A Ground Support Design Strategy for Deep Underground

Mines Subjected to Dynamic-Loading Conditions

by

Philippe Napoléon René Morissette

A thesis submitted in conformity with the requirements

for the degree of Doctor of Philosophy
Department of Civil Engineering
University of Toronto

© Copyright by Philippe Morissette (2015)

 ProQuest Number: 10013996

All rights reserved

INFORMATION TO ALL USERS

The quality of this reproduction is dependent upon the quality of the copy submitted.

In the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if material had to be removed,
a note will indicate the deletion.

ProQuest 10013996

Published by ProQuest LLC (2016). Copyright of the Dissertation is held by the Author.

All rights reserved.

This work is protected against unauthorized copying under Title 17, United States Code
Microform Edition © ProQuest LLC.

ProQuest LLC.
789 East Eisenhower Parkway
P.O. Box 1346
Ann Arbor, MI 48106 - 1346
 A Ground Support Design Strategy for Deep Underground Mines
Subjected to Dynamic-Loading Conditions

Philippe Napoléon René Morissette

Doctor of Philosophy

Department of Civil Engineering

University of Toronto

2015

Abstract

In deep and high stress mines, managing the seismic risk is a very important technical challenge.

Large-magnitude seismic events that result in damage to excavations and ground support systems

are referred to as “rockbursts”. Under dynamic-loading conditions, ground support design is

influenced, to a degree, by a lack of full understanding of the mechanisms of action and

interaction of reinforcement and surface support elements. Consequently ground control

engineers rely mostly on empirical experience to select appropriate ground support systems.

These decisions are often based on limited field and technical data.

This thesis proposes a ground support design strategy, supported by high-quality field data, for

underground mines subjected to dynamic-loading conditions. Comprehensive data were collected

on rockbursts that occurred over 8 to 14 years at Creighton, Copper Cliff, and Coleman mines, in

the Sudbury Basin. The majority of pertinent information was obtained through on-site field

assessments, seismic system records, and numerical elastic stress modeling. A database was

constructed and validated comprising of 324 events documenting the performance of ground

support during rockbursts.

ii
A multi-variate statistical analysis was conducted to characterize the dynamic-load demand on

ground support systems. Mine excavations susceptible to rockbursts were identified in time and

space based on the anticipated magnitudes, the radius of influence of seismic events, and the

mining-induced stress conditions. The undertaken analysis resulted in the development of a

transition strategy from static to dynamic ground support requirements in seismically active

mines. It was observed that strong and stiff support systems, enhanced by the application of

shotcrete, were capable of mitigating relatively small magnitude strain burst events up to a

magnitude of 1.5 Nuttli. Beyond this magnitude threshold, or if shotcrete is not applied for either

logistic or economic reasons, the potential dynamic-load demand on the support must be

quantified in terms of the design magnitude and distance from the seismic event. Support

performance thresholds, quantified using the peak particle velocity, were presented in order to

assist in the selection of appropriate ground support systems to mitigate damage. Ground support

selection guidelines were developed for a range of excavation dimensions and stress conditions.

iii
 Acknowledgments
First and foremost, I would like to thank my supervisor and mentor, Dr. John Hadjigeorgiou, for
his support, guidance, and friendship. I am infinitely grateful for the confidence you placed in me
and for the opportunity you provided.

I would like to acknowledge Vale for the funding of this research and for providing full access to
the participating mines. Sincere gratitude is extended to the ground control personnel at Vale for
the technical support offered over the years. I am particularly grateful to Dr. Mike Yao, Dr.
Denis Thibodeau, Allan Punkkinen, Danielle Pelletier, Anneta Sampson-Forsythe, and Reddy
Chinnasane. Additionally, I would like to thank Dr. Will F. Bawden, Dr. Matthew Pierce, and
Dr. Murray Grabinsky for their valuable input. Dr. Joe Repka from the department of
mathematics at the University of Toronto provided valuable assistance in developing the
statistical approach.

To my friends and colleagues at the University of Toronto, thank you for being part of this
journey on the research island. Special thanks to Efstratios Karampinos, Marie-Hélène Fillion,
Luke Nicholson, Alex Turichshev, Dr. Ryan Veenstra, and Dr. Ben Thompson for sharing
knowledge and coffee on several occasions. I would like to particularly acknowledge former
research partner Grace Liang for her input and collaboration at the earlier stages of data
collection, analysis, and interpretation. It has been a great time with all of you guys!

Je tiens également à remercier ma famille et amis proches, en particulier mes parents, Luc et
Lynda, pour leurs encouragements et leur support inconditionnel au cours de ces années d’étude
et de recherche.

And finally, this adventure would not have been possible without the love, support, and patience
of my wife, Patricia Nikolski. Your firm conviction that I would eventually overcome this
ambitious challenge has been my greatest source of motivation and I will be eternally grateful for
all the love and encouragement.

iv
 Table of Contents
Abstract ........................................................................................................................................... ii

Acknowledgments.......................................................................................................................... iv

Table of Contents ............................................................................................................................ v

List of Tables .................................................................................................................................. x

List of Figures .............................................................................................................................. xiii

1
Introduction ................................................................................................................................ 1

1.1
Problem definition .............................................................................................................. 1

1.2
Thesis objectives ................................................................................................................. 7

1.3
Methodology ....................................................................................................................... 8

1.4
Thesis structure ................................................................................................................. 11

2
Creighton, Copper Cliff, and Coleman Mines ......................................................................... 14

2.1
Regional geology of the Sudbury area .............................................................................. 15

2.2
Local geology and mining methods .................................................................................. 16

2.2.1
Creighton Mine ..................................................................................................... 16

2.2.2
Copper Cliff Mine ................................................................................................. 19

2.2.3
Coleman Mine ....................................................................................................... 22

2.3
Predominant geological structures .................................................................................... 24

2.3.1
Creighton Mine ..................................................................................................... 24

2.3.2
Copper Cliff Mine ................................................................................................. 25

2.3.3
Coleman Mine ....................................................................................................... 27

2.4
Geotechnical data .............................................................................................................. 28

2.5
Summary ........................................................................................................................... 30

3
Mining Sequences and Ground Control Practice and their Influence on Rockburst
Frequency and Severity ............................................................................................................ 32

3.1
Creighton Mine ................................................................................................................. 33

v
 3.1.1
Stope design and seismic risk management .......................................................... 33

3.1.2
Evolution of ground support systems ................................................................... 35

3.1.3
Cable-bolting of large excavations ....................................................................... 37

3.1.4
Summary ............................................................................................................... 38

3.2
Copper Cliff Mine ............................................................................................................. 38

3.2.1
March 25, 2008 rockburst ..................................................................................... 39

3.2.2
September 11, 2008 rockburst .............................................................................. 40

3.2.3
Modifications to support practice and mining sequence at Copper Cliff Mine .... 42

3.3
Coleman Mine ................................................................................................................... 43

3.3.1
September 24, 2010 rockburst .............................................................................. 43

3.3.2
November 20, 2010 rockburst............................................................................... 44

3.3.3
April 6, 2011 rockburst ......................................................................................... 45

3.3.4
Adjustments to support practice at Coleman Mine ............................................... 48

3.4
Three mines, three stories ................................................................................................. 48

3.4.1
Typical causes of ground support failure under dynamic-loading conditions ...... 49

3.5
Summary ........................................................................................................................... 51

4
Conventional and Empirical Approaches Toward Ground Support Design for Rockburst
Conditions ................................................................................................................................ 53

4.1
Dynamic-load demand on ground support systems for a conventional support design
approach ............................................................................................................................ 54

4.1.1
Background on rockbursts and mine seismicity ................................................... 55

4.1.2
Magnitude scales ................................................................................................... 64

4.1.3
Ejection velocity and peak particle velocity scaling laws .................................... 65

4.1.4
Brittle rock mass failure and estimation of the extent of damage around mine
openings ................................................................................................................ 73

4.1.5
Challenges and limitations in forecasting the dynamic-load demand on ground

support systems ..................................................................................................... 90

vi
 4.2
Dynamic capacity of ground support systems for a conventional support design
approach ............................................................................................................................ 91

4.2.1
Extrapolation of static (pull) test results ............................................................... 94

4.2.2
Testing rig facilities .............................................................................................. 96

4.2.3
Simulated rockburst experiments .......................................................................... 98

4.2.4
Capacity of individual support elements vs. capacity of support systems: how
do we fill the gaps? ............................................................................................. 102

4.3
Empirical approaches toward the selection of ground support under dynamic-loading
conditions ........................................................................................................................ 106

4.4
Summary ......................................................................................................................... 112

5
Rockburst Data Collection and Validation ............................................................................ 114

5.1
Rockburst field-data collection campaign ...................................................................... 115

5.2
Pertinent variables used in passive monitoring of support performance under dynamic
loads ................................................................................................................................ 118

5.2.1
Predictor variables .............................................................................................. 119

5.2.2
Response variables .............................................................................................. 121

5.3
Validation of seismic event magnitudes ......................................................................... 130

5.3.1
Adjusted magnitude scales .................................................................................. 130

5.3.2
Validation of magnitude adjustments using seismic data analysis techniques

for Creighton’s data ............................................................................................ 136

5.3.3
Seismic risk and probability of rockburst at Creighton Mine ............................. 138

5.4
Ground motion levels ...................................................................................................... 140

5.5
Distance from large-scale geological structures ............................................................. 142

5.6
Mining-induced stresses.................................................................................................. 143

5.6.1
Stress modelling: approach and results ............................................................... 144

5.6.2
Stress modelling: comparison between mining-induced stresses, seismicity,

and rockbursts at Creighton Mine ....................................................................... 150

5.6.3
Limitations and interpretations of numerical models ......................................... 153

5.7
Summary ......................................................................................................................... 154

vii
6
Multivariate Characterization of Rockburst Field Data ......................................................... 157

6.1
Principal component analysis ......................................................................................... 157

6.1.1
Scaling................................................................................................................. 159

6.1.2
Mean-centering ................................................................................................... 159

6.1.3
Log-transforming ................................................................................................ 160

6.2
Distribution of collected rockburst variables and recommended transformations ......... 161

6.3
Analysis of rockburst field data using PCA .................................................................... 164

6.3.1
Interpretation of the loading scatter plot ............................................................. 165

6.3.2
Interpretation of the score scatter plots ............................................................... 167

6.4
Summary ......................................................................................................................... 176

7
Assessment of Ground Support Performance ........................................................................ 178

7.1
Conceptual capacity of reinforcement systems............................................................... 179

7.2
Overview of the performance of ground support systems .............................................. 184

7.2.1
Rate of tolerance of support performances ......................................................... 185

7.2.2
Quantifiable interpretations of support performance .......................................... 186

7.3
Performance of ground support systems under strain-burst and fault-slip conditions.... 189

7.3.1
Performance of ground support systems installed in the back ............................ 191

7.3.2
Performance of ground support systems installed in the walls ........................... 194

7.3.3
Summary ............................................................................................................. 196

7.4
Considerations of excavation size, mining-induced stresses, and rock mass quality in
interpreting support performance .................................................................................... 197

7.4.1
Rockburst classification based on the excavation conditions ............................. 197

7.4.2
Interpretation of support performance in consideration of excavation

conditions and seismic loading ........................................................................... 202

7.5
Summary ......................................................................................................................... 225

8
Ground Support Strategy for Burst-Prone Grounds ............................................................... 229

8.1
Assessment of the seismic risk ....................................................................................... 229

viii
 8.2
Support strategy for an identified seismic source ........................................................... 231

8.3
Support strategy based on mining-induced stresses........................................................ 235

8.4
Ground support recommendations for managing dynamic-loading conditions .............. 240

8.4.1
Supporting the back of mine excavations under dynamic-loading conditions ... 241

8.4.2
Supporting the walls of mine excavations under dynamic-loading conditions .. 245

8.4.3
Example applications of the empirical support-performance threshold charts ... 248

8.5
Extrapolation of support performance data..................................................................... 253

8.6
Summary ......................................................................................................................... 262

9
Conclusions ............................................................................................................................ 266

9.1
Motivations and originality of the research .................................................................... 266

9.2
Accomplishments and contributions ............................................................................... 269

9.3
Limitations and future work............................................................................................ 275

References ................................................................................................................................... 277

Appendix A: Unusual Occurrence Report for Groundfall/Rockbursts ....................................... 286

Appendix B: A Review of Dynamic Testing Rig Facilities ....................................................... 295

Appendix C: Evolution of Mining-Induced Stresses within Creighton Mine Deep ................... 307

ix
 List of Tables
Table 1–1
Number of rockbursts reported from each mine in the comprehensive database ....... 9

Table 2–1
Rock mass data for various geological domains at Creighton, Copper Cliff, and
Coleman mines .................................................................................................................... 29

Table 2–2
Far-field stress regime at Creighton, Copper Cliff, and Coleman mines ................. 29

Table 3–1
Occurrences and severity of rockbursts in large excavations characterized by a span
or wall height greater than 7.3 m at Creighton Mine .......................................................... 38

Table 4–1
Classification of seismic event sources, modified after Ortlepp (1997) ................... 61

Table 4–2
Summarized description of the three main rockburst types (Kaiser & Cai, 2013) ... 62

Table 4–3
Summary of ppv scaling laws determined by linear regression (ppv is in m/s and the
distance R is in m) ............................................................................................................... 65

Table 4–4
Load-displacement parameters for different types of reinforcement and surface
support elements (Kaiser et al., 1996) ................................................................................. 95

Table 4–5
Energy-absorption capacity, determined using drop tests, of some of the
reinforcement and surface support elements tested as part of simulated rockburst
experiments by Heal (2010) .............................................................................................. 104

Table 4–6
Energy-absorption capacity of ground support systems determined based on the
summation and weakest-link assumptions and compared with the results of simulated
rockburst experiments conducted by Heal (2010) ............................................................. 105

Table 5–1
Summary of the purpose of each section of WSN’s Unusual Occurrence Report on
Groundfall/Rockburst (WSN, 2010) ................................................................................. 115

Table 5–2
Predictor variables reported in the rockburst database ........................................... 119

Table 5–3
Adjusted Rock Damage Scale................................................................................. 122

xi
Table 5–4
Assessment of the support performance as a result of rockburst CR-1433 (August
18, 2012) at Creighton Mine using the WSN template (WSN, 2010) ............................... 123

Table 5–5
Support damage scale (SDS) (Kaiser et al., 1992).................................................. 126

Table 5–6
Analysis of variance for testing the significance of regressions on magnitude data
from Creighton, Copper Cliff, and Coleman mines, at a 0.05 significance level ............. 134

Table 5–7
One-sided test of hypotheses on sample’s mean testing H0: μ=0 and H1: μ>0 or μ<0
for seismic data from Copper Cliff and Coleman mines ................................................... 135

Table 5–8
Comparison of damage initiation thresholds in brittle rock masses ....................... 149

Table 7–1
Load and energy-absorption capacities of reinforcement elements employed in

classifying ground support systems ................................................................................... 179

Table 7–2
Categories of reinforcement systems installed at rockburst locations, in the back of
mine openings .................................................................................................................... 182

Table 7–3
Categories of reinforcement systems installed at rockburst locations, in the walls of
mine openings .................................................................................................................... 183

Table 7–4
Summary of the categories of ground support systems installed in the back of mine
openings and tested over ranges of excavation and dynamic-loading conditions ............. 204

Table 7–5
Summary of the categories of ground support systems installed in the walls of mine
openings and tested over ranges of stress and dynamic-loading conditions ..................... 216

Table 8–1
PPV-performance threshold of detailed ground support systems installed in the back
of excavations at Vale mines ............................................................................................. 244

Table 8–2
PPV-performance threshold of detailed ground support systems installed in the walls
of excavations at Vale mines ............................................................................................. 247

Table 8–3
Load-bearing and energy-absorption capacity of the D-Bolt ................................. 254

xii
Table 8–4
Conceptual load-bearing and energy-absorption capacities of two configurations
of 20 and 22 mm diameter D-Bolts based on laboratory testing results ........................... 255

xiii
 List of Figures
Figure 1–1
A production drill parked at 2,200 m depth was partially buried when material was
ejected from the wall on June 15, 2007 at Creighton Mine as a result of a 3.0 MN seismic
event ................................................................................................................................... 1

Figure 1–2
Conceptual functions of a ground support system under dynamic-loading

conditions, after (a) Kaiser et al. (1996) and (b) Kaiser and Cai (2013) ............................... 2

Figure 1–3
The engineering wheel of design (Stacey, 2009) ...................................................... 5

Figure 1–4
Learning process toward the development of appropriate support strategies to
mitigate rockbursts ................................................................................................................ 6

Figure 1–5
Schematic of the rockburst data collection and verification process applied at Vale
mines ................................................................................................................................. 10

Figure 2–1
Location of Sudbury with respect to other Canadian cities (Map data ©2015
Google) ................................................................................................................................ 14

Figure 2–2
Geological map of the Sudbury area showing the location of mine sites including
Creighton, Copper Cliff, and Coleman mines (modified from Roussell & Card, (2009)) .. 15

Figure 2–3
Cross-section of Creighton Mine (Andrews & Courchesne, 2009) and the
associated number of rockbursts per mine levels from 2000 to 2013 ................................. 17

Figure 2–4
Three-dimensional representation of the Creighton Deep 400 and 461 orebodies,
looking N-E (Tan, 2003) ..................................................................................................... 18

Figure 2–5
Open stopes mined in Creighton Deep as of September 2013 illustrated using
Map3D ................................................................................................................................. 19

Figure 2–6
Simplified geological map of the Copper Cliff Offset, showing the location of the
mine shafts and ore deposits of the Copper Cliff North and South mines (Rickard &
Watkinson, 2001) ................................................................................................................ 20

xiv
Figure 2–7
Longitudinal section of the 100, 900, and 890 orebodies, looking west (modified
from Rickard & Watkinson (2001)) and the associated number of rockbursts per mine level
from 2004 to 2013 ............................................................................................................... 21

Figure 2–8
Longitudinal section of the Coleman mine, looking North (Vale, 2011) and the
associated number of rockbursts per mine level.................................................................. 22

Figure 2–9
The Coleman mine’s Main orebody divided into three mining horizons: MOB1,
MOB2, and MOB3 .............................................................................................................. 23

Figure 2–10
Plan view of Creighton Mine’s 7400 level showing accesses to the 400 orebody
and prominent shear zones (Snelling et al., 2013)............................................................... 25

Figure 2–11
Plan view of Copper Cliff Mine’s 3400 level showing the 100 and 900 orebodies
and the predominant geological structures on site .............................................................. 26

Figure 2–12
Isometric view of Coleman Mine’s 4700 and 4810 levels showing the two bornite
veins mapped in the 153 orebody and seismic events associated with rockburst ME-168 on
November 23, 2012 ............................................................................................................. 28

Figure 2–13
Potential for tunnel instability and brittle rock mass failure at Creighton, Copper
Cliff, and Coleman mines based on the classification proposed by (Martin et al., 1999) ... 30

Figure 3–1
Cumulative tonnage displaced over time as a result of rockbursts at Creighton,
Copper Cliff, and Coleman mines and some of the associated large magnitude events (in
Nuttli) ................................................................................................................................. 32

Figure 3–2
Magnitude-time history analysis performed using seismic data of magnitudes 0.6
MN and greater ..................................................................................................................... 34

Figure 3–3
Evolution of the support systems installed in areas susceptible to rockbursts at
Creighton Mine with the support system employed from (a) 2006 to 2010 and (b) 2010 to
date ................................................................................................................................. 36

Figure 3–4 Correlations between the evolution of ground support systems and the frequency and
severity of rockbursts at Creighton Mine ............................................................................ 37

xv
Figure 3–5
Location of the 9281 stope and the 2.9 MN event which occurred following a
production blast on March 25, 2008 .................................................................................... 39

Figure 3–6
Damage to the 9280 top sill on the 3880 level of the 900 orebody at Copper Cliff
Mine, triggered by a recorded 2.9 MN seismic event on March 25, 2008 ........................... 40

Figure 3–7
Damage to the Copper Cliff Mine ramp as a result of a 3.8 MN seismic event on
September 11, 2008: (a) cross-section of the ramp between the 3500 and 3550 levels and
(b) photography of the 181 tonnes of material displaced where the ramp intersects the Trap
Dyke between the 3000 and 3050 levels ............................................................................. 41

Figure 3–8
Significant shotcrete cracking as the installed support system in the main
intersection of Copper Cliff’s 3200 level contained the bulk of the broken material on
September 11, 2008 ............................................................................................................. 42

Figure 3–9
Damage associated with a 2.6 MN event at Coleman mine on September 24, 2010
located on (a) the 4700 level, more specifically in (b) the cut 11/12 access and (c) 9/10
access ................................................................................................................................. 44

Figure 3–10
Major seismic events and differential stresses (MPa) associated with the
November 20, 2010 rockburst in the 4320-3 Access to the Cut 12 of the 4400 mining
horizon .............................................................................................................................. 45

Figure 3–11
View from the hanging-wall of the MOB showing the location of the 3.7 MN event
and the confinement on the ore/footwall contact generated by the major principal stress
(MPa) prior to mining of the 7760 stope, using Map3D ..................................................... 46

Figure 3–12
Chronology of events associated with the April 6, 2011 rockburst (Sampson-
Forsythe, 2011) .................................................................................................................... 47

Figure 3–13
Typical causes of ground support failure under dynamic loads and the associated
phases of the ground support cycle ..................................................................................... 50

Figure 4–1
Conceptual guidelines used to assess the anticipated rockburst damage
mechanisms and the required support system characteristics (Kaiser et al., 1996)............. 55

xvi
Figure 4–2
Reconciliation of rockburst damage mechanisms identified by (a) (Kaiser et al.,
1996) and (b) (Ortlepp, 1992) ............................................................................................. 56

Figure 4–3
Typical local rock mass failure mechanisms that potentially cause seismic events:
(a) fault movement, (b) stress change causing rock mass fracturing near excavation, (c)
stope overbreak, (d) contrast in material properties causing strain-bursting, (e) crushing of
mine pillars, (f) stress increase causing rock mass deformation (Hudyma, 2008) .............. 59

Figure 4–4
Frequency magnitude relations for two groups of events at an Australian mine
(Hudyma, 2008) ................................................................................................................... 60

Figure 4–5
Scatter plots of S-wave and P-wave energy for two clusters of seismic events
(Hudyma et al., 2003) .......................................................................................................... 60

Figure 4–6
Representations of ppv scaling laws by (a) McGarr et al. (1981), (b) Hedley (1992),
(c) Kaiser et al. (1996), and (d) Potvin et al. (2010). Magnitudes in (a) were converted from
the ML to MN using the conversion proposed by Kaiser et al. (1996). The light grey near-
field area in (c) represents possible variations of α values in determining the source radius
(Eq. 4–3). A mid-range α value of 0.835 was used to assess the near-field conditions in
(d). ................................................................................................................................. 67

Figure 4–7
Comparison of ppv scaling laws with lower and upper bounds representing event
magnitudes of 1.0 and 4.0 MN ............................................................................................. 68

Figure 4–8
Distributions of ppv modelled based on (a,b) realistic S-wave radiation patterns
from two synthetic seismic events with different slip directions and (c) based on a spherical
radiation pattern assumption (Potvin & Wesseloo, 2013) ................................................... 69

Figure 4–9
Increase in the stress wave amplitude from actual seismograms recorded 10 m from
a tunnel sidewall and on the tunnel sidewall (Durrheim, 2012) .......................................... 70

Figure 4–10
Variations of damage severity at Creighton Mine with respect to the ppv
(Morissette et al., 2012) ....................................................................................................... 71

Figure 4–11
Well-developed slabbing in a footwall drive pre-developed ahead of stoping in

fine-grained, massive, and very strong quartzite at a depth of 2,600 m (Ortlepp, 1997) .... 74

xvii
Figure 4–12
Complete axial stress-strain curves obtained in triaxial compression tests on
Tennessee Marble at various confining pressures (Wawersik & Fairhurst, 1970), converted
in metric units (Brady & Brown, 2004)............................................................................... 75

Figure 4–13
Stress-strain curves of (a) Westerly granite and (b) Frederick diabase in uniaxial
and triaxial compression (Wawersik & Brace, 1971) ......................................................... 76

Figure 4–14
Shear rupture mechanism: (a) common frictional and (b) frictionless model
proposed by Tarasov and Randolph (2011)......................................................................... 77

Figure 4–15
Scale of brittleness index k and k1 with characteristic shapes of complete stress-
strain curves (Tarasov & Randolph, 2011).......................................................................... 78

Figure 4–16
Energy balances typical of the failure process in class I and class II rocks (Potvin
& Wesseloo, 2013; Tarasov & Potvin, 2012)...................................................................... 79

Figure 4–17
Damage on Creighton Mine’s 7400 level associated with a 2.9 MN rockburst on
March 14, 2009 .................................................................................................................... 81

Figure 4–18
View of the completed test tunnel at the AECL-URL (a) and microseismicity that
followed blasting of the 17th round (b) (Martin, 1997) ....................................................... 82

Figure 4–19
Damage initiation threshold for Lac du Bonnet granite derived from stresses
computed for microseismic events recorded ahead of the tunnel face (Martin, 1997) ....... 83

Figure 4–20
Damage initiation threshold for Lac du Bonnet granite compared with stresses
computed for microseismic events associated with the formation of the notch (Martin,
1997) .............................................................................................................................. 84

Figure 4–21
(a) Rock mass failure criterion derived from back-analyses of pillar bursts and (b)
its application in forecasting the possible failure of a crown pillar (Wiles, 2005) .............. 85

Figure 4–22
The role of loading system stiffness (mine stiffness) and rock mass stiffness in the
type of rock mass failure (Kaiser et al., 1996) .................................................................... 86

Figure 4–23
(a) Back analysis of local energy release density (LERD) during pillar bursts and
(b) the assessment of energy released during the failure of a crown pillar ......................... 87

xviii
Figure 4–24
Calculated normalized depths of failure for rockburst occurrences at Creighton,
Copper Cliff, and Coleman mines, which typically exceed the applicability threshold
determined by Kaiser et al. (1996) ...................................................................................... 88

Figure 4–25
Schematic cross-section of an open stope intruded by a metasediment stringer and
representation of the failure mode (Simser & Andrieux, 1999) .......................................... 90

Figure 4–26
Representation of a dual stiffness system with stiff and soft-deformable
reinforcement/holding elements (Charette, 2012) ............................................................... 92

Figure 4–27
Important stiffness contrasts between (a) chain-link mesh and holding elements
(modified conebolts) and (b) between small-gauge welded-wire mesh and a reinforcement
element (rebar) (Simser, 2012) ............................................................................................ 93

Figure 4–28
Load-displacement curves for (a) five types of reinforcement elements (Kaiser et
al., 1996) and (b) three types of mesh (Tannant, 2004) ...................................................... 96

Figure 4–29
Interpretation of some of the results from the dynamic testing of surface support
performed by Ortlepp and Stacey (1997) (Hadjigeorgiou & Potvin, 2011) ........................ 97

Figure 4–30
(a) Bent rebars hanging from the back after rockburst ME-80 on October 15, 2004
at Coleman mine and (b) ripples cast into rebar by deforming rock mass during transient
vibration at Kidd mine (Counter, 2012) .............................................................................. 98

Figure 4–31
Conceptual plan and cross-sectional views of a simulated rockburst test site layout
(Heal, 2010) ......................................................................................................................... 99

Figure 4–32
Test wall setup used to compare the performance of split sets and masterseal on
the left, and split sets and mesh on the right (Heal, 2010) ................................................ 100

Figure 4–33
Average energy demand required to trigger S3-level damage to ground support
systems (Heal, 2010) ......................................................................................................... 101

Figure 4–34
Empirical stability classification developed for square tunnels in massive
quartzite in South Africa (Hoek & Brown, 1980) ............................................................. 107

xix
Figure 4–35
The rockburst damage potential (RDP), i.e. the product of the excavation
vulnerability potential (EVP) and the peak particle velocity (PPV), introduced by Heal
(2010) to forecast the severity of rockbursts ..................................................................... 109

Figure 4–36
The relationship between EVP and ppv for rockburst case studies from LaRonde
Mine (Turcotte, 2014) ....................................................................................................... 110

Figure 4–37
Rockbursts from the Kirunavaara mine plotted over the empirical chart proposed
by Heal (2010) to assess the potential for rockburst damage by relating the largest probable
ppv to the EVP (Woldemedhin & Mwagalanyi, 2011) ..................................................... 111

Figure 4–38
Example of application of the adjusted EVP to the analysis of support
performance at LaRonde Mine (Turcotte, 2014) ............................................................... 112

Figure 5–1
Schematic of the rockburst data collection and verification process applied at Vale
mines ............................................................................................................................... 117

Figure 5–2
Visual tool for the assessment of damage to ground support systems comprising
steel mesh, based on observations of support performance in Sudbury mines ................. 128

Figure 5–3
Visual tool for the assessment of damage to ground support systems comprising
plain or reinforced shotcrete, based on observations of support performance in Sudbury
mines .............................................................................................................................. 129

Figure 5–4
Cross-validation of magnitude estimates: comparison of raw magnitude data from
Creighton, Copper Cliff, and Coleman mines with magnitudes recorded by the GSC ..... 131

Figure 5–5
Cross-validation of magnitude estimates: adjusted magnitude data...................... 133

Figure 5–6
Magnitude-time history from January 2000 to September 2013 at Creighton Mine
using the adjusted magnitude ............................................................................................ 137

Figure 5–7
Frequency-magnitude relationships for (a) the seismic events recorded by the
HDDR system and (b) the events recorded by the Paladin. Magnitude ranges proposed by
Ortlepp (1997) are presented to assist in the interpretation of the seismic source
mechanisms. ...................................................................................................................... 138

xx
Figure 5–8
Probability of a seismic event of a given magnitude to result in (a) a rockburst, (b)
more than 10 tonnes of displaced material within the mine, and (c) more than 100 tonnes of
displaced material within the mine. The probability lines for the three range of damage
severity are plotted in (d). .................................................................................................. 139

Figure 5–9
Levels of ground motion estimated based on the magnitude and distance from the
epicentre of seismic events using the scaling law proposed by Hedley (1992) for the
rockbursts from Creighton, Copper Cliff, and Coleman mines......................................... 141

Figure 5–10
Frequency and severity of rockbursts in the presence of large-scale geological
structures ........................................................................................................................... 143

Figure 5–11
Map3D model of Creighton Deep: (a) side view of the 400 and 461 orebodies,
looking west and (b) isometric view showing the ore and shear zones............................. 145

Figure 5–12
Stress conditions associated with rockbursts at Creighton, Copper Cliff, and
Coleman mines, colour-coded (a) by mine and (b) by lithology ....................................... 146

Figure 5–13
Distributions of stress conditions at rockburst locations: (a) approximated normal
distribution of the logarithm of the differential stress to UCS ratio and (b) approximated
lognormal distribution of the differential stress to UCS ratio ........................................... 148

Figure 5–14
Visualization of SGM seismic events plotted over (a) a longitudinal view and (b) a
top view of the open stopes, mined-out as of September 2013, and rockbursts, colour-coded
using the RDS, plotted over (c) the longitudinal and (d) top views of the mined-out
stopes. ............................................................................................................................. 151

Figure 5–15
Mining-induced stresses and seismicity on (a) the 7400, (b) 7530, and (c) 7680
level of Creighton Mine..................................................................................................... 152

Figure 6–1
Two principal components forming a plane in the multidimensional variable space
on which observations can be projected and described in terms of their score (Eriksson et
al., 2006) ............................................................................................................................ 158

xxi
Figure 6–2
Conceptual representation of the application of unit variance scaling and mean-
centering, which results in 5 variables of equal length and centred at a mean value 0
(Eriksson et al., 2006) ........................................................................................................ 160

Figure 6–3
Distributions of quantitative variables used in the PCA on collected rockburst data
with the proposed data transformation .............................................................................. 162

Figure 6–4
Distributions of quantitative predictor variables following data transformation .. 163

Figure 6–5
The influence of adding principal components on the cumulative R2 and Q2
parameters.......................................................................................................................... 164

Figure 6–6
Score scatter plots of the collected rockburst data with the horizontal and vertical
axes corresponding to (a) PC1 and PC2, and (b) PC1 and PC3 ........................................ 165

Figure 6–7
Loading scatter plots of the collected rockburst data with the horizontal and vertical
axes corresponding to (a) PC1 and PC2, and (b) PC1 and PC3 ........................................ 166

Figure 6–8
Score plots defined by PC1 and PC2 with rockbursts colour-coded according (a)
the interpreted seismic source mechanism and (b) the adjusted Nuttli magnitude ........... 168

Figure 6–9
Score plot defined by PC1 and PC2 with rockbursts colour-coded according to the
mine sites ........................................................................................................................... 169

Figure 6–10
Score plot defined by PC1 and PC2 with rockbursts colour-coded according to the
differential stress to UCS ratio .......................................................................................... 170

Figure 6–11
Score plot defined by PC1 and PC2 with rockbursts colour-coded according to the
RDS ............................................................................................................................. 171

Figure 6–12
Increasing levels of displaced tonnage influenced by (a) increasing depth and (b)
increasing major principal stress for strain burst data ....................................................... 172

Figure 6–13
Increasing levels of displaced tonnage influenced by (a) increasing event
magnitude and (b) increasing PPV for fault slip data........................................................ 173

xxii
Figure 6–14
Score plots defined by PC1 and PC3 with rockbursts colour-coded according to
(a) the span of mine openings and (b) the longest reinforcement elements installed in the
backs ............................................................................................................................. 174

Figure 6–15
Score plot defined by PC1 and PC2 with rockbursts colour-coded according to the
RDS ............................................................................................................................. 175

Figure 7–1
Load-bearing and energy-absorption capacities of combinations of reinforcement

elements installed in (a) the back and (b) the walls of mine openings (legend presented in
Tables 7–1 and 7–2) .......................................................................................................... 181

Figure 7–2
Proportion of acceptable, tolerable, and intolerable performances of the ground
support systems and resulting RoT values ........................................................................ 187

Figure 7–3
Rockbursts filtered as strain bursts and fault slips with a high degree of confidence
for Creighton, Copper Cliff, and Coleman mines ............................................................. 190

Figure 7–4
Event magnitude associated with each individual damage location reported in the
comprehensive rockburst database .................................................................................... 190

Figure 7–5
Performance of ground support systems installed in the back of mine excavations
under dynamic-loading conditions triggered by strain-burst and fault-slip events ........... 191

Figure 7–6
Performance of ground support systems installed in the walls of mine excavations
under dynamic-loading conditions triggered by strain-burst and fault-slip events ........... 195

Figure 7–7
Grouping of rockburst data into 6 categories visualized on (a) the PC2-PC3 score
scatter plot and explained using (b) the PC2-PC3 loading scatter plot ............................. 199

Figure 7–8
Score scatter plots defined by PC2 and PC3 showing a negative correlation
between (a) the span of mine openings (in metres) and (b) the normalized differential
stress ............................................................................................................................... 199

Figure 7–9
Bar charts illustrating distributions of (a) excavation span, (b) normalized
differential stress, (c) normalized major principal stress, and (d) adjusted magnitude within
each identified rockburst groups ....................................................................................... 201

xxiii
Figure 7–10
The 6 identified rockburst groups represented using the interquartile range of the
excavation span and normalized major principal stress .................................................... 202

Figure 7–11
Rockburst group 1: performance of ground support systems that comprise (a) steel
mesh and (b) mesh-reinforced shotcrete, in the back of mine openings ........................... 206

Figure 7–12
Rockburst group 2: performance of ground support systems that comprise (a) steel
mesh and (b) mesh-reinforced shotcrete, in the back of mine openings ........................... 207

Figure 7–13
Rockburst group 3: performance of ground support systems that comprise (a) steel
mesh and (b) mesh-reinforced shotcrete, in the back of mine openings ........................... 209

Figure 7–14
Rockburst group 4: performance of ground support systems that comprise (a) steel
mesh and (b) mesh-reinforced shotcrete, in the back of mine openings ........................... 210

Figure 7–15
Rockburst group 5: performance of ground support systems that comprise (a) steel
mesh and (b) mesh-reinforced shotcrete, in the back of mine openings ........................... 211

Figure 7–16
Rockburst group 6: performance of one-pass ground support systems that
comprise (a) steel mesh and (b) mesh-reinforced shotcrete, in the back of mine
openings ............................................................................................................................ 212

Figure 7–17
Rockburst group 6: performance of enhanced ground support systems that
comprise (a) steel mesh and (b) mesh-reinforced shotcrete, in the back of mine
openings ............................................................................................................................. 213

Figure 7–18
Rockburst group 1-2: performance of ground support systems that comprise
mechanical bolts and either (a) steel mesh or (b) mesh-reinforced shotcrete, in the walls of
mine openings .................................................................................................................... 217

Figure 7–19
Rockburst group 1-2: performance of ground support systems that comprise
friction sets and either (a) steel mesh or (b) mesh-reinforced shotcrete, in the walls of mine
openings ............................................................................................................................. 218

xxiv
Figure 7–20
Rockburst group 1-2: performance of ground support systems that comprise rebars
or enhanced support and either (a) steel mesh or (b) mesh-reinforced shotcrete, in the walls
of mine openings ............................................................................................................... 220

Figure 7–21
Rockburst group 5-6: performance of ground support systems that comprise 35 or
39 mm friction sets and either (a) steel mesh or (b) mesh-reinforced shotcrete, in the walls
of mine openings ............................................................................................................... 221

Figure 7–22
Rockburst group 5-6: performance of ground support systems that comprise
46 mm friction sets and either (a) steel mesh or (b) mesh-reinforced shotcrete, in the walls
of mine openings ............................................................................................................... 223

Figure 7–23
Rockburst group 5-6: performance of ground support systems that comprise
friction sets, enhanced with yielding reinforcement or cables, and either (a) steel mesh or
(b) mesh-reinforced shotcrete, in the walls of mine openings........................................... 224

Figure 7–24
Cumulated number of rockbursts for support systems comprising mesh and
shotcrete (over mesh) in mine excavations of typical size (rockburst groups 2 and 6)..... 227

Figure 8–1
Recommended support strategy based on the largest expected magnitude and the
associated probability of rockburst .................................................................................... 229

Figure 8–2
Conceptual assessment of the ground support requirements, based on the largest
anticipated event magnitude and probability of rockburst, for two different mining
areas .............................................................................................................................. 230

Figure 8–3
Anticipation of static and dynamic-loading conditions based on the radius of
influence of seismic events ................................................................................................ 232

Figure 8–4
Magnitude, distance from the epicentre of seismic events, and severity of damage
(RDS) for the rockbursts that occurred at Creighton, Copper Cliff, and Coleman mines 234

Figure 8–5
Normalized differential stress and severity of damage for the rockbursts that
occurred at Creighton, Copper Cliff, and Coleman mines ................................................ 236

xxv
Figure 8–6
Mining-induced stresses, as of September 2013, and seismicity on (a) the 7400, (b)
7530, and (c) 7680 levels of Creighton Mine .................................................................... 237

Figure 8–7
Mining-induced stresses, as of September 2013, and rockbursts on (a) the 7400, (b)
7530, and (c) 7680 levels of Creighton Mine .................................................................... 238

Figure 8–8
PPV-performance thresholds of ground support systems in the back of mine
excavations (support systems are described in Table 8–1) ............................................... 243

Figure 8–9
PPV-performance thresholds of ground support systems in the walls of mine
excavations for lower stress (left: σ1/UCS = 0.2-0.4) and higher stress (right:
σ1/UCS = 0.4-0.6) conditions (support systems are described in Table 8–2) ................... 246

Figure 8–10
Example 1: Cross-intersection characterized by a span of 7.8 m ........................ 248

Figure 8–11
Application of the proposed support-selection guideline to design a ground

support system for the back of the intersection in Example 1 ........................................... 249

Figure 8–12
Comparison of cross and T intersection spans of 5.5 m wide tunnels intersecting
each other perpendicularly ................................................................................................ 250

Figure 8–13
Application of the proposed support-selection guideline to design a ground

support system for the back of the mine tunnel in Example 2 .......................................... 251

Figure 8–14
Application of the proposed support-selection guideline to design a ground

support system for the walls of the mine tunnel in Example 2 ......................................... 251

Figure 8–15
Application of the proposed support-selection guideline to design a ground

support system for the back of the drift in Example 3 ...................................................... 252

Figure 8–16
Conceptual load and energy-absorption capacities of D-Bolt configurations as

compared to other reinforcement configurations reported in the rockburst database ....... 256

Figure 8–17
Cable bolts preventing a fall of ground to extend in the Fresh-Air-Drift
intersection on the 7810 level at Creighton Mine, on August 16, 2005 ............................ 258

xxvi
Figure 8–18
A cable bolt contributing to limiting the blown portion of a wall on Creighton
Mine’s 7200 level on June 15, 2007 .................................................................................. 259

Figure 8–19
Performance of a mesh-over-shotcrete support system during a 2.7 MN seismic

event at Coleman Mine on May 12, 2011 ......................................................................... 260

Figure 8–20
Learning process toward the development of appropriate support strategies to
mitigate rockbursts ............................................................................................................ 262

Figure 8–21
Summary of the proposed ground support design strategy for mine excavations
subjected to dynamic-loading conditions .......................................................................... 263

Figure 9–1
Radius of influence of seismic events determined based on the collected and
verified rockburst data ....................................................................................................... 273

xxvii
 1

1 Introduction

1.1 Problem definition

The mining industry is facing a series of technological challenges to ensure the safe and
profitable operation of deep underground mines. Amongst those challenges, the management of
mining-induced stress changes is of great importance. Underground mines operating in deep and
high stress conditions generally have to deal with mining-induced seismicity, which can result in
the occurrence of rockbursts. A rockburst is a seismic event, which causes violent and significant
damage to a tunnel or the excavation of a mine (Ortlepp, 2001). Due to their sudden and violent
nature, rockbursts can have considerable economical impacts on a mining operation. Their
consequence can range from damage to underground excavations generating high rehabilitation
costs, loss of production, permanent loss of mining reserves, damage to mining equipment
(Figure 1–1), to injury and fatalities. Consequently, the selection of appropriate ground support
systems is of great importance.

Figure 1–1 A production drill parked at 2,200 m depth was partially buried when material was ejected from the wall
on June 15, 2007 at Creighton Mine as a result of a 3.0 MN seismic event
 2

In a ground support design, the reinforcement corresponds to elements such as rockbolts that are
installed internally to the rock mass. The surface support elements, on the other hand, are applied
to the surface of excavations externally to the rock mass and typically comprise shotcrete and/or
steel mesh. The failure of a brittle rock mass is associated with large rock dilation and may
further be subjected to large impact energy due to the release of stored strain energy or seismic
loading. Ground support systems designed to support static loads therefore may not be capable of
managing the imposed dynamic loads. Consequently, in burst-prone environments, the design of
ground support systems must account for the anticipated dynamic-loading conditions. This is
typically achieved by prescribing reinforcement and/or surface support elements that are
perceived to enhance the yielding capacity of the system.

A description of the conceptual functions of ground support systems under dynamic-loading

conditions was provided by Kaiser, McCreath, and Tannant (1996) and further refined by Kaiser
and Cai (2013) (Figure 1–2). Those conceptual functions consist of:

• reinforcing the rock mass;

• retaining the fractured rock;

• holding the elements that provide the retaining function; and

• connecting the elements that provide the holding and retaining functions.

Figure 1–2 Conceptual functions of a ground support system under dynamic-loading conditions, after (a) Kaiser et
al. (1996) and (b) Kaiser and Cai (2013)
 3

The reinforcing function aims at strengthening the rock mass and controlling the bulking process
by preventing fractures from opening and propagating. This function is typically achieved by
installing stiff reinforcement elements, such as resin-grouted rebars. The installation of surface
support aims at preventing unraveling of the fractured rock mass and, therefore, provides the
retaining function to the system. Kaiser and Cai (2012) advocated the use of mesh-reinforced
shotcrete or mesh over shotcrete in order to achieve a much superior retaining function under
rockburst conditions. The holding function is required to tie the retaining elements and the
contained loose rock back to stable ground to prevent gravity-driven falls of ground. Under
rockburst conditions, the holding function further aims at dissipating the dynamic energy
released by the failing rock mass. It is neither practicable nor economically possible to contain
severe rockburst damage by increasing the strength of the tunnel support. […] It is necessary to
introduce a degree of compliance or yielding (Ortlepp, 1992). The installation of yielding
reinforcement elements such as conebolts or high capacity friction bolts is promoted (Kaiser &
Cai, 2012; Ortlepp, 1992). The connecting function aims at ensuring that reinforcement and
surface support elements work as an integrated support system. The use of heavy-gauge straps,
large plates, or mesh squares can improve the connecting function and facilitate the load transfer
from the surface support to the reinforcement elements during the rock bulking process. The
connections are often identified as the weakest link in a ground support system (Potvin,
Wesseloo, & Heal, 2010).

Under rockburst conditions, reinforcement and surface support elements form an integrated
support system, which shares and transfers dynamic loads through connections until the
excavation surface stabilizes or until the system fails at its weakest link (Hadjigeorgiou &
Potvin, 2011). The notion of integrated system implies that the ground support capacity under
dynamic loads must be assessed as a system rather than as individual support elements (Potvin et
al., 2010). In the static design process, the demand on the support is matched to the capacity of
the reinforcement alone, with little consideration in the calculations given for the contribution of
the surface support (Potvin et al., 2010). This design philosophy is carried through the dynamic
design process as existing design procedures focus mostly on assessing the load and energy-
absorption capacities of the rock bolts (Kaiser & Cai, 2013). The failure of the rock mass
between the bolts and its impact on the performance of the whole support system is often not
considered in the design.
 4

In theory, the design of ground support systems for rockburst conditions is a straightforward
engineering process. Stacey (2009) introduced the conceptual wheel of design, inspired by a
sequential design methodology for rock engineering problems proposed by (Bieniawski, 1992)
(Figure 1–3). To achieve the design objective, i.e. to contain rockburst damage so as to promote
safety and limit disruption of operations, the following mechanisms must be known and
understood (Stacey, 2012):

• the mechanisms of loading;

• the mechanisms of rock and rock mass failure that the support is being designed to
combat;

• the mechanisms of action of the support elements and their interactions in the support
system.

If this information is known, and supported by sufficient data, a conventional factor-of-safety

approach to design is possible. Such approach was advocated by Kaiser et al. (1996), in which
the demand imposed on the support system was opposed to the capacities of the support elements
that compose the system. In this approach, both the demand and the capacity were assessed in
terms of load, displacement, and energy. It was recognized by Ortlepp (1992) that energy
considerations rather than stress and strength calculations should form the basis of a tunnel
support system. In the approach advocated by Kaiser et al. (1996), the determination of the
design dynamic-load demand was based on the location of a design seismic event, the selection
of a design magnitude, and the selection of appropriate scaling-law parameters for assessing the
design peak particle velocity and hence, the rock ejection velocity. The assessment of the support
design capacity was based on data from laboratory and field-testing of support elements and
complemented by theoretical evaluations (Stacey, 2012). The design procedure introduced by
Kaiser et al. (1996) had the merit of being backed by the theoretical concepts that should drive
the performance of ground support systems under dynamic-loading conditions. Nevertheless, it
relied on the important, and somewhat questionable assumption that both the anticipated
dynamic-load demand and the capacities of available support options are quantifiable by the
designer. It is recognized however, that rockburst still represent a considerable challenge in deep
underground mines almost 20 years later, as accidents resulting from rockbursts continue to
occur. This situation reflects that in practice, the determination of the dynamic-load demand and
capacity of support systems is problematic. A conventional engineering design approach is
 5

therefore not possible for ground support systems subjected to rockbursts given the design
indeterminacy (Stacey, 2012).

Figure 1–3 The engineering wheel of design (Stacey, 2009)

In practice, the design of ground support systems for managing dynamic loads is generally based
on experience and perceived performance. In most cases, employed design guidelines, or
suggested ground support systems such as the one presented in Figure 1–2b, do not appear to be
validated using quantifiable field data on the support performance. Hence, the support design at
its initial stage often represents a best engineering guess and the installation proceeds without
having quantified the two essential parameters that are required to develop a robust design,
namely capacity and demand. Consequently, in practice, there is great importance on ensuring
good quality control and quality assurance. It is only by monitoring the performance under
dynamic loads of the installed ground support systems that the design can ultimately be verified.
Since rockbursts are of unpredictable nature, they cannot be adequately instrumented a priori.
Therefore, it is predominantly through field observations that highly valuable elements of design
information can be extracted. In practice, it is through empirical experience that, typically,
engineers improve their understanding of the support performance under dynamic-loading
conditions and further review and upgrade their design. The following schematic is proposed to
illustrate the continuous learning process associated with ground support design (Figure 1–4).
 6

Design
• Dynamic-load demand
• Support capacity
(best estimate)

Performance / Installation
monitoring
• Instrumentation • Quality control
• Field observations • Quality assurance

Figure 1–4 Learning process toward the development of appropriate support strategies to mitigate rockbursts

The process illustrated in Figure 1–4 is related to the observational method, which was
formulated and developed as a systematic procedure by Karl Terzaghi (Peck, 1969). The method
promotes the monitoring of the performance or behaviour of an infrastructure or surrounding
ground as construction proceeds. As data become available, it is possible to assess the deviation
of the observational findings from those predicted on the basis of the working hypothesis (Peck,
1969). The design can hence be modified to suit the actual conditions. This approach, described
as a “learn-as-you-go method”, as led to considerable savings in time and financial resources in
the practice of soil mechanics and engineering geology. A collection of case histories was
reviewed in the original paper on the advantages and limitations of the observational method
(Peck, 1969).

Perhaps the most important distinction between the learning process illustrated in Figure 1–4 and
the observational method is the absence of preset remedial measures, i.e. ground support
systems, for every foreseeable deviation of the observed performance with respect to the
expectations. Ground control engineers, unfortunately, do not dispose of a set of guidelines on
the performance of various ground support systems under dynamic-loading conditions. It is
therefore not possible to make ground support recommendations, supported by sufficient field
data, for different thresholds of loading. Furthermore, rockbursts often involve the failure of the
ground support design at one or several of the employed ground support components, in which
 7

case, the service life of the affected excavation is put on hold until rehabilitation is completed.
For the observational method to be fully applicable in designing ground support systems capable
of managing dynamic loads, it is important to quantify the performance of ground support
systems under rockburst conditions based on quantifiable field data.

Forensic engineering is the application of engineering practices and principles to determine and
interpret the causes of damage to, or failure of, equipment, machines, and structures (Petty,
2013). Forensic analysis, in the context of rockbursts, is the assessment of the damage to an
excavation or its support with the purpose of identifying the cause(s) of failure and/or obtaining
or validating design parameters (Kaiser & Cai, 2013). Given their unpredictability in time and
space, it is neither feasible nor practical to adequately instrument rockbursts, a priori.
Consequently, little quantitative data on ground support performance during rockbursts has been
collected to date (Hadjigeorgiou & Potvin, 2007).

This thesis postulates that passive monitoring of real events, even in the absence of complete
instrumentation data, is a valuable tool to assess the performance of ground support systems.
Passive monitoring, based on reviews of historical rockburst data, can trace the evolution of
mining and support practice at a mine site and further assess their influence on the frequency and
severity of rockbursts. This type of analysis, based on observed improvements (or deteriorations)
in managing the consequences of mining-induced seismicity, can justify changes (or the need to
implement changes) in mining strategy and/or support practice. As more mines operate under
conditions of high stress, it is possible to establish a database of carefully documented case
studies and passive monitoring of support systems under dynamic-loading conditions
(Hadjigeorgiou & Potvin, 2007).

