of Geotechnical Engineering, 12.
Zaruba, Q. and Mencl, V. 1976. Engineering geology. Amsterdam: Elsevier.
Zhang, K. S., Wu, M. S. and Feng, R. 2005. Simulation of microplasticity-induced deformation in uniaxially strained ceramics by 3-D Voronoi polycrystal modelling, International Journal of Plasticity, 21, 801-834.
Zoback, M. D. and Zoback, M. L. 2002. State of stress in the Earth's lithosphere. In: William H. K. Lee, H. K. P. C. J. and Carl, K., eds. International Handbook of Earthquake & Engineering Seismology. International Geophysics. 81A, 559- 568.
245 Zoback, M. L. 1992. First- and second-order patterns of stress in the lithosphere: The World Stress Map Project, Journal of Geophysical Research: Solid Earth, 97, 11703-11728.
Zoback, M. L. and Zoback, M. D. 2007. Lithosphere stress and deformation. In: Schubert, G. R., ed. Treatise on Geophysics. Amsterdam: Elsevier, 1-48.
246