1.2 Thesis objectives

This thesis aims at improving our understanding of the performance of ground support systems
under dynamic-loading conditions. The primary objective is to develop a ground support design
strategy, supported by high-quality field data, for mine openings subjected to rockbursts. The
secondary objectives of this thesis are:

• To develop a strategy for collecting rockburst data and observations related to the
performance of ground support systems
 8

• To improve the process of reporting rockburst occurrences in Ontario mines based on site
recommendations and research experience

• To build a high-quality and comprehensive database of rockburst case studies

• To rank the capacity of various ground support systems using passive monitoring

• To identify the thresholds that initiate a transition from static to dynamic support
strategies

1.3 Methodology
This thesis investigated the performance of ground support systems through back analyses of
past rockburst events in three deep underground hard rock mines. Vale’s Creighton, Copper Cliff
(formerly Copper Cliff North), and Coleman mines were selected for this research due to their
long history of mining, seismicity, and rockburst, and the quality of their seismic data. All three
mines have dedicated ground control teams allowing for continuity in good ground control
practice. A variety of ground support systems have been used over time in anticipation or in
response to higher dynamic loads or more frequent rockbursts. Creighton Mine is the deepest of
the three mines, approaching 2,420 m depth, and has the longest history of mine seismicity and
rockbursts amongst operating Vale Sudbury Mines. As a result, Creighton was identified as the
primary site of interest in this thesis.

A comprehensive data collection campaign was undertaken at the three mine sites by the
candidate. Time spent on site allowed familiarization with the mining operations, the employed
seismic monitoring systems, and the ground support and quality control practice specific to each
site. A high-quality, site-specific database was constructed of 324 case studies of support
performance under dynamic loads (Table 1–1). The collection of rockburst data has covered
events from January 2000, January 2004, and January 2006 to September 2013 at Creighton (14
years), Copper Cliff (10 years), and Coleman (8 years), respectively. The rockburst database is
comprised of events that occurred in access galleries developed in hard rock. The performance of
support systems installed in vertical excavations and development located in backfill was outside
the scope of this research.
 9

Table 1–1 Number of rockbursts reported from each mine in the comprehensive database

Creighton Copper Cliff Coleman Total

Number of seismic events 123 18 68 209
Number of damage locations 184 35 105 324

A valuable source of information for this study was the Unusual Occurrence Reports for
Groundfall/ Rockbursts (WSN, 2010) completed by the mine personnel at the time of events.
These reports identified the areas affected by rockbursts, provided a qualitative assessment of the
performance of each reinforcement and surface support element, and were often complemented
by pictures of rockburst damage. In retrospect, the employed report format was best suited for
recording events that resulted in single damage location. Furthermore, some inconsistencies in
the reports and variations in the accuracy of the reported information were attributed to the
interpretation of ground control personnel. Consequently, the collected Unusual Occurrence
Reports for Groundfall/ Rockburst were reviewed on site with the assistance of the ground
control personnel. This level of interaction with the mine personnel contributed in improving the
confidence in the data significantly. Furthermore, the collected data were cross-validated using
available sources of information, e.g. the seismic monitoring and CAD systems employed at the
mines, the public Earthquake Database (National Resources Canada, 2015), ground control
layouts, external engineering reports, and site inspections. The data revision and validation
process, represented schematically in Figure 1–5, has contributed to increasing the reliability of
the database.

Over the time-period covered by this thesis, there has been an evolution of the mining sequence
and ground support practice at Creighton, Copper Cliff, and Coleman mines. By monitoring
changes in the extraction sequences, employed rock mass de-stressing techniques, and ground
support systems, their influence on the frequency and severity of rockbursts was assessed, retro-
actively. This was made possible by having collected the continuous rockburst history at the
three mine sites. As a result of this analysis, changes in mining strategies and ground support
practice were validated based on observed improvements in managing the consequences of
rockburst over time.
 10

Figure 1–5 Schematic of the rockburst data collection and verification process applied at Vale mines

The rockburst variables gathered in the database were analysed in order to assess their individual
influence on the occurrence and severity of rockbursts. The complete rockburst data was further
subjected to a multi-variate statistical analysis to evaluate relationships amongst the collected
variables and to identify clusters of rockbursts that occurred under comparable conditions. The
performance of ground support systems amongst these rockburst clusters was further
investigated in order to identify relevant trends for the development of a ground support strategy
for mine openings subjected to dynamic-loading conditions. Based on the collected data, it was
possible to identify the thresholds that initiate a transition from static to dynamic support and to
recommend appropriate ground support systems.

Beyond recommending suitable ground support systems, in order to complete the development
of a ground support strategy, it was necessary to further assess the location and the timing of the
installation. The demand on ground support systems can change over time and no single ground
support system may provide both the properties required in the short and long term (Villaescusa,
Player, & Thompson, 2014). Ground and excavation conditions that led to an increased
 11

probability of rockbursts were quantified based on the collected data and recommendations were
made accordingly.

Finally, the template used at the time for reporting rockburst events in Ontario mines was
updated based on site recommendations and research experience cumulated throughout this
thesis. The updated template was accepted in 2013 by Workplace Safety North (WSN), a Health
and Safety Ontario partner, and is presented in Appendix A.

1.4 Thesis structure

The thesis consists of nine chapters and three appendices. The template accepted by WSN for
reporting rockburst events in Ontario mines is presented in Appendix A. In Appendix B, a
review of drop test facilities used for testing ground support elements under dynamic-loading
conditions is presented. Results from those rigs are presented and were used in this thesis to
assess the capacity of ground support systems at a conceptual level. In Appendix C, numerical
stress modelling results for Creighton Mine are presented in relation with the observed mine
seismicity and rockbursts.

The following outline describes the structure of the thesis chapters:

• Chapter 1 – Introduction – The research motivations, objectives, and methodology are

introduced.

• Chapter 2 – Creighton, Copper Cliff, and Coleman Mines – The local geology and
employed mining methods are discussed for each mine site. The large-scale geological
structures known for influencing the ground and seismic conditions are reviewed.
Geomechanical properties, rock mass classification, and in-situ stresses are further
discussed.

• Chapter 3 – Mining Sequences and Ground Control Practice and their Influence on
Rockburst Frequency and Severity – The chapter elaborates on the evolution of mining
sequences and ground support practice at Creighton, Copper Cliff, and Coleman Mines.
Correlations are identified between this evolution and the frequency and severity of
rockbursts over time. Justifications for introducing new support technologies, identifying
limitations in the employed support designs, and managing the mining process over time
are demonstrated in quantifiable terms.
 12

• Chapter 4 – Conventional and Empirical Approaches Toward Ground Support Design for
Dynamic-Loading Conditions – A state of the art review of available literature on
conventional and empirical methods to the design of ground support systems under
rockburst conditions is conducted. Mine seismicity, brittle rock failure, dynamic testing
of ground support elements and ground support systems, and dynamic support practices
in deep underground mines, are amongst the topics covered in this literature review.

• Chapter 5 – Rockburst Data Collection and Validation – The collected rockburst

variables are reviewed in this section and preliminary correlations with the occurrence of
rockbursts are explored. The chapter covers the adjustment to the magnitude data in order
to correlate the local estimates to the Nuttli scale used by the Geological Survey of
Canada (GSC). The consistency in the results of numerical elastic stress analyses,
performed to determine the stress conditions at each rockburst location, is further
discussed.

• Chapter 6 – Multivariate Characterization of Rockburst Field Data – A multi-variate

statistical analysis is conducted on the collected rockburst data. The collected data are
explored using the principal component analysis method (PCA). Identified patterns
amongst the rockburst data are used to characterize the dynamic-load demand on support
systems during rockbursts. Variations in the severity of rockbursts are linked to variations
in parameters that affect the severity of the dynamic-load demand. The collected
variables that drove most of the variability in the first three components of the model are
carried forward in order to analyse their influence on the performance of ground support
systems.

• Chapter 7 – Assessment of Ground Support Performance – The ground support systems

reported in the rockburst-history database are categorized based on observed similitude
and preliminary knowledge on the performance of individual reinforcement and surface
support elements. The performance of each category of support systems is quantified
using the Rate of Tolerance, a ratio of support performance introduced in this chapter.
The assessment of support performance is refined further by dividing rockbursts into
clusters that were determined using PCA. Those clusters reflect primarily variations in
the seismic source mechanism and magnitude of seismic events, mining-induced stress
conditions, rock mass quality, and size of excavations. Ultimately, a conceptual
 13

interpretation of the peak particle velocity (ppv), determined as a function of the

magnitude of a seismic event and distance between the epicentre and damage locations, is
used to assess the performance threshold of ground support systems.

• Chapter 8 – Ground Support Strategy for Burst-Prone Grounds – A ground support

strategy is proposed for underground mines undergoing various levels of mining-induced
seismicity. A threshold based on the largest anticipated event magnitude is used to
determine the level of seismic risk and select an appropriate approach, whether static or
dynamic, toward ground support. The radius of influence of seismic events is assessed.
This parameter is used to prescribe the installation of enhanced ground support within
appropriate distances from identified seismically-active structures. The location and
timing of installation of enhanced support systems is further assessed based on the
mining-induced stress conditions. Ground support recommendations are made and are
supported by the observed ppv-performance thresholds. The improvement of ground
support practice going forward is discussed based on the measured capacity of recently
developed energy-absorbing rock reinforcement elements and the successes and
limitations observed in the performance of the support systems reported in the collected
rockburst data.

• Chapter 9 – Conclusions – The findings of the thesis are summarized and

recommendations for further work are presented.
 14

2 Creighton, Copper Cliff, and Coleman Mines

Creighton, Copper Cliff, and Coleman mines are located in the Greater Sudbury Area, in the
province of Ontario, Canada, about 400 km north of Toronto (Figure 2–1). The three mines are
operated by Vale, a Brazil-based metals and mining company. Both Vale and the Sudbury region
are renowned for their nickel production on the global scale. In 2011, 59.7 thousand tonnes of
finished Nickel was produced out of the Vale Sudbury operations. This result corresponded to
approximately 27.2% of the Canadian production of Nickel (219,612 tonnes) and 3.3% of the
world production (1,826,000 tonnes) (British Geological Survey, 2013).

Figure 2–1 Location of Sudbury with respect to other Canadian cities (Map data ©2015 Google)

This chapter provides the technical background on Creighton, Copper Cliff, and Coleman mines.
The first section provides a brief description of the regional geology of the Sudbury area. Then,
the local geology, configuration of orebodies, and employed mining methods are discussed for
each site. The third section covers the predominant geological structures that affect seismicity
and ground conditions at Creighton, Copper Cliff, and Coleman mines. Finally, the mechanical
 15

properties of rock units and in-situ stresses are presented and the general rock mass quality is
discussed.

2.1 Regional geology of the Sudbury area

Nickel in the Sudbury area is found in magmatic deposits located at the outer rim of the Sudbury
Basin (Figure 2–2). The Basin structure was formed by a meteoritic impact that occurred at the
contact of the early Proterozoic Huronian formation of the Southern Province and the Archean
plutonic rock of the Superior Province. The meteorite impact perturbed the crust and underlying
mantle, leading to the intrusion of the Sudbury Igneous Complex (SIC) along the crater wall
(Dietz, 1964). The crater was filled subsequently by the fallback breccias of the Onaping
formation and the argillites and turbidites of the Onwatin and Chelmsford formations (Cochrane,
1991). The SIC is dated at 1.85 Ga (Naldrett, 1989) and forms an elliptical ring of 60 km long by
27 km wide.

Figure 2–2 Geological map of the Sudbury area showing the location of mine sites including Creighton, Copper
Cliff, and Coleman mines (modified from Roussell & Card, (2009))
 16

Differentiation processes have resulted in the formation of three main zones within the SIC: the
lower zone norites, middle-zone of quartz-gabbro, and upper zone of granophyres (Cochrane,
1991). The copper-nickel sulphide deposits are usually found in the discontinuous sublayer unit
of the SIC located at the base of the lower zone norites. Pyrrhotite, pentlandite, and chalcopyrite
are the main sulphide minerals within the SIC.

The mineral deposits of the Sudbury area are divided into North Range, South Range, Offset, and
Fault-related deposits (Naldrett, 1984). Orebodies at Creighton, Copper Cliff, and Coleman
mines are classified as South Range, Offset, and North Range deposits, respectively. The North
and South ranges are differentiated based on their igneous petrology, the thickness of the norite
and gabbro units, the character of the footwall rocks and the metamorphic history (Andrews &
Courchesne, 2009). The footwall rocks of the North Range are composed of Archean granites,
gabbro and gneiss whereas the footwall rocks of the South Range consist of metasediments and
metavolcanic rocks intruded by the Creighton pluton and the Nipissing diabase. The Copper Cliff
Offset, extruding from the southern part of the SIC, crosscuts the granite of the Creighton Pluton,
the Huronian metavolcanic and metasedimentary formations, and the Nippissing Diabase.

2.2 Local geology and mining methods

2.2.1 Creighton Mine
Creighton Mine is located north of the town of Lively, about 20 km west of Sudbury. The
deposit is located within the Creighton embayment on the outer rim of the South Range of the
SIC. The most prominent geological units at Creighton Mine are (Malek, Suorineni, & Vasak,
2009):

1. Hanging-wall rocks composed of basal norite of the SIC;

2. Sublayer norite (the common host for the ore);

3. Quartz diorite offset dyke (usually barren);

4. Footwall rocks composed of the Paleoproterozoic Creighton granite, which intrudes the
lower Huronian metavolcanic and metasediment rocks.
 17

Most of the ore at Creighton mine is located at the contact between the granite-metagabbro
footwall and the norite hanging-wall (Figure 2–3). Orebodies typically have a high-grade
footwall with a gradational lower grade hanging-wall (Andrews & Courchesne, 2009). The
orebodies at Creighton are classified in three main ore types: disseminated sulphide in sublayer
norite, massive sulphide lenses at the norite and footwall contact, and massive sulphide lenses in
shear zones extending several hundred metres into the footwall (Cochrane, 1991). In Figure 2–3,
these ore types are represented in green, red, and orange, respectively.

Figure 2–3 Cross-section of Creighton Mine (Andrews & Courchesne, 2009) and the associated number of
rockbursts per mine levels from 2000 to 2013

The majority of mining reserves at Creighton Mine are concentrated in the Deep 400, 461, and
the planned 649 orebodies. The most economically important area of the mine, Creighton Deep,
consists of the economic mineralization located below the 6400 level (1,950 m). Currently, the
majority of mining reserves have been depleted down to the 7400 level (2,255 m) in the Deep
400 orebody and to the 7840 level (2,390 m) in the 461 orebody. The 400 orebody is the largest
of Creighton orebodies and accounts for most of the production. The 400 orebody strikes roughly
east-west and dips 70o to the North. At 2,350 m depth, the main 400 orebody breaks into the
 18

West and East parts (Figure 2–4). Lower grade mineralization of the 400 West orebody is
contained in the 400 trough feature hosted in the norite of the SIC. Sulphide mineralization is
concentrated along the basal contact with the granitic and gabbroic footwall rocks. The high-
grade sulphide mineralization of the 400 East orebody enters the footwall to the east of the
contact, thinning along the 1290 shear zone. The 461 orebody is hosted within a Sudbury breccia
unit in the granite and gabbro footwall rocks. The 461 orebody is sub-vertical and strikes toward
North-East, i.e. roughly perpendicular to the 400 orebody (Figure 2–4).

Figure 2–4 Three-dimensional representation of the Creighton Deep 400 and 461 orebodies, looking N-E (Tan,
2003)

Several mining methods have been used in the past at Creighton, e.g. blastholes and cut-and-fill
in the upper areas, and vertical retreat mining. The Deep 400 and 461 orebodies are currently
mined using a top-down, centre-out (or V-shaped) sequence in order to accommodate higher
levels of mining-induced stresses and seismicity. Despite the selection of an appropriate mining
sequence, rockbursts are still a frequent occurrence at Creighton Mine. The collected data
indicate that over 92% of rockbursts occurred within Creighton Deep, i.e. below the 6400 level,
between 2000 and 2013 (Figure 2–3).

The occurrence of rockbursts at Creighton appears to be influenced by the depth, spacing, and
maturity of mine levels. A noticeable increase in the total number of rockbursts is observed as
the mine depth approaches the 7400 level (2,255 m) (Figure 2–3). Open stopes varying from 53
 19

to 60 m height were mined at levels above 2,255 m depth, in the 400 orebody. Below the 7400
level, the spacing between top and bottom sills had been reduced to 40 m in anticipation of
higher stress conditions at greater depths and to better delineate the mineable reserves. Since
2005, the majority of stopes were mined below the 7400 level in Creighton Deep. Mining in the
461 orebody began in 2006. Recognizing the unfavourable orientation of the orebody with
respect to the major principal stress, the mine employed stopes with a design height of 26 m. The
open stopes mined within Creighton Deep as of September 2013 are represented (Figure 2–5).
This figure illustrates the evolution of the stope dimensioning as mining progressed to greater
depths.

Figure 2–5 Open stopes mined in Creighton Deep as of September 2013 illustrated using Map3D

2.2.2 Copper Cliff Mine

The Copper Cliff mine is located about 8 km west of Sudbury, in the town of Copper Cliff. The
mine is located within the Copper Cliff Offset, a structure that extends south from the outer
margin of the SIC. The Copper Cliff Offset begins as a funnel-shaped embayment and thins to a
quartz diorite dyke to the south (Cochrane, 1984). The dyke has an approximate width of 40 m
and dips vertically or steeply west, its strike varying from southeast to south. Along its strike, the
dyke crosscuts the granite of the Creighton Pluton, the Huronian metavolcanic and
 20

metasedimentary formations, and the Nippissing Diabase (Figure 2–6). The dyke is offset by the
Creighton Fault (080o, 75o) over 700 m, the No.2 Mine Cross Fault over 70 m, and the No.1
Mine Cross Fault over 65 m (Figure 2–5). The No.2 and No.1 cross faults strike northeast and
east-west, and dip 55o northwest and 40o north, respectively.

The quartz diorite dyke is the host of several orebodies, all characterized by an elongated and
steeply plunging pipe shape. Orebodies of the Copper Cliff Offset were historically mined by the
Copper Cliff North Mine (Pump Lake, Lady Violet, 205, 191, 178, 175, 138, 120, 114, 100/900,
890, 880) and the Copper Cliff South Mine (880, 865, 860, 850, 830, 810, 800, 790) (Figure 2–
6). Since the closure of the South Mine in 2009, the North Mine is now referred to as the Copper
Cliff Mine.

Figure 2–6 Simplified geological map of the Copper Cliff Offset, showing the location of the mine shafts and ore
deposits of the Copper Cliff North and South mines (Rickard & Watkinson, 2001)
 21

The majority of rockburst occurrences from 2004 to 2013 at Copper Cliff Mine were in the
vicinity of the 100 and 900 orebodies (Figure 2–7). Both orebodies are located in an inclusion-
rich zone within the centre of the quartz diorite dyke (Cochrane, 1984). The country rocks,
around the 100 and 900 orebodies, consist of amphibolites and metasediments. Sudbury breccia
occurs at the contact with the quartz diorite. The dominant geological structures are the 900 cross
fault, olivine diabase dyke, and quartz diabase (trap) dykes. Historical data demonstrated that
these structures were seismically active at several production levels of the mine (Hudyma &
Brummer, 2007).

Figure 2–7 Longitudinal section of the 100, 900, and 890 orebodies, looking west (modified from Rickard &
Watkinson (2001)) and the associated number of rockbursts per mine level from 2004 to 2013
 22

The 100 and 900 orebodies are exploited using vertical retreat mining (VRM) and slot and slash
methods with a bottom-up centre-out sequence. The life-of-mine plan accounts for mining down
to the 5200 level (1,585 m) in the 100 orebody and to the 4200 level (1,280 m) in the 900
orebody. Since the occurrence of a major rockburst in September 2008, the mining sequence has
been adjusted at Copper Cliff by postponing the extraction of the 900 orebody. This strategy
aimed at minimizing the seismic hazard associated with mining on both sides of the Trap Dyke
(Vale, 2010b).

2.2.3 Coleman Mine

The Coleman mine is located within the Levack belt, in the North Range of the Sudbury Basin.
The mineralization is typically associated with discontinuous zones of sub-layer norite beneath
the south-dipping North Range of the SIC. The massive-sulphide ore lenses are associated with
metamorphosed, re-crystallized, and brecciated gneiss and granite breccias (Vale, 2011). The
production consists of the Main, West, 153, and 170 orebodies (Figure 2–8). Rockbursts between
2006 and 2013 were mostly located in the Main and 153 orebodies with fewer incidents in the
recently exploited 170 orebody.

Figure 2–8 Longitudinal section of the Coleman mine, looking North (Vale, 2011) and the associated number of
rockbursts per mine level

Coleman’s Main orebody (MOB) is the vertical extension of the Fraser mine’s #1 orebody. The
boundary between the two mines is located at 1,070 m (3,500 ft) depth and the orebody extends
down to a 1,370 m (4,500 ft) depth. The MOB strikes east-west over a distance of approximately
300 m and dips south at 50° to 60°. The orebody flattens to less than 30° at depth. The ore
 23

thickness varies from 1.5 to 135 m with an average of 60 m. Up until December 2013, post pillar
cut-and-fill was the predominant mining method in the MOB, with open stope mining being used
for sill pillar recovery in the upper part of the orebody (MOB1). The mine is currently
transitioning from cut-and-fill to open stope mining in the lower MOB (MOB2 and MOB3) for
recovering sill pillars (Figure 2–9).

Figure 2–9 The Coleman mine’s Main orebody divided into three mining horizons: MOB1, MOB2, and MOB3

The 153 and 170 orebodies consist of a complex system of narrow copper-nickel-precious metal
veins located within a breccia zone. These two orebodies are primarily recovered using overhand
cut-and-fill, with underhand cut-and-fill being used for recovering sill pillars. Underhand cut-
and-fill is used for most of the sill pillar recovery; open stoping accounts for less than 10% of
mining in the 153 orebody. The 153 orebody is located 820 m northwest of the MOB. It has a
strike length of 365 to 460 m and extends from 1,100 to 1,550 m depth (3,600 to 5,080 ft). The
host structural zone of felsic gneiss, granite, and Sudbury breccia, strikes east-west and dips from
40o to 50o but flattens to 30o to 35o at 1,430 m (4,700 ft) depth. Individual ore veins vary in dip
from 30o to 70o and in thickness from a few centimetres up to 25 m.

The 170 orebody is located about 450 m southeast of the 153 orebody, east of the Bob’s Lake
Fault, and is interpreted to be the faulted extension of the 153 orebody (Vale, 2011). The 170
orebody is relatively flat with the dip increasing to the north-northeast. It extends vertically from
 24

1,490 to 1,770 m (4,900 to 5,800 ft) depth and continues up-plunge outside of Vale’s claim,
where it has been mined previously. The footwall and hanging-wall of the 170 orebody are
composed of granite gneiss and grey granite interbanded with mafic gneiss, granite breccia and
Sudbury breccias.

2.3 Predominant geological structures

Large-scale geological structures can have an important influence on ground conditions and
seismicity in deep underground mines. At Creighton Mine, for example, an increase in the level
of seismic activity observed between 2001 and 2003 was attributable to the mining of a footwall
extension of the 400 orebody which generated stress concentrations in the vicinity of the Plum
shear zone, one of the most seismically-active structures at the mine (Morissette, Hadjigeorgiou,
Punkkinen, & Chinnasane, 2014). The predominant geological structures at Creighton, Copper
Cliff, and Coleman mines are reviewed in this section.

2.3.1 Creighton Mine

Shear zones, also referred to as late-stage faults (Cochrane, 1991; Coulson, 1996), are the most
prominent geological structures identified across the footwall and orebodies of Creighton Deep.
These structures vary in thickness from centimetres to tens of metres and commonly exhibit a
mylonitic textural appearance (Vale, 2012a). Within the Creighton granite, shear zones are rich
in biotite whereas in metagabbro, they are typically altered with chlorite. Shear zones are also
frequently associated with quartz veins and minor calcite (Snelling, Godin, & McKinnon, 2013).
Twelve shear zones have been mapped within Creighton Deep and the majority of them have
been intersected at 2,255 m depth, on the 7400 level (Figure 2–10). Four families of shear zones
are identified (Snelling et al., 2013):
• southwest-striking shear zones of the 118 system;
• northwest-striking footwall-family shear zones;
• east-west-striking 1290/400E shear zones;
• and splays between faults of the 118 system.

Shear zones at Creighton Mine are described as being healed (Coulson, 1996). Consequently,
little evidence exists for supporting their seismogenic reactivation within the mine (Snelling et
al., 2013). Yet, shear zones are often perceived as responsible for the generation of large-
 25

magnitude fault-slip events by the mine operating personnel. Coulson (1996) suggested that
seismicity was a result of mining-induced stresses and that it was primarily related to the joint
fabric, rather than to the presence of shear zones. Through the analysis of seismic source
parameters and the application of fault plane solutions and stress tensor inversions, Snelling et al.
(2013) demonstrated that slip along mine-scale shear zones was not a dominant failure
mechanism within Creighton Deep. Alternatively, cracking of intact rock and slip along existing
fractures was suggested as a plausible seismogenic mechanism.

Figure 2–10 Plan view of Creighton Mine’s 7400 level showing accesses to the 400 orebody and prominent shear
zones (Snelling et al., 2013)

2.3.2 Copper Cliff Mine

The predominant geological structures at Copper Cliff Mine are the 900 cross fault, the quartz
diabase (trap) dykes, and the olivine diabase dyke (Figure 2–11). These structures have been
seismically active on several mine levels (Hudyma & Brummer, 2007).

The 900 orebody cross fault cuts through both the 100 and 900 orebodies striking at 100o and
dipping at 45o to the north. The fault intersects the south end of the 100 orebody on the 2430
level and the north end on the 3000 level, offsetting the quartz diorite dyke by up to 30 m. The
 26

fault material is described as strongly sheared with black biotite schist and minor carbonate mud
gouge (Vale, 2010a). The thickness of the 900 OB Cross Fault varies from approximately 3.6 m
to 4.6 m. Its total width of influence, however, is of approximately 15 m given the occurrence of
semi-vertical joint systems located up to 6 m on each side of the fault. Ground control problems
in the 100 orebody have been attributed to the presence of the 900 OB cross fault. Stopes had to
be abandoned due to excessive dilution on the 3000 level and the mining sequence had to be
reviewed from the 3050 to 3400 levels due to rockbursting (Vale, 2010a).

Figure 2–11 Plan view of Copper Cliff Mine’s 3400 level showing the 100 and 900 orebodies and the predominant
geological structures on site

The trap dykes are very fine to fine grained, strike westward and dip steeply toward north. A
seismically-active trap dyke is located between the 100 and 900 orebodies, and crosscuts the 900
orebody below the 3500 level. The dyke’s continuous shape changes at depth where it splits into
two limbs on the 3700 level. The trap dyke generated significant levels of seismic activity
between the 3000 and 3400 levels where it plots nearby the open stopes of the 900 orebody
(Hudyma & Brummer, 2007). The rockburst data collected at Copper Cliff mine further indicate
 27

that seismicity and strain bursting occurred on the 3710, 3800, and 4050 mine levels at locations
where development headings intersected the limbs of the trap dyke.

An olivine diabase (O.D.) dyke intersects the 100 orebody from the 3000 level to the 3500 level.
The dyke is medium to coarse grained, trend northwest, and dip steeply toward north. It is
intersected by mine development from the 2400 level to the 4200 level (Hudyma & Brummer,
2007). From 2004 to 2013 at Copper Cliff mine, the trap dyke and the 900 cross fault had a much
greater influence on rockburst occurrences, as compared to the O.D. dyke.

2.3.3 Coleman Mine

The main large-scale geological structures in the MOB area are the Fraser No.2 fault and the
Lunch Room fault. The Fraser No.2 Fault dips at 60o towards east with a dip direction of 080o.
The structure does not intersect any of the currently active mining areas; however, it has
generated fault-slip events of magnitude smaller than 2.0 MN during development and bulk
mining on the northeast side of MOB1 (Vale, 2011). A splay of the Fraser No.2 Fault was also
reported to generate bad ground conditions during mining on the east end of upper MOB1 (Vale,
2011). The Lunch Room fault dips at 50o to 70o towards east with a dip direction that ranges
from 070 to 080o. The fault is 5 cm to 1 m thick and slightly infilled with clay. The influence
zone of the Lunch Room fault ranges from 2 to 5 m, given the occurrence of parallel joints. Mine
development drives through the fault on nearly every cut of MOB2 and MOB3 (Figure 2–9).
Additional ground control and support measures have been required over time in the vicinity of
the Lunch Room fault, e.g. blasting shorter rounds and supporting excavations using rebars and
cable bolts.

The 153 orebody is intersected by a barren olivine diabase dyke from 1,340 to 1,555 m (4,400 to
5,100 ft) depth (Figure 2–8). The dyke dips at 70o to 75o with a dip direction of 065o and varies
in thickness from 6 to 60 m. The dyke’s elastic properties are similar to those of the host rock.
Ground control records suggest that the dyke did not have significant influence on the stability of
the excavations (Vale, 2011). Since 2012, increasing levels of seismic activity have been noticed
in a granite-gneiss pillar delimited by the olivine diabase dyke to the west-southwest and two
bornite veins to the north and east-south-east (Figure 2–12). The two bornite veins were
identified after noticing frequent seismic activity in the pillar area below the 4700 level
 28

(1,433 m). The veins were suspected to act like faults by the ground control personnel at
Coleman.

Figure 2–12 Isometric view of Coleman Mine’s 4700 and 4810 levels showing the two bornite veins mapped in the
153 orebody and seismic events associated with rockburst ME-168 on November 23, 2012

2.4 Geotechnical data

In this section, the intact rock properties and rock mass rating (RMR) values (Bieniawski, 1989)
are provided for the main geological units at Creighton, Copper Cliff, and Coleman mines
(Table 2–1). Those design values are used internally at each site. Values of uniaxial compressive
strength (UCS), Young’s modulus (E), Poisson’s ratio (υ), and RMR employed at Creighton
were obtained from Coulson (1996). In the calculation of the RMR, a joint orientation reduction
factor of -12 was assumed to represent the worst-case scenario of very unfavourable jointing
(Coulson, 1996). The far-field stress regime is further provided for each mine (Table 2–2). The
ground conditions at Creighton, Copper Cliff, and Coleman mines, are characterized as
moderately fractured to massive, based on the RMR (Figure 2–13). Based on this conceptual
diagram, the stress conditions at the three mines range from intermediate to high in-situ stress. At
Creighton, Copper Cliff, and Coleman mines, the observed ground and stress conditions suggest
the potential for brittle rock mass failure and movement of blocks (Martin, Kaiser, & McCreath,
1999).
 29

Table 2–1 Rock mass data for various geological domains at Creighton, Copper Cliff, and Coleman mines

Mine Orebody Structural domains UCS E υ RMR (1989)

(MPa) (GPa)
Creighton All Granite and gabbro 240 57.0 0.267 60-73
Norite 210 72.2 0.277 57-74
Massive sulphide 133 72.1 0.256 56-70
Copper 100 Metasediments 160 60 0.22 60 - 75
Cliff Amphibolite 220 60 0.22 60 - 75
Quartz diorite 140 40 0.22 60 - 75
900 Metasediments 160 60 0.22 60 - 66
Amphibolite 220 60 0.22 60 - 66
Quartz diorite 150 40 0.22 60 - 66
– Trap dyke 220 74 0.22 –
– Olivine diabase dyke 140 54 0.22 –
Coleman All Massive sulphide 130 20 0.30 55-75
Granite-gneiss 240 60 0.20 80-90
Mafic gneiss 280 65 0.20 75-90
Granite breccia 210 55 0.20 70-80
Levack/Sudbury breccia 300 55 0.20 70-80
Olivine diabase dyke 200 50 0.22 75-90

Table 2–2 Far-field stress regime at Creighton, Copper Cliff, and Coleman mines

Principal stress Trend Plunge Magnitude [MPa] Magnitude [psi]

(Z = depth in m) (z = depth in ft)
Major (σ1) 270o 10o 10.35+0.0421*Z 1501+1.86*z
Creighton

Intermediate (σ2) 000o 00o 8.69+0.0330*Z 1260+1.46*z

Minor (σ3) 090o 80o 0.0290*Z 1.28*z
Major (σ1) 078o 10o 10.82+0.0407*Z 1570+1.80*z
Copper
Cliff

Intermediate (σ2) 168o 00o 8.69+0.0326*Z 1260+1.44*z

Minor (σ3) 258o 80o 0.0292*Z 1.29*z
Major (σ1) 127 0 10.86+0.0407*Z 1575+1.80*z
Coleman

Intermediate (σ2) 217 0 8.69+0.0326*Z 1260+1.44*z

Minor (σ3) 0 90 0.0292*Z 1.29*z
 30

Figure 2–13 Potential for tunnel instability and brittle rock mass failure at Creighton, Copper Cliff, and Coleman
mines based on the classification proposed by (Martin et al., 1999)

2.5 Summary
In this chapter, Creighton, Copper Cliff, and Coleman mines were introduced in their geological
and mining context. The information was complemented by a review of the large-scale
geological structures present at each mine. These structures were often associated with the
generation of mining-induced seismicity over the time period covered by this thesis. The chapter
concluded with the presentation of the mechanical properties of rock units and in-situ stresses,
 31

and with the assessment of the potential for brittle rock mass failure, as a function of the stress
conditions and rock mass quality.

This chapter presented background information on the ground and stress conditions, which is
valuable in order to differentiate the mines that are covered by this thesis and assess their
rockburst potential. The influence of mine depth on rockburst occurrences was demonstrated by
comparing Creighton Mine to two shallower mines: Copper Cliff and Coleman. At Creighton,
there was a recognizable influence of mine depth and maturity of mine levels on the number
rockburst occurrences (Figure 2–3). The collected data suggested that the mine has managed the
increasing number of rockbursts per mine level at great depth, i.e. below the 7400 level. There
was a correlation between the start of the observed decrease in the number of rockburst per mine
level and the depth at which the mine started employing stopes of smaller height. The decrease in
the number of rockburst per mine level below the 7400 level was observed despite the mining of
two orebodies at greater depths (Figure 2–5). The influence of mining sequences and ground
support practice on the frequency and severity of rockbursts at Creighton, Copper Cliff, and
Coleman mine is examined in greater detail in the following chapter.
 32

3 Mining Sequences and Ground Control Practice and

their Influence on Rockburst Frequency and Severity

In the collected rockburst data, the severity of each rockburst was assessed using visual estimates
of the displaced tonnage reported by the ground control personnel at the time. The evolution of
the frequency and severity of rockbursts can be represented by the cumulative displaced tonnage
over time, given that the complete rockburst history has been collected over the studied time
period (Figure 3–1).

Figure 3–1 Cumulative tonnage displaced over time as a result of rockbursts at Creighton, Copper Cliff, and
Coleman mines and some of the associated large magnitude events (in Nuttli)

The frequency of large magnitude seismic events (> 2.0 MN) and the associated damage have
been much higher at Creighton, as compared to both Copper Cliff and Coleman mines. This is
reflected by the steady increase of the cumulative displaced tonnage at Creighton (Figure 3–1).
Copper Cliff and Coleman mines have been occasionally affected by very severe rockbursts,
which generated displacements of substantial amounts of material. However, at Copper Cliff and
Coleman, the rate of displaced tonnage has been observed to be relatively low between severe
 33

events. Such distinction in the evolution of the frequency and severity of rockbursts might reflect
the influence of the increasingly high-stress conditions faced at Creighton as the mining
progressed to greater depths. The trend observed at Creighton Mine further suggests a gain of
experience in managing high-stress and burst-prone conditions over time. Since 2004, Creighton
has experienced the least amount of rock displaced due to rockbursts, despite being the deepest
of the three mines and the one that faced rockbursts the most frequently. In the following
sections, the evolution of the frequency and severity of rockbursts at Creighton, Copper Cliff and
Coleman mines is analysed by exploring correlations with the evolution of mining and ground-
support practice at the three sites.

3.1 Creighton Mine

3.1.1 Stope design and seismic risk management
There has been an evolution in the stope design at Creighton over time. Part of this evolution
consisted in reducing the spacing between top and bottom sills in anticipation of higher stress
conditions and to better delineate the mineable reserves below the 7400 level. Reducing the stope
height at Creighton has likely contributed in preventing an escalation of seismic activity as the
mine progressed to greater depths. This was reflected by the evolution of the rate of seismicity,
which was assessed from January 2000 to September 2013 using the magnitude-time history
analysis technique (Figure 3–2). This technique is described in detail by Hudyma and Potvin
(2010).

On the magnitude-time history chart, a relatively constant rate of seismicity was observed
between two labour interruption periods, which lasted from April to August 2003 and from July
2009 to July 2010, respectively. These two periods are identified by grey-shaded areas, within
which decay of seismic activity is representative of production interruption. Between these two
periods, a slight decrease in the rate of seismicity was noticed starting late 2005 and could be
attributed to the mining of smaller stopes in the 400 orebody. Starting in 2005, the majority of
open stopes developed within Creighton Deep were located below the 7400 level.
 34

Figure 3–2 Magnitude-time history analysis performed using seismic data of magnitudes 0.6 MN and greater

On the other hand, a noticeable increase in the rate of seismicity at Creighton occurred starting
September 2001, and accelerating in September 2002. Seismicity during this period was
exacerbated by mining a footwall extension of the 400 orebody between the 7000 and 7200
levels. Mining of these larger stopes generated stress concentrations in the vicinity of the
seismically active Plum shear zone. This was discussed in greater detail by Morissette et al.
(2014).

The magnitude-time history analysis demonstrated that, although mining has influenced
seismicity at Creighton Mine, the seismic hazard has been adequately managed. This was
reflected by the relatively constant rate of mining-induced seismicity over time, despite mining at
greater depth and within the unfavourable stress conditions associated with the 461 orebody. The
magnitude-time history analysis corroborated the engineering decision of adjusting the stope
design in anticipation of higher stress environments in the 400 and 461 orebodies. This case
study exemplified that in deep underground mines, the role of engineering is not to eliminate
seismicity but to manage it.
 35

3.1.2 Evolution of ground support systems

Ground support practices at Creighton Mine evolved as mining progressed to greater depths. The
mine began to install modified cone bolts (MCB) in conjunction with #0 gauge straps as part of
its “enhanced” ground support system in December 2004. In February 2005, a 2.0 m long, post
galvanized, 46 mm diameter friction set (FS-46) was introduced as part of the wall support
system. This bolt replaced the 1.7 m long, 35 mm diameter friction set and its predecessor, the
39 mm version (Punkkinen & Yao, 2007). As of November 2006, the minimum ground support
standard consisted of a diamond pattern of 2.4 m long resin rebars and mechanical bolts in the
back and 2.0 m long FS-46 in the walls. Reinforcement elements were installed on a 1.2 m x
1.5 m diamond pattern, overlapping #4 gauge galvanized welded-wire mesh down to floor level.
Shotcrete was frequently applied over the mesh to provide further surface support. In areas of the
mine susceptible to rockbursts, the support standard was enhanced by adding 2.4 m long
modified cone bolts and #0 gauge straps (Malek, Trifu, Suorineni, Espley, & Yao, 2008)
(Figure 3–3a).

The use of mechanical bolts at Creighton Mine was discontinued in June 2010 in response to
corrosion issues and inadequate performance under dynamic loads. In September 2010, MCB33
(modified cone bolts for installation into 33 mm diameter boreholes) became part of the primary
support system below the 7810 level, in areas where enhanced support is prescribed. MCB33 are
installed along with resin rebars on a dense, 1.2 m x 1.0 m, diamond pattern. #0 gauge mesh
squares (0.3 m) are installed in order to enhance the connection between the #4 gauge screen and
the reinforcement elements and to protect the surface support from being damaged by the cutting
action of the plates (Figure 3–3b). The support system in the walls was enhanced further with the
addition of modified cone bolts and #0 gauge squares or straps in areas susceptible to rockbursts.
Shotcrete, however, is no longer part of a minimum support standard as it was observed that it
could not effectively manage higher dynamic loads and often resulted in cracking and spalling,
which repeatedly called for rehabilitation. Shotcrete, however, is still used regularly on lower
walls and pillar noses of bottom sills to prevent damage caused by production equipment or to
rehabilitate damaged mine openings. Large diameter inflatable bolts are often used in
rehabilitation and in areas where the presence of highly fractured rock mass does not facilitate
the use of resin-grouted bolts.
 36

(a) (b)

Figure 3–3 Evolution of the support systems installed in areas susceptible to rockbursts at Creighton Mine with the
support system employed from (a) 2006 to 2010 and (b) 2010 to date

Ground control personnel at Creighton Mine monitored the performance of individual

reinforcement and surface support elements over time in order to identify limitations in the
employed support systems. Following observations of rebar failures in the threaded portion of
the tendon through 2011 and 2012, the rebar nut was modified with a spherical seat to
accommodate a dome washer plate (Vale, 2012b). The dome assembly provides for effective
installation of the rebar without risking damage to the threaded portion of the bolt upon
installation in often unavoidable angular orientations. For several years, the mine successfully
employed FS-46 bolts with a crimp design bushing. As the production front adversely loaded
pillars with the mining progression to increasing depths, observations of crimp failures began to
surface in 2011-2012. These repeated failures suggested the limitation of this bolt configuration
for the high-stress bottom sills of the 461 orebody and motivated the adoption of a welded-ring
design. Pull-tests conducted on site demonstrated that the capacity of the ring was enhanced from
10-11 tons to 17-18 tons with the welded design (Vale, 2012b). Furthermore, as part of
continuous efforts to explore new support strategies, the mine was, as of September 2013,
investigating the performance of the D-Bolt on its 7910 level.

Adjustments to the ground support systems employed under burst-prone conditions at Creighton
Mine were adopted pro-actively or in reaction to the identification of ‘weakest links’, which
affected the performance of the system. The chronology of this evolution correlates with the
observed decline in the rate of rockburst damage over time (Figure 3–4).
 37

Figure 3–4 Correlations between the evolution of ground support systems and the frequency and severity of
rockbursts at Creighton Mine

3.1.3 Cable-bolting of large excavations

Cable bolting at Creighton Deep is performed systematically and in a timely manner, prior to the
installation of mine services, in all intersections where development headings are greater than
5 m x 5 m. In practice, the mine cable-bolts all excavations having a span greater than 7.3 m,
which corresponds to 3 times the length of the primary reinforcement elements, as per the
support standard reviewed in June 2005. Cable bolts are also employed where geological
structures, high walls, dynamic-loading conditions, or ground conditions warrant, upon request
of the ground control department (Vale, 2012b). Double 16 mm (5/8”) plated cables are typically
installed on a 2.1 m x 2.1 m pattern. For excavation spans smaller than 12 m, the pair of cement-
grouted cables installed within 5 cm diameter drill holes consists of a 6.4 m long bulged cable
and a 5.5 m long plain strand cable.

A systematic review of rockburst occurrences at Creighton Mine indicated that in large

excavations (span or wall height > 7.3 m), the installation of cable bolts tended to enhance the
overall performance of the employed support system (Table 3–1). This was reflected by the
increased severity of damage, represented by the reported displaced tonnage, in areas where
cable bolts were not part of the support system. The advantage of using cable bolts can be
 38

attributed to their capacity of tying the support back to stable ground due to the provided
additional length. It may, furthermore, be attributed to the softer behaviour of cable bolts as
opposed to other reinforcement elements. Bulged cables provide an immediate stiff load
response, which is desirable in highly fractured ground, whereas the plain strand cables are
capable of withstanding moderate dynamic-loading conditions (Hutchinson & Diederichs, 1996).

Table 3–1 Occurrences and severity of rockbursts in large excavations characterized by a span or wall height
greater than 7.3 m at Creighton Mine

2000-2005 2006-2013
Reported Number of Tonnage displaced Number of Tonnage displaced from
installation of rockbursts from behind the support rockbursts behind the support
cable bolts (tonnes) (tonnes)
Yes 5 227 11 73
No 9 2021 4 122

3.1.4 Summary
The majority of changes in stope dimensions and support practice at Creighton Mine were
initiated between 2004 and 2005. Adjustments to the enhanced ground support system employed
in areas susceptible to rockbursts started in late 2004 and early 2005 with the introduction of the
MCB and the FS-46. Further changes were adopted in 2010 and 2012 aiming at eliminating
‘weakest links’ in the employed ground support systems. The evolution of the frequency and
severity of rockbursts at Creighton, depicted in Figure 3–4, indicates strong correlations with
these adjustments. As ground support practice was adjusted for deep and high stress conditions,
with the introduction of yielding reinforcement elements and a more consistent practice of
installing cable bolts in large openings, a significant decline in the rate of rockburst damage was
observed.

3.2 Copper Cliff Mine

Copper Cliff and Coleman mines, although not being as deep and seismically active as
Creighton, have encountered very severe, although sporadic rockbursts since 2004 and 2006. At
Copper Cliff Mine, the most severe rockbursts occurred on March 25 and September 11, 2008.
 39

3.2.1 March 25, 2008 rockburst

On March 25th, 2008, a recorded 2.9 MN seismic event generated over 635 tonnes of displaced
material on the 3880 top sill level of the 900 orebody. The rockburst was associated with the
mining of the 9281 stope. The stope was located between the two limbs of the trap dyke. The
2.9 MN event occurred following a production blast and plotted in the southern part of the dyke,
in the near vicinity of the stope (Figure 3–5).

Figure 3–5 Location of the 9281 stope and the 2.9 MN event which occurred following a production blast on March
25, 2008

The damage areas in the vicinity of the 9281 stope and the 3910 sill drift were supported using a
combination of 1.8 m long mechanical bolts and rebars in the back and 1.7 m long 39 mm
friction sets (FS-39) in the walls, with the possibility of mechanical bolts having been installed as
well. The surface support consisted of #6 welded-wire mesh overlapped with plain shotcrete. In
intersections, 6.4 m long cable bolts were installed on a 2.1 x 2.1 m pattern. Pictures of the most
severely damaged areas revealed a complete collapse of the surface support and failure of several
friction sets and mechanical bolts found in the muck pile (Figure 3–6).
 40

(a) (b)

Figure 3–6 Damage to the 9280 top sill on the 3880 level of the 900 orebody at Copper Cliff Mine, triggered by a
recorded 2.9 MN seismic event on March 25, 2008

3.2.2 September 11, 2008 rockburst

The September 11, 2008 rockburst was the result of a series of 10 seismic events that ranged
from 1.2 to 3.8 MN and occurred from 07:21 to 08:06 (Yao, Chinnasane, & Harding, 2009). The
series of events was triggered by the crown blast of the 94561 stope in the upper 100 orebody
(3050 to 3200 level) at 07:21. The seismic events resulted in damage to mine excavations from
the 2700 level down to the 3710 level. The rockburst displaced an estimated total of 2,100 tonnes
as the most prominent damage mechanism was interpreted as seismic shakedown due to the
3.8 MN event (Suorineni & Vasak, 2008). The most severe damage, estimated to 1,360 tonnes,
occurred in the mine ramp between the 3500 to 3550 levels (Figure 3–7a). The reported depth of
failure extended far beyond the primary reinforcement, varying from 3 to 6 m. The other three
most severely damaged areas consisted of the section of the ramp between the 3000 to 3050
levels (Figure 3–7b), the return air drift on the 3500 level, and the 3710 level footwall drift. The
damage to mine excavations and support systems in these three areas was localized at the
intersection with the trap dyke and was estimated at 181, 363, and 91 tonnes, respectively.

The ramp, as well as the majority of excavations affected by the September 11, 2008 rockburst,
were supported using a diamond pattern of 1.8 m long mechanical bolts and rebars in the back
and 1.8 m long mechanical bolts in the walls. The surface support in the damage locations
generally consisted of #6 welded-wire mesh. Plain shotcrete, however, was applied over the
mesh on the 3200 (Figure 3–8) and 3710 levels. Although the damage was typically less severe
 41

in areas where shotcrete was applied, the employed support system generally did not perform
satisfactorily against seismic shaking during the September 11, 2008 rockburst (Suorineni &
Vasak, 2008).

[ft]

(a)

(b)

Figure 3–7 Damage to the Copper Cliff Mine ramp as a result of a 3.8 MN seismic event on September 11, 2008:
(a) cross-section of the ramp between the 3500 and 3550 levels and (b) photography of the 181 tonnes
of material displaced where the ramp intersects the Trap Dyke between the 3000 and 3050 levels
 42

Figure 3–8 Significant shotcrete cracking as the installed support system in the main intersection of Copper Cliff’s
3200 level contained the bulk of the broken material on September 11, 2008

3.2.3 Modifications to support practice and mining sequence at Copper

Cliff Mine
The significant levels of rehabilitation caused by the September 11, 2008 rockburst prompted a
revision of the support practice at Copper Cliff Mine. Since 2008, many of the support elements
successfully used at Creighton Mine have been introduced on site. At the present time, the
minimum ground support standard employed at Copper Cliff consists of #4 welded-wire mesh
installed with 1.8 or 2.4 m long resin rebars in the back, depending on the size of the opening,
and 1.7 m long 39 mm friction sets in the walls. The reinforcement is installed on a 1.2 x 1.5 m
diamond pattern. When burst-prone conditions are anticipated, the 39 mm friction sets are
replaced by 46 mm friction sets in the design. Shotcrete is applied over the mesh and 2.4 m long
MCB are installed in conjunction with #0 gauge straps on a 1.2 x 1.8 m pattern (Chinnasane,
Yao, Landry, & Paradis-Sokoloski, 2012). Since 2008, the mining sequence has been adjusted by
postponing the extraction of the 900 orebody in order to minimize the seismic hazard associated
with mining on both sides of the Trap Dyke (Vale, 2010b). Finally, preconditioning of rock
masses became standard practice when developing in the vicinity of the Trap Dyke.

From a ground control view, the use of de-stress blasting in development headings and
adjustments to the mining sequence and ground support systems have been beneficial to Copper
Cliff Mine. Since the September 11, 2008 events, only eight rockbursts have occurred at the
 43

mine. These rockbursts resulted in 91 tonnes of cumulated displaced rock material from nine
mine locations (Figure 3–1). Among these damage locations, only four withstood damage to the
installed support system, i.e. in the remaining locations, the broken material was displaced from
unsupported areas such as lower walls or development faces. Since 2008, the mine has been able
to significantly reduce the rockburst hazard associated with production blasting. In effect, six of
the eight rockbursts since 2008 were associated with development activities and three of them
occurred while progressing through the Trap Dyke.

3.3 Coleman Mine

At Coleman Mine, the most severe series of rockbursts occurred from September 2010 to April
2011. The three rockbursts that were the most damaging to the support are reviewed in this
section.

3.3.1 September 24, 2010 rockburst

On September 24, 2010, a 2.6 MN seismic event displaced approximately 181 tonnes from two
accesses to the Block 2 of the 153 orebody, on the 4700 mine level (Figure 3–9a). The seismic
event was located in the footwall of the 153 orebody, approximately 55 m from the resulting
damage. About 172 tonnes were displaced from the back of the 11/12 access near the
intersection, at the location of a narrow bornite stringer.

The 2.4 m long rebars and mechanical bolts installed in the back of both the Cut 11/12 and Cut
9/10 accesses were heavily corroded (Figure 3–9b,c). Consequently, the bolts were not effective
in holding the damaged ground. The #6 gauge mesh-reinforced shotcrete was also severely
damaged during the event. Cable bolts installed in the back of the intersections, however, were
very effective in preventing further damage to extend outside of the accesses.
 44

Figure 3–9 Damage associated with a 2.6 MN event at Coleman mine on September 24, 2010 located on (a) the
4700 level, more specifically in (b) the cut 11/12 access and (c) 9/10 access

3.3.2 November 20, 2010 rockburst

An estimated 360 to 450 tonnes was displaced from the 4320-3 Access on the 4400 level of the
153 orebody on November 20, 2010. A development blast had been fired at 05:07 that day in the
4320-3-Access Footwall Drift and had triggered a recorded 2.9 MN seismic event, which plotted
in the vicinity of the blast. The 2.9 MN event occurred at 05:20 and was followed two minutes
later by a 2.0 MN event which plotted in the vicinity of the ore contact. The succession of events
therefore suggested that the 2.9 MN event had triggered a slip along the contact. The damage was
located in the access to the Cut 12, at the intersection of the ore/footwall contact, about 45 m
south-east of the blast. Most of the broken material was displaced from the back of the
 45

excavation and extended up to 3 m deep, beyond the 2.4 m length of the installed resin-grouted
rebars (Figure 3–10).

Figure 3–10 Major seismic events and differential stresses (MPa) associated with the November 20, 2010 rockburst
in the 4320-3 Access to the Cut 12 of the 4400 mining horizon

Numerical elastic stress modelling indicated that the footwall drift where the development blast
had been fired was likely undergoing stress changes due to mining of the sill pillar between the
4400 and 4250 mining horizons. Greater stress concentrations were located in the immediate
footwall of the 153 orebody, at the Cut 12 elevation (Figure 3–10). It is therefore likely that,
prior to the 2.9 and 2.0 MN seismic events, the rock mass at the damage location was highly
fractured. The mining-induced seismicity observed the morning of November 20, 2010 would
have, consequently, contributed in shaking the broken material and resulted in exceeding the
load-bearing capacity of the installed ground support system. A rehabilitation plan released after
the event requested the installation of MCB or Yielding Swellex in conjunction with #0 gauge
straps as well as a second pass of cable bolts in the back of the Cut-12 intersection.

3.3.3 April 6, 2011 rockburst

Mining of the 7760 secondary pillar in the narrow west end of the MOB1 generated extensive
seismic activity in late 2010 and early 2011. The first and final (crown) blasts in this stope were
fired on December 17, 2010 and April 4, 2011, respectively. During this period, five seismic
 46

events were recorded with a magnitude greater than 2.0 MN. Rockbursts have occurred on the
3511 top sill level, in the vicinity of the 7760 stope, on January 26, March 18, and April 6, 2011.
These rockbursts were associated with 2.9, 3.4, and 3.7 MN seismic events, as reported by the
GSC. The April 6, 2011 rockburst was the most severe of the three events, resulting in a total of
about 2,360 tonnes displaced from six stope accesses on the top sill level. Given the orientation
of the major principal stress, roughly perpendicular to the trend of the MOB in this area, high
stress conditions were observed within the 7760 stope and generated high confinement on the
ore/footwall contact (Figure 3–11). It is our interpretation that mining of the 7760 stope
contributed to ‘unclamping’ this discontinuity, which in turn resulted in the occurrence of large
magnitude fault-slip events.

Figure 3–11 View from the hanging-wall of the MOB showing the location of the 3.7 MN event and the confinement
on the ore/footwall contact generated by the major principal stress (MPa) prior to mining of the 7760
stope, using Map3D

The 3.7 MN event plotted at the footwall contact of the MOB and occurred at 02:34 on April 6,
2010, 45 hours after the crown blast. It had been reported that the blast did not break through the
crown and that a 10 m thick pillar remained at the top of the stope. Consequently, it was
suspected that the footwall of the 7760 stope was still under high stress at that time. At 17:06 on
April 5, in the 07 Slot, a development blast was fired at the ore/footwall contact, about 60 m
north-east of the 7760 stope. This development blast has most likely triggered the fault-slip event
along the ore/footwall contact, which was characterized by a 3.7 MN magnitude. Following the
 47

3.7 MN event, seismicity migrated toward the west abutment of the MOB1, indicating that
stresses had been diverted from the 7760 stope area (Sampson-Forsythe, 2011). A chronology of
the events is presented (Figure 3–12).

Figure 3–12 Chronology of events associated with the April 6, 2011 rockburst (Sampson-Forsythe, 2011)

The excavations, where damage resulting from the 3.7 MN event was located, were characterized
by a 6.1 to 8.2 m span. Furthermore, the ground was highly fractured due to the presence of high-
stress levels and the occurrence of severe seismicity during the mining of the 7760 stope. The
ground support system installed at the time of the event consisted primarily of 2.4 m and 1.8 m
long resin rebars, in the backs and walls, respectively, and #6 gauge welded wire-mesh. In the
7760 access, 7.3 m span, the support was enhanced with the installation of 2.4 m long MCB and
#0 gauge straps. On April 6, 2011, the majority of damage extended beyond the length of the
reinforcement. The installation of cable bolts, in these relatively wide accesses, could have
limited the consequences of seismic shaking, prevented unravelling of the fractured rock mass,
and resulted in improved performance of the support systems.

Following the event, the installation of enhanced support was prescribed in the west abutment of
the MOB1 in order to manage high-stress levels and promote the stability of mine openings in
this area, as mining of the sill pillar progresses to the west (Sampson-Forsythe, 2011). This
 48

decision was prompted by the migration of seismicity toward the west abutment, which was
indicative of stress migration following the mining of the 7760 stope.

3.3.4 Adjustments to support practice at Coleman Mine

Since the April 6, 2011 3.7 MN event, there has been a significant reduction in excavation
damage due to rockbursts at Coleman Mine (Figure 3–1). This can be partly attributed to the
introduction of a yielding support system. The current practice in burst-prone ground conditions
consists of enhancing the primary support system (composed predominantly of resin rebars and
#6 welded-wire mesh) using #0 gauge straps and either Yielding Super Swellex or D-Bolts. The
Yielding Super Swellex is usually preferred at Coleman due to its ease of installation, in areas
where (a) the ground is significantly fractured in the immediate vicinity of excavations and
seismic shakedown is anticipated, (b) older excavations have previously experienced large
magnitude seismic events, and (c) excavations have a shorter service life. The D-Bolt and #0
gauge straps are used in newer development headings and are installed immediately after the
installation of the primary support. The demonstrated performance of the support systems at
Coleman from April 2011 to September 2013 corroborates the adjustments made to the support
practice during this time period (Figure 3–1).

3.4 Three mines, three stories

The evolution of the frequency and severity of rockbursts was reviewed for Creighton, Copper
Cliff, and Coleman mines. Creighton Mine provided an excellent example of the evolution of
ground support systems as mining progressed to higher stress environments. Correlations were
identified between the improved performance in dealing with dynamic-loading conditions at
Creighton and changes in mining and ground support practice since 2005. Significant mining-
induced seismicity is a more recent occurrence at Copper Cliff and Coleman mines. This is
predominantly attributed to the maturity of the mines and the mining of the sill pillars of the
100/900 orebodies at Copper Cliff and the narrow sill pillars of the 153 orebody at Coleman. The
high severity rockbursts of 2008 at Copper Cliff, and 2010-2011 at Coleman necessitated an
immediate intervention in the ground support systems as opposed to the continuous evolution
observed at Creighton. Consequently, lessons from Creighton Mine provided a useful template.
 49

This section of the thesis highlighted the importance of addressing both ground control and
production concerns in designing mining sequences under high-stress conditions. The occurrence
of the September 11, 2008 rockburst at Copper Cliff and the April 6, 2011 rockburst at Coleman
were arguably attributable to issues related to the extraction sequence. Mining both the 100 and
900 orebodies on each side of the Trap Dyke at Copper Cliff contributed to the 3.8 MN rockburst.
At Coleman, the 7760 stope was used as a secondary pillar in order to allow the mining of open
stopes west of the post pillar cut-and-fill area. When mining of the 7760 stope began, the
footwall contact was clamped due to the high major principal stress. Severe mining-induced
seismic events occurred as the stress along the contact was released due to production blasting. It
is recognized that mining in burst-prone ground conditions requires a trade-off between
production requirements and ground control.

3.4.1 Typical causes of ground support failure under dynamic-loading

conditions
This section of the thesis provided the background for a discussion on ground support practices
applicable to dynamic-loading conditions encountered in deep and high stress mines. As part of
the review of support performance over time, typical causes of ground support failure under
dynamic-loading conditions were identified. These causes of support failure have been assigned
to three phases of the ground support cycle: the design; the installation, quality control and
quality assurance processes; and the performance under dynamic loads (Figure 3–13).

In the design phase, the selection of yielding reinforcement elements in conjunction with strong
connecting elements and the systematic installation of cable bolts in large excavations, improves
the performance of the ground support systems. The rate of rockburst severity, as represented in
Figure 3–1, was significantly reduced following the introduction of yielding reinforcement
elements at Creighton, Copper Cliff, and Coleman mines. Furthermore, since 2005, part of
Creighton Mine’s success in managing the ejection of large volumes of rock was due to the
systematic approach of cable-bolting large excavations.

Installation and quality control reviews at Creighton identified cases of premature damage to
certain reinforcement elements. This led the mine to implement a series of measures to minimize
early damage to the threaded portion of rebars and to the rings of friction sets. These measures
were further implemented at all Vale mines in the Sudbury area. Finally, corrosion was identified
 50

as an important factor leading to the degradation of ground support systems, as it severely

affected the support performance during the September 24, 2010 rockburst at Coleman Mine.

Figure 3–13 Typical causes of ground support failure under dynamic loads and the associated phases of the ground
support cycle

The performance of ground support systems was assessed through specific rockburst case studies
from Copper Cliff and Coleman mines. It was demonstrated that ground support systems cannot
manage dynamic loads when the depth of the damage zone exceeds the length of the installed
reinforcement elements. Furthermore, some reinforcement elements, such as the mechanical
bolts, have been found to be inadequate in managing dynamic loads and have been eliminated
from the ground support standards employed at the mines.

Finally, based on the experience at Creighton Mine, it would appear that the effectiveness of
shotcrete is diminished beyond a certain threshold of loading. Shotcrete can be perceived as
being unable to manage dynamic loads due to its high stiffness and fundamentally brittle
behaviour. Shotcrete loose has become a major issue in high stress mines under both static and
dynamic loads (Counter, 2012). Nevertheless, shotcrete is capable of keeping the ground tight by
limiting rock mass dilation as opposed to mesh, which is passive. As a result, shotcrete is
capable, to a certain extent, to preserve a laminated beam and maintain confinement around
reinforcement elements (Simser, 2012). The 3.8 MN seismic event at Copper Cliff Mine,
 51

indicated that the use of shotcrete could be effective in preventing large seismic shakedowns
triggered by remote seismic events. Recently, some high stress mines have adopted a mesh-over-
shotcrete approach in order to better manage dynamic loads (Counter, 2012; Punkkinen &
Mamidi, 2010; Simser, 2012). Such approach allows the shotcrete to keep the ground tight
whereas the mesh can better absorb high levels of kinetic energy and accommodate larger
deformations, hence providing a safety net for when the shotcrete becomes severely damaged.
There is a need to investigate the performance of surface support elements, as part of a system, in
order to better define a strategy for the use of mesh and shotcrete under dynamic-loading
conditions. This is addressed in the Chapter 7 of the thesis.

3.5 Summary
The objective of this chapter was to address the evolution of rockbursts at Creighton, Copper
Cliff, and Coleman mines and to link this evolution to the employed mining sequence and the
evolution of ground support practice. The chapter has reported on lessons learned over time in
managing dynamic-loading conditions at the three mines. By considering the evolution of the
frequency and severity of rockbursts, the justifications for introducing new support technologies,
identifying limitations in the employed support designs, and managing the mining process over
time were demonstrated in quantifiable terms, testifying of the quality of the collected rockburst
data. Empirical experience, developed over the analysis of rockburst case studies, has provided
valuable elements of design information, which have contributed to the implementation of
successful support strategies at Creighton, Copper Cliff, and Coleman mines over time. The
lessons learned in managing dynamic-loading conditions were transferable in these cases from
one mine site to another.

This chapter has demonstrated that the employed ground support systems affect the frequency
and severity of rockburst occurrences in deep underground mines. An important objective of this
thesis is to quantify the thresholds of performance of different ground support options. By
expressing, in quantifiable terms, the conditions under which each individual rockburst occurred,
the performance of ground support systems can be assessed in consideration of the severity of
the event. A literature review on elements of ground support design that affect the dynamic-load
demand and the capacity of ground support systems is presented in Chapter 4. The problem of
ground support design for managing burst-prone conditions is explored through the existing
 52

conventional and empirical design approaches. The analysis of the rockburst data collected and
reviewed at Creighton, Copper Cliff, and Coleman mines resumes in Chapter 5.
 53

4 Conventional and Empirical Approaches Toward

Ground Support Design for Rockburst Conditions

A conventional approach toward the design of ground support systems for excavations subjected
to rockbursts implies a quantification of the capacity of the available support options and of the
anticipated dynamic-load demand on the support. The capacity of ground support systems to
manage dynamic loads requires knowledge of the capacity of the employed reinforcement and
surface support elements. Most importantly, it requires the designer to be able to quantify the
impact of the interactions amongst support elements on the system’s performance.
Unfortunately, there is little data available on the dynamic interaction of components employed
within a ground support system (Roth, Cala, Brändle, & Rorem, 2014). To quantify the design
dynamic-load demand on the support, the anticipated rockburst damage mechanism(s) must be
defined. The volume of rock and velocity of ejection the system is designed to combat must
further be estimated. It is only by quantifying the capacity and the demand, i.e. the two essential
parameters required to enable a robust rock support design (Stacey, 2012), that a conventional
factor-of-safety approach is possible.

An empirical approach originates in or is based on observation or experience, often without due

regard for system and theory (Merriam-Webster, 2008). An approach toward ground support
design for dynamic-loading conditions based on passive monitoring of support performance
belongs into this category. Passive monitoring can be a useful tool in challenging the perceptions
with regards to support performance under dynamic loads.

This chapter provides a state-of-the-art literature review on the theoretical concepts that were
developed to enable a conventional design approach for ground support systems subjected to
rockbursts. The chapter further explores empirical work that was performed on rockburst data
over time. Limitations associated with both approaches are discussed further. This chapter aims
at extracting valuable information from the theory and both the conventional and empirical
approaches toward ground support design. The information highlighted in this chapter forms the
basis for the passive monitoring approach advocated in this thesis.
 54

4.1 Dynamic-load demand on ground support systems for a

conventional support design approach
Perhaps the most comprehensive conventional approach to designing ground support for
dynamic-loading conditions was the one that resulted from the Canadian Rockburst Research
Program 1990-1995 (Kaiser, McCreath, & Tannant, 1996). The five following steps summarized
this approach:

1. Identify the anticipated rockburst damage mechanism(s);

2. Identify the anticipated damage severity;
3. Identify the required support system characteristics to combat the anticipated damage
mechanism and severity;
4. Determine the performance characteristics of available support elements;
5. Match the available design characteristics with the required support functions to assemble an
appropriate support system.

The proposed guidelines for the identification of the damage mechanism, severity, and support
system characteristics were often dependent on the peak particle velocity, which is a conceptual
function of the magnitude and location of the seismic event with respect to the damage
location(s) (Figure 4–1). This section of the thesis discusses the assessment of the dynamic-load
demand on the support as part of a conventional support design approach. The discussion begins
by providing a background on the terminology associated with rockbursts and mine seismicity. It
further explores the concepts of ejection velocity and brittle failure of rock masses, which are
important in order to quantify the dynamic-load demand on ground support systems in a factor-
of-safety approach.
 55

(a) (b)

Figure 4–1 Conceptual guidelines used to assess the anticipated rockburst damage mechanisms and the required
support system characteristics (Kaiser et al., 1996)

4.1.1 Background on rockbursts and mine seismicity

4.1.1.1 Rockburst damage mechanisms

Under rockburst conditions, the demand placed on a ground support system, in terms of load,
displacement, and energy, depends on the rockburst damage mechanism, the severity of the
damage (thickness of failed rock), and on the ejection velocity of the rock, if any (Kaiser et al.,
1996). Three types of rockburst damage mechanisms were defined (Figure 4–1a):

• Rockfall due to seismic shaking;

• Bulking due to fracturing; and

• Ejection due to seismic energy transfer.

A prior rockburst damage classification proposed by Ortlepp (1992) was composed of (i) inertial
displacement and arch collapse, which may relate to the rockfall category, (ii) strain bursting and
laminar buckling, which may relate to the bulking category, and (iii) ejection. A reconciliation of
these two classifications of rockburst damage mechanisms is presented (Figure 4–2).
 56

Figure 4–2 Reconciliation of rockburst damage mechanisms identified by (a) (Kaiser et al., 1996) and (b) (Ortlepp,
1992)
 57

Under the rockfall mechanism, loose volumes of rock, stable under static conditions due to
arching forces and/or to the holding capacity of the support system, are accelerated by an
incoming seismic wave. This phenomenon is referred to as seismic shaking and it may trigger
rockfalls if it causes gravity forces to exceed the holding capacity of the installed support system.
This rockburst damage mechanism may result in the displacement of large volumes of rock,
especially from the back of excavations. Drift intersections and wide-span openings are
particularly vulnerable to this type of damage (Kaiser et al., 1996). Seismic shaking triggering
instabilities in the walls of mine excavations was described as inertial displacement by Ortlepp
(1992). This damage mechanism referred to the acceleration of a loose slab of rock in the
opposite direction to the rock mass movement, due to its inertia.

Bulking due to fracturing is an increase of the rock mass volume due to the development and
opening of new fractures. This damage mechanism occurs when the stresses near an excavation
suddenly exceed the rock strength or as a result of structural instability, e.g. when a slab of rock
buckles into an excavation. Under bulking due to fracturing, the bulk of material may remain in
place if the excavation is adequately supported or if the liberated strain energy is entirely
consumed by the fracturing and deformation process (Kaiser et al., 1996). On the other hand, any
excess energy may be transferred to the failing material and result in rock ejection. Highly-
stressed rock masses are susceptible to this rockburst damage mechanism.

Rock ejection due to seismic energy transfer results directly from the energy being transferred
from a distant seismic source to the wall of an excavation. Highly-jointed or stress-fractured rock
masses are susceptible to this damage mechanism (Kaiser et al., 1996). The extent of damage to
the installed support systems under such mechanism depends on the size of the ejected rocks and
the ejection velocity. Kaiser et al. (1996) proposed that the ejection velocity is equal to the peak
particle velocity at the surface of the opening. Damage caused by ejection due to seismic energy
transfer is suspected to be uncommon in Canadian mines; nevertheless, this damage mechanism
can be difficult to differentiate from rock bulking due to fracturing based on field observations
only (Kaiser et al., 1996).

4.1.1.2 Seismic source mechanisms

The nature of the seismic source that triggers a rockburst has a direct influence on the damage
mechanism. The seismic source mechanism is the rock mass failure mode that causes the release
 58

of seismic energy (Hudyma, 2008). Two types of tremor are generally observed: those directly
connected with mining operations and those strongly influenced by local geology and tectonics
(Gibowicz, 1990).

Two sets of techniques are commonly used to assess the source mechanism of seismic events:
waveform techniques and inferred techniques. Waveform techniques involve individual analysis
of ground motion waveforms collected at a number of seismic receivers for a particular event
and determining information about the energy radiation pattern for that event (Hudyma, 2008).
Waveform techniques include moment tensor inversion and first motion analysis, which are
reviewed by Hudyma (2008). It was recognized that the applicability of these techniques is quite
limited in mining given the requirement for high quality digital waveforms from triaxial sensors
and focal coverage of seismic sources.

Inferred techniques on the other hand use time and space trends in seismic source parameters to
infer seismic source mechanisms. For example, the seismic source mechanism may often be
deducted by spatial plotting of high precision seismic data, e.g. close to a geological structure,
within a mine pillar, or near production stopes or development blasts (Hudyma, 2008).
Conceptual zones of seismogenic rock mass failure are identified (Figure 4–3). The time of
occurrence of seismic events may further suggest the seismic source mechanism, e.g. seismicity
directly associated with stress changes due to mine blasting tend to follow mine blasts. On the
other hand, large magnitude seismic events typically associated with geological structures and
shear/slip mechanisms tend to correlate less with blasting times.

The dominant seismic source mechanism within a population of seismic events may also be
inferred using the frequency-magnitude relation. Variations of the b-value, i.e. the slope of the
linear portion of the frequency-magnitude chart, may suggest a change of seismic source
mechanism (Figure 4–4). Fault-slip related seismicity is typically characterized by a low b-value,
often less than 0.8 (Hudyma, 2008).
 59

Figure 4–3 Typical local rock mass failure mechanisms that potentially cause seismic events: (a) fault movement,
(b) stress change causing rock mass fracturing near excavation, (c) stope overbreak, (d) contrast in
material properties causing strain-bursting, (e) crushing of mine pillars, (f) stress increase causing rock
mass deformation (Hudyma, 2008)
 60

Figure 4–4 Frequency magnitude relations for two groups of events at an Australian mine (Hudyma, 2008)

The ratio of S-wave to P-wave energy is strongly related to the seismic source mechanism
(Hudyma, 2008). Seismic events associated with fault movements generate significantly more
energy in the S-wave than in the P-wave. Consequently, the ratio of S-wave to P-wave energy is
typically greater for shear-related events (generally greater than 10) than for non-shear events
(generally within the range of 1 to 3). Scatter plots of S-wave and P-wave energy are presented
(Figure 4–5). The cluster of seismic events on the left contains 63% shear related events with S:P
ratios greater than 10 whereas the cluster on the right contains only 21% shear related events
(Hudyma, Heal, & Mikula, 2003) .

Figure 4–5 Scatter plots of S-wave and P-wave energy for two clusters of seismic events (Hudyma et al., 2003)
 61

A classification by Ortlepp (1997) suggested correlations between the mechanism of seismic

events and the magnitude. Five categories of seismic events were identified along with a
suggested range of magnitude for each category, using the Richter scale. The classification by
Ortlepp (1997) was modified by including the corresponding range of Nuttli magnitudes
(Table 4–1). A conversion from the Richter scale to the Nuttli scale proposed by Kaiser et al.
(1996) was employed in this table. This conversion is explained in section 4.1.2.

Table 4–1 Classification of seismic event sources, modified after Ortlepp (1997)

Seismic Postulated First motion from Magnitude Rock

event source mechanism seismic records Richter Nuttli burst
Strain Superficial spalling with Usually undetected; -0.2 – 0.0 0.3 – 0.5
bursting violent ejection of fragments could be implosive

Self-initiated
Buckling Outward expulsion of pre- Implosive 0.0 – 1.5 0.5 – 2.0
existing larger slabs parallel to
surface of opening
Face crush/ Violent expulsion of rock from Mostly implosive, 1.0 – 2.5 1.5 – 3.0
pillar burst tunnel face or pillar sides complex
Shear Violent propagation of shear Double-couple 2.0 – 3.5 2.5 – 4.0

Remotely-
rupture fracture through intact rock shear

triggered
mass
Fault-slip Violent renewed movement on Double-couple 2.5 – 5.0 3.0 – 5.5
existing fault or dyke contact shear

The seismic source mechanism may coincide with the description of the rockburst damage
mechanism. This is the case when the rock involved in generating the seismic event becomes
part of the damage material, e.g. in strain bursting, buckling, face-crushing or pillar bursting-type
events. The face crush / pillar burst category is interpreted as an intermediate case where smaller
surfaces of slipping, perhaps along alignments of induced extension fracturing, form shear
surfaces which intersect in a somewhat conjugate manner ahead of a stope face or traverse, in
hourglass fashion, right through a pillar (Ortlepp, 1997).

Shear rupture and fault-slip mechanisms are both characterized by shearing movements, either
along an incipient or existing plane of failure. This type of event is more likely to occur in
mature, large-scale mining operations at considerable depth (Ortlepp, 1997). Seismic events
 62

characterized by a shear rupture or fault-slip mechanism typically involve release of strain

energy from a large volume of rock. Consequently, the corresponding magnitude can be very
large. Damage triggered by such events is typically remote from the seismic source location;
however, it is possible that the slip surface daylights into an opening, in which case the extent of
damage could be considerably larger.

The description of rockbursts in Ontario mines tends to follow a simplified version of the
classification by Ortlepp (1997). The main rockburst types are typically strain burst, pillar burst,
and fault-slip burst (Hedley, 1992; Kaiser et al., 1996; Kaiser & Cai, 2013; WSN, 2010). A
description of each rockburst type was provided by Kaiser and Cai (2013) and is summarized
(Table 4–2).

Table 4–2 Summarized description of the three main rockburst types (Kaiser & Cai, 2013)

• The stress near the excavation wall exceeds the rock mass strength and causes
damage.
Strain burst

• The energy stored in rock surrounding the excavation is a function of the strength of
the rock mass near the excavation, the post peak stress-strain slope of the failing
rock mass, and the local stiffness of the loading system, called mine stiffness.
• Strain bursts may be characterized as self-initiated, seismically triggered, or
dynamically loaded.
• The strain energy stored in a pillar and the surrounding rock is released.
Pillar burst

• The amount of energy is a function of the rock pillar strength, its post peak load-
displacement slope, and the local mine stiffness.
• If only the skin of a pillar fails, the rockburst should be classified as a mining-
induced strain burst. This recommendation was followed in this thesis.
• The energy radiated from the seismic source results in a dynamic stress wave that
Fault-slip burst

causes damage to excavations and support

• The energy and the related stress change is a function of the event intensity
(magnitude, seismic moment, etc.) and distance from the source.
• Strain energy stored in the rock near the location where damage is triggered or
caused by the stress wave may also influence the extent of damage
 63

Kaiser and Cai (2013) recommended dividing strain bursts into three categories depending on
whether the damage is exclusively related to the stored strain energy around the excavation or if
it is affected by the energy released by an associated remote seismic event. It was suggested that
self-initiated strain bursts occur when the damage is only related to the stored energy;
consequently, the seismic event and the damage are co-located. Seismically-triggered strain
bursts are triggered by a remote seismic event, however, the damage is not or only vaguely
related to the dynamic stress change. If a strain burst is strongly influenced by the energy
radiated from a remote seismic source, it is referred to as a seismically-triggered and
dynamically-loaded strain burst. In such case, the extent of the damage is significantly magnified
by the remote seismic event and is therefore related to both the stored energy and the magnitude
of the event.

In practice, the three sub-classifications of strain bursts proposed by Kaiser and Cai (2013)
complicate the assessment of the rockburst mechanism. The provided description can arguably
lead to confusion as there is appearance of overlaps between the definitions, e.g. between a fault-
slip burst and a seismically-triggered and dynamically-loaded strain burst. Furthermore, it is
arguably impossible to make a distinction between a self-initiated strain burst and a seismically-
triggered strain burst based on damage observations only. The more conventionally used
classification of rockburst mechanisms, i.e. strain burst, pillar burst, or fault slip, or perhaps an
even simpler classification, e.g. self-initiated or remotely-triggered rockbursts, is advisable for
the sake of consistency.

4.1.1.3 Practical application of the rockburst terminology

The update to the strain-burst description provided by Kaiser and Cai (2013) was perhaps
necessary to provide a hypothesis on why large-magnitude fault slip events often result in
damage to mine openings triggered by smaller aftershock seismic events. Nevertheless, this
update is not perceived to provide more clarity to the already varied and somewhat over-lapping
definitions of rockburst damage mechanisms, seismic source mechanisms, and rockburst types. It
is important to understand that most definitions correspond to concepts only: most of them were
not validated using field evidence and are not practical for rockburst classification during field
inspections. For practical purposes, it is perhaps relevant to rely on the fundamental description
 64

of a rockburst, i.e. whether the damage was self-initiated or remotely-triggered, following the
reduced version of the seismic event classification provided by Ortlepp (1997) (Table 4–1).

The review of rockburst field data from Creighton, Copper Cliff, and Coleman mines suggested
that ground control personnel tended to focus on distinguishing strain bursts from fault-slips.
This may be adequate to understand the basic mechanisms of dynamic loading that applied on
the support and perhaps a more complex classification would not provide any practical
advantage. To provide justification to a basic yet practical interpretation of a rockburst
mechanism, it can be relevant to refer to the recorded magnitude (Ortlepp, 1997) and the S-wave
to P-wave energy ratio (Hudyma, 2008) if high-quality seismic data are available. The
interpretation of the rockburst mechanism may be complemented by assessing the timing and
location of the triggering seismic event(s) with respect to mine infrastructures or geological
features (Hudyma, 2008).

4.1.2 Magnitude scales

The magnitude is the most common parameter used in mining to quantify the energy released by
a seismic event. In mining, the magnitude is typically assessed using either the Richter or the
Nuttli scales. The Richter scale (Eq. 4–1a) was originally developed for seismicity in California,
whereas the Nuttli scale (Eq. 4–1b) was developed for Central and Eastern North America. Both
scales provide a logarithmic interpretation of the amplitude signal, i.e. one magnitude unit
difference between two signals corresponds to a factor ten difference in the amplitude. The Nuttli
(MN) is the primary local magnitude scale used by the Geological Survey of Canada (GSC) for
mines located in the Canadian Shield. The Richter local magnitude scale (ML) is used for mine
seismicity in most of the world outside of north-eastern America. Over the range of interest for
rockburst, i.e. magnitudes 1.5 to 4.0, the Nuttli scale typically returns values 0.3 to 0.6 units
higher than the Richter scale for the same event (Hedley, 1992). Kaiser, et al. (1996) used a
difference of 0.5 to convert magnitudes from the Richter to Nuttli scale in Ontario mines.











(4–1)

where, KW = magnification of a Wood-Anderson seismograph at period T (Hedley, 1992).







 (4–2)

 65

where, D = epicentral distance to source, in km;

A = half the maximum peak to peak amplitude in the S-phase;

K = instrument magnification factor; and

T = time period, in seconds (Hedley, 1992).

4.1.3 Ejection velocity and peak particle velocity scaling laws

The ejection velocity during a rockburst is perceived to be related to the peak particle velocity
(ppv). For engineering purposes, Kaiser et al. (1996) recommended that the ejection velocity of
the broken material may be set equal to the ppv at the opening. As an outgoing stress pulse
travels away from a seismic source, its amplitude decreases due to geometrical spreading,
inelasticity, and scattering (Mendecki, 2008). Several published guidelines quantify this
phenomenon as a function of the magnitude and distance from the seismic source. PPV scaling
laws from McGarr, Green, & Spottiswoode (1981), Hedley (1992), Kaiser et al. (1996), and
Potvin et al. (2010) were considered for the purpose of this review (Table 4–1). Those guidelines
were site specific and involved different measuring equipment at different time.

Table 4–3 Summary of ppv scaling laws determined by linear regression (ppv is in m/s and the distance R is in m)

References Richter scale Nuttli scale Recommended

parameters
McGarr et al. 













(1981)





Hedley (1992) 











Kaiser et al.



















a = 0.5







(1996)

C = 0.2-0.3


Potvin et al.








































(2010)























α = 0.53–1.14
 66

In establishing a ppv scaling law, McGarr et al. (1981) conducted a least-square regression on
seismic data from the East Rand Propriety Mines in South Africa and from major earthquakes,
mostly from California, using the Richter scale. Hedley (1992) provided average fits as well but
employed the Nuttli scale and data from Ontario underground mines only, i.e. Creighton,
Strathcona, Denison, and Quirke mines.

Kaiser et al. (1996) complemented data from McGarr et al. (1981) with data from Brunswick, El
Teniente, and Creighton mines. Kaiser et al. (1996) proposed a conversion from Nuttli to
Richter, i.e. MN = ML + 0.5. The parameters “a” and “C” used by Kaiser et al. (1996) were
calibrated for a 90 to 95% confidence limit and were recommended for design purposes only.
Given that 90 to 95% of the data used by Kaiser et al. (1996) were supposedly below the
regression line, this relationship would, in theory, provide significant overestimates of the ppv if
it was applied to the back-analysis of rockbursts. This limitation was recognized in more recent
work (Kaiser & Cai, 2013).

Kaiser et al. (1996) identified a range for the near-field which was estimated to be once to twice
the source radius R0. Within this volume of rock that surrounds the epicenter of a seismic event,
ground motions are expected to be extremely non-uniform. Hence, Kaiser et al. (1996)
recommended that the design scaling law should not be applied when:











 (4–3)

where, α = 0.53 to 1.14.

The equations used in the scaling laws by McGarr et al. (1981), Hedley (1992), and Kaiser et al.
(1996) tend towards infinite peak particle velocities at the seismic source, i.e. at distance R = 0.
Hedley (1992) suggested that the ppv should plateau and depend only on the event magnitude at
close distances from the seismic source and that his equation should be modified accordingly.
Potvin et al. (2010) modified the ppv equation from Kaiser et al. (1996) by adding the source
radius to the denominator to account for the saturation of the ppv in the near-field.

The four ppv scaling laws listed in Table 4–1 are represented graphically (Figure 4–6). The four
scaling laws are further compared on the same chart by plotting the ppv-distance curves for
MN = 1 and MN = 4 (Figure 4–7).
 67

Figure 4–6 Representations of ppv scaling laws by (a) McGarr et al. (1981), (b) Hedley (1992), (c) Kaiser et al.
(1996), and (d) Potvin et al. (2010). Magnitudes in (a) were converted from the ML to MN using the
conversion proposed by Kaiser et al. (1996). The light grey near-field area in (c) represents possible
variations of α values in determining the source radius (Eq. 4–3). A mid-range α value of 0.835 was
used to assess the near-field conditions in (d).

Despite the average fit of the seismic data considered by McGarr et al. (1981) and the 90 to 95%
confidence limit regression used by Kaiser et al. (1996), the two scaling laws result in relatively
close ppv estimates, especially for high magnitude events (Figure 4–7). The ppv scaling law by
Potvin et al. (2010) matches the one by Kaiser et al. (1996) but saturates as it enters the near-
field. The ppv scaling law by Hedley (1992) is characterized by a significantly different
 68

behaviour; it provides much higher ppv estimates as it approaches the near-field but on the other
hand returns smaller estimates as the distance from the seismic source increases over 20 to 30 m.
These observations may reflect the variability in the ppv measurements that were obtained at
different locations over different periods of time.

Figure 4–7 Comparison of ppv scaling laws with lower and upper bounds representing event magnitudes of 1.0
and 4.0 MN

4.1.3.1 Limitations of ppv scaling laws

Scaling laws over-simplify the phenomenon of seismic-wave propagation by neglecting the
influence of the radiation pattern and the interaction with wave-transmission modifiers. A
spherical S-wave radiation pattern is assumed while applying ppv scaling laws. During a fault-
slip event, however, maximum ground motions associated with the S-wave occur at 0 and 90o
from the slip direction. The application of scaling laws results in overestimating the ppv in any
other direction. The influence of the radiation pattern assumption and slip direction on the ppv
are represented (Figure 4–8). Both the colour and size of the points are scaled according to the
ppv.
 69

Figure 4–8 Distributions of ppv modelled based on (a,b) realistic S-wave radiation patterns from two synthetic
seismic events with different slip directions and (c) based on a spherical radiation pattern assumption
(Potvin & Wesseloo, 2013)

The presence of mine excavations, geological features, and lithological contacts may result in the
reflection and refraction of stress waves as they radiate from the seismic source. Reflection and
refraction is also accentuated by the presence of discontinuity planes. Potvin and Wesseloo
(2013) suggested that the reflection and refraction phenomena may result in local superposition
of stress waves which in turn may contribute to high localized dynamic loads.

Ground motions obtained from scaling laws, especially from those designed based on an upper
confidence limit regression, should serve as design inputs only and should not compare directly
with field observations (Kaiser et al., 1996; Kaiser & Cai, 2013; Potvin & Wesseloo, 2013). For
design purposes, the ejection velocity may be set equal to the ppv at the opening (Kaiser et al.,
1996). Nevertheless, the possibility for the ppv to be amplified by a factor two at the surface of
an excavation was acknowledged (Kaiser et al., 1996). On the other hand, Durrheim et al. (2012)
suggested that the amplitude of a seismic wave doubles at a free surface and that the resulting
amplification of the ppv is within the range of 4 to 10 (Figure 4–9).

The site response, defined as the ratio of the measured peak ground velocity to the peak ground
velocity inferred from seismic data, was calculated for 9686 seismic events at Tau Tona, Kloof,
and Mponeng gold mines in South Africa (Milev et al., 2002). Seismic events with a maximum
ppv of 3 m/s were recorded and site responses as high as 25 were measured. The site response
was found to attenuate with the hypocentral distance from the seismic source. The maximum site
responses were associated with source radii of between 5 and 30 m and ppv lower than
100 mm/s. Higher ppv values were found to have little influence on the site response.
 70

Figure 4–9 Increase in the stress wave amplitude from actual seismograms recorded 10 m from a tunnel sidewall
and on the tunnel sidewall (Durrheim, 2012)

The radiation pattern, reflection and refraction phenomena, and site response being taken into
consideration, it is argued that it is next to impossible to estimate the ground motions generated
by an actual seismic event, even if the source mechanism is properly defined and the event
location is accurate (Kaiser & Cai, 2013). Furthermore, if the ppv is perceived to correlate with
the ejection velocity, and hence with the severity of damage, rockburst field data from the
Creighton Mine did not support such a correlation (Morissette, Hadjigeorgiou, & Thibodeau,
2012). In Figure 4–10, ppv values were calculated using the scaling law from Hedley (1992),
which was based on data from Ontario mines, including Creighton, and employed the Nuttli
magnitude scale. The severity of damage was assessed based on the displaced tonnage.
 71

Figure 4–10 Variations of damage severity at Creighton Mine with respect to the ppv (Morissette et al., 2012)

4.1.3.2 Proposed approach to the application of ppv scaling laws

In assessing the influence of a seismic event on the severity of rockburst damage, the ppv
provides a practical representation of two parameters that are perceived to influence the
dynamic-load demand on the support, namely the magnitude and the distance from the seismic
source. The ppv obtained from scaling laws, however, contributes to the impression that the
parameter is measurable and somehow reliable and that, therefore, it can be used to back
calculate the level of kinetic energy the support had to dissipate during a rockburst. The previous
section demonstrated that there are major flaws in the interpretation and application of ppv
scaling laws. The ppv obtained from a scaling law should not be expected to correspond to the
real ppv, whether because the employed scaling law purposefully aims at overestimating the ppv
or simply because of the mechanisms of reflection and refraction that influence the propagation
of seismic waves. The lack of correlation between ppv estimates and measurements was
demonstrated (Milev et al., 2002).

Consequently, engineers, from researchers to designers, should not rely on scaling laws to assess
the velocity of ejection as part of forensic analyses of support performance. Although it is
convenient to encapsulate the magnitude and the distance from the seismic event into one
parameter, the use of ppv scaling laws could possibly be reconsidered.
 72

Each scaling law employs a term that is referred to as the scaled distance (SD). Scaling of
distance is used in blasting in order to forecast peak particle velocities at a distance R from the
source i.e. a blast hole. Since the energy in blasting is proportional to the explosive weight (W),
the two following approaches are commonly used (Dowding, 1996):





Both the cube and square-root approaches are employed to compare field data and to predict the
attenuation or decay of ppv (Dowding, 1996). The use of the cube root is interpreted as a
dimensionless parameter, since the explosive weight is directly related to its volume. However,
the square root is more traditionally employed in blast designs as it typically considers long
charges of explosives, such as blast holes, where the hole’s diameter and the density of explosive
are constant (Dowding, 1996). The cube root, on the other hand, is more appropriate for
spherical charges.

The cube-root version of the scaled-distance equation was employed by Hedley (1992) to
characterize peak particle velocities generated by seismic events in Ontario mines. Hedley
(1992) proposed the following approach to predict the maximum displacement, velocity, or
acceleration of particles at a distance R from a seismic source characterized by a magnitude M,
recorded in Nutlli. Since the magnitude is a logarithmic scale, it is used as the exponent of 10.



























































































































where, u = peak particle velocity, acceleration, or displacement

n = attenuation or decay factor

K = constant.

Hedley (1992) calibrated his guideline using seismic data recorded at Creighton and Strathconna
mine, in the Sudbury area, and Denison and Quirke mines, in Elliot Lake. The least-square best
 73

fit of peak particle velocities generated by seismic events (in m/s) was obtained using K = 4 and
n = -1.6 (Table 4–3).

Given that the scaled distance concept is at the origin of the four scaling laws presented
previously (Table 4–3), it could be convenient to refer to it in order to assess the dynamic-load
demand on the support during rockbursts. The scaled distance can be employed in order to
identify relations between event magnitude, distance from seismic source, and severity of
damage, for example. The fact that the scaled distance is dimensionless would not affect the
correlations amongst these three parameters, if there were any. Nevertheless, the dimensionless
parameter would have the advantage of not carrying any incorrect physical meaning, which
could easily be misinterpreted in the design.

If there is a perceived advantage in using the ppv, it is of great importance to specify the
employed scaling law, given the differences observed amongst the proposed relations (Figure 4–
7). Furthermore, it is of equal importance to ensure that the ppv is simply interpreted as a mean
to combine the magnitude and the distance into one empirical parameter. Regardless of the
employed scaling law, little significance should be attributed to the velocity units.

For the analysis of support performance under rockburst conditions it may appear convenient to
reduce the magnitude and the distance to one parameter, either the ppv or the scaled distance.
Nevertheless, by doing so, the result is an important loss of information. Both the magnitude and
the distance appear to be more practical parameters to employ for support design purposes
around known seismically-active structures. Furthermore, the magnitude itself can provide more
information than the ppv or the scaled distance about the mechanisms of loadings that apply on
ground support systems. A low magnitude seismic event in the near vicinity of a mine opening
and a larger but remote seismic event would likely result in different damage mechanisms
although they could be characterized by the same ppv or scaled distance.

4.1.4 Brittle rock mass failure and estimation of the extent of damage
around mine openings
Support design for static-loading conditions in hard rock is largely based on limit equilibrium
analyses and on the use of empirical design techniques (Hadjigeorgiou & Charette, 2001). Limit
equilibrium analyses typically involve the calculation of a factor of safety where the load-bearing
 74

capacity of a support system is compared to the load acting on it. Under rockburst conditions,
however, the energy released by the development and expansion of new fractures or the kinetic
energy associated with the acceleration of a loose volume of fractured rock must be dissipated by
the support and therefore must be taken into account in the support design. Seismicity in
underground mines is a direct response to stress changes. Mining-induced stress conditions are
therefore associated with both the cause of rockbursts, i.e. seismicity, and the consequences by
defining the intensity and the extent of the fracture zone around mine openings.

In order to assess the dynamic-load demand posed by rockbursts on ground support systems, it is
important to understand the process that leads to brittle failure in hard rock. This type of failure,
often described as spalling or slabbing, involve the development of tensile fractures growing in
the direction of the maximum principal stress, parallel to the excavation boundary. Ortlepp
(1997) demonstrated this phenomenon through several case studies from underground mines in
South Africa (Figure 4–11).

Figure 4–11 Well-developed slabbing in a footwall drive pre-developed ahead of stoping in fine-grained, massive,
and very strong quartzite at a depth of 2,600 m (Ortlepp, 1997)

In this section, the brittle failure process is explored by first discussing failure of rock specimen
in a laboratory environment. Subsequently, practical approaches to quantify the potential for
brittle failure and estimating the depth of failure around underground excavations based on the
stress conditions are reviewed.
 75

4.1.4.1 Brittle failure of intact rock specimens

Brittle failure is defined as a sudden loss of strength following little or no plastic deformation
(Brady & Brown, 2004; Hoek & Brown, 1980). Alternatively, a ductile behaviour is described as
a sustained permanent plastic deformation without any loss of load-bearing capacity. Most rock
fails in a brittle manner at the confining pressures encountered in civil and mining engineering
applications (Hoek & Brown, 1980). An increase in confining stresses may, however, trigger an
increase of the peak strength as well as a transition from a brittle to a more ductile behaviour.
Wawersik and Fairhurst (1970) illustrated this phenomenon using triaxial compression tests on
Tennessee Marble specimen (Figure 4–12). The increase in peak strength as a result of
increasing confinement was explained by the effect of confining stresses inhibiting the
development of extension fractures (Hudson & Harrison, 1997). The transition towards a ductile
behaviour was attributable to a decrease in dilation as a result of increasing confining pressure
(Brady & Brown, 2004).

Figure 4–12 Complete axial stress-strain curves obtained in triaxial compression tests on Tennessee Marble at
various confining pressures (Wawersik & Fairhurst, 1970), converted in metric units (Brady & Brown,
2004)

The transition from brittle to ductile behaviour along with increasing confinement is exhibited by
rocks of relatively low hardness such as marble (Wawersik & Fairhurst, 1970) or sandstone
(Tarasov & Randolph, 2011). Hardness can be estimated on the basis of the average elastic
modulus (Tarasov & Randolph, 2011). The transition from brittle to ductile behaviour, however,
 76

is typically not achieved in hard rock such as Westerly granite and Frederick diabase (Wawersik
& Brace, 1971) or dolerite (Tarasov & Randolph, 2011). Such hard rock may, on the contrary,
exhibit an increasingly brittle behaviour as the level of confinement is increased (Figure 4–13).
An explanation to the observed increasing rock embrittlement was provided by (Tarasov &
Randolph, 2011).

Figure 4–13 Stress-strain curves of (a) Westerly granite and (b) Frederick diabase in uniaxial and triaxial
compression (Wawersik & Brace, 1971)

A shear rupture can propagate in its own plane due to the growth of short tensile cracks in the
direction of the maximum applied load. The interaction and coalescence of tensile cracks is
facilitated under increasing confinement; the extension of failed sites is progressively eliminated
and the interaction of small extension fractures results in the formation of a macro-shear plane.
Numerical approaches have demonstrated in fact that the peak strength of a rock sample is a
function of the elastic generation of extension strain and tensile crack accumulation in which
friction plays no part (Diederichs, 2003; Fang & Harrison, 2002).

Under sufficient confinement, the tensile cracks form an echelon of blocks described as the
‘book-shelf’ structure (Tarasov & Randolph, 2011). The common representation of the shear
rupture mechanism consists of the collapse of the book-shelf structure during its rotation as
 77

faulting propagates (Figure 4–14a). Frictional resistance τf resulting of the block collapse
substitutes gradually the cohesive strength τcoh and the initial resistance caused by the front
blocks. In accordance with this model, an increase in confinement generates an increase in
rupture energy and decrease in brittleness (Tarasov & Randolph, 2011).

Since shorter cracks are being created under high levels of confinement, Tarasov and Randolph
(2011) suggested that if made of strong material, shorter blocks, instead of collapsing, could
rotate and hence behave as ‘hinges’ (Figure 4–14b). In the rotation process, within the second
half of the fan structure, i.e. where the blocks have tilted beyond the vertical axis, active forces
assisting the fault displacement are created. These forces, referred to as negative shear resistance
(Figure 4–14b), are responsible for the increase in brittleness of the rock material under high
confinement (Tarasov & Randolph, 2011).

Figure 4–14 Shear rupture mechanism: (a) common frictional and (b) frictionless model proposed by Tarasov and
Randolph (2011)

Levels of brittleness can be described using the following conceptual scale (Figure 4–15). The
classification includes behaviours typical of class I rocks, i.e. ductile to transitional, and of class
II rocks, i.e. brittle to super-brittle. Class I behaviour is characterized by ‘stable’ fracture
propagation, in the sense that work must be done on the sample to effect further reduction in
load-bearing ability. In contrast to ‘stable’ fracture development, failure for class II rock
 78

behaviour is self-sustaining, i.e. the elastic strain energy stored in the sample when the applied
stress equals the compressive strength is sufficient to maintain fracture propagation until the
specimen has lost virtually all strength (Wawersik & Fairhurst, 1970). From a graphical
perspective, rocks exhibiting class I and class II behaviours differ as they are characterized by a
negative and positive post-peak modulus M, respectively.

Figure 4–15 Scale of brittleness index k and k1 with characteristic shapes of complete stress-strain curves (Tarasov
& Randolph, 2011)

The post-peak behaviour of the same rock type may evolve from a class I to class II behaviour,
and vice-versa, depending on the confinement conditions (Wawersik & Brace, 1971). In Figure
4–13a, the unconfined sample of Westerly granite exhibited a class II behaviour. Under low
confinement, i.e. 500 and 1,450 psi (3.4 to 10.0 MPa), the post-peak behaviour transitioned
toward class I. Above 1,450 psi (10.0 MPa), the Westerly granite failed in a very unstable
manner as it exhibited a class II behaviour (Wawersik & Brace, 1971).

The brittleness of the rock influences the level of energy released in the post-peak region as
opposed to the energy consumed in the fracturing process. A conceptual energy balance is
presented for class I and class II behaviour rocks (Figure 4–16). From left to right, the energy
balance is represented at peak conditions, at an intermediate post-peak stage, and at failure. For
both class I and class II rock behaviours, the red and grey areas represent the elastic strain energy
stored in the rock sample and the energy consumed in the post-peak region, respectively. The
yellow area is the fraction of elastic strain energy that is not consumed by the fracturing process
 79

and is therefore released as seismic, kinetic, or thermal energy. The release of excess energy is
exclusive to class II rocks.

Figure 4–16 Energy balances typical of the failure process in class I and class II rocks (Potvin & Wesseloo, 2013;
Tarasov & Potvin, 2012)

The brittle failure of intact rock accompanied by a sudden release of energy, seismic and/or
kinetic, is typical of strain bursts. In this section, it has been demonstrated that only class II rocks
are susceptible to violent releasing of energy in the post peak region and hence, it would appear
that only class II rocks are susceptible to strain-bursting. For softer class I rocks, increasing the
level of confinement can generate a transition from a semi-brittle to a more ductile behaviour,
hence leading to more energy being consumed in the fracturing process and allowing the failed
mass of rock to maintain a higher load-bearing capacity. From a ground support point of view,
increasing the level of confinement in the surrounding rock mass when developing in class I
rocks may appear like a sound strategy. This can be achieved by installing strong and stiff
reinforcement and surface support elements such as rebars and mesh-reinforced shotcrete.
 80

On the other hand, it was demonstrated that increasing the level of confinement in class II rocks
may result in an increased rock embrittlement. Under high levels of confinement, the post-peak
behaviour of Westerly granite and dolerite have transitioned from brittle to super-brittle (Tarasov
& Randolph, 2011), leading to more energy being released during the failure process. These
observations would suggest that at the mine opening scale, in class II rocks, increasing the level
of rock mass confinement may result in increasing the load bearing capacity at the peak
conditions; however, if failure occurs, the level of energy released could be considerably more
severe.

The granite at Creighton and Coleman mines is characterized by a UCS of 240 MPa, similar to
the 259 MPa (~37,600 psi) UCS of the Westerly granite (Wawersik & Brace, 1971). In the
collected rockburst case studies, over 72% of the damage locations occurred within the granite
and 94% of those locations were under a minimum principal stress greater than 10 MPa (1,450
psi). Numerical elastic stress analyses that led to these values are discussed in Chapter 5. The
obtained range of confinement in the rockburst data is similar to the σ3 values under which the
Westerly granite was tested (Wawersik & Brace, 1971). The evolution of the post-peak
behaviour of the Westerly granite, based on the stress-strain curves obtained in the laboratory
(Figure 4–13a), suggest that the large majority of the collected rockbursts case studies occurred
within brittle class II rocks capable of releasing important amounts of energy during failure.
Under those circumstances, it is important to consider whether deliberately increasing the
confinement in the surrounding rock mass by using strong and stiff reinforcement elements
represents an adequate support strategy from an energy-released perspective.

The concepts developed in the laboratory would suggest that a strong and stiff support system
would enhance the capacity of the rock mass to support itself under higher stress levels;
however, when failure, i.e. a rockburst, occurs, the event could potentially be significantly more
violent due to the release of greater amounts of energy. Consequently, the resulting level of
damage would be more severe. In class II rocks, a support system capable of accommodating
displacements and absorbing the energy progressively released over time, as opposed to
deliberately storing it within the rock mass, would appear to be the most appropriate strategy.

The stiffest ground support system reported in the rockburst database consisted of shotcrete
arches installed in the ‘4680 cross-cut’ on Creighton Mine’s 7400 level. On March 14, 2009, a
 81

2.9 MN event caused important damage on the 7400 level (Figure 4–17). The location where the
arches were installed endured the most severe rock displacement as 210 tons (190 tonnes) were
lifted from the floor, damaging the wall portion of the arches. Further 22 tons (20 tonnes) were
reportedly ejected from the west wall of the drift. Given its high stiffness, the employed ground
support system was not capable of containing the 2.9 MN rockburst on March 14, 2009 and the
event resulted in very severe damage. It is possible, based on the behaviour of class II rocks
described in this section, that a softer system capable of absorbing energy and accommodating
rock displacements would have led to a reduction in damage severity.

Figure 4–17 Damage on Creighton Mine’s 7400 level associated with a 2.9 MN rockburst on March 14, 2009

The arches were built in the ‘4680 cross-cut’ to provide long term support to an excavation vital
to the mine. Arches, however, were never used at Creighton where rockbursts were specifically
anticipated. The current support strategy in excavations susceptible to rockbursts involves the
use of yielding reinforcement elements (33 mm modified conebolts) in the first pass and #4
welded-wire mesh as surface support, the use of shotcrete being reduced in burst-prone grounds
at Creighton since 2010.
 82

4.1.4.2 Brittle failure of rock masses in situ

The behaviour of intact rock, most importantly in the post-peak region, has been reviewed in the
previous section. Class II rocks can fail in a very brittle and unstable manner whereas work must
be done on class I rocks to effect a reduction in load-bearing capacity. For both types of rock
behaviour, the result in terms of energy released during failure differs. In this section, brittle rock
mass failure around underground excavation is investigated in order to identify the stress
conditions that drive the failure and further anticipate the depth of failure around openings.

4.1.4.2.1 AECL-URL Lac du Bonnet granite

Martin (1997) investigated the brittle rock mass failure at the Atomic Energy of Canada Limited
(AECL) Underground Research Laboratory (URL) located in Lac du Bonnet granite.
Microseismicity was recorded during the development of a circular tunnel in the direction of the
intermediate principal stress (σ2). Seismic records and field observations provided evidences of
the formation of a notched shape at the location of the maximum tangential stress (Figure 4–18).

Figure 4–18 View of the completed test tunnel at the AECL-URL (a) and microseismicity that followed blasting of the
th
17 round (b) (Martin, 1997)
 83

Agreement between three-dimensional elastic stress analyses results and measured stresses ahead
of the tunnel advance suggested that the rock mass ahead of the tunnel face behaved in an
essentially elastic manner (Martin, 1997). Three-dimensional elastic stress analyses were used
therefore to assess the stress conditions at the location of microseismic events ahead of the tunnel
face. A damage initiation threshold similar to the one observed in laboratory for intact rock
samples was identified (Figure 4–19).

Figure 4–19 Damage initiation threshold for Lac du Bonnet granite derived from stresses computed for microseismic
events recorded ahead of the tunnel face (Martin, 1997)

The formation of the notched shape was associated with strong non-linear deformations.
Consequently, elastic analyses provided only the maximum stress possible for each microseismic
event location. The results demonstrated that slabbing occurred at stresses well below the long-
term strength of rock samples, determined in the laboratory (Figure 4–20).
 84

Figure 4–20 Damage initiation threshold for Lac du Bonnet granite compared with stresses computed for
microseismic events associated with the formation of the notch (Martin, 1997)

The constant deviatoric-stress criterion presented in Figure 4–19 was found to provide a
reasonable estimate of the maximum depth of damage around the test tunnel, with 70-75 MPa
corresponding to about 0.31-0.33σc. Re-expressed in terms of the Hoek-Brown parameters, the
constant deviatoric-stress criterion is equivalent to setting the parameter m at 0 and the parameter
s at 0.11 (Martin et al., 1999):







The parameter m is related to the angle of internal friction whereas s is related to the rock mass
cohesive strength. Field and laboratory observations provided evidences that peak cohesion and
friction were not mobilized together (Martin et al., 1999). The brittle failure process, up to the
critical damage stress or long-term strength σcd, is typically dominated by a loss of intrinsic
cohesion of the rock mass. Therefore, for the analysis of brittle failure only, it is justified to
 85

ignore the frictional strength component and to use a strength envelope based solely on cohesion
(Martin et al., 1999).

4.1.4.2.2 Pillar bursts at Creighton Mine

A series of pillar bursts that occurred on three cut-and-fill levels at Creighton Mine during the
mid-1980s were analysed in order to define an empirical failure criterion (Wiles, 2005). The
major and minor principal stresses in each pillar at the time of failure were assessed using the 3D
elastic boundary element package Map3D. A strength envelope was fit to the data by linear
regression and confidence intervals were provided, Figure 4–21a. The resulting best-fit rock
mass failure criterion was given by an intercept UCS of 123.6 MPa and a slope of 4.08. The
failure criterion was further used to predict the failure of a crown pillar several years later at
Creighton mine. The stresses within the pillar were assessed for several pillar thickness
scenarios, Figure 4–21b. The extrapolation of the results suggested that the crown pillar would
fail between 16 and 51 m thickness with a 90% confidence and at 38 m thickness on average.

Figure 4–21 (a) Rock mass failure criterion derived from back-analyses of pillar bursts and (b) its application in
forecasting the possible failure of a crown pillar (Wiles, 2005)

The previous analysis demonstrated that failure was susceptible to occur under a certain pillar
thickness but did not forecast whether failure would be violent, i.e. result in a rockburst, or not.
The potential for rock mass deformation is strongly influenced by factors attributable to the mine
geometry. According to Kaiser and Cai (2013), burst indices that focus entirely on intact rock
properties are doomed to fail in predicting burst potential because the dominant driving factor,
i.e. the local system stiffness or deformation potential, cannot be measured by such parameters.
The loading system stiffness is a measure of rate of energy release as the non-linear strains
 86

accumulate (Wiles, 2007). Large values of loading system stiffness are associated with relatively
slow releases of energy and progressive damage whereas with small values, the energy is
released at a much higher rate (Wiles, 2007). When the stiffness of the loading system, i.e. the
mine stiffness, is softer than the unloading stiffness of the volume of failing rock, failure is
expected to be sudden and violent (Figure 4–22). The area located below the rock mass stiffness
curve in the post peak is the amount of energy that would be consumed in generating new
fractures. The amount of energy released as kinetic energy is the area above the rock mass
stiffness curve and below the mine stiffness curve.

Figure 4–22 The role of loading system stiffness (mine stiffness) and rock mass stiffness in the type of rock mass
failure (Kaiser et al., 1996)

The loading system stiffness was related the local energy release density (LERD) (Wiles, 2005).
The LERD was obtained as the area under the load-deformation curve divided by the pillar
volume:




 

(Wiles, 2005) 


The LERD provided information on the amount of energy that would be released by the rock
mass if a pillar was to fail. In assessing the probability of failure of a crown pillar (Figure 4–
21b), Wiles (2005) demonstrated that energy would be released at a sufficient rate to induce a
rockburst for a pillar width of 59 m or less (Figure 4–23). Hence, since failure was expected to
occur between 16 and 51 m thickness with a 90% confidence, the failure of the crown pillar, if it
was to occur, was expected to be violent and generate rockbursts. The actual failure of the crown
pillar, which occurred at 35 m thickness and resulted in eight rockbursts of magnitude 2.0 to 3.6,
supported the interpretation provided by Wiles (2005).
 87

Figure 4–23 (a) Back analysis of local energy release density (LERD) during pillar bursts and (b) the assessment of
energy released during the failure of a crown pillar

4.1.4.3 Depth of failure

The concepts presented in section 4.1.4.2 can be applied in order to determine the likelihood of
rock mass failure around a tunnel and if the failure is prone to be violent. The use of the
maximum tangential stress was advocated by Kaiser et al. (1996) in order to assess the depth of
failure around tunnels. The maximum tangential stress (σmax) at the wall of a circular opening in
elastic ground can be estimated using the Kirsch solution (Brady & Brown, 2004). The following
demonstration shows that σmax = 3σ1-σ3:

σ3 = p

θ
σ1 = Kp

























 88

For non-circular openings, stresses can be approximated by circumscribing an equivalent circular

tunnel and by applying Eq. 4–11 (Kaiser et al., 1996; Kaiser & Cai, 2013). Based on Eq. 4–11,
the depth of failure df is dependant on both the ratios of σmax to σc and on the size of the opening,
represented by the tunnel radius a.


















This approach was used to assess the normalised depth of failure in massive or good quality rock
(Q’ > 4) and was valid for ratios of df/a lower than 0.6 (Kaiser et al., 1996). Application of the
methodology at Long-Victor mine, however, found that the estimated depths of failure were very
large and quite unlike those observed (Mikula, 2012). The conclusion reached by Mikula (2012)
is supported by the rockburst data from Creighton, Copper Cliff, and Coleman mines analysed
as part of this thesis where the calculated df/a ratios typically exceeded 0.6 (Figure 4–24). These
results suggested that the methodology proposed by Kaiser et al. (1996) for estimating the depth
of failure may be limited in deep and high stress mines where high values of σmax are observed.
The mining-induced stress results used for generating Figure 4–24 are discussed in Chapter 5.

Figure 4–24 Calculated normalized depths of failure for rockburst occurrences at Creighton, Copper Cliff, and
Coleman mines, which typically exceed the applicability threshold determined by Kaiser et al. (1996)

4.1.4.4 Influence of geological structures on rockbursts

The examples of brittle intact rock or rock mass failure discussed previously assumed
homogeneity of the rock medium. In a mining environment, the rock mass consist of a
heterogeneous medium. Heterogeneities such as large-scale geological structures may promote
 89

seismicity or enhance the potential for rock mass failure. Simser and Andrieux (1999) identified
three types of structural environments associated with burst-prone grounds at Brunswick Mine,
in Eastern Canada:

• Lithological contacts with brittle rocks;

• Stiffness contrast between adjacent rock types causing squeezing out of the softer
medium;

• Discrete geological structures.

The first source of seismicity and rockburst was associated with a soft-brittle-soft material
sequence. At Brunswick Mine, porphyry dykes and massive sulphide materials were much stiffer
than the metasediment host rocks. Because it had both lower UCS and Young’s modulus,
metasediments tended to fail first under increasing load whereas the stiff material showed little
deformation and accumulated large amounts of energy. The stiff and brittle material would fail
under much higher loads and would release some of the energy accumulated during the
deformation process in a sudden and violent manner. Seismicity generated as a result of this
material configuration would usually stop when the brittle material had failed sufficiently to
redistribute the load on a volume of rock able to support it. It was observed that the equilibrium,
however, was quite precarious and that seismicity would usually restart with subsequent mining.
When the brittle material had failed enough to shed the stresses away from the mining area,
ground control problems would typically evolve from being seismicity-driven to gravity-driven
with the broken rock mass starting to unravel.

A brittle-soft-brittle material configuration would cause ground stability problems as the softer
material would tend to squeeze out of the brittle matrix. The problems would occur under high
stresses when a metasediment stringer quickly failed into an opening as it deformed much more
than the surrounding massive sulphides. The failure of the softer rock in the middle would
destabilize the more competent ground on both sides, leading to brittle failure (Figure 4–25).
Failure would typically propagate upward until a stable arched back shape was reached, until the
process choked itself by filling the excavation with failed material, or until the entire
metasediment stringer had failed. To prevent this type of failure mechanism, it was
recommended to simply avoid undercutting waste intrusions or to undercut them with the
smallest possible span.
 90

Figure 4–25 Schematic cross-section of an open stope intruded by a metasediment stringer and representation of
the failure mode (Simser & Andrieux, 1999)

Seismicity and rockburst associated with discrete geological structures such as dykes, local
faults, or geological contacts, typically results from sudden slippage along the discontinuity.
Discontinuities are usually locked by the confining stresses; however, modification to the stress
state may allow slippage to occur. This tends to happen when the principal stress direction
becomes oblique to the discrete feature, hence cancelling the clamping effect and driving the
failure, and/or when the shear strength of the asperities is exceeded, resulting in the release of
large amounts of stored energy. The largest magnitude events at Brunswick Mine were
associated with this mechanism.

4.1.5 Challenges and limitations in forecasting the dynamic-load

demand on ground support systems
A methodology for quantifying the dynamic-load demand on ground support systems was
introduced by Kaiser et al. (1996). The steps consisted of identifying the anticipated rockburst
damage mechanism(s) followed by identifying the anticipated damage severity. None of these
steps is simple and there are many assumptions associated with each one of them.

Most design guidelines involve the application of the ppv obtained from scaling laws. This
implies that support designs, if validated by a conventional factor-of-safety approach, are
typically conceived for facing rockbursts that are triggered by remote seismic events and that the
support system is aimed at dissipating the kinetic energy associated with the acceleration of loose
blocks. There does not seem to be a clear methodology allowing the calculation of forces
 91

involved in a strain burst and the load transferred onto the support. Even if such methodology
existed and was validated, the choice between designing against bulking due to fracturing,
ejection due to seismic energy transfer, or rockfall due to seismic shaking is no simple task. The
type of demand may change over time and along that do the properties of a ground support
system required in the shorter and longer term.

The determination of the demand necessitates knowledge of the velocity and direction of
ejection, and of the mass of rock ejected during the rockburst. In employing ppv scaling laws,
despite their limitations, as a mean to calculate a design velocity of ejection, the designer faces
great challenge in determining the magnitude and location of a design seismic event. Techniques
such as the frequency-magnitude analysis may assist in defining the largest magnitude expected
in a mine or an area of a mine. However, it is not guaranteed that a large magnitude event located
further away from an excavation will necessarily generate more damage than a smaller
magnitude located in the near vicinity. Kaiser et al. (1996) proposed an equation to estimate the
volume of fractured rock that can be involved in a rockburst; however, it was observed that the
anticipated depth of failure was much larger than and quite unlike those observed at Long-Victor
mine (Mikula, 2012). Furthermore, this procedure was not supported by the range of normalized
depth of failures obtained for rockbursts at Creighton, Copper Cliff, and Coleman mines.

According to Stacey (2012), the magnitude and location of a potential seismic event, the ejection
velocity resulting from the design seismic event, the direction of action of the ejection force, and
the mass of rock involved in the ejection are not known with sufficient confidence. Hence, it is
clear that the demand parameter cannot be defined satisfactorily for rock support design
purposes using a conventional design approach (Stacey, 2012).

4.2 Dynamic capacity of ground support systems for a

conventional support design approach
Under dynamic-loading conditions, a ground support system must be capable of absorbing the
amount of released energy and withstanding potentially large deformation whilst maintaining its
load-bearing capacity. Hence, the ideal support system for excavations subjected to dynamic
loading has high deformability, high strength, and thus, high-energy absorption capacity (Kaiser
et al., 1996).
 92

The dynamic capacity of a support system is dependent on the connectivity and compatibility of
its reinforcement and surface support elements. Connectivity refers to the capacity of a system to
transfer the dynamic load from an element to another, e.g. from the reinforcement to the surface
support through plates and terminating arrangements (split set rings, nuts, etc.), or from a
reinforcement/holding element to others via the surface support. Compatibility is related to the
difference in stiffness amongst support elements. Load transfer may not take place appropriately
when there are strong stiffness contrasts within a ground support system. Case studies revealed
premature failures of stiffer elements prior to utilising the full capacity of more deformable
elements within the same system (Simser, 2007). The response of a dual stiffness system to
heavy loading conditions was exemplified by Charette (2012) with soft deformable bolts
accommodating larger displacements and stiff reinforcement elements offering a stronger
resistance to deformation (Figure 4–26). These two behaviours within the same support system
may result in the failure of weaker or less deformable elements, e.g. plates, nuts, individual mesh
wires, or shotcrete.

Figure 4–26 Representation of a dual stiffness system with stiff and soft-deformable reinforcement/holding elements
(Charette, 2012)

Compatibility between reinforcement and surface support elements is also of primary

importance. If the rockbolts have significant yielding capability, the performance of the overall
support system will be enhanced only if the surface support has sufficient strength to cause
yielding of the rockbolts to occur (Stacey & Ortlepp, 2004). The ability of the surface support to
 93

transfer loads to the reinforcement elements may be critical to the ultimate success of the ground
support system under dynamic-loading conditions (Simser, 2007). Failure by a soft chain-link
mesh to transfer the load to modified conebolts installed in conjunction with #0 gauge straps is
illustrated (Figure 4–27a). Another type of compatibility failure is represented as a bolt that is
too stiff (or is it the surface support that is too weak?) becomes disconnected from the rest of the
system (Figure 4–27b). In this case, the grouted tendon preserved its reinforcing function,
however, it’s holding function has terminated due to the loss of connection with the surface
support.

Figure 4–27 Important stiffness contrasts between (a) chain-link mesh and holding elements (modified conebolts)
and (b) between small-gauge welded-wire mesh and a reinforcement element (rebar) (Simser, 2012)
 94

From a design perspective, it is important to understand that the dynamic-load capacity of a

ground support system depends not only on the capacity of its reinforcement elements but also,
and perhaps most importantly, on their compatibility with other elements of the system and on
the strength of the connections. The failure of one component of the support system usually leads
to the failure of the system (Stacey & Ortlepp, 2004).

Unfortunately, design procedures for rock support design focus mostly on checking how much
load a rock bolt can carry, or how much energy the rock bolt can dissipate (Kaiser & Cai, 2012).
Over the years, testing of ground support elements have focused mostly on individual
reinforcement and surface support elements rather than on whole support systems. Nevertheless,
although the information that resulted from multiple dynamic (drop) tests over time cannot be
extrapolated to predict the dynamic capacity of ground support systems, it is still of use for
design purposes (Stacey, 2012). Drop tests provide essential information on the behaviour of
support elements under tensile dynamic loads and allow for direct comparison. The capacity of
support elements and ground support systems is reviewed in this section. The section covers the
extrapolation of static (pull) test results, testing rig facilities, and simulated rockburst
experiments.

4.2.1 Extrapolation of static (pull) test results

Kaiser et al. (1996) described the capacity of reinforcement and surface support elements using
the following parameters (Table 4–4):

• peak load capacity (Lp);

• ultimate load capacity (Lu);

• displacement at peak load (dp);

• ultimate displacement capacity (du);

• energy absorption at peak load (Ep);

• ultimate energy absorption capacity (Eu).

 95

Table 4–4 Load-displacement parameters for different types of reinforcement and surface support elements
(Kaiser et al., 1996)

Lp Lu dp du Ep Eu
[kN] [kN] [mm] [mm] [kJ] [kJ]
Reinforcement elements
19 mm resin-grouted rebar 120-170 na 5-10 10-30 na 1-4
16 mm, 2 m mechanical bolt 70-120 na 10-40 20-50 na 2-4
16 mm grouted smooth bar 70-120 50-100 5-10 50-100 na 4-10
16 mm cable bolt 160-240 na 5-10 20-40 na 2-6
16 mm, 4 m debonded cable 160-240 na 10-20 30-50 na 4-8
Split Set bolt 50-100 40-80 10-30 80-200 na 5-15
Yielding Swellex 80-90 80-90 10-20 100-150 na 8-12
Yielding Super Swellex 180-190 180-190 10-20 100-150 na 18-25
16 mm Cone bolt 90-140 90-140 10-20 100-200 na 10-25
Surface support elements
#9 gauge welded-wire mesh 12-18 na 100-150 125-175 0.5-1.0 1-4
#6 gauge welded-wire mesh 24-28 na 125-175 150-225 1.5-2.5 4-6
#4 gauge welded-wire mesh 34-42 na 150-200 175-250 2.5-4 6-9
#9 gauge chain-link 32-38 na 400-450 >400-450 3-4 10-12
Expanded metal 24-26 na 500-650 >500-650 6-7 8-10
#6 mesh-reinforced shotcrete 45-55 na 70-90 100-150 3-5 6-9
Steel fibre-reinforced shotcrete 20-30 na 30-40 na <1 2-3

At the time, this information was obtained predominantly from static (pull) tests performed on
support elements. A convenient way to determine the energy absorption capacity of support
elements was to calculate the area under the load-displacement curve (Figure 4–28). Although
more accurate information on the energy absorption capacity can be obtained since the
development of highly instrumented dynamic testing rigs, the values presented in Table 4–4 are
still used as rough guidelines in the industry.
 96

Figure 4–28 Load-displacement curves for (a) five types of reinforcement elements (Kaiser et al., 1996) and (b)
three types of mesh (Tannant, 2004)

4.2.2 Testing rig facilities

The use of a controlled laboratory facility to test the behaviour of ground support elements
subjected to dynamic loading is desirable from the point of view of the repeatability of the tests
and the relatively low operating cost (Potvin et al., 2010). Hadjigeorgiou and Potvin (2007,
2011) consolidated the information available on nine dynamic testing rigs that have been used
historically for testing the dynamic behaviour of reinforcement and surface support elements.
Out of these nine rigs, the CanmetMINING and Western Australian School of Mines (WASM)
facilities are the only ones in operation currently. These two rigs are used predominantly for
testing the performance of reinforcement elements. Several dynamic tests have been conducted
at the SRK drop weight test facility during the 1990s, particularly on surface support elements.
Given the amount of results that have been published over time from these three rigs, a review of
the SRK, CanmetMINING, and WASM rigs is provided in Appendix B and results on the energy
absorption capacity of reinforcement and surface support elements are presented. The Geobrugg
testing rig was commissioned more recently with the purpose of testing integrated ground
support systems. This rig is also covered in Appendix B.

The reliability and repeatability of dynamic tests conducted in drop-test facilities are desirable to
compare the performance of reinforcement and surface support elements under dynamic loads.
An example of application, which resulted from an extensive review of dynamic tests by
Hadjigeorgiou and Potvin (2011), is presented in Figure 4–29.
 97

Figure 4–29 Interpretation of some of the results from the dynamic testing of surface support performed by Ortlepp
and Stacey (1997) (Hadjigeorgiou & Potvin, 2011)

One of the main shortcomings of dynamic testing rigs, however, is the poor replication of the
rockburst phenomenon through the employed loading techniques. The single pulse generated
during drop tests does not duplicate the low frequency vibrations of large seismic events
(Hadjigeorgiou & Potvin, 2011). Furthermore, since reinforcement and surface support elements
are generally tested separately, testing rigs cannot simulate the interaction between the rock mass
and all the components of a support system. Only the Geobrugg testing facility in Switzerland
has come close to replicate the surface support-reinforcement interactions by testing the
performance of integrated support systems as opposed to individual elements. However, in all rig
configurations, support tendons are tested under axial loading, i.e. the acceleration is in line with
the tendon. Frequent field observations of bent bolts following rockbursts (Figure 4–30) have
demonstrated that the application of dynamic loads on tendons in situ is more complex than what
is being simulated in standard drop tests (Counter, 2012).
 98

Figure 4–30 (a) Bent rebars hanging from the back after rockburst ME-80 on October 15, 2004 at Coleman mine
and (b) ripples cast into rebar by deforming rock mass during transient vibration at Kidd mine (Counter,
2012)

4.2.3 Simulated rockburst experiments

Studies of in-situ rockburst damage can only be carried out after the fact. When relying on such
data, researchers have no control over the location and nature of the seismic source, often leading
to ambiguous results (Heal, 2010). It is possible to gauge the performance of complete ground
support systems, incorporating reinforcement, surface support, and connections using controlled
blasting techniques. Simulated rockbursts are destructive tests, which have been generally
conducted in operating underground mines. The primary objective of those tests is to submit a
range of ground support systems to strong ground motion in the environment in which they are
generally utilized. Blasts are carefully designed to generate ground motions that simulate a
mining-induced seismic event.

Hadjigeorgiou and Potvin (2007) have reviewed seven series of simulated rockburst experiments
performed in Canada, Australia, and South Africa. The influence of rapidly expanding gasses
generated by explosives has been a major concern in many of the attempted simulated
rockbursts. Gas expansion during blasts results in a high frequency shockwave, which attenuates
over time whereas seismic events are typically initiated by a relatively low frequency
compressive wave followed by a larger amplitude shear wave. This limitation cannot be
overcome; the only alternative is to calibrate the results of simulated rockbursts against recorded
rockburst case studies (Hadjigeorgiou & Potvin, 2007).
 99

A large number of simulated rockbursts would have to be performed in a variety of ground

conditions in order to generate robust empirical guidelines on the capacity of support systems
subjected to dynamic loads. The number of tests that can be performed in practice, however, is
limited given the complicated, time-consuming, and costly logistics of simulated rockburst
experiments. The repeatability of simulated rockbursts is also questionable. Consequently, a
comparison between tests may be misleading (Hadjigeorgiou & Potvin, 2007).

According to Potvin et al. (2010), very few simulated rockburst tests have given quantitative
results useable for a factor-of-safety design approach of ground support subjected to dynamic
loads. The series of tests conducted by Heal (2010) at the University of Western Australia are the
only published simulated rockburst test results interpreted in terms of energy absorbed by the
support systems (Potvin et al., 2010). Heal (2010) has performed simulated rockburst tests on ten
ground support systems. Six test sites and fourteen individual blasts have been used.

The setup used by Heal (2010) consisted of horizontal blast holes drilled parallel to a test wall
(Figure 4–31). A 10 m wall section was typically used for each test, with two ground support
systems installed over 5 m sections (Figure 4–32). Spacing of the reinforcement elements
typically varied between 1.2 and 1.5 m. This setup allowed for direct comparisons of support
performance under very similar dynamic-loading and ground conditions (Hadjigeorgiou &
Potvin, 2007).

Figure 4–31 Conceptual plan and cross-sectional views of a simulated rockburst test site layout (Heal, 2010)
 100

Figure 4–32 Test wall setup used to compare the performance of split sets and masterseal on the left, and split sets
and mesh on the right (Heal, 2010)

Instrumentation used by Heal (2010) consisted of observation holes for borehole camera logging,
one triaxial geophone per support panel, and a set of rockbolt-mounted and surface-mounted
geophones. Data were recorded by a 16 channel Impulse seismic monitoring system and an 8-
channel Ground Motion Monitor (GMM). Detailed mapping of the walls was done before and
after each blast using three-dimensional photogrammetry. Fully-digitised three-dimensional
images were used to accurately measure deformation of the rock mass and support systems and
identify areas of rock bulking and ejection. All simulated rockbursts were recorded using a
digital video camera.

The tests were conducted using three consecutive blasts. Blasts #1, #2, and #3 were designed to
generate a peak particle velocity (ppv) of 0.5 m/s, 1.5 m/s, and 5 m/s, respectively. Consecutive
blasts were considered to reflect the repetitive loading experienced in seismically active mines
(Hadjigeorgiou & Potvin, 2007). Packed emulsion was used instead of ANFO to maximise the
release of shock energy while minimising the effects of rapidly expanding gasses on the test site.
The highest ppv achieved throughout testing was 3.5 m/s. It has been suspected that the use of
packaged emulsion prevented a uniform detonation through the charge column, hence preventing
higher ppvs from occurring (Heal, 2010).
 101

A damage assessment procedure was developed by Heal (2010). The energy demand per surface
area was calculated using the ppv measured by surface-mounted geophones and the damage
thickness assessed using three-dimensional photogrammetry. The severity of damage to the
support was interpreted using the support damage scale rating proposed by Kaiser et al. (1992).
The rating values varied between S0 (conditions unchanged) and S5 (severe damage to support).
The rating S3 (moderate damage to support) applied to observations of local loss of support
functionality that would likely require rehabilitation of the support system. The average energy
demand leading to S3-level damage was plotted for the ten tested ground support systems
(Figure 4–33). Heal (2010) further compared simulated rockburst data to data from rockburst
case histories using the factor-of-safety approach described in Potvin et al. (2010). A relatively
good correlation between the results was found.

Figure 4–33 Average energy demand required to trigger S3-level damage to ground support systems (Heal, 2010)

Practical conclusions were drawn from Heal’s work on simulated rockbursts. First, the test
results demonstrated that plain shotcrete, fibre-reinforced shotcrete, and thin spray-on liners were
not suitable on their own as surface support in rockburst conditions. The tests demonstrated the
consistently better performance of support systems that included mesh over those that did not.
Secondly, the use of energy-absorbing reinforcement elements was found to enhance the
 102

capacity of support systems subjected to dynamic loads. For example, the addition of Conebolts
improved the capacity of the ground support system comprising SS47 split sets and mesh.

Finally, the transfer of dynamic loads to reinforcement elements was directly influenced by the
strength of the surface support. Both mesh and Tunnel Guard, a cementitious mixture combined
with a polypropylene fibre additive (South African Mining and Engineering Supplies, 2006),
have been tested with 2.4 m SS47 split sets and 3 m grouted Conebolts. The ground support
system comprising mesh has been able to absorb more energy than the one with Tunnel Guard.
Heal (2010) suggested that dynamically capable support elements such as Conebolts and
dynamic cables could perhaps have their full capacity engaged during a rockburst when strong
surface support is used.

4.2.4 Capacity of individual support elements vs. capacity of support

systems: how do we fill the gaps?
Simulating rockbursts using blasting appears to be the most realistic replication of seismically
induced dynamic loads on ground support systems. Heal (2010) supported this statement by
identifying similitudes between the capacity of support systems calculated based on simulated
rockburst experiments and from actual rockburst data. Simulated rockbursts, despite the issues
related to gas expansion, capture the interaction between the reinforcement and surface support
elements and the rock mass. However, limitations associated with the high costs and complicated
logistic complicates the testing of several ground support systems. Furthermore, the repeatability
of the tests is often questionable. There would be significant advantages from a design point of
view to be able to estimate the capacity of ground support systems based on drop tests results. In
this section, different conceptual approaches to calculate the capacity of ground support systems
are employed and results are compared to the capacities obtained using simulated rockburst
experiments. The capacity of the following support systems tested by Heal (2010) and
represented in Figure 4–33 is addressed:

• Split sets and fibrecrete

• Split sets and mesh

• Conebolts and mesh with fibrecrete

• Split sets and mesh with fibrecrete

• Split sets and mesh with conebolts

 103

4.2.4.1 Summation of energy-absorption capacities

The approach in estimating the energy-absorption capacity of ground support systems consists of
assuming that the energy-absorption capacity of each support element adds to the system. This is
similar to calculating the total resistance of a series electrical circuit, where the total resistance is
the sum of all the resistances present in the circuit. This approach has been employed in previous
work for estimating the capacity of ground support systems (Albrecht, 2005). The following
equation was used:








































































































where ei is the energy-absorption capacity of a support element installed per unit area of support
and A is the unit area of support.

In this calculation, a 1.2 m x 1.2 m bolting pattern was considered in order to be consistent with
the testing setup used by Heal (2010). For the support system comprising of split sets and
conebolts, it was assume that on a pattern of four bolts, two were split sets and two were
conebolts. The surface support reported as mesh by Heal (2010) was assumed to be #6 welded-
wire mesh and the capacity of mesh with fibrecrete was taken as the one of #6 mesh-reinforced
shotcrete. Given the inherent variability in the drop tests data from different testing sites, the
energy-absorption capacities of surface support elements were taken as is, regardless of the size
of the tested panel, either 1.0 x 1.0 m or 1.2 x 1.2 m panels. The capacities of the individual
support elements used in the calculations are presented (Table 4–5). The estimated energy-
absorption capacities of ground support systems are presented (Table, 4–6).
 104

Table 4–5 Energy-absorption capacity, determined using drop tests, of some of the reinforcement and surface
support elements tested as part of simulated rockburst experiments by Heal (2010)

Support Energy Reference Comments

element capacity
(kJ/m2)
39 mm friction 12 kJ Ortlepp and The split set bolt survived 7.7 and 11.6 kJ
sets Stacey (1998) impacts and failed under 15.4 kJ.
Modified 32 kJ Player (2008) The bolt was tested under 30 to 35 kJ input
conebolts energy (single drop).
Shotcrete with 10 kJ Ortlepp and Tested under 10.3 to 20.6 kJ impacts. Upper
30 mm Stacey (1997) limit of performance capability against a
Dramix steel single dynamic impulse on a 1.0 m x 1.0 m
fibre panel: 20 kJ. A 50% energy loss in the system
was assumed (Hadjigeorgiou & Potvin, 2011)
1
#6 welded- 5 kJ Ortlepp and Upper limit of performance capability
wire mesh Stacey (1997)1 against a single dynamic impulse on a 1.0 m
Kaiser et al. x 1.0 m panel: 10 kJ. A 50% energy loss in
(1996)2 the system was assumed (Hadjigeorgiou &
Potvin, 2011)
2
4-6 kJ, based on load-displacement curves
obtained from pull tests (1.2 m x 1.2 m)
(Tannant, 2004)
#6 mesh- 11 kJ (Tannant & Drop tests conducted on 1.2 m x 1.2 m
reinforced Kaiser, 1997) panels. Moderate damage to the support
shotcrete occurred between 9 and 12 kJ impact energy.

4.2.4.2 Weakest link

The energy absorption capacity of ground support systems can be estimated based on the
assumption that the ground support works as an integrated system which is only as strong as its
weakest link (Potvin et al., 2010). Based on this assumption, the total energy-absorption capacity
of a support system corresponds to the energy-absorption capacity of its weakest element. This is
similar to the total resistance of a parallel electrical circuit, which never exceeds the value of its
smallest resistance.
 105

The estimated values of energy-absorption capacity of ground support systems based on the
summation and weakest-link assumptions are presented (Table 4–6). Those values are compared
to the capacities obtained through simulated rockburst experiments (Heal, 2010).

Table 4–6 Energy-absorption capacity of ground support systems determined based on the summation and
weakest-link assumptions and compared with the results of simulated rockburst experiments conducted
by Heal (2010)

Support Reinforcement Surface System’s energy capacity (kJ/m2)

system (1.2 m x 1.2 m) support Summation Weakest link Simulated
rockbursts
1 Split sets Fibrecrete 20 10 1.5
2 Split sets Mesh 15 5 4.4
3 Conebolts Mesh with 37.7 11 5.4
fibrecrete
4 Split sets Mesh with 21 11 6.3
fibrecrete
5 Split sets and Mesh 23.2 5 7.6
conebolts

4.2.4.3 Limitations in estimating the energy-absorption capacity of

ground support systems
There is an important discrepancy between the conceptual energy-absorption capacity of support
systems estimated based on drop-test data and the capacity obtained through simulated rockburst
experiments. The weakest link assumption would appear to be the most conservative scenario
whereas the summation assumption would be the most optimistic. Nevertheless, results from
simulated rockbursts tend to range below the capacity obtained based on the weakest-link
approach. This may suggest that the actual weakest link in a support system is not the
reinforcement or surface support element that present the least energy-absorption capacity, but
perhaps the connections. Connections between surface support and reinforcement elements are
rarely tested during dynamic tests. Only the Geobrugg facility, by testing full-scale support
systems, allows this simulation.
 106

Furthermore, by comparing the performance of system #3 and system #4 (Table 4–6), it was
observed that in situ, under dynamic loads generated by blasting, the split set performed better
than the conebolt. In those tested ground support systems, both types of bolts were paired with
the same surface support element, i.e. mesh and fibre-reinforced shotcrete. This result does not
match the trend obtained in the laboratory, where conebolts are capable of absorbing higher
levels of kinetic energy. The capacity of the split set highly depends on the interaction between
the surface of the bolt and the rock mass. The replication of this level of interaction in the
laboratory is questionable. Hence, it is possible that the dynamic-test results on split sets are
misleading. This may also be the case for the drop test results on shotcrete. The capacity of fibre-
reinforced shotcrete panels tested by Ortlepp and Stacey (1997) did not appear to contribute
significantly to the capacity of the support systems during simulated rockbursts, despite their
relatively high energy-absorption capacity identified during impact testing.

This exercise has demonstrated that it is not possible to calculate the energy-absorption capacity
of ground support systems realistically based on the performance of individual support elements
under impact tests. As it was previously stated by Stacey (2012), whilst data are available on
individual support components, knowledge of the capacities of rock support systems, from
theoretical calculations or from practical testing programmes, is absent. Since neither the
demand imposed on the support, nor the capacity of support systems under dynamic loading, can
be defined, a conventional engineering design approach is therefore not possible given current
understanding.

4.3 Empirical approaches toward the selection of ground support

under dynamic-loading conditions
Empirical approaches were developed over time to assist with the selection of ground support
systems subjected to dynamic-loading conditions. An early approach based on the estimated
stress conditions was developed for square tunnels developed in massive quartzite, in South
Africa (Hoek & Brown, 1980). Various support systems employed at the time were
recommended for different ratios of far-field maximum principal stress (σ1) to unconfined
compressive strength of the surrounding rock (σc) (Figure 4–34). The following
recommendations were made at the time:
 107

• σ1/σc = 0.1: stable unsupported tunnel

• σ1/σc = 0.2: minor sidewall spalling

• σ1/σc = 0.3: severe sidewall spalling

• σ1/σc = 0.4: heavy support required

• σ1/σc ≥ 0.5: possible rockburst conditions

Figure 4–34 Empirical stability classification developed for square tunnels in massive quartzite in South Africa (Hoek
& Brown, 1980)

The Q system (Barton, Lien, & Lunde, 1974) is often employed for the design of ground support
systems in mining. In the calculation of the rock mass quality index Q, the stress reduction factor
(SRF) is employed. In concept, in competent rock, mild to heavy rockburst conditions are
associated with ratios of unconfined compression strength to major principal stress (σc/σ1) lesser
than 5. Barton et al. (1974) originally associated this range of strength to stress ratios to SRF
values varying between 5 and 20. The SRF assessment was reviewed further for rock stress
problems in competent rock, with the SRF range associated with strength to stress ratios lower
than 5 varying from 5 to 400 (Barton, 2002). The SRF is the only parameter that composes the Q
index that addresses the potential for rockburst and its assessment, based on the updated SRF
values, is quite subjective.
 108

The Q index and the equivalent dimension of underground openings, i.e. a function of the size
and purpose of excavation, define the support guideline proposed by Barton et al. (1974).
Nevertheless, although this guideline is employed in mining, very few case studies that
contributed to its definition were actually associated with temporary mining excavations, i.e. the
excavations that are the most susceptible to rockbursts. The majority of case studies used to
develop the Q system were from civil engineering works, e.g. water tunnels for hydropower,
storage rooms, portals, water treatment plants, road and railway tunnels, etc. Consequently, the
recommended support systems were mostly representative of the construction industry as
opposed to mining. Furthermore, given the very few number of temporary mine openings
amongst the case studies (2 out of 191) and the time at which the Q system was developed, the
validity of the recommendations for burst-prone ground conditions are highly questionable.

A more recent approach was developed by Heal (2010) based on 254 rockburst case studies from
11 Australian and Canadian mines. Heal (2010) introduced the excavation vulnerability potential
parameter (EVP) to characterize the susceptibility of mine excavations to rockbursts based on the
characteristics of the opening. The EVP accounts for the stress conditions at the rockburst
location (E1), the capacity of the installed ground support system (E2), the excavation span (E3),
and the presence of seismically active geological structures (E4) (Eq.4–13).

EVP = (Damage Initiation Factor) x (Depth of Failure Factor)

= (E1/E2) x (E3/E4) (Heal, 2010) (4–13)

The rockburst damage potential (RDP) was obtained by multiplying the EVP by the ppv obtained
using the scaling law proposed by Kaiser et al. (1996). The RDP parameter aims at forecasting
the severity of a rockburst, based on the characteristics of the opening and the dynamic-loading
conditions generated by a seismic event, which is reflected by the ppv (Figure 4–35).

Despite the relatively good correlations observed in the rockburst data presented by Heal (2010),
important limitations affect its applicability when it comes to selecting appropriate ground
support systems to mitigate rockbursts in a seismically-active mine. First of all, the E1
parameter, which was presented as the ratio of the mining-induced σ1, obtained using numerical
modelling, to the UCS of the surrounding rock was in fact, in many of the data analysed by Heal
(2010), the ratio of the far-field σ1 to the UCS. Any criterion defined using the far-field stress
 109

conditions does not allow the identification of high-stress or burst-prone areas on a given mine
level.

Figure 4–35 The rockburst damage potential (RDP), i.e. the product of the excavation vulnerability potential (EVP)
and the peak particle velocity (PPV), introduced by Heal (2010) to forecast the severity of rockbursts

Furthermore, the application of a ppv scaling law that was based on an upper-bound confidence
limit regression was criticized (Kaiser & Cai, 2013). It was argued that this ppv guideline was
strictly developed for design purposes and was not to be expected to reflect the dynamic-load
demand on the support as part of forensic analyses. Furthermore, the availability of seismic data
in Heal’s work was questionable. It appeared that the location of seismic events that triggered
rockbursts was often assumed and not validated using seismic monitoring data. Finally, the
capacities of ground support systems were roughly quantified using a scale of 2, 5, 8, 10, and 25.
A simple classification like the one employed by Heal (2010) is unlikely to capture the
complexity of the loading mechanisms that may apply on the support.

The EVP approach developed by Heal (2010) was applied to rockburst case studies from
LaRonde Mine located in the Abitibi region of Québec, Canada (Turcotte, 2014). Rockbursts
from LaRonde Mine were not part of Heal’s rockburst catalogue; therefore, this was an
independent assessment of the reliability of the EVP parameter (Figure 4–36).
 110

Figure 4–36 The relationship between EVP and ppv for rockburst case studies from LaRonde Mine (Turcotte, 2014)

A large proportion of rockbursts at LaRonde Mine were characterized by ppv smaller than 1 m/s,
compared to the rockburst data from Heal’s catalogue. The important proportion of larger ppv
values in Heal’s dataset is reflective of the several seismic events interpreted as strainbursts,
which were assumed to have occurred within 5 m from the damage as opposed to being verified
using seismic monitoring data. On the chart from LaRonde Mine (Figure 4–36), little correlation
is observed between the severity of damage, ppv, and EVP. The dependence of rockburst
severity on the EVP is therefore questionable at LaRonde Mine.

A similar trend was obtained at the Kirunavaara Mine (Woldemedhin & Mwagalanyi, 2011).
Several damage occurrences were observed to have occurred under very low ppv of less than
50 mm/s. In the assessment of damage severity as a function of the EVP and the ppv, the upper
limit scaling law proposed by Kaiser et al. (1996) was employed in order to be consistent with
Heal’s work. The empirical diagram showing the severity of rockburst damage at the
Kirunavaara Mine over the ranges of damage severity forecasted by Heal (2010) is presented
(Figure 4–37). All rockbursts plotted in the R1 zone, which corresponds to no observed damage
to the support (Heal, 2010). The rockbursts recorded at the Kirunavaara Mine, however, were
characterized by damage levels 2 to 5, which suggest the displacement of 0.5 m3 up to 50 m3 of
material. This assessment demonstrates the inability to predict rockburst severity using of the
EVP and ppv at the Kirunavaara Mine.
 111

Figure 4–37 Rockbursts from the Kirunavaara mine plotted over the empirical chart proposed by Heal (2010) to
assess the potential for rockburst damage by relating the largest probable ppv to the EVP
(Woldemedhin & Mwagalanyi, 2011)

One of the limitations of the EVP-ppv approach is the inability to account for variations in the
recorded damage mechanisms. All rockburst data collected by Heal (2010) were processed
processed together, without consideration of whether the seismic events that triggered damage
were of strain-burst or fault-slip nature. By dissimulating the magnitude parameter within the
ppv, important information on the seismic source and damage mechanisms is lost.

At LaRonde Mine, buckling is a frequent failure mechanism during rockbursts. This failure
mechanism is strongly influenced by the steeply dipping schistosity, parallel to the orebody. This
is reflected by the important number of wall failures (51) as opposed to back failures (8), which
occurred between 1999 and 2013 at the mine. Turcotte (2014) adapted the EVP index to account
for the ground conditions and dominant failure mechanism encountered at LaRonde Mine. The
E5 parameter, which multiplies the EVP, was introduced (Turcotte, 2014). This parameter
reflects the angle of intersection of the schistosity by the drift where the rockburst occurred. By
adjusting the EVP to account for the dominant rockburst damage mechanism identified on site,
the correlation with the ppv was strongly improved (Figure 4–37).
 112

Figure 4–38 Example of application of the adjusted EVP to the analysis of support performance at LaRonde Mine
(Turcotte, 2014)

The approach employed at LaRonde Mine demonstrated that empirical approaches are often data
and site specific. In order to be applicable to the design of ground support strategies, they need to
be verified using local field data and fine-tuned locally.

4.4 Summary
Chapter 4 consisted of a literature review, which aimed at providing an adequate background on
parameters related to the dynamic-load demand generated by rockbursts and on the dynamic
performance of ground support systems. Theoretical concepts aiming at understanding and
quantifying the dynamic-load demand on ground support systems were presented. Furthermore,
methods employed to assess the energy-absorption capacity of ground support systems, i.e.
extrapolation of static test results, dynamic testing rig facilities, and simulated rockbursts, were
presented. It was concluded that in the context of rockbursts, both the demand and the capacity
cannot be quantified given current understanding. Consequently, a conventional factor-of-safety
approach is not possible for the design of ground support systems aiming at mitigating
rockbursts.
 113

Early empirical approaches for the selection of ground support systems subjected to dynamic
loading conditions (Barton et al., 1974; Hoek & Brown, 1980) were discussed. A more recent
approach by Heal (2010) was more elaborate and supported by more rockburst case studies;
hence, it has been critically reviewed. Limitations due to the employed parameters and the
quality of the data have been identified and the applicability of the proposed methodology has
been questioned. The work of Turcotte (2014) demonstrated that in order to be applied with
confidence, empirical design methods must be verified using local field data and fine-tuned
accordingly.
 114

5 Rockburst Data Collection and Validation

Chapter 4 made a strong case demonstrating that a conventional approach to the design of
ground support systems exposed to dynamic-loading conditions is not suitable given the level
design indeterminacy. It was argued further that the quantity and quality of the data employed in
developing the existing empirical design approaches were deemed questionable. The Canadian
Rockburst Research Handbook, a practical reference on support design for burst-prone ground
conditions, was published in 1996. Almost 20 years later, there is no reason why the design of
ground support systems in deep underground mines should rely on concepts that are not
validated using high quality field data. It is possible, given the experience developed over time in
managing dynamic-loading conditions in Canadian mines, to use existing rockburst data to
conduct passive monitoring of support performance.

This research distinguishes itself by the quality and quantity of the collected rockburst data. Case
studies of support performance under dynamic loads were gathered over the period of May 2010
to September 2013 in three deep underground mines of the Sudbury Basin. A rockburst database
was constructed of 209 seismic events, which led to 324 damage locations at Creighton, Copper
Cliff, and Coleman mines. Focusing on three deep underground mines contributed to increasing
the quality of the collected data by getting access to the local resources, including the
experienced ground control personnel at the mine sites. In-depth and consistent analyses of
quantitative parameters such as the magnitude and mining-induced stresses were possible due to
the understanding of the mining sequence developed over the time spent at the mines. Time-
continuity in the collected data allowed the identification of trends in the evolution of rockburst
frequency and severity in relation to the evolution of ground support practice. In summary, the
developed rockburst database is comprised of high quality and verified data and is complete for
the time-period covered by this research.

The information reported in the rockburst database is presented in this chapter. Some of the
information was taken directly from the “Unusual Occurrence Reports on
Groundfall/Rockbursts” (WSN, 2010) whereas other parameters had to be obtained through
 115

subsequent analyses. The process of data collection and validation is presented in this chapter.
This information will be carried over the next chapter, which discusses the multivariate statistical
analysis performed on the collected rockburst data.

5.1 Rockburst field-data collection campaign

A comprehensive rockburst-data collection campaign was undertaken at Creighton, Copper Cliff,
and Coleman mines from May 2010 to September 2013. The collected information was
reviewed, validated, and further stored in a comprehensive rockburst database. An invaluable
source of information on site has been the “Unusual Occurrence Reports for Groundfall/
Rockburst” (WSN, 2010) prepared by the mine personnel at the time. Ontario mines are in the
legal obligation, under the Occupational Health and Safety Act of Ontario, to fill these reports
when a rockburst occur causing damage to equipment or the displacement of more than five
tonnes of material (WSN, 2010). The WSN report template, however, has been further used at
Creighton, Copper Cliff, and Coleman mines to internally document least severe rockburst
occurrences. In the reports, the areas affected by rockbursts were identified and an assessment of
the resulting damage was provided. Pictures of the damage to the support and mine level plans
indicating the damage location(s) were complementary sources of information typically attached
to the reports. Five sections of the WSN report template were used to report the information
relevant to rockburst investigations. Table 5–1 describes the purpose of each section.

Table 5–1 Summary of the purpose of each section of WSN’s Unusual Occurrence Report on
Groundfall/Rockburst (WSN, 2010)

Report section Purpose

1. General To identify the mine site where the incident occurred, the date and time of
the incident, and to provide a general description of the incident.
2. Workers To provide information about the location of workers with respect to the
event and any injury suffered, if applicable.
3. Description of To provide information about the location of the occurrence, the damage
Occurrence sustained, rock mass characteristics and the failure mode.
4. Ground To provide information concerning ground support systems used in or
Support surrounding the incident area(s) and the performance of support elements.
5. Rockburst To report on mine seismicity associated with the incident and on the
Section apparent rockburst mechanism.
 116

The WSN template imposed a level of consistency over time in the type of information reported
as part of rockburst site-investigations. This template was used consistently and appropriately at
the three mines, therefore enabling data comparison in this study. The WSN report template was
best suited for reporting damage to a unique mine location, typically resulting from a strain burst.
This limitation had an effect on the quality of the data when reporting on large-magnitude fault-
slip events that typically resulted in damage to several mine locations. For such events, variations
in the quality and accuracy of information depended on the experience of the ground control
personnel at the time. Consequently, interpretations related to the seismic source mechanism and
the ground support performance, for example, were sometime subjected to personal
interpretations.

All reported rockbursts were investigated during the undertaken data collection. The information
was verified and validated in order to ensure consistency throughout the database. Given the
limitations in the employed report template, the level of assistance brought by the experienced
ground control personnel at the mines is acknowledged. On many occasions, especially in
investigating rockbursts that resulted from large-magnitude seismic events, it was necessary to
clarify the location of the damage within the mine and the ground support elements that
comprised the installed support systems. Throughout the verification process, complementary
information obtained on site by interviewing the mines’ ground control personnel was of
invaluable importance. In order to increase the confidence in the accuracy of the collected
rockburst data, the information gathered from the WSN reports was further cross-referenced and
complemented using other available sources of information. Those other sources of information
consisted of:

• Local seismic monitoring systems;

• National Earthquake Database (National Resources Canada, 2015);

• 2D and 3D mine surveying layouts;

• Geological mapping layouts;

• 3D numerical elastic stress modelling;

• Ground support standards;

• Technical reports;

• Site inspections.
 117

The data collection and verification process, depicted in Figure 5–1, resulted in a database of
rockburst case studies that were verified and validated. For the purpose of this thesis, the
“Unusual Occurrence Reports for Groundfall/ Rockburst” provided a starting point to investigate
the performance of ground support systems under dynamic-loading conditions. Although some
minor inconsistencies were identified and further addressed through the verification process, the
WSN reports provided an excellent overview of the rockburst occurrences. Most importantly,
those reports were the only documents in which field observations of ground support
performance under rockburst conditions were consistently reported. The quality of the data and
the quantity of variables reported in the database were increased significantly by consulting the
complementary sources of information listed previously. A more comprehensive database of
rockburst case studies has resulted from this process.

Figure 5–1 Schematic of the rockburst data collection and verification process applied at Vale mines

In reviewing over 200 “Unusual Occurrence Reports for Groundfall/Rockburst”, limitations of

the WSN template employed at the time were identified. In order to improve the process of
reporting rockburst occurrences in Ontario mines, an updated version of the template was
proposed based on site recommendations and research experience. The template benefited from
 118

added sections for reporting on multiple damage locations. This aimed at a greater level of clarity
in the reported displaced tonnage, installed ground support systems, etc. The section on the
displaced rock tonnage was divided in order to make a distinction between volumes of rock that
were contained and uncontained by the installed ground support systems as well as the material
displaced from unsupported ground, e.g. lower walls. Furthermore, since several guidelines on
estimating the dynamic-load demand on the support involve the calculation of the ppv using
scaling laws, reporting the 3D coordinates of both the seismic event associated with the
rockbursts and the damage locations was deemed relevant. A list of attachments to the report
(pictures, mine level plans, etc.) was also perceived as an important addition to the template. The
proposed template from this research was accepted and adopted by WSN on November 28, 2013
and is provided in Appendix A. It is believed that the updated WSN template is best suited for
reporting on damage associated with rockbursts as opposed to falls of ground that occurred under
static conditions.

5.2 Pertinent variables used in passive monitoring of support

performance under dynamic loads
In the developed rockburst database, pertinent information was organized based on whether the
collected data were of quantitative or qualitative nature. The WSN reports provided a consistent
description of the rockbursts recorded over time. The reports further provided the essential
information on the installed ground support elements and their performance. Nevertheless, few
of the variables employed to characterize the dynamic-load demand on the support were directly
extracted from the WSN reports. As previously discussed and illustrated (Figure 5–1),
complementary sources of information were accessed in order to extract and validate the
majority of the quantitative variables.

In the subsequent multivariate statistical analysis performed on the collected rockburst data, the
investigated variables are categorized as predictor or response variables. Predictor variables (xi),
also referred to as explanatory variables, are used to explain the variations in dependent or
response variables (yi) (Umetrics, 2008). In this section, predictor and response variables
employed to characterize the dynamic-load demand on the support and the severity of rockburst
damage are discussed.
 119

5.2.1 Predictor variables

Table 5–2 reports the predictor variables that were expected to influence the support
performance during rockbursts. Variables from the first two categories were related to the
dynamic-load demand on the support. Those variables were classified based on whether they
defined the seismic or excavation conditions. Seismic conditions consisted of the seismic event
magnitude, its proximity to the damage, the interpreted seismic source mechanism, and the
resulting damage mechanism. A relevant indicator of the damage mechanism was whether the
seismic event generated localized damage or damage to multiple mine locations. Based on the
revision of over 200 WSN reports at Creighton, Copper Cliff, and Coleman mines, there was a
valid perception that seismic events that generate mine-wide damage tend to initiate seismic
shaking, as opposed to bulking due to fracturing. Stress and ground conditions, as well as the
presence of geological structures and the dimension of openings, contributed to what was defined
as the excavation conditions.

Table 5–2 Predictor variables reported in the rockburst database

Categories Quantitative Qualitative

Seismic conditions • Nuttli magnitude (MN) • Interpretation of the seismic
• Distance between seismic source mechanism (strain burst,
source and damage fault slip)
• Damage to single or multiple
locations
Excavation • Depth of mine opening • Large-scale geological structure
conditions • Differential stress normalized to intersecting the damage location
(stress, ground and the UCS at the damage location (fault/shear zone, dyke)
dimensions)
• Distance from the nearest large-
scale geological structure
• Rock Mass Rating (1989)
• Dimension of mine opening
(span, wall height)
Ground support • Length of reinforcement • Installed reinforcement and
elements surface support elements
 120

The last category of variables was related to the capacity of the installed ground support systems.
This category accounted for the type of installed reinforcement and surface support elements as
well as the length of the reinforcement. A brief description of each collected variable is provided.

Magnitude: The seismic source associated with each rockburst was verified using the mines’
seismic data. The local seismic monitoring systems employed at Creighton, Copper Cliff, and
Coleman mines have been calibrated to the Nuttli scale used by the Geological Survey of Canada
as part of this study.

Distance: The distance between the epicentre of the seismic event and the damage was
interpreted as a straight line and calculated using the damage and seismic event coordinates,
extracted from the CAD and seismic monitoring systems, respectively. An inherent assumption
in this approach is that seismic events are considered as point sources by the employed seismic
monitoring systems. This may be appropriate for smaller events but may not always be the case
for large magnitude seismic events, which are usually slips along planes of weakness, such as
faults, rather than point sources. The practical implication is a lower confidence in the calculated
locations of large fault-slip events as opposed to smaller and medium size events. The inherent
variability in the collected seismic data, due to the employed assumptions, is readily
acknowledged. Taking this into consideration, the distance parameter, in complement to the
magnitude of seismic events, provided relevant and useful information on the level of seismic
loading faced by the support installed at the damage locations.

Seismic source mechanism: The seismic source mechanism is predominantly based on the
interpretation of the ground control personnel, as reported in the WSN reports. This
interpretation was not typically validated using advanced seismic data analysis methods due to
the limited confidence in the advanced parameters recorded by the local microseismic systems,
e.g. the S:P energy ratio. This characterisation was mostly based on the recorded magnitude, the
location of the seismic event with respect to the damage and geological structures, the nature of
damage observed, and whether the event triggered damage to a single or multiple locations.
Typically, the terms strain bursts and fault slips were used loosely in classifying the seismic
events and/or rockbursts, and essentially established a distinction between self-initiated and
remotely-triggered rockbursts.
 121

Depth: The depth of mine openings affected by rockbursts was accurate and verified for each
damage location using the mines’ CAD system. This parameter represented the most notable
distinction between the three mines. It further provided relevant information on the far-field
stress conditions.

Normalized differential stress (NDS): Seismicity and rockbursts are a direct consequence of
mining; therefore, in conducting passive monitoring of support performance, it was important to
quantify the mining-induced stress conditions. Mining-induced stresses at the time and location
of rockbursts were obtained through 3D numerical elastic stress modelling and quantified using
the differential stress normalized to the UCS, i.e. (σ1-σ3)/UCS. Numerical stress modelling and
the relevance of the NDS parameter are discussed further in the chapter.

RMR: The Rock Mass Rating (Bieniawski, 1989) provides a very broad characterization of the
rock mass quality for each geological unit where rockbursts have occurred.

Geological structures: The distance between the nearest large-scale geological structure and the
damage locations was taken into consideration by referring to geological mapping data. The type
of structure, whether dyke, fault or shear zone, was reported when it intersected the damage
location.

Dimensions of mine openings: The dimensions of mine openings subjected to rockbursts were
verified using the CAD system and reported in order to illustrate the potential for large falls of
ground, typically associated with larger openings. Since variations in the excavation span were
considerably more important than variations in wall height, the span was used in order to
describe the dimensions of mine openings in the statistical analysis presented in the following
chapter.

Ground support elements: The type of reinforcement and surface support elements installed at
each rockburst location were reported. This section was complemented by information on the
length of reinforcement elements and the installation pattern.

5.2.2 Response variables

Response variables were recorded in order to characterize the severity of the rockburst damage.
Quantitative variables characterized the damage in terms of displaced tonnage whereas
 122

qualitative variables reflected the performance of individual reinforcement elements or of the

system.

The displaced tonnage was extracted from the WSN reports directly. There was a need, however,
for a consistent approach in recording rock damage. A five-category rock damage scale (RDS)
has been developed to describe the level of damage caused by rockbursts (Kaiser, Tannant,
McCreath, & Jesenak, 1992). This approach was somewhat limited in that it used as a threshold
for its most severe damage category a displacement of 10 tonnes. Since 1992, when this damage
scale was proposed, there have been numerous incidents resulting in significantly higher levels
of damage necessitating a scale that was more sensitive to these variations. This was addressed in
the work of Heal (2010) who adjusted the rock damage scale (RDS) from Kaiser et al. (1992) to
account for larger events and grouped together events that resulted in no damage or limited
damage. The damage scale suggested by Heal (2010) was used in this work but reference to the
original grouping by Kaiser et al. (1992) was provided (Table 5–3).

Table 5–3 Adjusted Rock Damage Scale

Adjusted RDS Summarized description RDS

(Heal, 2010) (Kaiser et al., 1992)
R1 No damage or slight damage to excavations. Small R1, R2
shards or loose displaced.
R2 Minor damage. Less than 1 tonne displaced from R3
supported excavations or 2 tonnes from unsupported
excavations.
R3 Considerable damage to excavations. Less than 10 R4
tonnes displaced.
R4 Severe damage to excavations. Less than 100 tonnes R5
displaced.
R5 Extreme damage to excavations. 100 tonnes displaced or R5
more.

Information on the ground support performance under dynamic loads was reported in the
“Ground Support” section of the WSN reports (Table 5–4). The employed reinforcement
 123

elements were specified as well as their location (back or walls), the length of the tendon, the
installation pattern, and whether failure was observed or not. Damage to reinforcement elements
was reported by checking the “failed” or “beyond” column. The “beyond” category applied when
rock movement extended beyond the length of the installed reinforcement elements (WSN,
2010).

Table 5–4 Assessment of the support performance as a result of rockburst CR-1433 (August 18, 2012) at
Creighton Mine using the WSN template (WSN, 2010)

Tendon / Location Pattern

Type Length Failed Beyond
Dowel Walls Back Wide Long
Mechanical
5/8” TBE X 8’ 4’ 5’ X
rockbolts
Resin rebar ¾” X 8’ 4’ 5’ X
Friction
FS-46 X 6.5’ 4’ 2.5’
stabilizers
Swellex

Cable bolts

MCB MCB33 X 8’ 4.5’ 5’

Opening Location Depth or Cracked or

Type Broken Failed
Liner Walls Back Length Bulged

Mine screen #4 ga X X 5’ x 11’

Shotcrete Plain Dry X X 3” X X X

Straps (mesh) #0 ga X 1’ x 1’
Used to
Support Deformed Broken Failed
Other System
Walls Back

Comments Regarding Effectiveness of Support Systems: The enhanced support installed in the 7200L Aux
ramp was effective in preventing greater damage in the area. The support on 7220L also served to contain 75lbs
of rock within the screen.

The employed surface support elements, e.g. mine screen, shotcrete, mesh straps, etc, was
specified further in the reports. The excavation surface on which the support was installed, the
dimensions, e.g. size of mesh sheets or thickness of shotcrete, and the level of damage suffered
by the surface support elements were reported as well. The damage to surface support elements
 124

was characterized as “cracked or bulged”, “broken”, or “failed”. A section of the WSN report
was available for the ground control personnel to write comments regarding the effectiveness of
the employed support systems. This section was useful for external reviewers as it often provided
more accurate explanations about the support performance.

Although the interpretation of the suggested level of damage can be somewhat subjective, the
following descriptions were generally accepted on site:

• Bulged: Applies to liner or surface support (e.g. welded-wire mesh, shotcrete) and
indicates deformation of the liner that contained the rock, without breaking or failing.

• Broken: Indicates broken strands or cracked shotcrete but the rock mass is contained or
partially contained.

• Failed: The liner has not contained any of the moving rock mass. Bolts or plates were
snapped and broken but part of the tendon was still anchored in the rock mass.

• Beyond: Applied only to reinforcement elements when the failed or displaced rock mass
extended beyond the length of the tendon.

The collected reports on rockburst occurrences at Creighton, Copper Cliff, and Coleman mines
provided useful information on the performance of individual ground support elements. Most
relevant, however, was the assessment of the performance of the system as a whole.
Observations reported in the WSN reports, often complemented with pictures of the damage to
the support, were used in order to rank the performance of ground support systems based on the
Support Damage Scale (SDS). This procedure was proposed by Kaiser et al. (1992) and
employed explicitly by (Mikula, 2012). The SDS uses a ranking system similar to the RDS,
which comprises six categories, S0 to S5. These categories were based on descriptions of
damage to the reinforcement elements, mesh, and shotcrete (Table 5–5). The SDS is a qualitative
assessment of the damage whereas the RDS is more quantitative by nature. Kaiser et al. (1992)
applied these procedures to 18 seismic events and the resulting damage at 48 locations in
Sudbury mines between 1985 and 1991.

Mikula (2012) summarized the support damage scale (Kaiser et al., 1992) into steel support
damage (S0-S5) and shotcrete/fibercrete damage categories (SC0-SC5). Mikula (2012) described
 125

ratings 0-2 as acceptable damage, 3 as tolerable damage, and 4-5 as intolerable damage and
provided the following descriptions:

• Acceptable damage (rating 0 to 2) – The damage is sufficiently limited that it needs not
be repaired but it may be repaired if weakening or fatigue is suspected.

• Tolerable damage (rating 3) – The damage requires a level of repair or rehabilitation, but
the repair efforts and costs can be tolerated.

• Intolerable damage (rating 4 to 5) – The damage is so extensive that the repair effort and
cost, if contemplated, would be prohibitively high.
 126

Table 5–5 Support damage scale (SDS) (Kaiser et al., 1992)

Damage General Description Support Damage Shotcrete Damage

Level
S0 Conditions unchanged. No new damage or loading. No new damage or loading.
S1 Support undamaged No damage to any support component. Shotcrete shows new cracks,
but first signs of very fine or widely
distress detectable. distributed.
S2 Slight damage to Plates and wooden washers on some rock Shotcrete cracked, minor
support. bolts are deformed, showing loading. flakes dislodged.
Loading clearly evident Individual strands in mesh broken. Shotcrete is clearly taking
but full functionality Mesh bagged but retains material well. load from broken rock mass
maintained. (mostly drummy).
S3 Moderate damage to Plates, wooden washers, and wood blocking Shotcrete fractured, often
support. on rock bolts are heavily deformed, showing debonded from rock and/or
Support shows significant loading; bolts heads may be reinforcement.
significant loading and sucked into rock. Major flakes possibly
local loss of Mesh torn near bolt heads with some strands dislodged.
functionality; retaining broken and mesh torn or opened at Holding elements mostly
function primarily lost overlapping edges. intact.
(except in laced or Moderate bagging of mesh and isolated
shotcreted areas). failures of rockbolts.
Cable lacing performs well.
S4 Substantial damage to Mesh is often torn and pulled over rock bolt Shotcrete heavily fractured
support. plates; if it did not fail, it is substantially and broken, often separated
More extensive loss of bagged (at capacity). from the rock mass with
retaining and holding Many rock bolts failed. pieces lying on the ground or
functions (except for Rock ejected between support components. hanging from reinforcement.
lacing systems) Cable lacing is heavily loaded with bagged Connections to holding
mesh. elements often failed or
holding elements failed
locally.
S5 Severe damage to Most ground support components broken or For damage level S5,
support. damaged. shotcrete fails to be functional
Support retaining, Most rock bolts fail and rock peels off cable and the left-hand column
holding, and bolts. applies.
reinforcing functions Shotcrete non-functional.
failed. Mesh without cable lacing heavily torn and
damaged.
Cable lacing systems heavily stressed and
often failed.
Notes: 1) The damage indicators listed in this table describe damage that is new and was caused by the
rockburst. If the observer cannot ascertain that the damage was inflicted by the rockburst then the
damage should be ignored for the purposes of damage classification.
2) One or more damage scales may be observed in the same area and should be recorded separately.
3) Rock and support damage levels need not correspond.
4) Because the function of shotcrete support is somewhat different and more complex than for other
support systems, a separate column of indicators is provided over the range of S0 to S4. It is
important to record where shotcrete is present and when it has been used to determine the support
damage level.
5) Failure of rock bolt applies to failure of nut, plate, anchor, or shank.
 127

The descriptions proposed by Mikula (2012) provide a conceptual assessment of support

damage. The importance of these schemes, for practical purposes, is the proposed distinction
between tolerable and intolerable damage. In referring to acceptable or tolerable levels of support
damage, it is implied that the displaced material has been contained by the support, for the most
part. The precise distinction between acceptable and tolerable damage is quite subjective. These
levels of damage are interpreted as successful performance of the support, regardless of the
possible need for rehabilitation. Intolerable damage implies that the safety of mining equipment
or personnel could have been compromised by the support not being capable of containing the
displaced material. Such levels of damage correspond to unsuccessful performance of the
support and the need for rehabilitation is practically certain.

Following a review of the rockburst case studies at Creighton, Copper Cliff, and Coleman mines,
a template providing a visual tool for damage assessment was constructed for the purpose of this
thesis (Figure 5–2, 5–3). This template, based on the SDS guidelines for steel mesh and shotcrete
damage (Kaiser et al., 1992), was applied in assessing the severity of damage to the ground
support systems in the collected rockburst case studies. In underground mines affected by
seismicity, ground control personnel can further use this template as a reference to assess the
level of damage to the support during site investigations. It is believed that the developed visual
assessment tool can contribute to removing some of the subjectivity of assessment and further
improve consistency in the interpretation of support performance. The template can be applied
further as a training tool to assist less experienced ground control personnel in their assessment.
 128

S0–No damage S1–First signs of distress

CR-1418 CR-1416

S2–Loaded, plates deformed, mesh bagged but S3–Heavily loaded, bolts strongly deformed,
functional mesh bagged, torn or open at overlapping
edges, few strands broken,
ME-167 CR-1441

S4–Major damage, broken bolts, mesh failed S5–Complete failure of support components
or bagged to capacity, rock ejected
CR-1419 NM-117

Figure 5–2 Visual tool for the assessment of damage to ground support systems comprising steel mesh, based on
observations of support performance in Sudbury mines
 129

SC0–No damage SC1–First cracks in shotcrete

NM-120 CR-1436

SC2–Shotcrete cracked and loaded SC3–Shotcrete fractured, debonding, some

fragments

CR-1399 ME-126

SC4–Shotcrete heavily fractured, large pieces SC5–Shotcrete non-functional

fallen
CR-1421 NM-115

Figure 5–3 Visual tool for the assessment of damage to ground support systems comprising plain or reinforced
shotcrete, based on observations of support performance in Sudbury mines
 130

5.3 Validation of seismic event magnitudes

Creighton, Copper Cliff, and Coleman mines are equipped with state-of-the-art seismic
monitoring systems, which provide good coverage in the vicinity of the mines’ infrastructures.
This has allowed the collection of high quality seismic data for the purpose of this thesis.
Microseismic systems record arrival times and locations of seismic events and estimate source
parameters such as seismic energy, seismic moment, local magnitude, S:P energy ratio, etc.
Macroseismic systems, also referred to as strong ground motion (SGM) systems, are used for
recording the magnitude of large seismic events with greater accuracy. The SGM systems at
Creighton, Copper Cliff, and Coleman mines consist of a single or a combination of tri-axial
geophones installed at distant locations from the mining areas. The employed sensor
configurations allow for capturing magnitudes larger than 0.4 MN at Creighton, 1.0 MN at Copper
Cliff, and 0.1 MN at Coleman with a high level of confidence. SGM systems at the three sites are
calibrated with the Geological Survey of Canada (GSC); therefore, the magnitude is interpreted
using the Nuttli scale.

5.3.1 Adjusted magnitude scales

Seismic events that resulted in damage to excavations at Creighton, Copper Cliff, and Coleman
mines were typically recorded by the SGM systems. At Coleman, however, 24 seismic events
associated with rockbursts between 2006 and 2009 were only recorded by the microseismic
system. The magnitude estimates from the SGM systems employed at the three mines and from
the microseismic system at Coleman were cross-validated by comparing seismic events recorded
both locally and at the GSC. Large-magnitude seismic events recorded by the GSC were
obtained through the National Earthquake Database (National Resources Canada, 2015). The
objective of this validation process was to ensure that magnitudes recorded at the three different
sites and reported in the rockburst database were comparable to each other. This comparison is
presented for each seismic monitoring system in the rockburst database (Figure 5–4). The
collected raw data were primarily fitted using linear regression, however, when a distinct
structure was observed in the residuals, polynomial regression was used to improve the quality of
the fit.
 131

Figure 5–4 Cross-validation of magnitude estimates: comparison of raw magnitude data from Creighton, Copper
Cliff, and Coleman mines with magnitudes recorded by the GSC

At Creighton Mine, the employed SGM system was upgraded from a Hyperion Digital Drum
Recorder (HDDR) to a Paladin 24-bit recorder in 2008. The accuracy of event magnitudes
recorded at Creighton Mine was assessed separately for the two SGM systems (Figure 5–4a,b).
 132

At Coleman Mine, it was justified to analyse the magnitude events recorded from 2006 to 2011
separately from those recorded in 2012 and 2013, as different trends were observed in the data
(Figure 5–4e,f). A calibration of the SGM system completed in January 2012 at Coleman was
suspected to be the source of this variation.

The regression relationships presented in Figure 5–4 were used to adjust the raw magnitude data
(Figure 5–5). The significance of these adjustments was assessed using the analysis of variance
(ANOVA) technique (Montgomery & Runger, 2007). For these adjustments to be validated and
accepted, the regression models had to differ significantly from the 1:1 dashed reference line
(Figure 5–4). The reference lines that intersect the charts’ origin represent an ideal calibration of
the local seismic monitoring system with the GSC. In order to apply the ANOVA technique, the
response variable y was considered as the difference between the local and GSC magnitudes.
This led to a new graphical representation which implied that a perfectly calibrated seismic
monitoring system at one of the mines would have its recorded magnitude data aligned on the x-
axis, as all recorded magnitudes would be equal to the GSC Nuttli magnitude. Furthermore, a
slope equal to 0 would characterize the regression line fitted to the raw magnitude data from such
a well-calibrated seismic monitoring system.

The test statistic for the significance of regression used in ANOVA aims at determining whether
a linear relationship exists between the response variable y and a subset of the regressor variables
x1, x2, …, xk (Montgomery & Runger, 2007). In this analysis, the GSC magnitude is the regressor
variable x1 but it can also be expressed as x2 = x12, e.g. in a polynomial regression. Any
regression model that is linear in parameters, i.e. the β’s in the generalized equation
y = β0 + β1x1 + β2x2 + … + βkxk, is a linear regression model, regardless of the shape of the
surface it generates. The hypotheses tested using ANOVA are:
• H0: β1 = β2 = … = βk =0,
• H1: βj ≠ 0 for at least one j.

A failure to reject H0 reflects the absence of a significant linear relationship between the
regressor variable(s) and the response variable. On the other hand, rejection of H0 infers that at
least one of the regressor variables contributed significantly to the regression model. For a
simple linear regression, i.e. y = β0+β1x, this implies that the slope of the fitted regression line,
i.e. β1, is significantly different than 0.
 133

Figure 5–5 Cross-validation of magnitude estimates: adjusted magnitude data

The test statistic for ANOVA is presented (Montgomery & Runger, 2007):









































































































































 134

H0 is rejected if the computed value of the test statistic is greater than f ,k,n-p, obtained from a F-
α

distribution table. In this context, α is the level of significance, i.e. the probability of rejecting
the null hypothesis given that it is true, k is the number of regressor variables xj, n is the number
of observation used in the regression, and p is the number of parameter βj. The level of
significance was set at 0.05 for this analysis. The procedure for testing the significance of
regressions for each local seismic monitoring system is presented in the ANOVA table (Table 5–
6).

Table 5–6 Analysis of variance for testing the significance of regressions on magnitude data from Creighton,
Copper Cliff, and Coleman mines, at a 0.05 significance level

Mine Seismic Source of Sum of Degrees of Mean F0

System Variation Squares Freedom Square
Regression 1.281 1 1.280 13.18
(2000-07)
HDDR

Error 8.260 85 0.097 F0.05,1,85 = 3.95

Creighton

Total 9.541 86 H0 is rejected

Regression 3.010 2 1.505 48.16
(2008-13)
Paladin

Error 1.188 38 0.031 F0.05,2,38 = 3.24

Total 4.198 40 H0 is rejected
Regression 0.000 1 0.000 2.67E-03
Copper

SGM
Cliff

Error 1.563 19 0.082 F0.05,1,19 = 4.38

Total 1.563 20 H0 is not rejected
Regression 0.584 1 0.584 4.34
seismic
Micro-

Error 1.346 10 0.135 F0.05,1,10 = 4.96

Total 1.9303 11 H0 is not rejected
Regression 0.673 2 0.336 14.29
(2006-11)
Coleman

SGM

Error 0.259 11 0.024 F0.05,2,11 = 3.98

Total 0.932 13 H0 is rejected
Regression 0.035 1 0.035 1.63E-01
(2012-13)
SGM

Error 2.373 11 0.216 F0.05,1,11 = 4.84

Total 2.408 12 H0 is not rejected
 135

H0 was successfully rejected in the regressions performed on the HDDR and Paladin data at
Creighton Mine, and in the regression performed on the SGM events that occurred prior to 2012
at Coleman Mine. The null hypothesis could not be rejected, at a 0.05 level of significance, for
the SGM events recorded at Copper Cliff Mine (2004-2013) and Coleman Mine (2012-2013).
The null hypothesis was not rejected either for the regression performed on the microseismic
events at Coleman. Consequently, the regression models fitted on the seismic data from the SGM
system at Copper Cliff and from the microseismic and SGM (2012-2013) systems at Coleman
were not considered as valid. As an alternative, student tests of hypotheses on samples’ mean
(Montgomery & Runger, 2007) were performed to verify whether these seismic monitoring
systems significantly over or underestimated the magnitude of seismic events with respect to the
GSC, using a 0.05 significance level. The results from those tests (Table 5–7) demonstrated that
the HDDR system employed at Copper Cliff Mine significantly overestimated magnitudes by
0.4 MN, on average. The microseismic system at Coleman Mine significantly underestimated
magnitudes by 1.7 MN, on average. Even if the collected seismic data indicated that the SGM
system at Coleman overestimated the magnitudes by 0.2 MN on average between 2012 and 2013,
this was not considered significant, as the null hypothesis could not be rejected.

Table 5–7 One-sided test of hypotheses on sample’s mean testing H0: μ=0 and H1: μ>0 or μ<0 for seismic data
from Copper Cliff and Coleman mines

Copper Cliff Coleman Coleman SGM

Microseismic (2012-13)
Average 0.429 -1.682 0.189
Standard deviation 0.280 0.411 0.448
n 21 12 13
T0 7.026 -14.163 1.523
t0.05,n-1 1.725 1.796 1.771
Result H0 is rejected H0 is rejected H0 cannot be rejected

The magnitude of seismic events recorded by the local seismic monitoring systems were adjusted
using the regression equations (Figure 5–4), when H0 was rejected (Table 5–6). Magnitudes
recorded by the SGM system at Copper Cliff were decreased by 0.4 MN; those recorded by the
 136

microseismic system at Coleman were increased by 1.7 MN (Table 5–7). Magnitudes recorded by
the SGM system at Coleman Mine in 2012 and 2013 were not adjusted since the null hypothesis
could not be rejected in any of the test statistic performed on the data. These adjustments resulted
in the adjusted magnitude data (Figure 5–5). These adjustments were applied to all magnitudes
reported in the rockburst database.

5.3.2 Validation of magnitude adjustments using seismic data analysis

techniques for Creighton’s data
At Creighton Mine, the ground control personnel report large magnitude seismic events recorded
by the SGM system in a retrievable database. This practice ensures that magnitudes and
coordinates of past large-magnitude events remain accessible, even after archiving the seismic
data as part of the system’s maintenance. The current practice consists of reporting all seismic
events that were associated with rockbursts or characterized by a magnitude greater or equal to
0.8 MN. This has allowed access to complete seismic records of large magnitude seismic events
from 2000 to 2013.

The full seismic records were used to verify the influence of the magnitude adjustments on the
seismic data through the application of seismic data analysis techniques. A magnitude-time
history analysis was conducted in consideration of the two seismic monitoring systems used at
Creighton Mine: the HDDR from 2000 to May 2008 and the Paladin starting May 2008 (Figure
5–6). The validation of the magnitude adjustments was supported by the absence of significant
variation in the rate of seismicity starting May 2008. Any variation in the rate of seismicity
coinciding with the change of seismic monitoring system at Creighton would have reflected a
questionable calibration of the two systems. A threshold of 0.6 MN was set in order to prevent
the results and the interpretation to be affected by the different sensitivity levels of the two
seismic monitoring systems or by changes in the thresholds for reporting large magnitude events
over time at Creighton.

The frequency-magnitude relation further supported the validated magnitude adjustments by

showing little variation in the slope of the linear portion of the curve, i.e. the b-value, and in the
x-axis intercept of the linear projection of the curve, i.e. the a/b ratio (Figure 5–7). The a/b ratio
is an indicator of the seismic hazard and a predictor of the largest expected event (Hudyma &
Potvin, 2010).
 137

Labour interruptions

Figure 5–6 Magnitude-time history from January 2000 to September 2013 at Creighton Mine using the adjusted
magnitude

At Creighton Mine, the largest recorded event was a 4.1 MN, which occurred on November 29,
2006. The charts composed of the HDDR and Paladin data have resulted in comparable x-axis
intercepts of 4.0 and 4.1 MN, which demonstrated consistency in the two adjusted magnitude
scales. The charts demonstrated that managing mining-induced seismicity remains an important
challenge at Creighton Mine given the relatively high magnitudes recorded on site. The b-value
is reflective of the predominant seismic source mechanism(s). A low b-value (lower than 1)
suggests that a significant proportion of seismic events are characterized by a fault-slip
mechanisms. Higher b-values, in the range of 1.2 to 1.5, tend to be associated with seismicity
that results from local stress changes, e.g. induced by mine blasting (Hudyma & Potvin, 2010).
Very similar b-values were observed between the adjusted HDDR and Paladin data, i.e. 1.0 and
0.9 respectively. These b-values, not much lower than 1, suggest that there is a combination of
seismic source mechanisms in the data, which is explicable given that the study focused on the
entire mine as opposed to specific areas. Nevertheless, the b-values are reflective of a substantial
proportion of seismic events of shearing nature at Creighton Mine. Based on the magnitude
ranges typically associated with various source mechanisms (Ortlepp, 1997), a proportion of
 138

1.3% of all the seismic events reported in this analysis were characterized by a magnitude
representative of shear rupture or fault slip mechanism.

Figure 5–7 Frequency-magnitude relationships for (a) the seismic events recorded by the HDDR system and (b)
the events recorded by the Paladin. Magnitude ranges proposed by Ortlepp (1997) are presented to
assist in the interpretation of the seismic source mechanisms.

5.3.3 Seismic risk and probability of rockburst at Creighton Mine

Both the complete records of large magnitude SGM events and rockbursts have been collected
for the time period of January 2000 to September 2013. It was possible to assess the probability
of a seismic event of a given magnitude to result in damage to mine openings at Creighton by
calculating the ratio of the number of rockbursts to the number of seismic events, for each 0.1
magnitude increment. This was possible thanks to the performed magnitude adjustments, which
ensured that magnitudes recorded by the HDDR and Paladin systems were comparable. The
results are presented (Figure 5–8). The ratios of number of rockbursts to number of seismic
events were calculated for:
a. all rockbursts;
b. rockbursts that generated more than 10 tonnes of displaced material throughout the mine;
c. rockbursts that generated more than 100 tonnes of displaced material throughout the
mine.
 139

Figure 5–8 Probability of a seismic event of a given magnitude to result in (a) a rockburst, (b) more than
10 tonnes of displaced material within the mine, and (c) more than 100 tonnes of displaced
material within the mine. The probability lines for the three range of damage severity are
plotted in (d).

The regression lines (Figure 5–8) are all characterized by a very good fit (0.89 < R2 < 0.93) and
are interpreted as the probability of seismic events to result in damage within the mine. Given the
time-continuity in the collected seismic data, it was further possible to assess the typical return
period of large-magnitude seismic events (Figure 5–8d). The combination of probability of
rockburst and return-period of seismic events provides a practical tool to assess the seismic risk
at Creighton Mine. Figure 5–8 characterizes indirectly the effectiveness of ground support
practice employed over time at Creighton in managing dynamic loads generated by large-
 140

magnitude events. Based on this chart, a magnitude 1.0 MN has a 1.9% probability of generating
a rockburst and a 0.8% probability of generating more than 10 tonnes of displaced material
within the mine. A magnitude 2.0 MN has a 10.8% probability of generating a rockburst, a 5.2%
probability of generating more than 10 tonnes of displaced material, and a 3.4% probability of
resulting in over 100-tonne damage.

The rockburst probability lines have practical applications for characterizing the increase in the
level of risk associated with an increase in event magnitude. When planning the installation of
ground support systems designed for managing dynamic loads, it is important to identify areas
where seismicity presents a significant risk on the stability of mine openings and on the safety of
mine personnel. For support design purposes, magnitude thresholds can be identified using
Figure 5–8 and further used to assess the geospatial distribution of seismic events that present a
justifiable level of risk within the mine.

5.4 Ground motion levels

The application of ppv scaling laws in estimating the levels of ground motion generated by
seismic events was discussed in Chapter 4. Important limitations of ppv scaling laws were
identified. In particular, the assumption of a spherical radiation pattern and the disregard of
seismic-wave reflection and refraction phenomena can lead to important discrepancy between
the actual and predicted levels of ground motion. Furthermore, the amplification of seismic
waves in the vicinity of excavations is difficult to quantify. As a result, wide ranges of
amplification factors have been proposed in the literature.

Despite the discrepancies between measured and calculated ground motion levels, the ppv is a
convenient approach to represent two important parameters in the assessment of the dynamic-
load demand on ground support systems: the magnitude and the distance from the epicentre of
seismic events. Based on experience in collecting and reviewing large volumes of rockburst data,
it was perceived that large-magnitude seismic events had resulted in damage over considerably
larger distances at Creighton, Copper Cliff, and Coleman mines. It was important, given the high
quality of the collected seismic data, to verify and quantify this perception. The application of a
ppv scaling law was deemed appropriate for this analysis. The scaling law developed by Hedley
(1992) was selected for two important reasons: 1) the guideline was based on data collected in
underground mines from Elliot Lake and Sudbury, in northern Ontario, including Creighton
 141

Mine, and 2) the guideline was based on a best fit of the collected seismic data, as opposed to an
upper limit regression. A guideline based on a best fit of the seismic data was interpreted to be
more appropriate to conduct back analysis of rockbursts as opposed to a guideline based on an
upper limit regression, which could, in theory, be more directly applicable for design purposes, if
previously validated.

Figure 5–9 demonstrates that the trend in the rockburst data from Creighton, Copper Cliff, and
Coleman mines is captured adequately by the selected ppv scaling law. It is possible to quantify
the relationship between the estimated levels of ground motion and the occurrence of rockbursts
using the percentiles. The 10th, 25th, 50th, 75th, and 90th ppv percentiles are represented (Figure 5–
9).

Figure 5–9 Levels of ground motion estimated based on the magnitude and distance from the epicentre of seismic
events using the scaling law proposed by Hedley (1992) for the rockbursts from Creighton, Copper
Cliff, and Coleman mines

The 10th and 90th percentiles of the rockburst data indicate that the majority of damage to mine
excavations and ground support were triggered by ppv levels ranging from 29 to 725 mm/s.
Those values are conceptual and may not reflect the actual levels of ground motion that were
associated with the damage. Nevertheless, the scaling law from Hedley manages to captures the
variability in the rockburst data and its application for design purposes should not be excluded.
The collected data indicate that larger magnitude events tended to result in damage over wider
distances. This trend is well captured by the employed ppv scaling law.
 142

Using the collected rockburst data, it is possible to verify the damage thresholds that were
originally proposed by Hedley (1992), i.e.:

• Falls of loose ground: 50 mm/s

• Fracturing of intact rock: 300 mm/s

• Severe damage: 600 mm/s

The 10th to 90th ppv percentile range is comparable to the range of damage threshold suggested
by Hedley (1992). However, if fracturing of intact rock and severe damage were expected to
occur starting at 300 and 600 mm/s, a larger proportion of the collected rockburst would have
been expected to be characterized by higher ppv levels. Only close to 25% of the collected
rockburst data were associated with ppv levels higher than 300 mm/s. As it was previously
discussed in Chapter 4, the observed level of damage is not expected to correlate directly with
the conceptual level of ppv. Other parameters, including the employed ground support system,
must be considered in forecasting the severity of rockbursts. Nevertheless, the assessment of the
ground motion levels in the collected rockburst data, represented in Figure 5–9, indicate that the
ppv scaling law from Hedley (1992) can potentially be applied to determine the likelihood of
rockburst occurrence at Creighton, Copper Cliff, and Coleman mines.

5.5 Distance from large-scale geological structures

The large-scale geological structures present at Creighton, Copper Cliff, and Coleman mines
were discussed in Chapter 2. In conducting passive monitoring of support performance, it was
relevant to report the presence of such structures in the vicinity of rockburst locations. In order to
be consistent in reporting the presence of structures, an arbitrary distance threshold was defined.
When large-scale geological structures were mapped within 100 m from the damage locations,
the distance from the damage was consistently reported in the rockburst database. The type of
structure, whether fault, shear zone, dyke, or seismically-active lithological contact was also
reported along with the distance. A histogram was constructed of the number of rockbursts
located within distance increments of large-scale structures (Figure 5–10). The data were colour-
coded using the Rock Damage Scale (RDS) in order to represent the evolution of rockburst
severity as the distance from structures increases.
 143

Figure 5–10 Frequency and severity of rockbursts in the presence of large-scale geological structures

The number of rockbursts decreased exponentially as the distance from the nearest structure
increased (Figure 5–10). Such a decrease over distance was primarily observed within the first
35 m. Amongst rockbursts that occurred within 100 m from a large-scale structure, 69% occurred
within the first 15 m. It is of particular importance to note that 71% of the severe rockbursts
characterized by RDS 4 and 5 occurred within 15 m from a large-scale geological structure.
Figure 5–10 demonstrates that the presence of large-scale geological structures has a significant
influence on the frequency and severity of rockbursts. The collected rockburst data suggest that
within 15 m from a large-scale geological structure, the likelihood of facing dynamic-loading
conditions is increased significantly. This result can have practical implications for ground
support design in deep and high stress mines by contributing to the identification of areas that
warrant the installation of enhanced yielding support.

5.6 Mining-induced stresses

In order to develop a strategy for the design of ground support capable of managing dynamic-
loading conditions, it is important to account for the mining-induced stresses. Previous empirical
guidelines that were developed for assisting in the selection of ground support or for forecasting
the severity of rockbursts were only based on the far-field stress conditions (Heal, 2010; Hoek &
Brown, 1980). The location of mine seismicity varies in time and space, as mining progresses
and as the stress conditions are being affected by the mining sequence. In order to control the
costs associated with ground support, it is important to target the areas where the stress
conditions warrant investing in the installation of yielding support. This cannot be achieved
 144

using the far-field stresses only. A valid and consistent approach has to be used in order to
develop a realistic representation of the mining-induced stress conditions at Creighton, Copper
Cliff, and Coleman mines.

5.6.1 Stress modelling: approach and results

Creighton, Copper Cliff, and Coleman mines are complex three-dimensional environments and
over the years, rockbursts have occurred on several mine levels. For the current study, a 3D
numerical package was required in order to generate a representation of the mines that was as
accurate as possible. Furthermore, the capacity to generate stress and deformation results in a
reasonable amount of time was important in selecting a numerical package given the number of
modelling steps associated with the 14 years of mining at Creighton, 10 years at Copper Cliff,
and 8 years at Coleman being covered in this study. In dealing with a three-dimensional space,
numerical models using a linear elastic response of material, as opposed to more complex non-
linear behaviours, result in significantly less computation time (Andrieux, Brummer, Li, &
O'Connor, 2007). In the past, forecasting the brittle failure of rock masses around mine openings
has been successfully achieved using elastic models (Diederichs, 2003; Martin, 1997; Martin et
al., 1999; Wiles, 2005). In geomechanics, simple or parsimonious models that use the simplest
possible concepts and smallest possible number of inputs can provide great ranges of explanation
(Hammah & Curran, 2009).

The Map3D boundary element stress analysis package was selected for this study. The software
has been employed for many years at Creighton Mine (Diederichs, 2003; Tan, 2003; Wiles &
MacDonald, 1988; Wiles, 1998) and is still used on a regular basis at Vale Sudbury mines.
Map3D can handle large-scale three-dimensional problems and model the elastic response of
multiple materials to mining. The Fault-Slip version of Map3D further allows the representation
of planar discontinuities and is capable of conducting plastic fault-slip stress analyses.

For the purpose of this thesis, numerical models of the 400 and 461 orebodies at Creighton, 100
and 900 orebodies at Copper Cliff, and MOB and 153 orebodies at Coleman were shared by the
mines. These models have been valuable design tools used on site for mine planning and stope
sequencing and were calibrated based on field observations (Berruar, Yao, & Sampson-Forsythe,
2012; Malek et al., 2008; Suorineni & Vasak, 2008). The model of Creighton Deep, which
included both the stopes of the 400 and 461 orebodies, was extended, for the purpose of this
 145

thesis, to include the stopes mined on upper levels, between the 6600 and 7200 levels (Figure 5–
11a). The model was modified further to include the prominent shear zones that intersected
openings in the ore and footwall of Creighton Mine (Figure 5–11b). In the numerical model of
Copper Cliff Mine, the trap dyke, which intersects the 100-900 orebody, was represented (Figure
2–7). For Coleman Mine, the Lunch Room Fault, which intersects the West part of the Main
Orebody, was part of the provided model (Figure 2–9). The rock mass elastic properties
employed in these models were specified in Chapter 2.

Figure 5–11 Map3D model of Creighton Deep: (a) side view of the 400 and 461 orebodies, looking west
and (b) isometric view showing the ore and shear zones

The primary objective of numerical stress modelling in this analysis was to quantify the
influence of mining on stress-redistributions around the stoping areas. At Creighton Mine, shear
zones, particularly those located in the immediate footwall of the 400 orebody, have been
associated with several rockburst occurrences between 2000 and 2013. Three-dimensional
numerical models used at Creighton Mine in recent years, whether linear or non-linear (Malek et
al., 2008; O’Connor, Cotesta, Brummer, & Thibodeau, 2010), did not explicitly account for the
presence of shear zones. In the elastic model employed as part of this study, shear zones mapped
on two mine levels or more have been represented for visualization purposes only; the selected
mechanical properties of shear zones aimed at minimizing the effect of these structures on the
mining-induced stresses. The shear zones were therefore modelled using an elastic behaviour and
their elastic properties, i.e. the bulk (K) and shear (G) moduli, were calculated using the Young’s
 146

modulus (E) and Poisson’s ratio (υ) of Creighton’s footwall granite using the following
analytical equations:




































































































































































































































The Map3D models of Creighton, Copper Cliff, and Coleman mines were used to assess the
stress conditions at the time and location of damage to mine excavations reported in the
rockburst database. Based on previous work involving the use of elastic models in underground
mining (Diederichs, 2003; Martin, 1997; Martin et al., 1999), the differential stress was used in
order to investigate the rock-mass behaviour at the three mines. The normalized differential
stress (NDS) was plotted (Figure 5–12).

Figure 5–12 Stress conditions associated with rockbursts at Creighton, Copper Cliff, and Coleman mines, colour-
coded (a) by mine and (b) by lithology
 147

The mining-induced stress data from the three mines overlapped relatively well on the chart
(Figure 5–12a), suggesting an appreciable level of consistency in the results. There was also a
relatively good overlap of NDS for various rock types characterized by different UCS, as
discussed in Chapter 2. The NDS in the massive sulphide rocks was typically higher than in the
other rock types. This was both attributed to the high stress conditions generally reported in the
near vicinity of the mined-out areas in the employed elastic model and to the lower UCS of
massive sulphides, as compared to other rock types. The major principal stress (σ1) values were
typically higher at Creighton Mine, as opposed to Coleman and Copper Cliff, and this was
reflective of the greater depth of the mining operation.

The distribution of NDS can be approximated using a lognormal distribution, which corresponds
to the normal distribution of the logarithm of the NDS (Figure 5–13). In a perfectly normal
distribution, 67% and 95% of the data are within one and two standard deviations from the
average, respectively (Figure 5–13a). In the elastic stress results, the average stress conditions
associated with rockbursts consisted of (σ1-σ3) = 0.25*UCS. By generalizing the elastic stress
results using a lognormal distribution, 67% of rockbursts can be expected to occur between
differential stresses of 0.15 and 0.39 times the UCS and 95% between differential stresses of
0.10 and 0.63 times the UCS (Figure 5–13b).

In previous studies, stress levels associated with the initiation of damage to brittle rock masses
were identified using different numerical modelling approaches and validation methods. Martin
(1997) and Diederichs (2003) employed three-dimensional elastic stress models to analyse stress
redistributions around a circular tunnel in competent granite and open stopes that were mined
between the 6600 and 6700 levels at Creighton Mine, respectively. Using acoustic emissions,
Martin (1997) proposed a damage initiation threshold corresponding to a NDS of 0.33. This
threshold was supported by field observations of a notched shape developing in the roof of the
tunnel. Diederichs (2003) monitored damage in production boreholes using camera surveys and
compared the elastic stress results to the extent of the delineated damage zone. This resulted in
the identification of a damage initiation threshold corresponding to a NDS range of 0.40 to 0.50.
 148

Figure 5–13 Distributions of stress conditions at rockburst locations: (a) approximated normal distribution of the
logarithm of the differential stress to UCS ratio and (b) approximated lognormal distribution of the
differential stress to UCS ratio

Using a non-linear modelling approach for Creighton Mine, O’Connor et al. (2010) found that
the 60th percentile of differential stress values associated with microseismicity in the Creighton
granite (UCS = 240 MPa) corresponded to 92.9 MPa, on average, with a standard deviation of
9.5 MPa. A damage initiation threshold of NDS = 80-100 MPa, i.e. ≈ 0.33-0.42 times the UCS,
was proposed. From experience in calibrating models against mining-induced seismicity,
O’Connor et al. (2010) indicated that damage generally initiates at differential stresses equivalent
to 0.3 to 0.5 the UCS and that the majority of smaller magnitude events (microseismicity) are
expected within this region. The stress levels associated with the initiation of damage to brittle
rock masses (Diederichs, 2003; Martin, 1997; O’Connor et al., 2010) are compared to the stress
levels associated with the collected rockburst case studies (Table 5–8).
 149

Table 5–8 Comparison of damage initiation thresholds in brittle rock masses

References (σ1-σ3)/UCS Numerical modelling Validation method

approach
Martin (1997) 0.33 3D linear elastic Acoustic emissions
Diederichs (2003) 0.40-0.50 3D linear elastic Borehole camera
surveys
O’Connor et al. 0.33-0.42 3D non-linear Microseismicity
(2010)
Morissette (2015) 0.15-0.40 3D linear elastic Rockburst case studies

The differences in the rock mass constitutive behaviours assumed in linear and non-linear
models is of great importance. Their implications ought to be taken into consideration in
interpreting mine-scale numerical modelling results. Nevertheless, the general damage initiation
stress range suggested by O’Connor et al. (2010) based on experience appears to be applicable
regardless of the numerical modelling approach, whether linear or non-linear. The range of
differential stresses obtained for the collected rockburst case studies, defined as one standard
deviation below and above the NDS average, overlaps with the existing damage initiation
thresholds. The average value, i.e. NDS = 0.25, however, is arguably low compared to the other
thresholds. A point must be made that the investigated phenomena, either the occurrence of
microseismicity or the development of a damage zone in the immediate vicinity of open stopes,
differed from the cases of rockburst occurrences considered in this study. Whereas other work
has investigated the propagation of damage within the rock mass, this analysis has focused on the
incidence of damage to supported entry mine openings triggered by mining-induced seismicity.
Various problem scales were also considered as the field case studies varied from a single tunnel,
to the open stopes on one mine level, or to microseismicity and rockbursts occurring at the mine
scale and over extensive time periods.

The field data and stress analysis undertaken as part of this thesis challenge some of the
perceptions, based on conceptual relations regarding the interaction of stress and rockbursts.
Although previous studies listed in Table 5–8 do not compare directly, it appears that it would be
a mistake in interpreting linear elastic numerical models to attribute the occurrence of rockbursts
 150

only to the very high stress areas of a mine. The reality is that stress analysis models, irrespective
of the degree of sophistication, provide only a partial picture of the onset of rockbursts. A failure
criterion employed at Coleman Mine specifies that major ground control issues such as
rockbursts occur at NDS of 0.5 to 0.7 in the massive sulphide (ore) and of 0.7 to 1.0 in the
granite host rock (Berruar et al., 2012). When comparing to the NDS values obtained for all the
validated rockburst case studies considered in this study, it is questionable whether such failure
criterion has been validated using quality field data.

5.6.2 Stress modelling: comparison between mining-induced stresses,

seismicity, and rockbursts at Creighton Mine
Seismicity in underground mining reflects the rock mass response to mining-induced stress
changes. The collected SGM seismic data, and more specifically the occurrence of seismic
events in relation to the mining sequence, can be visualized using Map3D. A relatively well-
defined pattern of SGM seismic events was observed in the footwall of Creighton Mine (Figure
5–14). This concentration of seismic events indicates that significant levels of rock mass damage
were sustained in the immediate footwall of the 400 orebody as a result of stress-redistributions
around the stoping areas. A dashed line defines the outside limit of this damage zone (Figure 5–
14a,b). The relationship between rockbursts and mining-induced seismicity is self-evident as the
majority of rockbursts plotted within the boundary of the damage zone (Figure 5–14c,d).

Given these observations, it was pertinent to verify the mining-induced stresses generated by the
linear elastic model through comparison with mine seismicity. This study focused on seismic
activity susceptible to generate rockbursts; therefore, only the SGM seismic events were used.
Based on the calculated probability of rockbursts as a function of the magnitude (Figure 5–8), it
is important in assessing the level of seismic hazard to pay particular attention to the location of
seismic events greater than magnitude 1.0 MN. This study focused on the mine footwall, more
specifically south of the 400 orebody and east of the 461 orebody, given that most of the
permanent development at Creighton is located in this area. This is consequently where the
majority of rockbursts have occurred between 2000 and 2013 and also where the seismic events
were located with greater accuracy by the seismic monitoring system due to the possibility of a
better 3D areal coverage by the sensors. In the hanging-wall, the accuracy of the location of
SGM seismic events was questionable given the limited number of sensors installed in this
geological unit due to limited access.
 151

Figure 5–14 Visualization of SGM seismic events plotted over (a) a longitudinal view and (b) a top view of
the open stopes, mined-out as of September 2013, and rockbursts, colour-coded using the
RDS, plotted over (c) the longitudinal and (d) top views of the mined-out stopes.

The analysis conducted for data-validation purposes has focused on the 7400, 7530, and 7680
mine levels located at depths of 2255 m, 2295 m, and 2340 m. Detailed mining-induced stress
results overlapped with mine seismicity over periods of two years are presented in Appendix C.
The model was able to capture the high stress zones in the footwall with relatively good
agreement with the observed seismicity, particularly on the east side of the 461 orebody. By
considering the differential-stress contours generated for each mine level in 2013 and the seismic
events from 2000 to 2013, it was observed that seismicity occurred predominantly in areas where
the differential stress was greater than 60 to 72 MPa (Figure 5–15). These stress levels
corresponded to 0.25 to 0.30 times the UCS of the footwall granite (240 MPa).
 152

Figure 5–15 Mining-induced stresses and seismicity on (a) the 7400, (b) 7530, and (c) 7680 level of Creighton Mine
 153

It is quite obvious that the model did not capture all the mechanisms that lead to seismic activity
in the Creighton Mine footwall, particularly with respect to the shear zones. It is common on the
three considered mine levels to observe seismicity diverging from the 0.25-0.30 NDS contour
and being controlled by a geological structure instead. The collected seismic data and numerical
modelling results suggested that in the presence of shear zones in the Creighton Mine footwall,
seismicity occurred generally in locations where the differential stress had exceeded 0.20 times
the granite UCS. The main observations reported from Figure 5–15 are summarized:

• 7400 level: Seismicity typically occurred east of the 461 orebody and extended into the
footwall of the Plum shear following the 0.25 NDS contour, and reappeared in the
vicinity of the Grizzly shear zone following the 0.20 NDS contour;

• 7530 level: Seismicity occurred between the 461 orebody and the Plum shear zone, which
coincided with the 0.30 NDS contour, and extended further along the Grizzly shear zone,
at the border of the 0.20 NDS contour;

• 7680 level: Seismicity has typically followed the 0.25 NDS contour in the footwall of the
402 shear zone and has occurred further between the Plum shear and a possible extension
of the Grizzly Shear, within the 0.20 NDS contour.

Using a three-dimensional linear elastic stress modelling approach, mining induced stresses were
calculated for the time and location of each reported rockburst in the comprehensive database.
The obtained NDS results have shown an appreciable level of consistency, which justified the
application of this parameter in characterizing the ground conditions associated with rockbursts.
The applicability of a guideline based on the differential stress was investigated by comparing
mining-induced stresses on the Creighton Mine 7400, 7530, and 7680 levels to the damage zone
defined by the occurrence of SGM seismic events between 2000 and 2013. The observed level of
correlations between the differential stresses and mining-induced seismicity validated the stress
results that were used to characterize each damage location reported in the rockburst database.

5.6.3 Limitations and interpretations of numerical models

Every numerical model comes with a range of assumptions, whether it affects the geometry (e.g.
2D vs. 3D), the excavation sequence, the material properties, or the constitutive laws and
behaviour. In the present analysis, it was recognized that the three-dimensional geometry and the
mining sequence were of vital importance. Nevertheless, it is recognized that the analysis could
 154

be improved going forward if a well calibrated model that may use a different constitutive
relation, including a non-linear constitutive law, could arguably better capture the behaviour of
the rock mass. It is possible that this process, if undertaken over the time frame and the sequence
of development, may provide an indication going forward.

The present analysis used a 3D numerical model that relied on an elastic analysis. The
constructed model is original in that it provides a first application of calibration and correlation
with an extensive and time-continuous database of rockburst events. In the past, numerical
elastic models have indicated that they have been calibrated adequately with respect to rock mass
damage or seismicity (Bruneau, Hudyma, Hadjigeorgiou & Potvin, 2003; Diederichs, 2003;
Karampinos, Simser & Hadjigeorgiou, 2012; Martin, 1997; Martin et al., 1999; Wiles, 2005).
The use of non-linear models have been successful in other applications capturing certain
elements of the behaviour by linking it to mine seismicity (Beck, Reusch & Arndt, 2008;
O’Connor et al., 2010).

A limitation of the employed numerical model is that, comparing the results to other elastic
models, the absolute values of stress magnitudes will probably be different. As indicated in this
study, the elastic models were calibrated with rockburst incidents and provided confidence going
forward based on the observed conditions in the constructed rockburst database. It is recognized
that if the analysis is undertaken using another model with other assumptions, including non-
elastic behaviour, the resulting stress values could be significantly different. The practical
significance of this is if in the future the recommendations are based on a non-elastic value, this
will probably result in a potentially different threshold. This should be taken into consideration
for future work when establishing site-specific guidelines using different numerical tools.

5.7 Summary
This chapter has reported on the rockburst field-data collection campaign undertaken at
Creighton, Copper Cliff, and Coleman mines. The rockburst database developed as part of this
study is significant given the time-continuity in the collected data. Furthermore, each observation
and parameter reported in the database has been thoroughly reviewed and validated, either by
cross-referencing using various sources of information available on site, by conducting
interviews with the mines’ personnel, or by conducting in-depth statistical analysis on the
collected quantitative parameters, e.g. the event magnitudes and mining-induced stresses.
 155

A database has been constructed of 209 seismic events and the resulting damage to 324 mine
locations. Each damage location consisted of a rockburst case study where the installed ground
support was subjected to dynamic-loading conditions. The reliability of the collected rockburst
data was demonstrated in this chapter by undergoing a thorough data-validation process. The
collected magnitude data have been adjusted based on a calibration of the local seismic
monitoring systems employed at Creighton, Copper Cliff, and Coleman mines. The significance
of these adjustments was tested further through the application of ANOVA and of Student tests
of hypotheses on samples’ mean. The data were further validated through the application of
seismic data analysis techniques on the magnitudes recorded by the two seismic monitoring
systems employed at Creighton over time. The level of confidence developed in the collected
and adjusted seismic data was further validated by demonstrating the possibility of assessing the
probability of rockburst based on the event magnitude. It was demonstrated that at Creighton
Mine, a 1 MN event has a 2% probability of generating damage to mine openings whereas a 2 MN
event has an increased probability of 10% of generating damage. This assessment was possible
only due to the time-continuity and the completeness of the seismic event and rockburst datasets.
The obtained probabilities of rockbursts based on the event magnitude justified focusing on the
seismic events recorded by the SGM systems, as opposed to considering microseismic events in
the analysis.

The collected SGM events were imported into the numerical model in order to validate the
employed three-dimensional linear elastic numerical modelling approach. Correlations were
identified between the obtained stress profiles and the rock mass damage zone represented by the
location of mining-induced seismicity. The patterns reported in this analysis have confirmed the
model’s reliability in characterizing the ground conditions associated with mining-induced
seismicity in the vicinity of the Creighton Mine footwall development. This has provided
confidence in the stress results obtained for the time and location of rockbursts reported in the
database.

The comprehensive rockburst database, along with the detailed observations and verified
parameters, was used in Chapter 6 as part of a multivariate statistical analysis. This analysis
focuses on identifying patterns amongst the collected rockburst data, which are representative of
the various levels of dynamic-load demand imposed on the installed ground support systems at
Creighton, Copper Cliff, and Coleman mines. This statistical analysis allows the extraction of
 156

important information from the collected data, which is pertinent to the development of a ground
support design strategy for mine openings subjected to dynamic-loading conditions.
 157

6 Multivariate Characterization of Rockburst Field Data

Chapter 5 described the process of collecting and validating rockburst data from Creighton,
Copper Cliff, and Coleman mines. This has allowed the development of a comprehensive
documented database, which comprised both a large number of observations, i.e. 324 case
studies of support performance under dynamic loads, and variables, e.g. seismic event
magnitude, rock mass quality (RMR), and mining-induced stresses. Multivariate data contain
much more information than univariate data. As a result, gaining insight into the system or
process studied is a challenge. In conducting passive monitoring of ground support performance,
one of the objectives is to assess the influence of the collected variables on the dynamic-load
demand imposed on the support. This is a complex engineering problem, which cannot be solved
by simply looking at the table of collected data. The data must be analysed so that the
information in the data can be expressed in a comprehensible way (Eriksson et al., 2006). A
multivariate characterization is a necessary first step in the investigation.

The primary objective of the analysis discussed in this chapter was to reduce the dimensionality
of the studied data. The secondary objective consisted of identifying relevant trends amongst the
selected predictor variables for describing variations in the dynamic-load demand amongst the
collected rockburst case studies. This was undertaken using a multivariate statistical method
capable of handling both the collected numerical and categorical data and of coping with missing
data. The principal component analysis (PCA), a projection method, was selected for this
analysis.

6.1 Principal component analysis

PCA is a multivariate analysis technique whose goal is to reduce the dimensionality of a large
number of interrelated variables (Michailidis, 2007). This is achieved by fitting lines, planes, and
hyper-planes that approximate the data as well as possible in the least square sense (Eriksson et
al., 2006). Such geometries are defined by a new set of variables called principal components
(PCs), which are obtained as linear combinations of the original variables (Abdi & Williams,
2010). An important advantage of PCA is that all the variables from the model are considered in
 158

defining the PCs. The first principal component is characterized by the largest variance. The
remaining principal components are computed under the constraint of being orthogonal to the
first principal component and to each other, and to have the largest possible inertia (Eriksson et
al., 2006). Hence, PCA produces orthogonal and mutually uncorrelated linear combinations that
aim at explaining as much common variation as possible in the studied dataset.

PCA is a multivariate projection method: the approach consists of projecting each observation on
planes formed by two vectors, with each vector corresponding to a PC. In Figure 6–1, a three-
variable data set was reduced to two variables, PC1 and PC2. The structures of the data set can
be visualized by projecting each observation on the resulting plane. The coordinate values of
each projection, in the sub-space defined by PC1 and PC2, are referred to as scores (Eriksson et
al., 2006). On a score plot, observations close to each other have similar properties whereas those
apart from each other are dissimilar with respect to the studied variables. In studying the
performance of ground support systems under dynamic-loading conditions, observations close to
each other on a score plot would reflect rockbursts that occurred under similar conditions.

Figure 6–1 Two principal components forming a plane in the multidimensional variable space on which
observations can be projected and described in terms of their score (Eriksson et al., 2006)

PCA was selected for this study due to its capacity to treat both categorical and numerical data.
Furthermore, the technique can cope with missing data and multicollinearity since all the
variables are considered while fitting the PCs. Multicollinearity occurs when variables are
approximate linear functions of other variables (Eriksson et al., 2006) and are therefore highly
correlated together. Multiple linear regression (MLR), for example, cannot treat such data
 159

appropriately. An important limitation of PCA, however, is the assumption of linearity, which

implies that variables can only be related through linear relationships. PCA can be considered
mainly as a descriptive technique and arguably, there is no need for explicit distributional
assumptions (Abdi & Williams, 2010). However, pre-treatment of the data can address the
limitation of linearity and often make the difference between a useful model and no model at all
(Eriksson et al., 2006). Scaling of data, mean-centering, and log-transforming are common form
of data pre-treatments applied in PCA in order to transform the data into a form suitable for
analysis. These pre-treatments are discussed below.

6.1.1 Scaling
In scaling, the range covered by variables is adjusted in consideration of the variance. A variable
that covers a large numerical range has a large variance whereas a variable with a small
numerical range has a small variance. Since the PCs are fit under the constraint of capturing the
maximum variance, variables that cover a large numerical range are more likely to be
represented in the model. Normalizing the data ensures that all numerical variables make the
same contribution to the model (Eriksson et al., 2006). This is achieved by multiplying each
variable column Xk in the data matrix by the inverse standard deviation (1/sk) calculated for each
variable column. This is referred to as unit variance scaling.

6.1.2 Mean-centering
Mean-centering improves the interpretability of the model by centering the data at the origin.
This second data pre-treatment consists of calculating the average of each variable and then
subtracting it from the data. In Figure 6–1, the red point represents the average vector, which has
been subtracted to the data in order for the average point to coincide with the origin. A
conceptual representation of the effect of unit variance scaling and mean-centering on a data set
composed of 5 variables is presented (Figure 6–2).
 160

Figure 6–2 Conceptual representation of the application of unit variance scaling and mean-centering, which results
in 5 variables of equal length and centred at a mean value 0 (Eriksson et al., 2006)

6.1.3 Log-transforming
Although a level of structure in the residuals can be tolerated, PCA, given the assumptions
underlying unit variance scaling and mean-centering, works best with normally distributed data.
Normalizing a skewed data set stabilizes the variance of the residuals and effectively reduces the
number of outliers. It also contributes to linearizing a non-linear response-factor relationship
(Eriksson et al., 2006). Several transformations can be used to approximate normality, however,
the log-transform is the most frequently applied one. Lognormal distributions are indeed often
encountered in natural systems, particularly when the variables studied have natural minimum
values (Eriksson et al., 2006). The logarithm of a variable characterized by a positively skewed
distribution may not necessarily be perfectly normal, but it is usually much closer to normality
than the untransformed data. Approaching normality typically results in improving the level of fit
of the PCA model.

An appropriate tool for identifying the need for data transformation is the histogram plot. Based
on the shape of the distributions observed, transformations other than log-transform could be
relevant, e.g. negative logarithm, square root, inverse, or power transformations. In studying the
dynamic-load demand on ground support systems during rockbursts, the first step of the multi-
variate statistical analysis consisted of plotting the distribution of each quantitative variable and
 161

identifying those that required transformation in order to approach normality. This step is
discussed in the next section.

6.2 Distribution of collected rockburst variables and

recommended transformations
A list of collected rockburst variables was presented in Chapter 5. The distributions of
quantitative predictor variables (adjusted Nuttli magnitude, distance between seismic source and
damage, depth of mine opening, (σ1-σ3)/UCS, distance from nearest geological structure, RMR,
excavation span, maximum length of reinforcement elements) are presented using histogram
plots (Figure 6–3). On the histograms, the vertical axis refers to the number of recorded cases of
support performance under dynamic loads reported in the database. Proposed data
transformations are noted on the right side of the charts, where appropriate.

Amongst the studied predictor variables, it was interpreted that the adjusted Nuttli magnitude,
the depth of mine openings, the RMR, and the maximum length of reinforcement elements
installed in the back did not require any transformation. The normality of the distribution of
Nuttli magnitude values was deemed satisfactory (Figure 6–3a). Depth, RMR, and reinforcement
length, although of quantitative nature, are not continuous variables; hence they could be
interpreted as categorical variables. Therefore, transforming these three variables was not
interpreted as relevant. The distribution of depth values is bi-modal, with the data from
Creighton Deep being dissociated from the rockbursts that occurred on upper levels of
Creighton, Copper Cliff, and Coleman mines (Figure 6–3c). The RMR categories corresponded
to the average values typically employed for design purposes at the mines and were quite tightly
distributed (Figure 6–3f). The maximum length of installed reinforcement elements in the back
reflected the typical length of primary reinforcement, i.e. 1.8 m and 2.4 m, and the more
occasional installation of cable bolts (Figure 6–3h).
 162

Log

(a) (b)

Log

(c) (d)

Square
root

(e) (f)

Log

(g) (h)

Figure 6–3 Distributions of quantitative variables used in the PCA on collected rockburst data with the proposed
data transformation
 163

The distributions of distance between the seismic event and damage location, normalized
differential stress, and excavation span were, to different degree, positively skewed (Figure 6–
3b,d,g). These variables were adjusted using a log-transformation in order to better approach
normality (Figure 6–4). The distance between the damage and the nearest geological structure
was also positively skewed, however, given the important number of “zero” values in the data, a
log-transform was not appropriate. Data were instead corrected by employing a square-root
transformation. The square root transformation did not completely eliminate the skewness of the
distribution but it increased its tightness (Figure 6–4c). The damage locations directly intersected
by geological structures appeared as outliers to the otherwise relatively normal distribution. This
outlier is worth keeping in the analysis however since it may be associated with a different
damage mechanism. The proposed transformations to the quantitative predictor variables ensured
that the rockburst data set was best suited to run PCA. The constructed PCA model and the
results generated are discussed in the next section.

(a) (b)

(c) (d)

Figure 6–4 Distributions of quantitative predictor variables following data transformation

 164

6.3 Analysis of rockburst field data using PCA

A PCA model was constructed using the predictor variables listed in Chapter 5 (Table 5–2). The
Simca-P+ software (Umetrics, 2009) was employed to conduct this analysis. For each principal
component added to the PCA model, the software provided quantitative information related to
the quality of the model. The correlation coefficient R2 measures the quality of the fit by
indicating the percentage of variation explained by the model. The maximum R2 value
achievable is 1, which indicates a perfect fit. The level of fit typically increases along with
increasing model complexity, i.e. along with an increasing number of principal components. The
predictive ability of the model is quantified by the parameter Q2, which measures the percentage
of variation of the training set of variables predicted by the model, as estimated by cross-
validation. A Q2 value greater than 0.5 suggests good predictivity. In the PCA model constructed
using the collected rockburst field data, the effect of increasing the number of principal
components on the quality of the fit and the predictive ability of the model was represented
(Figure 6–5).
































2 2
Figure 6–5 The influence of adding principal components on the cumulative R and Q parameters

The objective of this analysis was to reduce the dimensionality of the collected rockburst data
while adequately summarizing the X-variables. A three-component model was perceived to
achieve this objective without accentuating the loss of predictivity due to increasing model
complexity. The three-component model was characterized by a R2 value of 0.631 and a Q2
value of 0.193. The level of fit achieved by the model was deemed acceptable given the
 165

relatively important proportion of X-variables that were not continuous in the model. The
predictive ability of the model on the other hand was quite low, as implied by the Q2 value. This
indicates that there is an important level of unexplained variation in the collected data and that
the observed pattern may not be applicable to characterize an independent set of rockburst data.

6.3.1 Interpretation of the loading scatter plot

Score plots were used to visualize the patterns that characterized the rockburst data. The
rockbursts were projected on the planes defined by two of the first three principal components
and their positions were functions of the collected variables (Figure 6–6). The scores correspond
to the rockburst coordinates in the space defined by the principal components. In order to explain
the influence of the collected variables on the positioning of rockbursts on the score plots, the
principal component loadings were analysed. The loadings are the regression parameters that
multiplied each variable in the definition of the principal components. Variables characterized by
a high absolute loading value in a given component are responsible for a greater proportion of
the variability captured by the component, as opposed to variables characterized by a loading
value closer to zero. The loadings are visualized using the loading scatter plot (Figure 6–7).
Loading scatter plots are an indispensable complement to explain the position of the collected
observations, i.e. rockbursts, on the score scatter plot.

Figure 6–6 Score scatter plots of the collected rockburst data with the horizontal and vertical axes corresponding
to (a) PC1 and PC2, and (b) PC1 and PC3
 166

Figure 6–7 Loading scatter plots of the collected rockburst data with the horizontal and vertical axes
corresponding to (a) PC1 and PC2, and (b) PC1 and PC3

On the loading scatter plot, variables located close to each other contribute to the same
information and are therefore positively correlated. On the other hand, variables located at
opposite sides of the plot origin, in diagonally opposed quadrants, are negatively correlated
(Eriksson et al., 2006). Variables such as magnitude and strain burst, for example, are inversely
correlated in PC1. This suggests that seismic events interpreted as strain bursts were typically
characterized by smaller magnitudes. The magnitude and depth variables, however, form an
angle close to 90o with both loading plots’ origin. This indicates that these two variables did not
exhibit significant correlations in the reduced variable space defined by the first three
components. Hence, in the collected data, it would appear that rock mass damage, triggered by
large magnitude events, was recorded at both shallower and greater depths.

The first principal component was successful in capturing the variability related to the
characteristics of the seismic events associated with rockbursts. A rockburst is a seismic event
that resulted in damage to an underground excavation. Variables such as the magnitude, the
distance between the seismic source and the damage, the interpretation of the seismic source
mechanism, and the generation of damage to single or multiple locations, plotted the furthest
along the PC1 axis. The loadings indicated that seismic events interpreted as fault slips by the
ground control personnel, correlated well with larger magnitudes and greater distances between
the centroid of seismic sources and damage locations. In most cases, fault slip events resulted in
damage to multiple locations in the mines. This result was consistent with the collected rockburst
 167

case studies, which often reported on damage to multiple mine locations as a result of seismic
shaking triggered by large and remote seismic events. On the other hand, strain bursts tended to
generate damage to single mine locations. This is consistent with the definition of a strain burst,
which consist of a localized and brittle rock mass failure caused by the rock mass strength being
suddenly exceeded by the stress level. In the collected rockburst data the damage resulting from
strain bursts was typically closer to the seismic source and was generally caused by a smaller
magnitude event.

The second principal component captured variations in depth, stress conditions, and rock mass
quality (RMR). To a certain extent, PC2 made a distinction between damage locations
intersected by a dyke and those intersected by a fault or a shear zone. Variations in the span of
mine excavations were represented in PC2 but were better captured in PC3. The loadings in PC3
demonstrated that the installation of longer reinforcement elements was correlated to the size of
mine openings. This was predictable since the ground control standards for the installation of
cables at Creighton, Copper Cliff, and Coleman mines are primarily based on the span of mine
excavations.

6.3.2 Interpretation of the score scatter plots

The position of the investigated variables on the loading scatter plots, in terms of distance and
angle from the origin, provides important information on the collected data. Variables located the
furthest from the origin in a given direction have the strongest influence on the positioning of
rockburst data on the score plot along the same direction. The variables that explained the
greatest proportion of the variability captured by the first three components were outlined in the
previous section. Their influence on the patterns observed amongst rockburst data on the score
plots is explored in this section. The interpretation provided in this section has significant
implications on the design of ground support systems in seismically-active mines.

6.3.2.1 First and second principal components

The first important pattern identified amongst the collected rockburst data was based on the
interpretation of the seismic source mechanism. This led to the strong definition of two clusters
of rockbursts located apart from the origin along the PC1 axis, with strain bursts on the left and
fault-slip events on the right (Figure 6–8a). The remaining events that were not strongly
 168

influenced by PC1 comprised of both fault slip and strain burst events. This third cluster was
located along the vertical axis, which represents PC2.

Figure 6–8 Score plots defined by PC1 and PC2 with rockbursts colour-coded according (a) the interpreted
seismic source mechanism and (b) the adjusted Nuttli magnitude

The strong influence of the event magnitude on the interpretation of the seismic source
mechanism is well represented (Figure 6–8b). There was a transition from the lowest to the
highest magnitudes on the score plot, with the highest magnitudes being located further east.

With PC2 being defined predominantly by non-continuous variables such as depth, RMR, and
major geological structures, this component led to a classification of rockburst case studies that
was mine-specific. This was illustrated by colour-coding the score plot based on the mine sites
where rockbursts occurred. In Figure 6–9, the majority of rockbursts from Creighton Mine were
characterized by a positive PC2 loading value. Rockbursts from Copper Cliff and Coleman
mines, which occurred typically closer to the surface and in ground characterized by higher rock
mass quality, had negative PC2 loading values.
 169

Figure 6–9 Score plot defined by PC1 and PC2 with rockbursts colour-coded according to the mine sites

The stress conditions expressed in terms of the differential stress to UCS ratio also had a
considerable influence on the patterns observed on the score plot. Its impact was not as strong as
the one of the magnitude but it still generated a perceivable division between low and high stress
environments. On the score plot defined by PC1 and PC2, mine openings under lower and higher
stress conditions were attracted toward the southwest and northeast corners of the chart,
respectively (Figure 6–10). This led to the interpretation that regardless of the depth of the mine
operation, seismic events interpreted as fault slips typically generated damage to mine openings
that were under higher stress conditions, as opposed to strain bursts. The reason for that may be
that under higher differential stress to UCS ratios, the rock mass becomes highly fractured. The
fractured or loose material is more susceptible to a shakedown damage mechanism, triggered by
a remote, large magnitude seismic event. It was further observed on the score plot that mine
openings subjected to rockbursts at Creighton Mine were typically affected by higher values of
differential stress to UCS ratio, as opposed to Copper Cliff and Coleman. This is reflective of
Creighton Mine operating at much greater depths and under high mining-induced stresses.
 170

Figure 6–10 Score plot defined by PC1 and PC2 with rockbursts colour-coded according to the differential stress to
UCS ratio

The trends that were identified using PC1 and PC2 were useful in analysing the distribution of
rockburst damage severity. Rockbursts on the score plot defined by PC1 and PC2 were colour-
coded according to the RDS, a damage classification based on the displaced tonnage (Heal,
2010; Kaiser et al., 1992). The RDS was introduced previously, in Chapter 5. At first, an
important degree of variation in RDS values was observed over the score scatter plot (Figure 6–
11). Given the explained variations in depth, stresses, seismic event magnitude and source
mechanism, a direct correlation between damage severity and a combination of those variables
may be inappropriate. Hence, based on the collected and verified rockburst data, it was not
possible to accurately predict the severity of rockbursts using only variables that were interpreted
as being reflective of the dynamic-load demand on the support. This interpretation challenged
previous guidelines that proposed conceptual relationships between ppv or the level of stress and
the severity of rockburst damage (Hedley, 1992; Kaiser et al., 1996). This is an important
observation, which is supported by validated rockburst field data.
 171

Figure 6–11 Score plot defined by PC1 and PC2 with rockbursts colour-coded according to the RDS

Comparing the distribution of damage severity within various zones of the score plot can provide
certain information on the support performance at the three mines. Amongst the strain burst data,
the RDS indicated that these events caused greater damage at Creighton Mine. A greater
proportion of level-2 damage (green) was observed in the southwest part of the score plot, which
comprised strain bursts from Coleman and Copper Cliff mines, predominantly. This pattern
suggested that at increasing depth, there is a tendency for strain bursts to increase in damage
severity (Figure 6–12a). The increase in depth and its impact on damage severity is well reflected
by the major principal stress at the damage locations (Figure 6–12b).

Based on the PC1-PC2 score plot, fault slip events in the rockburst database had generated
greater damage at shallower depth, as opposed to strain bursts. Damage categories 4 and 5 were
proportionally more present as a result of fault slip events at Copper Cliff and Coleman mines.
At Creighton, a more important proportion of damage levels 1, 2, and 3 were reported as a result
of fault slips. This can appear as contradictory, however, it may reflect the impact of the
evolution of ground support systems over time. Creighton Mine has dealt with mine seismicity
and its consequences on excavation stability for a long time. Hence, the ground support strategies
have evolved progressively in order to account for the limitation in the observed performance of
 172

the support systems. At Copper Cliff and Coleman mines, large magnitude events are a more
recent challenge and their occurrence in more recent years has necessitated a direct intervention
in the ground support systems following severe rockbursts. The evolution of ground support
practice at Creighton, Copper Cliff, and Coleman mines was discussed in detail in Chapter 3.

Figure 6–12 Increasing levels of displaced tonnage influenced by (a) increasing depth and (b) increasing major
principal stress for strain burst data

The largest magnitude fault slip events at Copper Cliff and Coleman mines have occurred much
closer to the mine infrastructures, as compared to the largest events at Creighton. At Creighton,
an important proportion of large magnitude events occur deep in the mine hanging-wall. This
was the case for the November 29, 2006 4.1 MN fault slip event. Although the largest seismic
event that has occurred at Creighton Mine generated damage over several mine levels, the level
of severity was much lower than for the September 11, 2008 3.8 MN event at Copper Cliff or the
April 6, 2011 3.7 MN event at Coleman. When associated with a fault slip mechanism, larger
magnitude events have the potential to generate more severe damage (Figure 6–13a). However,
in assessing the dynamic-load demand resulting from fault slip events, the collected and verified
rockburst data demonstrated that it is important to take the distance between the seismic source
and mine openings into consideration. The peak particle velocity (ppv) obtained using scaling
laws reflects both the magnitude and the distance from the seismic source. Using the scaling law
proposed by Hedley (1992), the influence of the event magnitude and distance from the seismic
 173

source on the severity of damage, reflected by the displaced tonnage, was demonstrated (Figure
6–13b).

Figure 6–13 Increasing levels of displaced tonnage influenced by (a) increasing event magnitude and (b) increasing
PPV for fault slip data

Large magnitude fault slip events can generate damage over large distances within a mine. The
magnitude and the distance from the seismic source, both reflected in the ppv derived from
scaling laws, affect the dynamic-load demand on the support. The rockbursts defining the upper
bound in Figure 6–13b demonstrated that increasing damage severity can result from increasing
levels of ground motions. Nevertheless, there is an important degree of variation in the damage
severity per increment of ppv. This can be attributed to the limitations of ppv scaling laws
discussed in Chapter 4, particularly the assumption of a spherical radiation pattern. Variations in
damage severity per increment of ppv can also be attributed to the variations in the installed
ground support systems. Ground support systems designed in consideration of the anticipated
dynamic loading and seismic shaking conditions generated by large magnitude seismic events
can provide improved performances under those conditions. This is supported by the distribution
of RDS values amongst fault slip events on the PC1-PC2 score plot and by the evolution of
ground support systems at Creighton, Copper Cliff, and Coleman mines.
 174

6.3.2.2 First and third principal components

The score plot defined by PC1 and PC3 tended to isolate rockbursts that occurred in larger
openings north of the PC1 axis (Figure 6–14a). There was a clear correlation between the span of
mine openings and the length of the longest reinforcement elements installed (Figure 6–14b).

Figure 6–14 Score plots defined by PC1 and PC3 with rockbursts colour-coded according to (a) the span of mine
openings and (b) the longest reinforcement elements installed in the backs

The score plot demonstrated that shorter 1.8 m bolts were installed in smaller mine openings,
where large magnitudes were not anticipated. Smaller openings supported using short 1.8 m bolts
are mainly located in the third quadrant, characterized by negative PC1 and PC3 loadings (Figure
6–14b). Moving to the right on the score plot, as the potential for large magnitude events
increased, the mines tended to employ longer 2.4 m long bolts. In larger excavations, in order to
manage the increased static-load demand, the mines installed cable bolts as part of their standard
practice. Excavation spans larger than 7.3 m (24 ft) typically warrant the installation of cables at
Creighton, Copper Cliff, and Coleman mines. On the score plot, the majority of excavation spans
greater than 7.5 m were plotted in the first and second quadrant, where PC3 loadings are positive
(Figure 6–14a). The correlation between large spans and the installation of cable bolts is self-
 175

evident by comparing the location of large openings to the location of excavations supported
using cable bolts on the score plot (Figure 6–14b).

The installation of cable bolts appears to have contributed to improved performances of ground
support systems under dynamic loads triggered by fault slip events. On the score plot defined by
PC1 and PC3, a greater proportion of level 1 (blue) and level 2 (green) damage was observed
amongst the rockbursts that plotted in the first quadrant, as opposed to those that plotted in the
fourth quadrant (Figure 6–15). The influence of cable bolts in mitigating damage triggered by
strain bursts, however, was less obvious when considering the events that plotted west of the
PC3 axis, in the second quadrant.

Figure 6–15 Score plot defined by PC1 and PC2 with rockbursts colour-coded according to the RDS

Variations in the performance of ground support systems comprising cable bolts can be
attributed to the variation in damage mechanisms associated with fault slips and strain bursts. In
the occurrence of fault slip events, larger excavations are more susceptible to seismically-
induced rockfalls. Such damage occurs as fractured rock mass material, stable under static
conditions beforehand, becomes accelerated by an incoming seismic wave, hence causing forces
to overcome the capacity of the support system (Kaiser et al., 1996). The backs of mine openings
 176

are particularly susceptible to this damage mechanism. Supporting excavations’ back using cable
bolts is a sound strategy to ensure that ground support systems are tied back to stable ground and
that their holding capacity is preserved under seismic shaking. This was supported by the
collected rockburst data that plotted in the first quadrant of the score plot and indicated an
important proportion of minor damage levels 1 (blue) and 2 (green) (Figure 6–15).

The damage mechanism associated with strain bursts, on the other hand, is typically referred to
as bulking due to rock mass fracturing (Kaiser et al., 1996). In concept, the seismic source and
the damage coincide at the boundary of the mine opening. The additional reinforcement length
provided by cable bolts may not represent a direct advantage in those cases, as the damage tends
to be more superficial and localized. Furthermore, the back of mine openings, where cable bolts
are typically installed, are not more susceptible to bulking due to fracturing than the excavation
walls. As a result, fewer large-span openings have been affected by strain bursts, as opposed to
fault slip. The distribution of damage severity, quantified using the displaced tonnage, did not
differ significantly when strain bursts occurred in larger or smaller mine openings (Figure 6–15).

6.4 Summary
In this chapter, the justification for the application of PCA in studying patterns amongst the
collected rockburst data was presented. The collected variables were transformed when it was
deemed relevant in order to maximize the quality of the fit of the PCA model. A general
interpretation of the correlations in the collected rockburst variables was provided using the
loading scatter plots. The reported correlations, positive or negative, were useful in interpreting
the score scatter plots.

The score scatter plots provided more specific information on the patterns that characterized the
collected rockburst data. The strongest characterization of the data was based on the correlation
between seismic source mechanisms and event magnitudes, which was best captured by PC1.
The most relevant classification in PC2 was based on the depth of mine openings and normalized
differential stress, obtained through numerical elastic stress analyses. In PC3, most of the
variability was explained by the span and maximum length of reinforcement elements. Both
these variables were highly correlated.
 177

In investigating the influence of these predictor variables on the severity of rockbursts, each
event projected on the score plots was colour-coded using the RDS. It was identified that both
lower magnitude strain burst events and larger magnitude fault slip events can result in
significant levels of damage. However, strain bursts tended to result in more severe
consequences when mine openings were located at greater depth and subjected to higher stress
conditions. This was particularly the case at Creighton Mine. The levels of damage that resulted
from fault slips at Creighton Mine however, were typically less severe than at Copper Cliff and
Coleman. The proactive evolution of ground support systems at Creighton Mine, which was
reviewed in Chapter 2, had potentially an important impact on this result. As part of this
evolution, the mine’s commitment to installing cable bolts systematically in large excavations
had a demonstrated positive impact in managing the consequences of fault slip events.

This chapter identified the important trends observed in the collected and validated rockburst
data. More insight was developed on the factors that influence the dynamic-loading conditions at
Creighton, Copper Cliff, and Coleman mines, e.g. the magnitude, depth and stress conditions,
size of excavations, evolution of ground support practice, etc. The role of ground support
systems in managing dynamic-loading conditions is evident. In reviewing the performance of
ground support systems, it is important to consider the seismic source mechanism associated
with the event, as it is indicative of the damage mechanism. The severity of damage resulting of
strain bursts appeared to be dependent on the depth and stress conditions the mine opening is
subjected to. On the other hand, the severity of damage triggered by fault slip events was
influenced by the magnitude and distance between seismic events and mine openings. Increasing
levels of ground motion (ppv) generated by fault slip events can result in increasing severity of
damage to mine openings.

The following chapter reviews the performance of the installed ground support systems under the
conditions defined by the variables that led to the most relevant trends on the score scatter plots.
Chapter 6 was certainly very important in designing the analysis presented in Chapter 7.
 178

7 Assessment of Ground Support Performance

In Chapter 6, patterns in the collected rockburst data were identified using PCA. The focus was
on variables that were associated with the dynamic-load demand on ground support systems.
Most of the variation in the collected data was explained by variables related to mine seismicity,
e.g. the event magnitude, distance between the seismic source and mine openings, and seismic
source mechanism. The 2nd and 3rd principal components captured a lesser amount of the
variation, which was primarily driven by the stress conditions, the size of mine openings, and the
length of installed reinforcement elements. The length of rock reinforcement elements was
considered along with the excavation span as they together provided an indication of the static-
load demand and the capacity of the support system to retain large volumes of fractured rock.
The installed ground support systems and their underlying capacities, however, were not
explicitly considered in this analysis.

It is not possible to reliably derive the capacity of ground support systems to manage dynamic
loads using analytical solutions. This is attributed to a limited understanding of the mechanisms
of action and interaction of the support elements under dynamic loading conditions. This was
discussed in Chapter 4. Drop testing of rock reinforcement elements, nevertheless, provides
useful information on the anticipated performance of the support. This information was not
available prior to 1996, at the time the Canadian Rockburst Support Handbook (Kaiser et al.,
1996) was published.

This information was used in this chapter to calculate a conceptual capacity, defined in terms of
load and energy per surface units of support, for the ground support systems reported in the
rockburst database. Support systems were ranked based on their conceptual capacities and
expected performance under rockburst conditions. A grouping of ground support systems was
performed based on similarities in the conceptual load-bearing and energy-absorption capacities.
This classification resulted in a reduction of the number of ground support systems reported in
the rockburst database. In this chapter, the performance of each category of ground support
systems is assessed through a range of dynamic-loading conditions. Dynamic-loading conditions
 179

are defined as a function of seismicity, size of mine excavations, stress conditions, and rock mass
quality.

7.1 Conceptual capacity of reinforcement systems

In the collected WSN reports analysed for Creighton, Copper Cliff and Coleman mines, the
reinforcement elements installed at the damage locations were reported along with their
installation patterns. Using this information, load-bearing and energy-absorption capacities per
square metre of support were calculated for each reported configuration of reinforcement
elements. The load and energy capacities of individual reinforcement elements were obtained
from static (pull) or dynamic (drop) testing campaigns (Table 7–1).

Table 7–1 Load and energy-absorption capacities of reinforcement elements employed in classifying ground
support systems

Reinforcement Load Energy References

elements capacity capacity
(kN) (kJ)
Mechanical bolt 115 2.2 (Doucet & Voyzelle, 2012; Mansour Mining, 2015)
Resin-grouted rebar 170 14 (Doucet & Voyzelle, 2012; Mansour Mining, 2015)
Modified Conebolt 160 30 (Doucet & Voyzelle, 2012; Mansour Mining, 2015;
Player, Thompson, & Villaescusa, 2008)
D-Bolt (20 mm) 210 45 (Doucet & Voyzelle, 2012; Li & Doucet, 2012; Li,
D-Bolt (22 mm) 250 56 2012; Normet, 2014)

Swellex PM12 bolt 110 4.6 (Atlas Copco, 2013; Ortlepp & Stacey, 1998)
Swellex Mn12 bolt 110 15 (Atlas Copco, 2013; Bureau & Doucet, 2013)
35 mm friction set 89 5.0 (Kaiser et al., 1996; Mansour Mining, 2015)
39 mm friction set 102 7.7 (Mansour Mining, 2015; Ortlepp & Stacey, 1998)
46 mm friction set 145 15 (Mansour Mining, 2015; Player, Villaescusa, &
Thompson, 2009)
Plain strand cable 265 18 (Mansour Mining, 2015; Ortlepp & Stacey, 1998;
Player et al., 2008)
 180

The reported load capacity corresponded to the tensile strength or maximum load whereas the
energy capacity corresponded to the maximum impact energy. The references provided in Table
7–1 consisted typically of dynamic-testing laboratories and bolt manufacturers’ catalogues.
Under both static and dynamic conditions, there was important variability in the reported
capacities of reinforcement elements. Several references were therefore taken into considerations
and mid-range values have been selected as design capacities for the purpose of this analysis.
Table 7–1 is an update from the capacities of rock reinforcement elements provided in the
Canadian Rockburst Support Handbook (Kaiser et al., 1996). It includes more recently
developed rock bolts and reflects the actual dynamic performance of reinforcement elements
under impact tests.

An example calculation of the capacity of a rock-reinforcement configuration is provided,

following the approach of Kaiser et al. (1996) and using data from Table 7–1. The support
system considered comprises a diamond pattern of mechanical bolts and rebars, each bolt being
installed on a 1.22 m x 1.52 m (4’ x 5’) pattern. Modified conebolts are installed in a second
pass, on a 1.37 m x 1.37 m (4.5’ x 4.5’) pattern.












Load-bearing capacity:










































Energy-absorption capacity:


































This procedure was applied for each support system installed in the back and walls of mine
excavations reported in the rockburst database. The conceptual capacities of all configurations of
reinforcement elements that have encountered significant levels of damage (SDS ≥ 2) are
represented graphically in Figure 7–1, where the horizontal axis represents the load-bearing
capacity of the system and the vertical axis, the energy-absorption capacity.

A positive linear correlation is observed between load-bearing and energy-absorption capacities

of reinforcement configurations (Figure 7–1). The points corresponding to load and energy
capacities of reinforcement configurations employed in the back of mine openings are located
further apart from each other on the chart as compared to reinforcement configurations installed
in the walls. This reflects the more diverse selection of reinforcement elements employed in the
walls at Creighton, Copper Cliff, and Coleman mines over time. In addition to mechanical bolts
 181

and rebars, 35, 39, and 46-mm diameter friction sets have been employed to reinforce the walls.
This has resulted in smaller increments of conceptual support capacity between the different
support systems in the walls. Primary reinforcement in the back of mine openings generally
consisted of mechanical bolts and/or rebars. The installation of friction sets in the excavations’
backs is not standard practice in tunnels developed in hard rock at Vale Sudbury mines.

Figure 7–1 Load-bearing and energy-absorption capacities of combinations of reinforcement elements installed in
(a) the back and (b) the walls of mine openings (legend presented in Tables 7–1 and 7–2)

Ranking and grouping of reinforcement configurations was done based on the observed
similarities in the conceptual load-bearing and energy-absorption capacities (Figure 7–1). Ten
categories of reinforcement configurations have been identified for the backs and walls of mine
excavations subjected to dynamic-loading conditions. These categories are described in Tables
7–2 and 7–3. The number of rockburst case studies varies from a reinforcement category to
another. These variations are representative of the ground support standards employed over time
and rehabilitation work.
 182

Table 7–2 Categories of reinforcement systems installed at rockburst locations, in the back of mine openings

Reinforcement Symbols Descriptions N. of rockbursts

categories (Fig. 7–1) (SDS ≥ 2)
B1
Mechanical bolts 7
 Swellex PM12 2
B2 Mechanical bolts, 2nd pass of cables 3
B3
Mechanical bolts and rebars 66
B4 Mechanical bolts and rebars, 2nd pass of cables 22
B5
Rebars 25
B6  Mechanical bolts and rebars, 2nd pass of rebars or 2
 Mechanical bolts and rebars, 2nd pass of FS-46 1
B7 Rebars, 2nd pass of cables 6
B8
nd
Mechanical bolts and rebars, 2 pass of MCB 8
B9 Mechanical bolts and rebars, 2nd pass of MCB 1
and cables
B10
Rebars and MCB 7
Total 150

The configurations of reinforcement elements installed in the back can be divided into three
general schemes:

1. One pass of rockbolts

2. Two passes of rockbolts; the second pass being typically installed in conjunction with
heavy-gauge straps

3. One pass of rockbolts followed by a second pass of cable bolts

Ground support systems comprising of a single pass of stiff reinforcement elements such as
mechanical bolts and/or rebars are limited to less than 200 kN/m2 and 15 kJ/m2 from the installed
rockbolts. The replacement of mechanical bolts by rebars in the support design and the
introduction of yielding reinforcement elements add to both the load and energy-absorption
 183

capacity of the system. Load capacities in excess of 260 kN/m2 and energy capacities beyond
35 kJ/m2 can be achieved, at a conceptual level, with the conjunct installation of resin-grouted
rebars and modified conebolts. A second or third pass consisting of the installation of cable bolts
contributes to increasing the energy capacity of the system, however, the impact of cable bolts is
considerably reflected more in the added load-bearing capacity. This increase in capacity is
represented by the offset between circles and triangles of the same colour (Figure 7–1a).

Table 7–3 Categories of reinforcement systems installed at rockburst locations, in the walls of mine openings

Reinforcement Symbols Descriptions N. of rockbursts

categories (Fig. 7–1) (SDS ≥ 2)
W1
Mechanical bolts; or 35

Mechanical bolts and 35 or 39 mm friction sets 8
W2  35 or 39 mm friction sets 84
W3
Mechanical bolts and rebars; or 4

Mechanical bolts, 2nd pass of Swellex MN12 1
W4  39 mm friction sets and rebars; or 3
 46 mm friction sets 18
W5
Rebars; or 6
 39 mm friction sets, 2nd pass of Swellex MN12 4
W6 35 mm friction sets, 2nd pass of cables 6
W7  35 mm friction sets, 2nd pass of MCB 1
W8 46 mm frictions sets, 2nd pass of cables; or 4
39 mm friction sets, tight cable-bolt pattern 1
W9
nd
46 mm frictions sets, 2 pass of MCB 7
W10
Rebars and MCB 3
Total 185

In Figure 7–1, a distinction is made between the installation of friction-type reinforcement

elements consisting of a hollow tube and solid-steel tendons. Reinforcement elements such as
friction sets or inflatable bolts, e.g. the Swellex bolt, are represented using a diamond shape.
Solid-steel tendons are represented using circular shapes. This distinction is of particular
 184

relevance in excavation walls where both types of tendons have been used in comparable
proportions. Using friction bolts as opposed to point-anchored or grouted bolts typically result in
a slight reduction of the load-bearing capacity, which is compensated by marginal gains in
energy-absorption capacity. This conceptual capacity offset is best represented in Figure 7–1b. In
excavation walls, the ease of installation of friction sets represents a considerable advantage in
terms of productivity. In burst-prone grounds, the additional capacity of 46 mm friction sets (FS-
46) is desirable from a ground control view. Installed in conjunction with yielding reinforcement
elements, a reinforcement system comprising FS-46 can, conceptually, reaches over 240 kN/m2
in load-bearing capacity and over 28 kJ/m2 in energy-absorption capacity.

This assessment of the load-bearing and energy-absorption capacity led to a grouping of

reinforcement systems employed in the back and walls of mine openings. The employed metrics
took into consideration both the bolt type(s) and the installation pattern(s). The calculated
capacities of reinforcement configurations were derived using a conceptual approach and the
pertinence of the resulting values should be considered as such. The obtained load-bearing and
energy-absorption capacities are conceptual values; hence their direct application as part of a
design methodology can be limited. Nevertheless, in this analysis, the purpose of the conceptual
capacity of reinforcement configurations was to perform a grouping of ground support systems.
The performance of each reinforcement category under rockburst conditions is assessed in the
following sections.

7.2 Overview of the performance of ground support systems

The assessment of the performance of ground support systems under dynamic loading conditions
is typically made by both qualitative and subjective interpretations. As a result, it is difficult to
develop an explicit understanding of the mechanisms of action and interaction of support
elements. In this study, support systems reported in the collected rockburst data have been
grouped based on the conceptual capacity defined based on the type of reinforcement elements
employed and the installation patterns. Ground support systems installed in the back and in the
walls have been classified separately in 10 categories. In the interpretation of the support
performance, the employed surface support elements, whether welded-wire mesh or mesh-
reinforced shotcrete, were taken into consideration. This led to a further division of ground
support systems, leading to 20 possible support categories for both the back and the walls.
 185

7.2.1 Rate of tolerance of support performances

The performance under dynamic-loading conditions of the identified categories of support
systems was assessed using the Support Damage Scale (SDS) (Kaiser et al., 1992) (Table 5–5).
This assessment was often supported with pictures of the damage locations (Figures 5–2, 5–3)
and reports on the performance of individual reinforcement and surface support elements.
Rockburst case studies were reviewed and validated and a SDS value ranging from 0 to 5 was
assigned to the ground support systems installed in the back and walls. Based on the description
of each SDS category, and the associated interpretations provided by Mikula (2012), SDS
rankings 4 and 5 were interpreted as intolerable levels of damage to the employed support
systems. SDS rankings 2 and 3 were interpreted as acceptable and tolerable levels of damage,
respectively. SDS rankings 0 and 1, on the other hand, were disregarded, as the severity of the
dynamic loads faced by the support could be interpreted as negligible.

In order to facilitate the interpretation of the ground support capacity to mitigate rockburst
damage, in a quantifiable manner, a rate of tolerance (RoT) was elaborated for the purposes of
this thesis. The RoT is defined as the ratio of acceptable and tolerable levels of damage to the
number of significant dynamic loads (rockbursts) faced by each category of support system
(Equation 6-1).

























(Eq. 7–1)







 
 










The quality of performance under dynamic loads of the tested support systems was interpreted
using the following RoT classification:

• 100-86: Excellent;

• 85-71: Acceptable;

• 70-56: Tolerable – adjustments to the system must be considered;

• 55-41: Unreliable – significant adjustments are mandatory;

• 40-0: Inappropriate – the system must undergo a complete redesign.

This quantifiable classification of the performance facilitated the comparison of ground support
systems as part of the subsequent analyses. It provided a clear distinction between ground
support systems that performed consistently well under specific conditions, and those that did
 186

not. Consequently, this classification can be used to identify the need to modify ground support
strategies and assist in the selection of ground support systems.

7.2.2 Quantifiable interpretations of support performance

The complete assessment of the performance of ground support systems under dynamic loading
conditions at Creighton, Copper Cliff, and Coleman mines is presented in Figure 7–2. The figure
consists of:
a) Ground support systems installed in the back and also comprising steel mesh;
b) Ground support systems installed in the back and also comprising mesh-reinforced
shotcrete;
c) Ground support systems installed in the walls and also comprising steel mesh;
d) Ground support systems installed in the walls and also comprising mesh-reinforced
shotcrete.

This assessment, presented in a histogram form, displays the proportion of acceptable, tolerable,
and intolerable damage suffered by each support category. The assessment further includes the
calculated RoT, which is presented as circles, colour-coded according to the RoT classification.
The number of rockburst case studies that lead to damage levels greater or equal to SDS 2 is
indicated in brackets below the horizontal axis, next to each support category.

In the back, the capacity of shotcrete to enhance the performance of support is quite noticeable.
Six ground support systems out of ten that comprised shotcrete, in the back, presented acceptable
or excellent performance quality (Figure 7–2b). Without shotcrete, only two support systems
achieved those thresholds of quality (Figure 7–2a). The common factor in the ground support
systems that comprised shotcrete and that resulted in RoT greater than 70 in the back was the
installation of reinforcement elements in a second pass. Those reinforcement elements were
either cable bolts or yielding elements, predominantly the MCB.

In the walls, the application of shotcrete appeared to have benefited the performance of rebars as
the RoT associated with systems W5 and W5s differed considerably: 20 as compared to 100,
respectively. Both support systems were reported in 5 rockburst case studies. For the other
combinations of reinforcement elements, the improvement of performances as a result of the
application of shotcrete over the walls was not as significant as in the back. Support systems
 187

comprising of two passes of reinforcement elements, however, tended to perform better, just as
they did in the back.

Figure 7–2 Proportion of acceptable, tolerable, and intolerable performances of the ground support systems and
resulting RoT values

Two attributes of shotcrete make its application desirable in a two-pass support system. First of
all, given its stiffness, shotcrete is capable of limiting rock mass dilation. As a result, shotcrete
can preserve a laminated rock beam and maintain confinement around reinforcement elements,
to a certain extent (Simser, 2012). The same outcome cannot be achieved using steel mesh,
which behaves in a passive manner. Secondly, by being a stiffer surface support element, a
 188

shotcrete membrane is better at transferring loads to the reinforcement elements installed in the
second pass, i.e. over the shotcrete. The load-transfer from the surface support to the
reinforcement elements or amongst reinforcement elements through the surface support is further
enhanced by strong connecting elements such as heavy gauge straps. At Creighton, Copper Cliff,
and Coleman mines, yielding support has often been installed in conjunction with #0 gauge
straps over the shotcrete.

The only combination of reinforcement elements that resulted in excellent performance quality in
the back, without the application of shotcrete, was support B7, which consisted of resin-grouted
rebars followed by a second pass of cable bolts. It is possible that anchoring the support system
back into stable ground using cable bolts contributed to maintaining the level of confinement
around the rebars by preventing rock-mass dilation. Hence, it is conceivable that, to some extent,
the benefits of using shotcrete can be reproduced through the installation of cables. Nevertheless,
this interpretation is supported only by the two case studies of support system B7. Support B4,
which comprised a first pass of mechanical bolts and rebars, as opposed to rebars only, and was
tested during 9 rockbursts, had a much lower RoT of 44. This drop in the performance could be
attributed to the lower capacity of mechanical bolts to manage dynamic loads, although support
systems B3 and B5 presented similar RoT values (37 and 39, respectively).

This interpretation highlights one of the weaknesses of damage assessment based on the RoT, i.e.
the previous assessment did not consider the conditions under which the support was installed
nor the severity of the dynamic load it had been subjected to. At Creighton Mine, as mining
progressed to greater depths and higher stress conditions, mechanical bolts were eliminated from
the support standard. The minimum support requirements adopted in 2010 at Creighton consisted
of a 1.2 m x 1.5 m pattern of resin-grouted rebars as opposed to the previous standard, which
consisted in a diamond pattern of mechanical bolts and rebars. Hence, it is very likely that
ground support systems composed of rebars only were subjected to higher levels of mining-
induced stresses and consequently, higher thresholds of mining-induced seismicity at Creighton.
These factors could have contributed to a more pronounced degradation of the support systems at
greater depths. This interpretation allows the questioning of the relatively high RoT values (83
and 67) achieved by the ground support systems that comprised mechanical bolts as the only
reinforcement elements (B1 and B1s). Support systems composed of mechanical bolts only were
practically never installed at Vale Sudbury Mines under high stress conditions.
 189

In interpreting the performance of ground support systems, it is important to account for the
factors that may have affected the degradation of the support, the damage mechanism, or the
dynamic-load demand. In the following section, a first categorization of the rockburst data is
made by assessing the performance of ground support systems under strain-burst and fault-slip
conditions.

7.3 Performance of ground support systems under strain-burst

and fault-slip conditions
The interpretation of support performance provided in section 7.2 was only based on the installed
reinforcement and surface support elements. It did not account for the dynamic-loading
conditions and damage mechanisms generated by seismic events. Although the assessment did
not capture the complete onset of rockbursts, it identified trends that can have important
applications for the design of ground support systems in deep and high stress mines. In
particular, the influence of shotcrete on the management of dynamic loads in the back of mine
openings and the importance of a multi-pass support system were recognized.

An important distinction between rockbursts that resulted from strain-burst and fault-slip seismic
events was made in Chapter 6. In this section, the performance of ground support is investigated
amongst the two differentiated types of rockbursts. The rockbursts considered for this analysis
were those that were associated with a strain-burst or fault-slip mechanism with the highest
degree of confidence. Referring back to the score plot defined by the first two principal
components, these rockbursts plotted the furthest from the origin along the PC1 axis. Theses
events were colour-coded according to the inferred seismic source mechanism (Figure 7–3). The
classification presented in Figure 7–3 was achieved by applying hierarchical clustering over the
first and second principal components’ scores. The ranges of magnitudes associated with the two
groups of rockbursts, i.e. strain bursts and fault slips, were represented using a histogram chart
(Figure 7–4). There is a clear magnitude distinction between the events interpreted as strain
bursts versus those interpreted as fault slips. A magnitude of 2.2 MN appears as an adequate
threshold to distinguish the nature of seismic events, based on the collected rockburst data.
 190

Figure 7–3 Rockbursts filtered as strain bursts and fault slips with a high degree of confidence for Creighton,
Copper Cliff, and Coleman mines

Figure 7–4 Event magnitude associated with each individual damage location reported in the comprehensive
rockburst database

The performance of ground support systems was investigated using the proportions of acceptable
(SDS 2), tolerable (SDS 3), and intolerable (SDS 4 and 5) levels of damage to the support. Those
proportions are represented in Figure 7–5 for the support systems installed in the back and Figure
7–6 for the support systems installed in the wall.
 191

7.3.1 Performance of ground support systems installed in the back

The RoT is used to compare support performance and is plotted for each support category
employed in the back (Figure 7–5). The figure presents the support performance under strain-
bursting conditions (a and b) apart from the performance that resulted from fault-slip events (c
and d).

Figure 7–5 Performance of ground support systems installed in the back of mine excavations under dynamic-
loading conditions triggered by strain-burst and fault-slip events

Under strain-burst conditions, Figure 7–5 (a) and (b) indicate that the installation of yielding
reinforcement elements in the back resulted in excellent performance of the support systems B8,
 192

B8s, B10, and B10s. Ground support system B8 is composed of a first pass of mechanical bolts
and rebars followed by a second pass of MCB, whereas support system B10 is composed of a
single pass of rebars and MCB. Those were the only support systems that resulted in a RoT of
100, with or without shotcrete. This would suggest that under strain-burst conditions, the
capacity of a support system to yield and absorb energy is the key design feature. The application
of shotcrete has resulted in enhanced performance of stiffer reinforcement systems, bringing
support systems B3 and B4 from the unreliable performance range to the tolerable range (B3s,
B4s). The application of shotcrete however can be costly and labour intensive and sometimes
logistically complicated in deep underground mines. In order to mitigate strain bursts, the
installation of yielding reinforcement would represent a more efficient investment of resources as
opposed to the application of shotcrete.

Under fault-slip conditions, Figure 6-5 (c) and (d) suggest that the support requirements are
different from under strain-bursting conditions. Without the application of shotcrete, several stiff
and yielding ground support systems behaved inadequately. However, ground support systems
B4 and B7, which together accounted for five rockburst case studies, were characterized by
acceptable to excellent performance quality. Those two support systems consist of a first pass of
stiff reinforcement elements followed by a second pass of cable bolts. Enhanced by the
application of shotcrete, support systems B4s and B7s maintained excellent levels of
performance. Similar quality of performance was reached by yielding support systems B8s and
B9s. Those two systems comprised a second pass of yielding bolts, installed over mesh-
reinforced shotcrete and in conjunction with heavy-gauge straps. It is demonstrated by
comparing the performance of support systems B3 and B5 to support systems B3s and B5s that
the application of shotcrete contributed significantly to the capacity of support systems to
manage dynamic loads triggered by fault-slip events. By applying shotcrete, the RoT scores of
these support systems have improved from 17 to 52, and from 0 to 50, respectively. The ground
support systems B3s and B5s however are still interpreted as unreliable under dynamic loads,
based on their recorded levels of performance quality. This emphasizes the importance of a
second pass of reinforcement elements capable of yielding or of holding large volumes of
fractured rock.
 193

Based on the information presented in Figure 7–5 (c) and (d), the recommended support
strategies to mitigate damage triggered by large-magnitude, remote fault slip events consist of
one of the two following approaches:

1. To enhance a primary support system consisting of stiff reinforcement elements such as

resin-grouted rebars and welded-wire mesh with the installation of long holding elements
such as cable bolts.
2. To enhance a primary support system consisting of stiff reinforcement elements such as
resin-grouted rebars and welded-wire mesh with the application of shotcrete followed by the
installation of yielding reinforcement elements in conjunction with heavy-gauge straps.

Strategy #1 reflects the importance of tying the ground support systems back to stable ground
when the mass of fractured rock contained in the back is dynamically loaded by an incoming
seismic wave. It is perceived that the installation of cable bolts can contribute in controlling rock
mass dilation and hence, preserve confinement around the primary reinforcement. The collected
rockburst case studies confirmed that such support system is capable of managing the seismic-
shaking conditions that result from fault-slip events. Given the similarity in the performance of
support systems B4, B7, B4s, and B7s, the application of shotcrete is not considered as a
necessity when the employed support system comprises cable bolts. However, shotcrete could
still be a valuable option when degradation of the support is observed and rehabilitation has to be
performed.

Strategy #2 is based on the energy-absorption principle. In order to facilitate the load transfer
between the surface support and the yielding reinforcement elements, the application of shotcrete
is recommended between the first and second passes of reinforcement. In this strategy, it is the
assignment of the shotcrete membrane to prevent rock mass dilation and to maintain the arching
forces created within the back of mine excavations. A shotcrete membrane applied over a stiff
support system, as opposed to the second-pass installation of cable bolts, is not capable of
adequately preventing damage triggered by seismic shaking. This is based on the results
presented in Figure 7–5. Given the high stiffness of shotcrete, it is recommended to enhance the
support system using a second pass of reinforcement elements. Reinforcement elements capable
of absorbing large amounts of kinetic energy through a yielding mechanism can perform well
under those conditions.
 194

Given the necessity of applying shotcrete between the two passes of reinforcement installation,
the second strategy can be more time consuming and labour intensive. Strategy #1 therefore
stands out as potentially the most cost efficient. The commitment toward a support strategy must
account for economical factors, equipment capabilities and acceptance by the mining crews.

7.3.2 Performance of ground support systems installed in the walls

In the walls, relatively few ground support systems have faced strain-bursting conditions (Figure
7–6a,b). None of the systems enhanced using a second pass of cables or yielding reinforcement
elements have been tested without the application of shotcrete (Figure 7–6a). The general trend
suggests that a single pass of mechanical bolts, rebars, or friction sets installed in conjunction
with welded-wire mesh is insufficient for mitigating strain bursts. Ground support systems
enhanced by the application of shotcrete and a second pass of cables or yielding tendons (W6s,
W9s), on the other hand, have resulted in excellent levels of performance quality through four
case studies.

Under dynamic loads generated by fault-slip events, support systems that comprised frictions
sets (W2 and W4) have resulted in better performances than support systems that comprised
mechanical bolts or rebars (W1 and W5). Combinations of 35-mm friction sets and MCB
installed along with welded-wire mesh (W7) have resulted in excellent performance quality. The
application of shotcrete, however, has contributed to a significant improvement of the
performance under seismic-shaking conditions. Support systems comprising rebars or 46-mm
friction sets have performed in an excellent manner through 12 case studies (W3s, W4s, W5s). It
is therefore coherent that combinations of 46-mm friction sets with cables or MCB (W8s, W9s)
have also resulted in an excellent performance quality under fault slip events.
 195

Figure 7–6 Performance of ground support systems installed in the walls of mine excavations under dynamic-
loading conditions triggered by strain-burst and fault-slip events

A support strategy to mitigate damage to the walls of mine openings that is triggered by fault slip
events could simply consist of increasing the stiffness of the support system by applying
shotcrete over mesh and FS-46. In areas where mining-induced stress changes and strain-burst
problems are anticipated, the collected data demonstrated that a stiff support system has limited
probability of success. In order to mitigate strain bursts, a stiff support system that comprises
shotcrete must be enhanced by the installation of yielding reinforcement elements in conjunction
with heavy-gauge straps. It is possible that a support system that comprises friction sets, ideally
FS-46, and welded-wire mesh only could manage dynamic loads induced by strain bursts or fault
 196

slips if yielding bolts and straps are installed over it. Nevertheless, this interpretation is based on
one fault-slip case study only (Figure 7–6c). The performance of system W7 and W9 under strain
bursts is hypothetical as it is based on extrapolation of the data.

7.3.3 Summary
The collected and reviewed rockburst case studies were grouped based on multivariate patterns
that were captured primarily by the first principal component. These patterns reflected variations
in the event magnitude and seismic source mechanisms. Rockbursts were classified as whether
strain bursts or fault slips and the performances of ground support systems were investigated
under these two categories of events. By making a distinction based on the seismic source
mechanisms, trends that differed from those of section 7.2 were identified. This has resulted in
support recommendations that can address more specific problems related to mine seismicity.

In the back, shotcrete may not be required to manage strain bursts, as long as the support system
comprises yielding reinforcement elements such as MCB. Under dynamic loads generated by
large magnitude fault-slip events, it is necessary to maintain confinement around the bolts by
preventing dilation of the rock mass. Hence, the support must contribute in maintaining the
arching forces created within the back of mine openings. This can be achieved by the installation
of cable bolts, which aims at tying the support back to stable ground, or to some extent by the
application of shotcrete. The use of shotcrete however results in a very stiff support system and it
is recommended to overcome this limitation by installing a second pass of yielding
reinforcement elements.

In the walls, increasing the stiffness of ground support systems through the application of
shotcrete has been a sound strategy to manage the consequences of fault slip events. This
strategy, however, had limited success under strain bursts. The installation of reinforcement
elements capable of absorbing energy was recommended over shotcrete. Both yielding bolts and
cable bolts are up to the task. The possibility of success of such yielding system without
shotcrete could difficultly be assessed given the very few strain burst case studies available.
However, based on patterns observed in the back and walls of mine openings, it is interpreted
that a support strategy that comprises welded-wire mesh can be successful under strain bursts
and fault slips when long-enough reinforcement elements, e.g. 2.0 m-long FS-46, are installed in
conjunction with yielding reinforcement elements and heavy-gauge straps.
 197

7.4 Considerations of excavation size, mining-induced stresses,

and rock mass quality in interpreting support performance
Section 7.3 provided a step forward in interpreting the performance of ground support systems
under dynamic-loading conditions by considering the inferred seismic source mechanism.
Different requirements of support elements were identified in order to prevent intolerable levels
of damage to occur under strain bursts and fault slips. By analysing these types of rockbursts
separately, the influence of parameters related to seismicity, predominantly the event magnitude
and the distance that separated the epicentre of the seismic source from the damage, was
indirectly accounted for. What have not been considered however were the conditions of the
excavations where damage occurred. In this section, a classification of rockbursts accounts for
the excavation size, mining-induced stress conditions, and rock mass quality.

7.4.1 Rockburst classification based on the excavation conditions

Hierarchical clustering was performed over the scores of the second and third principal
components. These two components captured variations in the excavations’ conditions whereas
the first principal component essentially captured variations in the dynamic-loading conditions
generated by the seismic events.

An agglomerative method of hierarchical clustering was used in this analysis. Agglomerative

methods, as opposed to divisive methods, proceed by a series of successive fusions of the n
individuals, i.e. rockbursts, into groups (Everitt, Landau, Leese, & Stahl, 2011). In this analysis,
the Ward’s method of clustering (Ward, 1963) was applied. In this method, the fusion of two
clusters is based on the size of an error sum-of-squares criterion. The objective at each stage is to
minimize the increase in the total within-cluster error sum of squares (Everitt et al., 2011).
According to Everitt et al. (2011), no one method can be recommended above all others.
However, although the Ward’s method may impose a spherical structure where none exist, it
often appears to work well (Everitt et al., 2011).

Hierarchical clustering based on the excavation conditions resulted in the identification of 6

rockburst groups (Figure 7–7a). The interpretation of the clustered score scatter plot is facilitated
by referring to the loading scatter plot, which indicates the loadings associated with PC2 on the
horizontal axis and PC3 on the vertical axis (Figure 7–7b).
 198

The loading plot reveals that an important part of the variation in the collected rockburst data, on
the plane defined by the second and third principal components, results from the excavation span
and the length of reinforcement elements. PC2 and PC3 captured a negative correlation between
the span of mine openings and the normalized differential stress (NDS). This pattern suggests
that in large openings such as intersections, damage have resulted from mining-induced
seismicity in areas where the stress conditions did not appear to be especially hazardous. On the
other hand, damage to smaller excavations has occurred typically under higher mining-induced
stress conditions. This negative correlation between span and normalized differential stress is
represented by colour-coding rockbursts according to these two parameters on the PC2-PC3
score scatter plot (Figure 7–8).
 199

Figure 7–7 Grouping of rockburst data into 6 categories visualized on (a) the PC2-PC3 score scatter plot and
explained using (b) the PC2-PC3 loading scatter plot

Figure 7–8 Score scatter plots defined by PC2 and PC3 showing a negative correlation between (a) the span of
mine openings (in metres) and (b) the normalized differential stress
 200

The excavation size and NDS presented negligible correlations with depth and rock mass quality
(RMR) (Figure 7–7b). Both the “depth” and “rock mass quality” variables appeared to be
negatively correlated to each other. This pattern reflects the more fractured nature of the rock
mass at Creighton Mine as compared to Coleman and Copper Cliff. The observed pattern along
this axis also involved correlations with the type of structures typically reported at each mine
sites, i.e. dykes at Copper Cliff (shallower) and shear zones at Creighton (deeper).

A series of box-and-whisker diagrams (Figure 7–9) suggested that it is possible to rank the
identified rockburst groups based on the excavation span (Figure 7–9a) and the stress conditions.
Two stress parameters were considered in order to characterize the rockburst groups:

• the normalized differential stress [(σ1-σ3)/UCS] (Figure 7–9b); and

• the normalized major principal stress [σ1/UCS] (Figure 7–9c).

In investigating the performance of ground support systems, it is important to relate the stress to
the strength of the intact rock, as it is perceived to provide a better indicator of the potential for
rockbursts as opposed to the stress or depth variables alone. Both stress parameters employed the
mining-induced stresses obtained through numerical modelling, which was discussed in Chapter
5. The PC2-PC3 loading plot indicates that the variable “depth” accounted for a greater portion
of the variability in the rockburst data, as compared to the NDS (Figure 7–7b). This observation
suggested that a better rockburst classification could be obtained using the major principal stress,
which is more directly influenced by the depth, as opposed to the differential stress. A
comparison of the box-and-whisker plots (Figure 7–9b,c) contributed to validating this
interpretation. The interquartile ranges (IQR) were better spread out across the chart in Figure 7–
9c, especially when considering rockburst groups 1, 2, 5, and 6.
 201

Figure 7–9 Bar charts illustrating distributions of (a) excavation span, (b) normalized differential stress, (c)
normalized major principal stress, and (d) adjusted magnitude within each identified rockburst groups

It was observed that variables associated with mine seismicity did not contribute significantly to
the identification of patterns on the PC2-PC3 score plot. This was reflected by the location of
these variables, closer to the origin of the loading scatter plot (Figure 7–7b). According to this
pattern, the 6 identified rockburst groups were expected to overlap in terms of event magnitude.
The box-and-whisker plot (Figure 7–9d) further supported this interpretation. The influence of
variables associated with mine seismicity was predominantly captured in PC1.

A ranking of rockbursts based on the excavation span and normalized major principal stress is
proposed (Figure 7–10). The 6 groups of rockbursts defined in PCA are represented on this chart.
The extent of each rockburst group is quantified using the interquartile range from the box-and-
whisker diagrams (Figure 7–9a,c). Given the negative correlation between the variables “depth”
and “RMR” identified on the loading scatter plot, this classification further accounts indirectly
for variations in the rock mass quality amongst Creighton, Copper Cliff, and Coleman mines.
 202

Figure 7–10 The 6 identified rockburst groups represented using the interquartile range of the excavation span and
normalized major principal stress

In this section of the thesis, the performance of installed ground support systems is investigated
within each rockburst group. This approach, therefore, accounts for the effect of the excavation
conditions on the support performance. This classification, however, do not account for the
seismicity that was associated with the damage. In the following section, the magnitude and
distance between seismic events and damage are used to assess the influence of seismicity on the
performance of ground support systems installed in excavations of various size and under diverse
stress conditions.

7.4.2 Interpretation of support performance in consideration of

excavation conditions and seismic loading
A multivariate grouping of rockbursts was presented in Figure 7–7a. This grouping was
quantified in a more practical and applicable manner by focusing on the excavation size and
stress conditions (Figure 7–10). This grouping of rockbursts is used in this section to assess the
performance of ground support systems under excavation conditions that are interpreted as
comparable. The performance of the identified categories of support systems presented in Table
7–2 for the backs and Table 7–3 for the walls is assessed for each rockburst group in
consideration of the magnitude of the seismic event and the distance between the epicentre of the
seismic source and the damage location.
 203

7.4.2.1 Performance of support systems in the backs of mine openings

A summary of the ground support systems installed in the back and tested under dynamic
loading conditions in each rockburst group is presented (Table 7–4). The tabulated summary
indicates the number of rockburst in each group, the RoT associated with each support system in
the rockburst groups, the range of magnitudes the support systems have been subjected to, and
both the largest magnitude and ppv that led to acceptable or tolerable support performance. The
ppv was estimated using a scaling law that was developed based on data recorded in Ontario
mines, including Creighton Mine (Hedley, 1992). The ppv scaling laws have been reviewed in
Chapter 4.

The comprehensive assessment of support performance in the back of mine excavations is

presented from Figure 7–11 to Figure 7–17. These figures consist of scatter plots defined by the
adjusted event magnitude on the horizontal axis and the distance between the epicentre of the
seismic source and the damage location on the vertical axis. Each data point represents the
dynamic-loading conditions faced by the installed ground support systems. The shape of the data
points differentiates the categories of ground support systems and the employed colour-code
represents the support performance. Colour-shaded points represent intolerable performance of
the ground support systems. Light grey and white-shaded points represent tolerable and
acceptable performance, respectively.

The range defined by the 10th and 90th percentiles of encountered ppv-values was plotted over
each scatter plots. This range, introduced previously in Chapter 5, represents the most typical
ground motion levels faced by the support in Creighton, Copper Cliff, and Coleman mines. For
support design purposes, ground support systems that performed in an acceptable or tolerable
manner at the 75th percentile ppv threshold can be prescribed to manage dynamic-loading
conditions. The 10th, 75th, and 90th percentile of the ppv are 29, 312, and 719 mm/s, respectively,
and are represented on Figure 7–11 to 7–17 using dashed lines.
 204

Table 7–4 Summary of the categories of ground support systems installed in the back of mine openings and tested over ranges of excavation and dynamic-loading
conditions

Rockburst Rockburst group Support systems tested Number of RoT Magnitude Largest Largest ppv
group description rockbursts range magnitude (mm/s)
(SDS ≥ 2) (Nuttli) (SDS 2 or 3) (SDS 2 or 3)
Group 1 • Very large B4 – Mechanical bolts and rebars, cables 3 67 2.4 – 3.0 3.0 NA
excavations B7 – Rebars, cables 2 100 1.2 – 2.8 2.8 152
• Moderate stress B2s – Mechanical bolts, cables 2 0 2.4 NA NA
conditions
B4s – Mechanical bolts and rebars, cables 1 100 2.1 2.1 96
B7s – Rebars, cables 3 100 0.9 – 3.7 3.7 1,024
Group 2 • Typical B3 – Mechanical bolts and rebars 8 13 1.0 – 3.8 1.0 63
excavation size B5 – Rebars 8 25 0.7 – 3.7 1.4 185
• Moderate to high B3s – Mechanical bolts and rebars 3 0 2.4 – 3.2 NA NA
stress conditions
B5s – Rebars 4 50 0.9 – 3.8 3.4 390
B10s – Rebars and MCB 1 100 1.7 1.7 104
Group 3 • Large to very B3 – Mechanical bolts and rebars 1 0 1.0 NA NA
large excavations B5 – Rebars 3 67 0.7 – 1.0 1.0 53
• High stress B8 – Mechanical bolts and rebars, MCB 1 0 3.2 NA NA
conditions
B3s – Mechanical bolts and rebars 2 0 2.7 – 3.0 NA NA
Group 4 • Narrow B1 – Mechanical bolts or Swellex PM12 2 50 0.8 – 1.2 1.2 35
excavations B3 – Mechanical bolts and rebars 6 50 -0.2 – 2.0 1.2 125
• Moderate to very B5 – Rebars 4 75 -0.2 – 1.7 1.7 612
high stress
B10 – Rebars and MCB 1 0 3.7 NA NA
conditions
B1s – Mechanical bolts or Swellex PM12 2 100 1.6 – 2.8 2.8 137
B3s – Mechanical bolts and rebars 4 25 1.0 – 1.4 1.0 14
B5s – Rebars 1 0 1.7 NA NA
 205

Group 5 • Typical to large B3 – Mechanical bolts and rebars 1 100 1.3 1.3 162
excavations B4 – Mechanical bolts and rebars, cables 6 33 0.9 – 2.4 2.0 312
• High stress B5 – Rebars 1 0 1.0 NA NA
conditions
B2s – Mechanical bolts, cables 1 0 0.9 NA NA
B4s – Mechanical bolts and rebars, cables 11 82 0.9 – 3.1 2.8 997
B7s – Rebars, cables 1 0 2.6 NA NA
B9s – Mechanical bolts and rebars, MCB and 1 100 2.6 2.6 2,824
cables
Group 6 • Typical B1 – Mechanical bolts or Swellex PM12 4 100 0.4 – 1.4 1.4 89
excavation size B3 – Mechanical bolts and rebars 22 41 0.4 – 4.1 3.1 2,765
• High to very high B5 – Rebars 2 0 0.5 – 1.3 NA NA
stress conditions
B6 – Mechanical bolts and rebars, rebars or FS-46 2 50 2.0 – 4.1 4.1 114
B8 – Mechanical bolts and rebars, MCB 2 50 0.9 – 3.2 0.9 122
B10 – Rebars and MCB 4 50 0.7 – 2.6 2.6 243
B1s – Mechanical bolts or Swellex PM12 1 0 2.6 NA NA
B3s – Mechanical bolts and rebars 19 68 0.5 – 4.1 4.1 1,519
B4s – Mechanical bolts and rebars, cables 1 100 2.4 2.4 167
B5s – Rebars 2 0 0.6 – 2.0 NA NA
B6s – Mechanical bolts and rebars, rebars or FS- 1 100 2.7 2.7 206
46
B8s – Mechanical bolts and rebars, MCB 5 100 0.5 – 3.2 3.2 633
B10s – Rebars and MCB 1 100 1.5 1.5 125
 206

Figure 7–11 Rockburst group 1: performance of ground support systems that comprise (a) steel mesh and (b)
mesh-reinforced shotcrete, in the back of mine openings

Rockburst group 1 comprises very large mine openings where the span have frequently exceeded
9 m. The majority of these openings were located within low to moderate stress conditions
(Figure 7–10). All these openings have been supported using cable bolts. Using steel mesh as
surface support, the installation of ground support system B4 led to intolerable levels of damage
whereas system B7 performed in an acceptable manner. The support system B7 was tested under
a magnitude 2.8 MN event which occurred 66 m away from the damage. The resulting ppv of
152 mm/s was not sufficient to generate significant damage to the support.
 207

Combined with shotcrete, both support systems B4s and B7s performed in an acceptable manner.
However, the support system B7s withstood much larger ppv levels that ranged from 467 to
1,024 mm/s. Those two events associated with magnitudes 2.7 and 3.7 MN, respectively, resulted
in ppv at the damage locations that exceeded the 75th percentile threshold. Support system B2s
resulted in intolerable damage at ppv levels slightly below the 75th percentile threshold (221 and
263 mm/s). Those performances highlight the limitations in the performance of mechanical bolts
under dynamic loads. Based on support performance results presented in Figure 7–11, the use of
resin-grouted rebars in conjunction with cable bolts is recommended in type 1 excavations.

Figure 7–12 Rockburst group 2: performance of ground support systems that comprise (a) steel mesh and (b)
mesh-reinforced shotcrete, in the back of mine openings
 208

Rockburst group 2 (Figure 7–12) comprises typical size excavations (5.0-7.0 m span) undergoing
moderate to high stress conditions. The ground support systems employed at rockburst locations
consisted typically of a single pass of stiff reinforcement elements, whether resin-grouted rebars
or a combination of mechanical bolts and rebars. Based on the employed support systems, the
collected data suggest that the mine personnel did not perceive type-2 excavations as presenting
a significant risk of rockburst at the time. Nevertheless, these excavations have been hit by
severe magnitudes that exceeded 3.0 MN. The majority of these large seismic events were
associated with large-scale seismically-active structures such as the Trap Dyke at Copper Cliff
Mine and the contact between the MOB and the granite footwall at Coleman Mine. Damage that
resulted from those events occurred in the near vicinity of those structures. This indicates that
despite the lower stress environment, ground support system must be adjusted in locations where
typical-size openings intersect seismically-active structures.

In areas where geological structures did not pose a problem, inconsistent performance of the
support systems has been observed. Despite that resin-grouted rebars withstood a 185 mm/s ppv
without shotcrete (system B5) and 390 mm/s with shotcrete (system B5s), failure of the system
has also been reported at lower ppv thresholds. These observations suggest that one-pass, stiff
support systems perform inconsistently when subjected to dynamic loads. Nevertheless, the data
suggest that the application of shotcrete, hence the construction of a strong and stiff support
system, can prevent the occurrence of strain bursts. This is implied by the few rockburst case
studies of magnitude lower than 2.2 MN that are plotted on Figure 7–12b. Three rockbursts in
Figure 7–12b were characterized by magnitude of 2.2 MN or less as opposed to 7 rockbursts in
Figure 7–12a.

Rockburst group 3 (Figure 7–13) comprises large to very-large openings that were under
moderate to very high stress conditions. This group includes broad ranges of excavation
conditions, however, the backs that were hit by rockbursts have in common the fact that they
were not supported using cable bolts. As a result, the performances of the support systems
reported in rockburst group 3 have been generally intolerable, despite the low ppv values that
were estimated based on the magnitude and the distance from the seismic event. The failure of
ground support systems in rockburst group 3 emphasizes the importance of supporting large
excavations using long reinforcement elements such as cable bolts.
 209

Figure 7–13 Rockburst group 3: performance of ground support systems that comprise (a) steel mesh and (b)
mesh-reinforced shotcrete, in the back of mine openings

Rockburst group 4 (Figure 7–14) comprises mainly narrow excavations characterized by a span
lower than 5.0 m. Stress conditions however vary considerably from moderate to very high.
Similarly to rockburst group 2, the support systems employed in rockburst group 4 consisted
typically of a single pass of stiff reinforcement elements. Used in conjunction with welded-wire
mesh, resin rebars (B5) have performed successfully under a 612 mm/s ppv. With the application
of shotcrete, it is mechanical bolts that performed the best in the back of mine openings
sustaining a ppv of 137 mm/s. These results are contradictory and indicate that one pass stiff
support systems perform inconsistently under dynamic-loading conditions. Nevertheless, in type-
 210

4 openings, a stiff support system comprising resin-grouted rebars and welded-wire mesh could
be appropriate to handle dynamic-loading conditions generated by strain-burst events.

Figure 7–14 Rockburst group 4: performance of ground support systems that comprise (a) steel mesh and (b)
mesh-reinforced shotcrete, in the back of mine openings

The rockburst group 5 (Figure 7–15) comprises events that occurred in high stress conditions and
in typical to large-size openings. Cable bolts were installed in the majority of these excavations.
The support system used most frequently was composed of mechanical bolts and rebars installed
on a diamond pattern in conjunction with mesh-reinforced shotcrete and a second pass of cable
bolts. Higher ppv thresholds were sustained in locations where the support system comprised
mesh-reinforced shotcrete as opposed to welded-wire mesh. Compared to the events reported in
 211

rockburst group 1, those in rockburst group 5 indicate that under high stress conditions, it
becomes highly recommended to support the excavations using mesh-reinforced shotcrete.

Figure 7–15 Rockburst group 5: performance of ground support systems that comprise (a) steel mesh and (b)
mesh-reinforced shotcrete, in the back of mine openings

Rockburst group 6 comprised typical-size excavations under high to very high stress conditions.
Out of the 139 rockbursts that comprised this group, 66 events generated significant levels of
damage to the support installed in the back of mine openings. In order to make more sense of the
information presented on the scatter plots, the performance of single-pass, stiff support systems
(Figure 7–16) was investigated apart from the one of multiple-passes, enhanced support systems
(Figure 7–17).
 212

Figure 7–16 Rockburst group 6: performance of one-pass ground support systems that comprise (a) steel mesh and
(b) mesh-reinforced shotcrete, in the back of mine openings

Stiff support systems that did not include shotcrete (Figure 7–16a) have resulted in inconsistent
performances under dynamic loads. Failure of support systems have occurred over the entire
range of ppv values. Some rockbursts have resulted in tolerable levels of damage (SDS 3),
however, there have been very few acceptable levels of damage (SDS 2). Used in conjunction
with mesh-reinforced shotcrete (Figure 7–16b), the same combinations of reinforcement
elements (B1, B3, B5) have generated improved performances under both strain bursts and larger
magnitude fault-slip events. The relatively few number of rockbursts reported with magnitudes
lower than 2.2 MN on Figure 7–16b as opposed to Figure 7–16a demonstrate that a strong and
stiff surface support can prevent the occurrence of strain bursts. The 9 cases of acceptable and
 213

tolerable levels of damage severity that resulted from magnitudes greater than or equal to 2.5
MN, as opposed to the 5 cases of intolerable support performance, indicate that a strong and stiff
surface support, under high to very high stress conditions, increases the capacity of the system to
manage seismic shaking.

Figure 7–17 Rockburst group 6: performance of enhanced ground support systems that comprise (a) steel mesh
and (b) mesh-reinforced shotcrete, in the back of mine openings

The addition of a second pass of yielding reinforcement elements or cables has a clear impact on
the capacity of support systems to manage dynamic loads in high to very high stress
environments (Figure 7–17). Without the application of shotcrete, two-pass support systems have
failed under dynamic loads when the ppv exceeded the 75% threshold. With the application of
 214

shotcrete, there has been only one rockburst where the ppv has exceeded the 75% threshold: the
system performed appropriately.

Based on Figures 7–16 and 7–17, two distinct approaches have been used to the design of
support systems for mine excavations subjected to very high stress conditions.

1. To increase the strength and stiffness of the support system using mesh-reinforced shotcrete
2. To increase the capacity of the support system to yield and absorb kinetic energy

The first approach can be successful in mitigating strain bursts and in managing the
consequences of seismic shaking. However, beyond a certain threshold of loading, the
performance of stiff support systems becomes inconsistent and therefore unreliable. This is
demonstrated in Figure 7–16. The application of shotcrete has proven to be labour intensive at
Creighton Mine as cracking and spalling often necessitated rehabilitation.

As part of the second approach, yielding support has provided consistently adequate
performances under the 75th percentile ppv threshold, both with and without the application of
shotcrete (Figure 7–17). Based on the collected rockburst data, it appears necessary in typical
size openings that are undergoing high to very high stress conditions to rely on yielding support
to manage the dynamic-loading conditions triggered by all ranges of event magnitudes. The
application of shotcrete between two passes of reinforcement can contribute to increasing the
level of confinement around the primary reinforcement and enhance the load transfer between
the surface support and the second-pass reinforcement. However, the application of shotcrete
depends on the economics and logistics of the mine. Based on the collected data, cable bolts
represent a way to avoid the application of shotcrete while ensuring that the support system is
properly tied to stable ground. The installation of cable-bolts in conjunction with yielding
reinforcement elements and welded-wire mesh could together form an economically and
technically feasible support strategy to manage dynamic loads in the back of mine excavations in
deep mining environments.

7.4.2.2 Performance of support systems in the walls of mine openings

The rockburst classification presented in Figure 7–10 was appropriate for the back since it
referred to the span of mine openings. In the walls, the critical dimension is the height of the
excavation. Since variations in height are not as sensitive as variations in span, the assessment of
 215

support performance in the walls accounted for variations in mining-induced stresses only.
Hence, as part of this analysis, the rockbursts data used to investigate trends in the performance
of ground support systems were those classified amongst:

• Rockburst groups 1 and 2;

• Rockbursts groups 5 and 6.

This section therefore compares the performance of ground support systems in the walls of mine
openings that were subjected to low and moderate stress conditions to those subjected to high
and very high ranges of normalized major principal stress.

Under low to moderate stress conditions, three types of primary reinforcement elements have
been employed: mechanical bolts, friction sets, and rebars. Ground support systems comprising
friction sets and rebars have been occasionally enhanced using yielding bolts such as the MCB.
Under high to very high stress conditions, ground support systems typically comprised friction
sets. Mechanical bolts were not reported within rockbursts groups 5 and 6 and the installation of
rebars was very limited. This trend demonstrates the potential limitations in the performance of
mechanical bolts under high stress. Furthermore, given the absence of grout, it is possible that
friction sets were preferred to rebars for installation in highly stressed and more fractured
ground. The selection of friction sets can also be justified economically given its affordability
and ease of installation. The configurations of reinforcement elements reported in the walls of
mine openings within rockburst groups 1-2 and 5-6 are reported in Table 6-5. The table further
reports on the ranges of magnitudes faced by the support, the largest magnitudes and ppvs that
resulted in acceptable or tolerable performances, and the calculated RoT for each category of
support. The analysis of support performance is extended further by introducing the magnitude
and distance between the epicentre of the seismic source and the damage. The performance of
ground support systems under seismic loads and stress conditions captured by rockburst groups
1-2 and 5-6 are represented from Figures 7–18 to 7–23.
 216

Table 7–5 Summary of the categories of ground support systems installed in the walls of mine openings and tested over ranges of stress and dynamic-loading
conditions

Rockburst Rockburst group Support systems tested Number of RoT Magnitude Largest Largest ppv
group description rockbursts range magnitude (mm/s)
(SDS ≥ 2) (Nuttli) (SDS 2 or 3) (SDS 2 or 3)
Group 1-2 • Low-moderate W1 – Mechanical bolts, mechanical and FS35/39 17 47 0.5 – 3.8 3.8 273
stress conditions W2 – FS35 or FS39 4 75 0.7 – 1.6 1.6 233
W4 – FS39 and rebars or FS46 2 50 2.9 – 3.7 2.9 622
W5 – Rebars 3 0 0.7 – 3.7 NA NA
W10 – Rebars and MCB 1 100 2.9 2.9 186
W1s – Mechanical bolts 7 14 1.2 – 3.8 1.2 212
W2s – FS35 or FS39 8 50 1.7 – 3.8 2.4 315
W4s – FS39 and rebars 1 0 3.2 NA NA
W5s – Rebars or FS39 and Swellex MN12 5 100 2.7 – 3.4 3.4 742
W9s – FS46 and MCB 2 100 1.5 – 2.5 2.5 851
Group 6 • High to very high W2 – FS35 or FS39 36 58 0.5 – 4.1 4.1 503
stress conditions W4 – FS46 9 44 0.7 – 2.6 2.6 389
W6 – FS35 and cables 1 0 1.9 NA NA
W7 – FS35 and MCB 1 100 2.8 2.8 587
W8 – FS46 and cables 1 100 2.0 2.0 115
W2s – FS35 or FS39 30 43 1.1 – 4.1 4.1 860
W4s – FS39 and rebars or FS46 9 67 0.6 – 2.7 2.7 528
W6s – FS35 and cables 5 60 1.2 – 2.9 2.8 118
W8s – FS46 and cables 4 100 1.2 – 2.6 2.6 747
W9s – FS46 and MCB 5 100 0.5 – 3.2 3.2 125
 217

Figure 7–18 Rockburst group 1-2: performance of ground support systems that comprise mechanical bolts and
either (a) steel mesh or (b) mesh-reinforced shotcrete, in the walls of mine openings

In low to moderate stress environments, support systems that comprised mechanical bolts were
somewhat unreliable at managing dynamic-loading conditions in the walls of mine openings
(Figure 7–18). Although there have been cases, typically at very low ppvs, where the resulting
damage to the support was acceptable or tolerable, intolerable performances of the support have
occurred over a wide range of ppvs. It may have been possible to prevent small magnitude
rockbursts through the application of shotcrete; however, when subjected to significant dynamic
loads, mechanical bolts did not performed in an acceptable manner. Therefore, in the walls of
 218

mine excavations, even under low to moderate stress conditions, mechanical bolts cannot be
recommended.

Figure 7–19 Rockburst group 1-2: performance of ground support systems that comprise friction sets and either (a)
steel mesh or (b) mesh-reinforced shotcrete, in the walls of mine openings

Similar to the mechanical bolt, the friction set is one of the most affordable reinforcement
elements used in mining and its installation is quite straightforward. However, the performance
of these two types of bolts under dynamic-loading conditions differed considerably. This
affected the performance of the support systems, as illustrated in Figure 7–19. In conjunction
with welded-wire mesh (Figure 7–19a), friction sets have performed satisfactorily up to high ppv
levels, near the 75th percentile threshold. In conjunction with shotcrete, the lowest magnitude
 219

event that triggered failure of the system was a 2.4 MN (Figure 7–19b). Magnitudes 1.7 MN and
2.0 MN have both resulted in tolerable performances of the system. Once again, the use of a
strong and stiff surface support contributed to preventing strain bursts. Failure of the system,
however, was typically triggered by ppvs that ranged close to the 75th percentile threshold. The
only exception was a large 3.8 MN event where the resulting ppv at the damage location was 82
mm/s. It is possible that under dynamic-loading conditions, the stiff shotcrete membrane has
prevented the sliding mechanism of friction sets to take action. This could explain the arguably
lower ppv damage threshold for shotcrete support as opposed to welded steel mesh. The
installation of friction sets is recommended under low to moderate stress conditions; however,
for improved performances under dynamic loads, it is preferable to avoid covering them with
shotcrete. For increased system stiffness, a more appropriate approach could be to install the
mesh and bolts over shotcrete.

Installed in conjunction with welded-wire mesh, rebars have failed under lower ppv thresholds,
as compare to the friction sets (Figure 7–20). Given that rebars are more costly and that their
installation is more time consuming, there doesn’t seem to be an inherent advantage in spending
additional resources on the installation of rebars in the walls. Under low to moderate stress
conditions, friction sets can achieve the same level of performance as rebars under dynamic
loads or even exceed it. With or without shotcrete, the use of yielding bolts, e.g. MCB or Swellex
MN12, have resulted in enhanced performance of the systems W10, W5s, and W9s. When the
reinforcement was installed in conjunction with mesh-reinforced shotcrete, no support failure has
been recorded, even beyond to the 90th percentile ppv threshold.

The support system W5s comprised either resin-grouted rebars or a combination of FS39 (1st
pass) and Swellex MN12 (2nd pass). Although these are two fundamentally different support
systems, the conceptual load-bearing and energy-absorption capacities, calculated in
consideration of the installation pattern, were quite similar. This is why these two reinforcement
configurations were grouped within the same category. Combining the installation of resin-
grouted rebars with mesh-reinforced shotcrete results in a very stiff support system. Such
strategy, over the last set of support performance charts, has proven to be effective in preventing
small magnitude strain bursts and in minimizing the level of damage triggered by seismic
shaking. In the second version of support system W5s, friction sets, mesh, Swellex MN12, and
straps are all installed over the shotcrete membrane. This is a distinctive approach in the sense
 220

that both the primary and secondary tendons can beneficiate from the capacity of the shotcrete
membrane to distribute the load over a wider area. Enhanced load and energy transfers can result
from this approach. Furthermore, the installation of mesh over the shotcrete can prevent
excessive degradation of the system by preventing the detachment of loose pieces of shotcrete,
hence maintaining the level of confinement in the periphery of the excavation as cracking
progresses. This system has been effective at managing high ppv levels under low and moderate
stress conditions. It would be interesting to test its performance under high stress conditions.
Unfortunately, the application of this system was limited to Coleman Mine.

Figure 7–20 Rockburst group 1-2: performance of ground support systems that comprise rebars or enhanced
support and either (a) steel mesh or (b) mesh-reinforced shotcrete, in the walls of mine openings
 221

Under high and very high levels of normalized major principal stress, friction sets were the only
reinforcement elements employed as part of a one-pass support system, in the walls of mine
excavations. This must testify of the reliability of the friction set in deep and high stress
conditions and furthermore, to the high level of productivity that can be achieved using this bolt
type in comparison to the mechanical bolt and the resin-grouted rebar. For rockbursts that were
part of groups 5 and 6, the performance of systems that comprised 35 and 39 mm friction sets
(Figure 7–21) was analysed apart from systems that comprised 46 mm friction sets (Figure 7–
22).

Figure 7–21 Rockburst group 5-6: performance of ground support systems that comprise 35 or 39 mm friction sets
and either (a) steel mesh or (b) mesh-reinforced shotcrete, in the walls of mine openings
 222

Although 35 and 39 mm friction sets have performed relatively well under low and moderate
stress conditions, their performance under high and very high stress has become more
inconsistent. Installed in conjunction with welded-wire mesh, FS35 and FS39 have provided
mixed performance, mostly between the 10th and 75th percentile ppv thresholds. In conjunction
with shotcrete, failure of the system W2s typically occurred at ppv greater than 100 mm/s.
Cracking of shotcrete under high stress conditions may have contributed in lowering the damage
threshold, as compared to low and moderate stress environments (Figure 7–19). In spite of this
potential limitation, Figure 7–21 demonstrates that the performance of small diameter friction
sets in the walls of mine excavation is limited and inadequate under high and very high stress
conditions.

The performance of support systems comprising 46 mm friction sets under high and very high
stress conditions is represented (Figure 7–22). Without the application of shotcrete, the collected
data may suggest that FS46 are less reliable than smaller diameter friction sets. This is
attributable to the greater proportion of intolerable damage observed in Figure 7–22a as opposed
to Figure 7–21a. Furthermore, the collected data may suggest that the performance of the FS46 is
fairly limited under magnitudes greater than 1 MN. This inconsistency between the performance
of FS46 and smaller diameter friction sets may only be attributable to the inherent variability in
the field performance of ground support systems. One element that may reflect an expected
improved performance due to the increased diameter of friction set bolts is the higher ppv
damage threshold that characterized rockburst data in Figure 7–22a as opposed to Figure 7–21a.

Employed in conjunction with mesh-reinforced shotcrete, the improved capacity and consistency
of the support systems that comprise larger diameter friction sets is noticeable. This
improvement is supported by the alignment of six cases of acceptable and tolerable performances
along the 75th percentile ppv threshold. The lowest magnitude that triggered failure of the
systems within the 10th to 90th percentile ppv interval was a 1.9 MN. This result is consistent with
the previous charts, i.e. shotcrete enhance the performance of support systems and prevent the
occurrence of smaller magnitude strain-burst events. Both small and large friction sets under
high and very high stress conditions have performed better when combined with shotcrete.
Maintaining the confinement around the friction sets may be an important factor in ensuring
adequate performance of the support system.
 223

Figure 7–22 Rockburst group 5-6: performance of ground support systems that comprise 46 mm friction sets and
either (a) steel mesh or (b) mesh-reinforced shotcrete, in the walls of mine openings

It was argued, in reviewing the performance of support systems installed in the back of mine
openings, that the preservation of the confinement around the primary reinforcement elements
could be achieved using cable bolts. There was a clear enhancement of the support performance
when cables were added to a system that comprised friction sets (Figure 7–23). This was
demonstrated by the performance of support systems W6, W8, W6s, and W8s. The support
system W8s has survived ppvs that exceeded the 75th percentile ppv threshold.
 224

Figure 7–23 Rockburst group 5-6: performance of ground support systems that comprise friction sets, enhanced
with yielding reinforcement or cables, and either (a) steel mesh or (b) mesh-reinforced shotcrete, in the
walls of mine openings

There has been also excellent performance of support systems enhanced by the installation of
yielding bolts, namely the MCB. As expected, because it has been repeated for all the support
performance charts for both the walls and the backs, all size of excavations and all stress
conditions, yielding support works in managing dynamic loads. If a mine is seismically active,
there is no reason why yielding support should not be part of the minimum support system. The
installation of reinforcement elements capable of yielding is the only way to achieve consistent
support performance under dynamic loads over typical ranges of ground motions (ppv) and event
magnitudes.
 225

7.5 Summary
In this chapter, ground support systems installed in the backs and walls of mine openings have
been regrouped based on their conceptual load-bearing and energy-absorption capacity. An
overview of the performance of support systems was provided using the Rate of Tolerance
(RoT), which was developed as the ratio of acceptable and tolerable performance of ground
support systems to the number of significant dynamic-loading events faced by the support.
Thresholds of support performance were defined based on the RoT, ranging from excellent,
acceptable, tolerable, unreliable, and intolerable. The performance of support systems that
comprised welded-wire mesh and mesh-reinforced shotcrete was analysed separately. Shotcrete,
cable bolts, and yielding reinforcement elements were common to all support systems that
provided acceptable to excellent performance under dynamic loads.

It was important to incorporate the dynamic-loading component in the interpretation of the

support performance. The rockbursts were split in two groups based on the inferred seismic
source mechanism. The performance of support systems was analysed using the RoT for strain
bursts and fault slips, separately. It was concluded that in the back, shotcrete was not always
necessary to manage the consequences of strain bursts, as long as the employed support system
comprised yielding reinforcement elements such as MCB. Under dynamic loads generated by
large magnitude fault-slip events, maintaining the confinement around the bolts by preventing
dilation of the rock mass was a necessary function of the support. In the collected data, this was
achieved by the installation of cable bolts, which aimed at tying the support back to stable
ground, and by the application of shotcrete. The use of shotcrete however results in a very stiff
support system and it was recommended, in order to overcome this limitation, to enhance the
system using a second pass of yielding reinforcement elements.

In the walls, increasing the stiffness of ground support systems through the application of
shotcrete has been a sound strategy to manage the consequences of fault slip events. This
strategy, however, had limited success under strain bursts, based on the fault slip and strain burst
classification. The installation of reinforcement elements capable of absorbing energy was
recommended over the shotcrete membrane. Both yielding bolts and cable bolts were observed to
perform adequately based on the RoT. The possibility of success of such yielding system without
shotcrete could difficultly be assessed given the very few strain burst case studies available.
 226

However, based on patterns observed in the back and walls of mine openings, it was interpreted
that a support strategy that comprised welded-wire mesh could be successful under strain bursts
and fault slips when long-enough reinforcement elements, e.g. 2.0 m-long FS-46, are installed in
conjunction with yielding reinforcement elements and heavy-gauge straps.

The classification of rockbursts based on the inferred seismic source mechanism brought the
interpretation of the support performance a step forward. It was only by classifying the rockburst
data based on the stress conditions and size of excavation that a complete picture of the
performance of each support systems could be drawn. This classification accounted indirectly for
the rock mass quality since there was a negative correlation between the variables depth and
RMR. The performance of the support systems was investigated within 6 groups of rockbursts, in
the back, and 2 groups, in the walls. The performance was reviewed under various stress and
ground conditions and in consideration of the magnitude, distance from the seismic event, and
the estimated ppv.

It has been extracted that under similar circumstances, stiff support systems behaved
inconsistently. Support systems that included yielding elements, although not always successful,
tend to perform adequately and to sustain higher ground motion levels. Support systems that
comprised shotcrete have been successful in mitigating the overall damage of strain bursts. On
the support performance charts, it was generally observed that damage to support systems that
comprised shotcrete tended to occur under higher magnitude thresholds as compared to support
systems that comprised mesh. Focusing on mine opening of typical size, i.e. rockburst groups 2
and 6, it is possible to state that support systems that comprise shotcrete over mesh, as opposed
to mesh only, are capable of mitigating strain bursts up to a magnitude of 1.5 MN. This is
supported by the lower rate of rockburst, represented by the dotted line, up to magnitude 1.4 to
1.6 MN for support systems comprising shotcrete (Figure 7–24). For support systems comprising
mesh, the increase in the rate of rockburst starts under a much lower magnitude of 0.2 MN. Based
on Figure 7–24, the installation of a strong and stiff surface support such as mesh-reinforced
shotcrete in typical size openings is justified for mitigating strain bursts that generate magnitudes
between 0.2 MN and 1.5 MN.
 227

Figure 7–24 Cumulated number of rockbursts for support systems comprising mesh and shotcrete (over mesh) in
mine excavations of typical size (rockburst groups 2 and 6)

Over 1.5 MN, it is recommended to reconsider the support strategy as increasing the support
stiffness may no longer provide the optimal solution. The introduction of cable bolts and/or
yielding reinforcement elements in the system becomes fully justified. Early in this analysis, the
installation of cables was proven to be necessary in large and very large excavations that
exceeded 7 m span. As the analysis progressed, it was observed that cables could benefit support
systems by preventing rock mass dilation while anchoring the support into stable ground. The
softer behaviour of cables, as opposed to rebars for example, makes it possible to accommodate
certain levels of displacements and released energy. The yielding concept and the installation of
longer reinforcement elements are part of the success in managing dynamic-loading conditions in
deep and high stress mines.

Chapter 7 has reviewed the performance of ground support systems through various levels of
data filtering. The lessons in this chapter were based on an analysis of documented rockburst
case studies over 32 years at three different mines. The completeness of the rockburst dataset
over the time period covered in this study represented an important advantage in interpreting the
performance of ground support systems. In the industry, assessments of support performance and
validations of support designs are often based on individual rockburst events. In this analysis,
perceptions related to the support performance were challenged and the interpretations were
 228

validated using verified rockburst data. The chapter demonstrated that ground support systems
behave differently under various levels of dynamic-load demand and damage mechanisms.

In Chapter 8, the knowledge developed on the performance of ground support systems is

integrated to the analysis of seismic, ground, and stress conditions that were characteristic of
rockburst occurrences in the collected data. The outcome is a ground support strategy that
prescribes 1) what type of support should be installed, 2) where in the mine should it be installed,
and 3) when in the life of the mine should it be installed.
 229

8 Ground Support Strategy for Burst-Prone Grounds

8.1 Assessment of the seismic risk

An important step in establishing a ground support strategy for a deep underground mine is to
evaluate the level of risk associated with mining-induced seismicity. Ground control engineers
must assess whether the recorded, or anticipated, seismicity presents a significant risk for the
stability of mine openings and the safety of the mining personnel and equipment. Based on the
collected, and reviewed, seismicity and rockburst data that covered a 14-year time span at
Creighton Mine, it is interpreted that a significant level of seismic risk is present in an
underground mine when magnitudes exceed or are anticipated to exceed 1.6 – 2.0 MN (Figure 8–
1). A 1.6 MN event, at Creighton Mine, resulted in a 5% probability of rockbursts and a 2.5%
probability of generating over 10 tonnes of displaced material. A 2.0 MN event resulted in a 10%
probability of rockbursts and a 5% probability of generating over 10 tonnes of displaced
material.

Figure 8–1 Recommended support strategy based on the largest expected magnitude and the associated
probability of rockburst
 230

The 1.6 – 2.0 MN range of magnitude appears to correlate with the transition from a strain burst
to fault slip mechanism identified in the collected data (Figure 7–4) and further supported by the
interpretations provided by Ortlepp (1997) on the seismic source mechanisms (Table 4–1). The
developed guidelines presented in Figure 8–1 recommend a zone of transition in ground support
practice, i.e. a zone where a mine should consider transitioning from a static to a dynamic ground
support strategy. Below 1.6 MN, the seismic risk is interpreted as sufficiently low and a dynamic
ground support strategy may be applied only in particular circumstances. It is critical that when
magnitudes are anticipated to exceed 2.0 MN, a mine must be prepared to manage rockbursts and
adopt a dynamic support strategy that can be applied in a systematic manner.

In practice, the largest anticipated seismic event at a given mine or in a specific area of a mine
can be determined using the frequency-magnitude relationship (Hudyma & Potvin, 2010). A
conceptual example of the assessment of ground support requirements is based on the frequency-
magnitude relationship obtained for two different mining areas (Figure 8–2).

Figure 8–2 Conceptual assessment of the ground support requirements, based on the largest anticipated event
magnitude and probability of rockburst, for two different mining areas

The largest anticipated magnitude in the population of seismic events represented by red
diamond shapes approaches 2.4 MN and therefore exceeds the 1.6 – 2.0 MN threshold. The
seismic risk in the area of the mine where the population of seismic events was located is
therefore sufficiently high to anticipate the occurrence of rockbursts. It would therefore be
 231

recommended that the mine adopt a ground support strategy aiming at managing dynamic loads
for this area. On the other hand, the level of risk associated with the population of seismic events
represented using blue squares is reasonably low. Rockburst may not be an important concern in
addressing ground support requirements for this area. Therefore, the level of ground support may
be specified in consideration of the static conditions, only. The seismic risk can vary in time and
space; hence, it is important in a deep underground mine to be aware of the evolution of the
seismic risk as mining progresses. This is a considerable challenge.

The guideline presented in Figure 8–1 was based on seismic data that were calibrated using the
Nuttli scale employed by the Geological Survey of Canada (GSC). Therefore, the guideline is
most pertinently applicable in Canadian underground mines where the employed seismic
monitoring system has been calibrated to the GSC. The design magnitude and the magnitude
threshold, between 1.6 and 2.0 MN, adopted at a particular mine for transitioning from a static to
a dynamic ground support strategy, should also be reflective of the risk tolerance of the operator.
The responsibility of ground control is to be able to communicate this risk to the management.

8.2 Support strategy for an identified seismic source

Larger magnitude events have a greater probability of causing damage to mine openings and to
further cause damage over larger distances. The collected rockburst data allowed the
determination of the radius of influence of seismic events, i.e. the distance from a seismic source
over which damage can be reasonably anticipated. The radius of influence was determined using
the estimated peak particle velocity (ppv). In this study, the ppv was estimated using the scaling
law proposed by Hedley (1992), which was developed based on data collected in Ontario mines,
in Canada.

In Chapter 5, iso-ppv lines corresponding to the 10th, 25th, 50th, 75th, and 90th percentiles of the
collected rockburst data were presented. The 10th and 25th ppv percentiles were employed in this
section in order to define a distance range, from the epicentre of seismic events, over which
dynamic-loading conditions can be anticipated. Based on the complete and verified rockburst
records from Creighton, Copper Cliff, and Coleman mines, it is justified to anticipate the
exposure of supported mine excavations to dynamic-loading conditions in areas where the ppv
exceeds 30 to 65 mm/s (Figure 8–3).
 232

Figure 8–3 Anticipation of static and dynamic-loading conditions based on the radius of influence of seismic events

The interpretation of the consequences of the ppv on the stability of mine excavations by Hedley
(1992) consisted of:

• Falls of loose ground occurring at 50 mm/s;

• Fracturing of intact rock developing at 300 mm/s; and

• Severe damage occurring at 600 mm/s.

The 50 mm/s ppv threshold specified by Hedley (1992) coincides with the 30 to 65 mm/s
rockburst initiation threshold derived based on the collected and verified rockburst data. This
result contributes further in validating the adjusted seismic data. Figure 8–3 can have direct
implications on quality control and quality assurance, for example, when planning site
investigations following the occurrence of seismic events of significant magnitudes. Based on
the radius of influence, ground control personnel can identify areas where the support was
potentially affected by dynamic-loading conditions. Mine excavations located within the radius
of influence of a seismic event can be targeted to assess the impact of dynamic loading on the
support integrity.

The application of Figure 8–3 for ground support design, however, can be quite limited. This is
because predicting both the location and the magnitude of a potential seismic event is a
significant challenge, which may not be possible to overcome. Nevertheless, when seismicity or
 233

a cluster of seismic events in a mine is specifically associated with an identified seismic source,
such as a fault or a dyke, Figure 8–3 can be used as a tool for the selection of ground support
systems. The radius of influence specific to individual seismically-active structures can be
assessed based on the level of seismic hazard related to the structure. The seismic hazard can be
interpreted as the largest anticipated magnitude event, determined based on the frequency-
magnitude relationship (Hudyma & Potvin, 2010).

For the population of seismic events represented using red diamond shapes (Figure 8–2), the
largest anticipated magnitude is close to 2.4 MN. Using the 30 to 65 mm/s ppv threshold (Figure
8–3), the damage that would result of a design 2.4 MN event, if any, could be anticipated within
83 to 134 m from the seismic source, at a 75% to 90% confidence level, respectively.
Consequently, if this population of seismic events was related to an identified seismically-active
structure, ground control personnel could apply Figure 8–3 to recommend the enhancement of
ground support systems installed within the 83 to 134 m vicinity. The ppv threshold employed
for assessing the radius of influence of an identified geological structure should reflect the risk
tolerance of the operator.

Figure 8–3 is a practical tool, derived from reviewed and validated rockburst field data, which
can assist ground control personnel in assessing the likelihood of rockbursts in the vicinity of
identified seismic sources. This tool can be used to prescribe appropriate support measures
within a design radius of influence. Figure 8–3 does not aim at predicting the severity of
rockbursts. The arguably random distribution of the RDS over the collected rockburst data
demonstrates that the severity of rockburst is not a direct function of the ppv, only (Figure 8–4).
Each rockburst reported in the rockburst database is represented in Figure 8–4 in terms of
magnitude and distance between the damage and the epicentre of the seismic event. The
rockbursts were colour-coded using the 5 levels of damage severity, according to the RDS
(Kaiser et al., 1992). Straight lines representing the PPV percentiles calculated using the scaling
law from Hedley (1992) are displayed over the rockburst data points. Those percentiles do not
suggest any direct relationship between increasing levels of ppv and increasing damage severity.
Furthermore, they do not support the threshold provided by Hedley for the occurrence of severe
damage, as the vast majority of the reviewed rockbursts occurred under ppv levels much lower
than 600 mm/s.
 234

Figure 8–4 Magnitude, distance from the epicentre of seismic events, and severity of damage (RDS) for the
rockbursts that occurred at Creighton, Copper Cliff, and Coleman mines

The conclusion supported by Figure 8–4 challenges previous guidelines that suggested
increasing levels of damage severity along with increasing ppv (Hedley, 1992) or prescribed
adjustments to the employed support systems directly based on the ppv levels (Kaiser et al.,
1996). It is evident that several factors can affect the severity of rockbursts, in conjunction with
the magnitude and proximity of seismic event(s), e.g. the size of excavations, mining-induced
stress conditions, capacity of installed ground support systems, etc. Such combined influence has
been demonstrated by Heal (2010) in developing the Excavation Vulnerability Potential (EVP).
In this thesis, the recommended approach toward support design is based on the individual
assessment of the ppv, excavation size, and mining-induced stresses. This approach, as opposed
to merging all relevant parameters into a single index, allows ground control personnel to
independently assess the influence of these parameters on the anticipated performance of the
designed ground support systems. The use of a single index is very attractive but can mask the
effect of several parameters.

In this section, the influence of the magnitude and distance from the seismic source on the
likelihood of rockburst was discussed. A practical tool was proposed to identify the radius of
influence of seismic events, which can be used to target mining areas where dynamic-loading
conditions can be anticipated, at a level of confidence varying between 75 and 90%. This has
 235

practical implications in determining the level of support required in the vicinity of identified
seismic sources. In section 8.3, the influence of mining-induced stresses on the likelihood of
rockbursts is discussed. The implications of the size of excavation and recorded support
performance on the selection of ground support systems are examined further in this chapter.

8.3 Support strategy based on mining-induced stresses

Attributing seismic activity to well-defined geological structures is a straightforward scenario for
the development of a ground support strategy. In reality, over seismically-active structures,
mining-induced seismicity is greatly influenced by mining-induced stresses. Hence, as opposed
to being concentrated near specific geological structures, seismic events may spread-out within a
geological unit, reflecting the rock mass transition towards its post-peak behaviour. Dissimulated
behind the brittle rock failure process, the direct influence of geological structures on mining-
induced seismicity may be difficult to recognize. In this context, the application of a guideline
singularly based on the ppv may not be practical for the development of a ground support
strategy.

In reviewing the mining-induced stress data extracted from 3D elastic numerical modelling, a
trend amongst rockbursts was defined by the differential stress normalized to the UCS (NDS)
(Figure 8–5). The NDS percentiles represented by straight lines in Figure 8–5 indicate the
percentage of rockbursts in the collected data that were associated with NDS values smaller than
the specified percentile threshold. For example, 75% of the rockbursts in the collected data
occurred in mine locations where the mining-induced stresses were quantified by a differential
stress smaller than 0.33 times the UCS of the intact rock. This means that in only 25% of the
rockburst locations the NDS was greater or equal to 0.33.

The 10th – 90th percentile interval is defined by NDS values of 0.13 and 0.46. This NDS range
represents the stress conditions typically associated with rockbursts at Creighton, Copper Cliff,
and Coleman mines. It was obtained by considering all rockbursts occurrences over the studied
time periods at those three mines. The 10th – 90th percentile NDS range extends towards
relatively low stress values, as compared to the somewhat expected rockburst differential stress
thresholds of 0.3 to over 0.5 times the UCS. What the collected data indicate is that it would be
mistaken, using an elastic model, to anticipate rockbursts only in high stress areas of a mine. In
fact, by reviewing the stress distributions on Creighton’s 7400, 7530, and 7680 levels, and
 236

comparing the results to the location of mine seismicity and rockbursts, it was observed that a
level of risk posed by seismic events must be considered as soon as the stress conditions
becomes altered by mining.

Figure 8–5 Normalized differential stress and severity of damage for the rockbursts that occurred at Creighton,
Copper Cliff, and Coleman mines

The stresses represented in Figures 8–6 and 8–7 are the result of the mine geometry at the end of
September 2013. On the 7400, 7530, and 7680 levels at Creighton Mine, the majority of seismic
events (Figure 8–6) and rockbursts (Figure 8–7) were contained within the 0.25 NDS green
contour line. This value is very close to the median and average NDS obtained from the
rockburst data. To some extent, and more specifically in areas where the ground conditions were
affected by a large-scale geological structure, particularly the 74 Grizzly and FAR shear zones,
seismicity and rockbursts have extended up to the 0.20 NDS cyan contour line. This value
corresponds approximately to the 25th NDS percentile obtained from the rockburst data.
 237

Figure 8–6 Mining-induced stresses, as of September 2013, and seismicity on (a) the 7400, (b) 7530, and (c) 7680
levels of Creighton Mine
 238

Figure 8–7 Mining-induced stresses, as of September 2013, and rockbursts on (a) the 7400, (b) 7530, and (c)
7680 levels of Creighton Mine
 239

The information presented in Figures 8–5 to 8–7 has important implications on the development
of a ground support strategy designed for mitigating the consequences of dynamic-loading
conditions. First of all, such strategy cannot be solely based on the assumption that burst-prone
areas in deep underground mines are only limited to areas of high stress levels. Rockbursts have
been proven, through the investigation of time-continuous data, to occur under ranges of stress
conditions varying from low to high. This is because rockbursts are complex phenomena and
their occurrence does not depend only on the principles of brittle rock failure triggered by the
surrounding stress environment.

In practice, the installation of enhanced ground support must be completed in advance of the
development of high stress conditions. Based on the stress results displayed in Figure 8–5, it is
recommended to proceed to the enhancement of ground support systems at the moment the NDS
approaches 0.20, i.e. approximately the 25th NDS percentile. The targeted mine locations must be
those where the stress levels are anticipated to exceed differential stress levels corresponding to
0.25 times the UCS. In areas where large-scale geological structures play a significant role in the
occurrence of mine seismicity and are located within 15 m from mine excavations, the
recommended NDS threshold is revised to 0.20. These NDS thresholds are supported by the
observations of mine seismicity and rockbursts on three deep and mature mine levels at
Creighton Mine (Figure 8–6, 8–7), and by the observed relationship between the proximity of
large-scale geological structures and the number of rockburst occurrences (Figure 5–10).

Figure 8–5 has contributed to the definition of a threshold for the installation of enhanced ground
support systems in mine locations where dynamic-loading conditions are anticipated. Based on
the numerical modelling results, there is no direct relationship between stress conditions and
rockburst severity in the collected data. Severe R4 and R5 level rockbursts have occurred over a
wide range of stress conditions (Figure 8–5). Based on the support performance reviewed in
Chapter 7, the employed surface support and reinforcement elements have a direct impact on
rockburst consequences.

Section 8.1 presented an approach, based on the anticipated event magnitudes, to determine
whether seismicity poses a significant risk to the stability of mine excavations and if a ground
support strategy should account for anticipated dynamic-loading conditions. In section 8.2, the
ppv, estimated using the magnitude and distance from the epicentre of seismic events and
 240

derived from empirical scaling laws, allowed the definition of the radius of influence of seismic
events. The radius of influence was proposed as a tool to plan the installation of enhanced
ground support systems in the surroundings of identified seismic sources. In the present section
8.3, an approach supported by rockburst field data led to the identification of the location and
timing of enhanced support installations. The three approaches presented in sections 8.1, 8.2, and
8.3 are part of a comprehensive ground support strategy aiming at managing the impact of
rockbursts in deep underground mines. It is possible to define the timing and location of
enhanced support installation based on the anticipation of dynamic-loading conditions. In order
to manage rockbursts, configurations of reinforcement and surface support elements must be
recommended and the threshold of success of the recommended support systems must be
assessed. This is discussed further in section 8.4.

8.4 Ground support recommendations for managing dynamic-

loading conditions
Section 8.1 has defined the magnitude threshold under which a mine should become concerned
about the occurrence of rockbursts. With anticipated magnitudes exceeding 1.6 to 2.0 MN, deep
underground mines should transition towards the implementation of ground support strategies
designed for mitigating dynamic-loading conditions. Sections 8.2 and 8.3 have developed on
establishing the locations and the timing under which such strategy should be implemented
underground, based on the location of seismically-active structures and mining-induced stresses.
The present section provides recommendations on the type of ground support systems that are
capable of managing the anticipated dynamic-loading conditions. This section therefore
completes the development of a ground support strategy aiming at mitigating rockbursts.

In Chapter 7, categories of ground support systems were defined and their performances were
assessed. Using the Rate of Tolerance (RoT), an overview of the performance of each category
of support installed in the back and in the walls was provided. It was apparent that in order to
manage the dynamic-load demand imposed even by lower magnitude strain burst events,
yielding ground support had to be introduced. Alternatively, it was possible to increase the
stiffness of the installed ground support systems, e.g. through the application of shotcrete. Up to
magnitude 1.5 MN, support systems comprising shotcrete have demonstrated to be capable of
mitigating strain bursts. Under the dynamic-load demand imposed by larger magnitude fault-slip
 241

events, it was extracted that under similar circumstances, ground support systems that are
exclusively stiff behave inconsistently. Ground support systems comprising mesh-reinforced
shotcrete and yielding reinforcement elements tended to perform adequately. The installation of
cable bolts as part of a second pass over mesh-reinforced shotcrete resulted in an appreciable
level of successful performances as well. With welded-wire mesh used as the only surface
support element, the installation of cable bolts proved to be essential to the achievement of
excellent RoT levels under fault slip events.

As a result of a multivariate statistical analysis performed on the collected rockburst data, all
events were grouped within specific categories, which were functions of the mining-induced
stresses and the size of mine openings. The performance of each ground support system or
categories of ground support systems was analyzed in consideration of the seismicity, which was
quantified using the magnitude, the distance between the damage and the epicentre of seismic
events, and the ppv extracted from scaling laws (Hedley, 1992). In this section, the performance
thresholds of the ground support systems tested under dynamic-loading conditions in the backs
and in the walls are presented. These performance thresholds, quantified using the ppv, are used
as part of the proposed ground support strategy in order to define appropriate ground support
systems for anticipated seismic and stress conditions and excavation size.

8.4.1 Supporting the back of mine excavations under dynamic-loading

conditions
In the Sudbury Basin, the major principal stress is sub-horizontal; hence, high-stress conditions
are typically encountered in the shoulders and back of mine openings. At Vale mines, yielding
reinforcement elements have first been implemented in the back of excavations and further
introduced in the walls as a second pass reinforcement system in conjunction with heavy-gauge
straps. In the evolution of ground support systems at Creighton Mine presented in Chapter 3, it is
in the back that the employed support systems have undergone the most significant adjustments.
Ground support installed in the back, furthermore, must account for the potential for gravity
driven failure, especially under large excavation spans. Cables are typically installed in the back
of intersections and other large mine openings in order to enhance stability under static
conditions. In seismically-active mines, rockfalls can be triggered by seismic shaking. Based on
the collected rockburst data, it was observed that cable bolts can contribute to the capacity of
ground support systems to manage dynamic-loading conditions.
 242

It is not possible to adequately quantify the dynamic-load demand on ground support systems, in
terms of energy per surface unit for example. However, it is clear that the span of mine openings,
the stress conditions, and the level of seismic loading ultimately play a role in the performance of
ground support systems. Using the collected rockburst data, support-performance thresholds
specific to categories of ground support systems were quantified in terms of conceptual ppv
(Figure 8–8). For the back, those support-performance thresholds were defined in consideration
of the span of mine openings and stress conditions. The performance thresholds of ground
support systems are presented further in a table format, providing additional information on the
employed reinforcement and surface support elements and conceptual load-bearing and energy-
absorption capacities (Table 8–1).

Figure 8–8 and Table 8–1 indicate the support systems that, based on the collected rockburst data
and available information on the capacity of individual reinforcement elements, can be
recommended for managing dynamic-loading conditions. Mechanical bolts, although they were
installed as part of ground support systems in the collected rockburst data, were excluded from
this guideline. The mechanical bolt has very limited energy-absorption capacity and its
performance is highly sensitive to environmental factors, e.g. corrosion, exposure to ground
vibrations, hardness of the rock, etc (Hadjigeorgiou & Charette, 2009). In this guideline, the stiff
reinforcement elements recommended as part of a first pass consist of resin-grouted rebars.

The guidelines indicate that cable bolts are mandatory in the back of large mine openings
(span > 7.0 m). The guidelines further emphasize the benefits of applying shotcrete between a
first pass of resin rebars and a second pass of cable bolts. Under low to moderate stress
environments, the application of shotcrete improved the ppv performance threshold from 150 to
725 mm/s. 725 mm/s was the largest threshold value attributed to the successful performance of
ground support systems as it consisted of the 90th percentile of the ppv estimated in the collected
rockburst data. The improvements to the performance of the ground support systems, attributed
to shotcrete, however, are limited under high stress conditions. The ppv-performance threshold
of support systems B7s was reduced from 725 to 350 under major principal stress levels
corresponding to 0.4 to 0.5 times the rock UCS. The addition of yielding reinforcement elements
in the second pass, as part of support system B9s, resulted in improved support performance
under high stress.
 243

Figure 8–8 PPV-performance thresholds of ground support systems in the back of mine excavations (support systems are described in Table 8–1)
 244

Table 8–1 PPV-performance threshold of detailed ground support systems installed in the back of excavations at
Vale mines

Span σ1/UCS Ground support systems Conceptual PPV-

reinforcement performance
capacity thresholds
Reinforcement Surface support Load Energy
(m) (kN/m ) (kJ/m2)
2
(mm/s)
Welded-wire 300 19 150
1st pass rebars mesh (B7)
0.2 – 0.4
2nd pass cables Mesh-reinforced 300 19 725
shotcrete (B7s)
> 7.0 1st pass rebars Mesh-reinforced 300 19 350
nd
2 pass cables shotcrete (B7s)
0.4 – 0.5 1st pass rebars Mesh-reinforced 380 35 725
2nd pass cables shotcrete (B9s)
and MCB
Welded-wire 180 15 70
mesh (B5)
0.3 – 0.4 Rebars
Mesh-reinforced 180 15 170
shotcrete (B5s)
Welded-wire 180 15 40
mesh (B5)
0.4 – 0.6 Rebars
5.0 – 7.0 Mesh-reinforced 180 15 135
shotcrete (B5s)
Welded-wire 270 31-36 420
1st pass rebars
mesh (B8/B10)
2nd pass MCB
0.4 – 0.6 Mesh-reinforced 270 31-36 725
or one-pass
shotcrete
rebars/MCB
(B8s/B10s)
< 5.0 0.4 – 0.6 Rebars Welded-wire 180 15 65
mesh (B5)

In typical (5.0 – 7.0 m) or narrow-span (< 5.0 m) excavations, a first recommended configuration
of reinforcement elements consists of a 1.2 m x 0.8 m (4’ x 2.5’) pattern of resin-grouted rebars,
 245

which can be installed in conjunction with welded-wire mesh or mesh-reinforced shotcrete. The
performance threshold of these support systems decrease as the stress conditions or the
excavation span increase. When higher conceptual ppv levels are anticipated, the installation of
yielding reinforcement elements becomes necessary. In typical excavations under high stress
conditions, the installation of yielding reinforcement elements as part of a second pass or within
a one-pass system resulted in increased performance thresholds. The performance threshold of
support systems B8 and B10 was 420 mm/s as compared to 40 mm/s for support system B5.
Similarly, support systems B8s and B10s were characterized by a 725 mm/s performance
threshold as compared to a 135 mm/s threshold for support system B5s.

In the charts of Figure 8–8, the grey-shaded areas represent uncharted territory, i.e. under the
specified stress conditions and range of excavation span, no support systems have demonstrated
both successful and consistent performances. In applying the empirical guideline presented in
Figure 8–8, if the anticipated dynamic-loading conditions lie in the grey-shaded area, it is
recommended to refer to the charts located either above i.e. larger span, or to the right, i.e. higher
stress conditions. Moving up or to the right on Figure 8–8 results in selecting the support based
on the recorded performance under span and stress conditions that were, most likely, more
difficult to manage. Consequently, the support selection made in consideration of conditions that
are worst than anticipated should result in an appropriate design.

8.4.2 Supporting the walls of mine excavations under dynamic-loading

conditions
The ppv-performance thresholds of ground support systems installed in the walls of mine
openings are presented (Figure 8–9). The reinforcement and surface support configurations
assessed in Figure 8–9 are explained in more detail in Table 8–2.

Mechanical bolts can be used in the walls under low to moderate stress conditions; however,
their threshold of performance is relatively low. The performance improves when mechanical
bolts are substituted by 35 or 39 m friction sets in the support design and improves even further
when 46 mm friction sets are employed. Under low to moderate stress conditions, shotcrete did
not appear to raise the ppv-performance threshold of ground support systems that comprised only
one pass of reinforcement. Consequently, supporting the walls using shotcrete is most likely
unnecessary under such conditions.
 246

The highest ppv-performance threshold was obtained by support systems W5s and W9s. Based
on the conceptual capacity, support systems W5s comprised either resin grouted rebars installed
in conjunction with mesh-reinforced shotcrete or a FS-39 pattern installed in conjunction with
mesh over a membrane of fibre-reinforced shotcrete and enhanced further by a second pass of
Swellex MN12 in conjunction with heavy-gauge straps. Although the conceptual load-bearing
and energy-absorption capacities were similar, the two systems performed differently under
dynamic loads, with the latter resulting in a 725 mm/s performance threshold. This level of
performance was matched by support system W9s, which consisted of a first pass of FS-46,
mesh-reinforced shotcrete, and a second pass of straps and MCB.

Figure 8–9 PPV-performance thresholds of ground support systems in the walls of mine excavations for lower
stress (left: σ1/UCS = 0.2-0.4) and higher stress (right: σ1/UCS = 0.4-0.6) conditions (support systems
are described in Table 8–2)
 247

Table 8–2 PPV-performance threshold of detailed ground support systems installed in the walls of excavations at
Vale mines

σ1/UCS Ground support systems Conceptual PPV-

reinforcement performance
capacity thresholds
Reinforcement Surface support Load Energy
(kN/m ) (kJ/m2)
2
(mm/s)
Welded-wire mesh 110-124 2-5 90
Mechanical (W1)
bolts Mesh-reinforced 110-124 2-5 90
shotcrete (W1s)
Welded-wire mesh 96-110 5-8 170
(W2)
FS-35/39
Mesh-reinforced 96-110 5-8 170
shotcrete (W2s)
0.2 – 0.4
Welded-wire mesh 156 12 500
(W4)
FS-46
Mesh-reinforced 156 12 500
shotcrete (W4s)
FS-39 and Mesh over fibercrete 168 16 725
Swellex MN12 (W5s)
1st pass FS-46 Mesh-reinforced 230-240 26-28 725
2nd pass MCB shotcrete (W9s)
Welded-wire mesh 96-110 5-8 50
(W2)
FS-35/39
Mesh-reinforced 96-110 5-8 100
shotcrete (W2s)
Welded-wire mesh 156 12 90
(W4)
0.4 – 0.6 FS-46
Mesh-reinforced 156 12 300
shotcrete (W4s)
Welded-wire mesh 228-270 16-28 400
1st pass FS-46
(W8/W9)
2nd pass MCB
Mesh-reinforced 228-270 16-28 400
or cables
shotcrete (W8s/W9s)
 248

Shotcrete had a more significant influence on the performance of ground support systems under
high stress conditions. In situations where the rock is significantly fractured, shotcrete appears to
control dilation, which results in a better confinement around the bolts. The ppv-performance
threshold of ground support systems in the wall was considerably reduced due to the higher
stress environment. Nevertheless, there was still a clear advantage in using larger diameter
friction set bolts in the walls. The installation of cable bolts or MCB as part of a second pass
contributed to increasing the ppv-performance threshold further.

8.4.3 Example applications of the empirical support-performance

threshold charts
The ppv-performance threshold of various support systems under ranges of stress conditions and
excavation dimensions were extracted from the collected rockburst data and presented in Figures
8–8 and 8–9. These empirical guidelines can be used to select appropriate ground support
systems for mitigating rockbursts. Example applications are presented in this section.

8.4.3.1 Tunnel intersection under moderate stress conditions

The first example consists of selecting a ground support system to support the back of the
intersection of two 5.5-m span tunnels (Figure 8–10). The span of the intersection is 7.8 m. The
tunnel is located in competent granite (UCS = 220 MPa), at a depth of 1,600 m. The major
principal stress at this location is estimated at 70 MPa. The ground support design should
accommodate a 1.5 MN event within the 30 m vicinity. From the seismic data, it is estimated that
a magnitude 3.0 MN could occur within 120 m.

Figure 8–10 Example 1: Cross-intersection characterized by a span of 7.8 m

 249

The major principal stress to UCS ratio for this location is 0.32. The intersection is therefore
under moderate stress conditions. The section of the guideline (Figure 8–8) employed to solve
this problem is found under span > 7.0 m and 0.2 ≤ σ1/UCS < 0.4 (Figure 8–11). Based on the
application of the proposed support-selection guideline, a support system comprising resin-
grouted rebar and welded-wire mesh enhanced by a second pass of cable bolts would be
adequate to support the cross-intersection under the specified dynamic-loading conditions (Table
8–1).

Figure 8–11 Application of the proposed support-selection guideline to design a ground support system for the back
of the intersection in Example 1

The length of the reinforcement elements with respect to the span of excavations is an important
factor to be considered in designing ground support to manage dynamic-loading conditions in the
back of mine openings. In order to promote adequate performance of the designed support
systems, it is critical that the employed reinforcement elements are long enough to hold the
fractured and loose material. Reducing the span of excavations can contribute to minimize the
potential amount of loose rock to be supported and hence, the demand on the support. It is
therefore recommendable in deep and high stress mines, where rockfalls are susceptible of
occurring as a result of seismic shaking, to avoid designing cross-intersections and instead to
promote the design of smaller T intersections (Figure 8–12).
 250

Figure 8–12 Comparison of cross and T intersection spans of 5.5 m wide tunnels intersecting each other
perpendicularly

8.4.3.2 Typical mine-opening under moderate stress conditions

For the second example, a typical mine tunnel, 6 m wide, is to be supported in consideration of a
design seismic event of a magnitude 3.2 MN, which could occur within a distance of 60 m from
the tunnel. The excavation is under moderate stress conditions (0.3 ≤ σ1/UCS < 0.4). Appropriate
ground support systems for the back and the walls of the tunnel are recommended using the
support selection guidelines presented in Figures 8–8 and 8–9.

For the back of the excavation, the ppv-performance threshold of support systems presented in
Figure 8–8 for typical span and moderate stress conditions fall short of the anticipated dynamic-
load demand (Figure 8–13a). Consequently, the procedure refers to excavations under high stress
conditions in order to select an appropriate support system for the designed dynamic-loading
conditions. Figure 8–13b indicates that the demand imposed by the design 3.2 MN event could be
adequately managed by ground support systems B8 and B10. Hence, for the back of the tunnel,
the recommended support system consists of either a diamond pattern of rebars and MCB or a
first pass of rebars followed by a second pass of MCB and heavy-gauge straps. Reinforcement
elements should be installed in conjunction with welded-wire mesh. The installation pattern
should be designed in consideration of the required conceptual load-bearing and energy-
absorption capacities of 270 kN/m2 and 31–36 kJ/m2, respectively.
 251

Figure 8–13 Application of the proposed support-selection guideline to design a ground support system for the back
of the mine tunnel in Example 2

The same procedure is applied for the selection of a ground support system for the walls of the
tunnel. Under moderate stress conditions, support systems W4 and W4s present a ppv-
performance threshold that is sufficient for managing the design dynamic-loading conditions
(Figure 8–14).

Figure 8–14 Application of the proposed support-selection guideline to design a ground support system for the walls
of the mine tunnel in Example 2

Since the application of shotcrete was not specifically recommended for the back, the proposed
ground support system for the walls consists of 46 mm friction sets installed in conjunction with
 252

welded-wire mesh. The installation pattern should be designed in consideration of the required
conceptual load-bearing and energy-absorption capacities of 156 kN/m2 and 12 kJ/m2.

The second example of application of the proposed support-selection guidelines resulted in the
suggestion of appropriate ground support systems for the back and the walls of a mine tunnel
under moderate stress conditions. Given the limited rockburst data and ground support systems
employed under similar stress conditions, reference to the performance of ground support
systems under high stress conditions was necessary in order to make a recommendable support
selection in the back.

8.4.3.3 Typical mine opening under high stress conditions

The third example consists of recommending a ground support system for the back of a drift,
5.5 m span, located under high stress conditions (0.4 ≤ σ1/UCS < 0.5). Severe seismic conditions
are anticipated and as a requirement, the support system must manage the dynamic-load demand
imposed by a 2.8 MN event that could be located as close as 15 to 20 m from the drift. The
dynamic-loading conditions anticipated as part of this design problem are represented in Figure
8–15.

Figure 8–15 Application of the proposed support-selection guideline to design a ground support system for the back
of the drift in Example 3

The design conditions stated in the problem statement exceed the ppv-performance threshold of
the most competent ground support systems (Figure 8–15). As a result, using the support-
 253

selection tool developed based on the collected and verified rockburst data, it is not possible to
recommend one of the tested ground support systems with a sufficient level of confidence. This
is because the specified design level of ppv, determined based on the magnitude and location of a
design seismic event, was either not represented sufficiently in the collected rockburst data or
that the studied ground support systems have not performed adequately or consistently at such
level.

In front of such design case, unsupported by high-quality field data, it is necessary to assess the
limitations of the proposed ground support systems and to recommend improvements that would
overcome those limitations. Based on laboratory test data on recently developed yielding
reinforcement elements, it is possible to increase the performance threshold of ground support
systems. This is part of the motivations of deep underground mines to explore new support
options. The extrapolation of the support performance data is discussed in section 8.5.

8.5 Extrapolation of support performance data

As mines become deeper, it is possible that the required ppv-performance thresholds exceed
those of the support systems covered in this chapter. Those support systems reflected the ground
support practice at Creighton, Copper Cliff, and Coleman mines from 2000 to 2013. Under
anticipated burst-prone conditions, the mines have typically relied on ground support systems
that were a combination of stiff and yielding elements. MCB were often installed as a second
pass, in conjunction with #0 gauge straps, over a first pass of rebars or mechanical bolts and
rebars. Based on the collected rockburst data, such approach is recommendable up to a ppv level
of 725 mm/s in the back of mine openings under high stress conditions. As the challenges posed
by mine seismicity increase, it will become necessary to design ground support systems that
exceed this performance threshold.

Over the last five years, bolt manufacturers have developed different concepts of yielding bolts
and some of them are gaining exposure in the mining industry. A highly promising
reinforcement element for managing dynamic-loading conditions is the D-Bolt (Normet, 2014).
The D-Bolt is an energy-absorbing rock bolt, which consists of a smooth steel bar with anchors
spaced along its length (Li, 2012). The smooth sections are de-bonded from the grout; this allows
the bolt to stretch under impact, as opposed to other bolt concepts that rely on a ploughing
mechanism to dissipate the kinetic energy. In practice, the stretching mechanism promoted by
 254

the D-Bolt results in a reinforcement element that is capable of absorbing high levels of kinetic
energy while maintaining the resulting displacements under tolerable levels. Under static
conditions, the D-Bolt’s capacity is comparable to the one of the rebar, as both are relatively stiff
and yield only under high loads. The yield load, tensile strength, and maximum impact energy
sustained by the 20 and 22 mm diameter D-Bolts in laboratory tests are presented in Table 8–3.
These properties are compared to those of the rebars, MCB, and cable bolts, which are part of the
support systems recommended in this study (Table 8–1).

Table 8–3 Load-bearing and energy-absorption capacity of the D-Bolt

Yield Tensile Maximum References

load strength impact energy
(kN) (kN) (kJ)
D-Bolt (20 mm) 150 210 45 (Doucet & Voyzelle, 2012; Li &
D-Bolt (22 mm) 190 250 56 Doucet, 2012; Li, 2012; Normet,
2014)
Rebar 130 170 14 (Doucet & Voyzelle, 2012;
Mansour Mining, 2015)
MCB 110 160 30 (Doucet & Voyzelle, 2012;
Mansour Mining, 2015; Player et
al., 2008)
Plain strand cable 235 265 18 (Mansour Mining, 2015; Ortlepp &
Stacey, 1998; Player et al., 2008)

Based on the test results presented in Table 8–3, it is possible to design a ground support system
using a single type of reinforcement elements that is stiff, strong, and capable of absorbing high
levels of kinetic energy. In concept, such system qualifies to manage both static and dynamic-
loading conditions encountered in deep and high stress mines. The conceptual load-bearing and
energy-absorption capacities of two patterns of D-Bolts are calculated using the tensile strength
and the maximum impact energy. Those parameters are used to provide an ultimate capacity,
which can further be compared to the capacity of other configurations of reinforcement elements
presented in Chapter 7 and referred to in this chapter (Tables 8–1 and 8–2). The conceptual
capacities of D-Bolts’ configurations are presented in Table 8–4. The bolting densities reflect a
 255

3-2-3 pattern, i.e. 8 bolts per 1.52 m x 3.35 m (5’ x 11’) mesh sheet, and a 4-3-4 pattern, i.e. 11
bolts per mesh sheet. A 0.30 m (1’) overlap between mesh sheets is considered in the
calculations. The 3-2-3 and 4-3-4 configurations result in 4 and 6 effective bolts per 3.7 m2
(40 ft2) of surface, respectively.

Table 8–4 Conceptual load-bearing and energy-absorption capacities of two configurations of 20 and 22 mm
diameter D-Bolts based on laboratory testing results

Bolting density Load capacity Energy capacity

(kN/m2) (kJ/m2)
3-2-3 226 48
1.08 bolts/m2
0.10 bolt/ft2
D-Bolt (20 mm)
4-3-4 339 73
1.61 bolts/m2
0.15 bolt/ft2
3-2-3 269 60
1.08 bolts/m2
0.10 bolt/ft2
D-Bolt (22 mm)
4-3-4 404 90
1.61 bolts/m2
0.15 bolt/ft2

The conceptual load-bearing and energy-absorption capacities presented in Table 8–4 are
considerably high when compared to the conceptual capacities of other ground support systems
reported in the collected rockburst case studies from Creighton, Copper Cliff, and Coleman
mines (Figure 8–16). Although the performance under dynamic-loading conditions of ground
support systems comprising D-Bolts have not been assessed as part of this study, the conceptual
capacities derived from laboratory testing results demonstrate that reinforcement elements
required to increase both the load-bearing and energy-absorption capacities of ground support
systems are available. Going forward, rockbolts similar in concept to the D-Bolt should be
integrated as part of dynamic ground support strategies.
 256

Figure 8–16 Conceptual load and energy-absorption capacities of D-Bolt configurations as compared to other
reinforcement configurations reported in the rockburst database

Figure 8–16 demonstrates clearly that a major step forward is taken in terms of ground support
capacity by implementing ground support designs using the D-Bolt as opposed to stiff
reinforcement elements or combinations of stiff and yielding reinforcement elements.
Substituting reinforcement elements for the D-Bolt, or other rockbolts similar in concept,
represents an important step, going forward, to improve the dynamic support practice beyond
their current levels. Empirical evidences of success in implementing ground support systems
comprising D-Bolts in deep and high stress mining environments, at depths exceeding 2,680 m,
have been discussed recently (Counter, 2012; Counter, 2014).

Another reinforcement element that might have been overlooked in ground support practices for
dynamic-loading conditions is the cable bolt. The installation of cable bolts is of primary
importance in large span openings such as intersections. In the collected data, ground support
systems comprising cable bolts have shown appreciable ppv-performance thresholds, especially
when installed over mesh-reinforced shotcrete (Figure 8–8). In interpreting the performance of
ground support systems using the RoT, the performance of support systems comprising cable
bolts was comparable to the one of support systems comprising both stiff and yielding
reinforcement elements, especially under dynamic loads generated by fault slip events (Figure 7–
5). During fault slips, support systems that comprised either mesh or mesh-reinforced shotcrete
have achieved comparable RoT levels. Under seismic shaking conditions generated by fault slip
events, it appeared that a critical feature of ground support systems was the length of the
 257

employed reinforcement elements; the reinforcement must be long enough to hold the loose
volume of rock, accelerated due to the incoming seismic wave. As compared to rock bolts, which
are of a typical 1.8–2.4 m length, the holding function, under seismic shaking, is best achieved
by cable bolts.

In practice, ground control personnel do not design ground support systems to mitigate whether
strain bursts or fault slips. In seismically-active mines where large magnitude seismic events are
susceptible to generate damage over large areas, a ground support system designed for managing
dynamic-loading conditions must be capable of containing strain bursts and to prevent large
rockfalls from occurring due to seismic shaking. In order to manage strain bursts, it is possible to
increase the stiffness of the support by applying shotcrete or to enhance the yielding capability of
the support systems using dynamic bolts. The application of shotcrete is costly and logistically
complicated. Furthermore, under higher thresholds of loading, the performance of shotcrete
becomes limited and the support does not perform as consistently. Although support systems
comprising mesh-reinforced shotcrete are capable of mitigating rockburst up to 1.5 MN, a
decrease in the level of performance, attributed to the progressive degradation of the stiff surface
support, is anticipated under higher magnitudes. The design of support systems based on the
installation of yielding reinforcement elements is therefore a more recommendable starting point.

Rockbolts similar in concept to the D-Bolts can contribute to managing a large range of
dynamic-loading conditions. A support system comprising D-Bolts or similar bolts and welded-
wire mesh and further enhanced by the installation of cables has the potential to provide
excellent performance under both strain burst conditions and seismic shaking conditions
triggered by remote large-magnitude seismic events. This interpretation is based on the collected
field observations on the performance of cable bolts and the conceptual assessment of the
capacity of D-Bolts’ support pattern.

The capacity of cable bolts to manage dynamic loads is supported by case studies presented in
Figures 8–17 and 8–18. The consequences of a recorded 2.6 MN event on August 16, 2005 at
Creighton Mine are illustrated in Figure 8–17. The event magnitude was adjusted to 2.0 MN in
this thesis. The damage consisted of about 210 tonnes of fractured granite and metagabbro
displaced from behind the intersection of the Fresh-Air drift on the 7810 level, 42 m away from
the epicentre of the event. The intersection was supported using a diamond pattern of mechanical
 258

bolts and rebars, installed in conjunction with welded-wire mesh, and a second pass of 6.40 m
long (21’) cable bolts. Signs of loading were observed on the cables, however, the 9.9 m span
intersection remained stable under ppv levels approximated to 115 m/s. The support installed
behind the last row of cables however completely collapsed (Figure 8–17). This example
illustrates the importance of installing long reinforcement elements to prevent rockfalls triggered
by seismic shaking.

Figure 8–17 Cable bolts preventing a fall of ground to extend in the Fresh-Air-Drift intersection on the 7810 level at
Creighton Mine, on August 16, 2005

Figure 8–18 shows damage to the wall of the Truck Loop on the 7200 level at Creighton Mine,
which resulted from a recorded 3.0 MN seismic event (adjusted to 2.6 MN) on June 15, 2007. An
estimated 27 tonnes was ejected from the wall and generated minor damage to a drill and a bolter
parked in the immediate vicinity. The epicentre of the seismic event was located approximately
18 m from the damage, generating ppv levels estimated at 997 mm/s at the damage location. As
indicated on the picture, the blown portion of the wall stopped where a plated cable bolt was
installed. This support performance case study suggests that, due to their capacity of anchoring
 259

the loose rock back to stable ground, cable bolts contributed to controlling rock mass dilation,
which as a result maintained confinement around the installed reinforcement elements and
resulted in enhanced support performance.

Figure 8–18 A cable bolt contributing to limiting the blown portion of a wall on Creighton Mine’s 7200 level on June
15, 2007

The two previous case studies demonstrated that cable bolts can prevent damage triggered by
seismic shaking, in the case of remote seismic events, and survive high levels of ppv. In large
openings, cable bolts must be installed in a timely manner, e.g. prior to the development of the
branches of an intersection, in order to meet the static-load demand. As part of a dynamic
support strategy in typical size openings, however, the timing of the installation of cable bolts
can be delayed, as cable bolts are primarily required once a fractured zone has developed around
the excavation. From a conceptual standpoint, an enhanced support strategy to face both strain
burst and fault slip events consists of installing D-Bolts, or rockbolts similar in concept, and to
further monitor the rock-support interaction as the load builds up. Cable bolts being stiffer than
the D-Bolts, they are installed as part of a delayed second pass, once a certain level of
 260

deformation has occurred. This approach is based on the concept of convergence-confinement

(Hoek & Brown, 1980). The delayed installation of cable bolts would prevent high levels of
initial loading to propagate within the cables as the rock mass fractures and dilate. Consequently,
this approach promotes the mobilization of the full capacity of the cable bolts under seismic
shaking conditions, which could occur later as the excavation matures.

There are clear benefits of using shotcrete to manage dynamic-loading conditions, as

demonstrated by the ppv-performance thresholds identified in this chapter. However, the
application of shotcrete in deep mines is costly and logistically complicated. Therefore, it is
justified to limit its application only in areas where it is necessary. In the evolution of ground
support systems proposed in this section, the application of shotcrete should be considered when
a support system comprising D-Bolts, cables, and mesh reaches its limit of success. Under high
thresholds of loading, shotcrete tends to crack and spall and requires frequent rehabilitation. A
pertinent strategy in burst-prone grounds would consist of applying shotcrete prior to installing
the bolts and mesh. Such strategy has resulted in successful performance at Coleman mine
during a 2.7 MN event. Despite the high ppv levels estimated at the damage location
(approximately 700 mm/s), the installed support system responded quite well even if only
relatively weak reinforcement elements were employed, i.e. 39 mm friction sets and Swellex
MN12 (Figure 8–19). The back was undamaged and wall bulking was generally well controlled
by the support. The only material ejection reported occurred at the mesh overlap in the shoulder
and was due to the failure of one split set. A plate and a split set ring were found in the muckpile.

Figure 8–19 Performance of a mesh-over-shotcrete support system during a 2.7 MN seismic event at Coleman Mine
on May 12, 2011
 261

Based on the case study illustrated in Figure 8–19, it is interpreted that the installation of D-
Bolts, or bolts similar in concept, and mesh over a shotcrete membrane would respond
adequately to very high levels of loading. Based on the discussion provided in this section, the
following evolution of ground support systems is proposed for managing dynamic loading
conditions in deep and high stress mines:

1. D-Bolts, or bolts similar in concept, and welded-wire mesh

2. D-Bolts, or bolts similar in concept, and welded-wire mesh with a second pass of cable bolts
3. D-Bolts, or bolts similar in concept, and welded-wire mesh installed over a shotcrete
membrane
4. D-Bolts, or bolts similar in concept, and welded-wire mesh installed over a shotcrete
membrane and a second pass of cable bolts

It is important to monitor the performance of the employed support systems. The transition from
a support system to the next should be initiated when limitations are observed following
rockbursts. Prior to redesigning the support system, it is possible to enhance the performance by
improving the type of mesh and the connections. For example, #4 gauge welded-wire mesh can
be used instead of #6 mesh. The cutting action of bolt plates on the mesh can be prevented by
installing #0 gauge mesh plates, which spread the load over a wider surface. Finally, #0 gauge
straps can enhance the load transfer amongst reinforcement elements and prevent excessive
bulking of the mesh. These additional support elements should be considered as a first step to
overcome observed limitations. The design of ground support systems is a continuous learning
process, as conceptually illustrated in Figure 8–20. Monitoring the support performance in situ is
the starting point to a pro-active ground support strategy.
 262

Design
• Dynamic-load demand
• Support capacity
(best estimate)

Performance / Installation
monitoring
• Instrumentation • Quality control
• Field observations • Quality assurance

Figure 8–20 Learning process toward the development of appropriate support strategies to mitigate rockbursts

8.6 Summary
In this chapter, a ground support design strategy for managing dynamic-loading conditions was
presented. This strategy was validated by the collected and reviewed rockburst data and is
summarized in Figure 8–21.

The Nuttli magnitude scale was employed throughout the ground support design strategy in order
to assess the probability of rockburst, the radius of influence of seismic events, and the ppv-
performance thresholds of ground support systems. This was the most appropriate scale to use in
this thesis given available data for calibration. It is recognized that the application of the Nuttli
magnitude scale is the standard reference for mine operations in Canada, while the Richter
magnitude scale is more accessible in other mining jurisdictions. The proposed guidelines can be
applied in other parts of the world provided a correlation is established between the Richter and
Nuttli scales. The magnitude conversion proposed by Kaiser et al. (1996) (MRichter = MNuttli - 0.5)
may be used in the proposed guidelines for approximating the magnitude using the Richter scale.
Given the empirical nature of the proposed ground support design strategy, irrespective of the
employed magnitude scale, it is strongly recommended that the ground control guidelines be
calibrated using local seismic data.

The ground support design strategy first established a distinction in terms of magnitude threshold
between mines where a static support strategy is appropriate and those where a dynamic support
 263

strategy is necessary. Based on the probability of rockburst, the recommended transition zone
consisted of the magnitude range of 1.6 to 2.0 MN and coincided with the transition from strain-
burst to fault-slip source mechanisms.

Figure 8–21 Summary of the proposed ground support design strategy for mine excavations subjected to dynamic-
loading conditions
 264

Based on the collected rockburst data, the radius of influence of seismic events has been defined.
The radius of influence can be applied in order to determine the areas around identified
seismically-active structures where the support must be enhanced in order to face dynamic-
loading conditions. Seismicity associated with geological structures is one part of the problem.
Stress redistributions around the orebodies mined at Creighton were compared to the seismicity
and rockburst locations. It was recommended that support systems capable of managing
dynamic-loading conditions should be installed in areas where the normalized differential stress
is anticipated to reach or exceed 0.25. In the vicinity of seismically-active geological structures,
more specifically when structures are reported within 15 m from mine excavations, the NDS
threshold was reduced to 0.20. It is recommended that the installation of enhanced support be
implemented prior to the NDS reaching 0.20. Using 3D elastic numerical modelling, it was
therefore possible to determine the adequate location and timing of installation of enhanced
support. The actual threshold values are based on applying the developed methodology to similar
conditions as in the database and using design tools calibrated as in the present study.

The recommended support systems depend on the size of the excavation and the stress
conditions. The support selection, based on the assessment of the performance, is further
influenced by the magnitude and location of seismic events. The reviewed rockburst database
demonstrated that a support design cannot be based simply on the magnitude. PPV-performance
thresholds have been identified for a range of ground support systems. This led to the
development of a practical tool that can assist ground control personnel in the selection of
ground support systems.

The obtained ppv-performance thresholds reflected the ground support practice over time at
Creighton, Copper Cliff, and Coleman mines, between 2000 and 2013, and did not include recent
developments in dynamic and high-strain capacity bolts. The mines in the database have
successfully introduced such dynamic reinforcement successfully since 2013. As mining
progresses to greater depths and higher stress conditions, the required performance threshold of
ground support systems is likely going to increase. Reinforcement elements relying on a
stretching mechanism for dissipating kinetic energy may provide a solution to match higher
levels of dynamic-load demand. This was demonstrated by calculating the conceptual capacity of
D-Bolt patterns as compared to the support systems reported in the rockburst database. Going
forward, this new ground support technology that was not employed at the time in the collected
 265

data should be included as part of dynamic ground support strategies. Cables and shotcrete, in
the collected rockburst data, have improved the performance of ground support systems and
consequently, raised the ppv-performance thresholds. Cables and shotcrete were recommended,
in conjunction with D-Bolts, or bolts similar in concept, as part of the following 4-step ground
support strategy for managing dynamic-loading conditions:

1. D-Bolts, or bolts similar in concept, and welded-wire mesh;

2. D-Bolts, or bolts similar in concept, and welded-wire mesh with a second pass of cable bolts;
3. D-Bolts, or bolts similar in concept, and welded-wire mesh installed over a shotcrete
membrane;
4. D-Bolts, or bolts similar in concept, and welded-wire mesh installed over a shotcrete
membrane and a second pass of cable bolts.

The practice of installing reinforcement elements and mesh over shotcrete is strongly
recommended where the need for a strong and stiff surface support element is identified. The
capacity of shotcrete can be limited under high thresholds of dynamic loading. Therefore, the
installation of mesh over shotcrete can accommodate larger levels of kinetic energy while
minimizing the exposure to loose fragments of shotcrete that can be ejected under dynamic
loads. The relatively minor damage observed as a result of a 2.7 MN event at Coleman Mine
(Figure 8–19) supported this approach.
 266

9 Conclusions

The final chapter of this thesis presents a summary of the work done as part of this research and
provides conclusions and recommendations for future work. Firstly, the motivation and
originality of the research are reiterated. Secondly, the contributions made to the analysis of the
support performance under dynamic-loading conditions and the design of ground support
systems for burst-prone mining environments are summarized. The most important conclusions
achieved as a result of this research are listed. Finally, the limitations of this thesis are outlined
and recommendations for future work on the performance of ground support systems under
dynamic-loading conditions are provided.

9.1 Motivations and originality of the research

Underground mines are operating at greater depths and under increasingly high stress conditions
in order to access high-grade mineral deposits. Typically, in deep and high stress environments,
seismicity is a direct consequence of mining. Managing the seismic risk in a heavy production
environment is a significant and very important technical challenge. Large-magnitude seismic
events are capable of causing severe damage to excavations and ground support systems, and as
a result, can compromise the workforce safety and mine profitability. Damage to excavations
and/or ground support, triggered by mining-induced seismicity, is referred to as a “rockburst” by
mine practitioners. In order to better manage seismicity and mitigate the associated risks, ground
control engineers have, during the past two to three decades, beneficiated from advancements in
the fields of geosciences and ground support. The main developments that contributed to
improving the mining practice at great depths consist primarily of 1) progress in seismic
monitoring, 2) numerical stress modelling, 3) yielding reinforcement elements, 4) re-entry
protocols, and 5) quality control programs.

Seismic monitoring systems are state-of-the-art tools used by ground control engineers to
monitor the rock mass response to mining and develop pro-active measures to avoid exposing the
workforce to adverse ground conditions. In recent years, there have been important advances in
the development of seismic data analysis techniques to assess and map the evolution of the
 267

seismic hazard within a mine (Hudyma, 2008; Kaiser, Vasak, Suorineni, & Thibodeau, 2005).
These techniques have contributed in maximizing the amount of information that can be
extracted from seismic monitoring systems. A seismic monitoring system that provides good
three-dimensional coverage of a mine infrastructure is an indispensable tool for mining in deep
and high stress conditions.

Since the mid-1980’s, the use of numerical stress modelling has gained significant importance in
mining engineering. Several deep underground mines currently run their own three-dimensional
elastic models of the mining sequence and use it to trace the evolution of mining-induced
stresses within the mine. Mining and geotechnical engineering consultants also provide a wide
range of numerical packages that can model complex geological features and material
behaviours. Over time, the use of numerical models, regardless of the complexity level, has
become very important in mine planning.

In the early 1990s, it was acknowledged that the containment of severe rockburst damage could
not be achieved by increasing the strength of ground support systems (Ortlepp, 1992). In order to
manage dynamic-loading conditions, ground support systems had to be designed in consideration
of the level of released kinetic energy. Reinforcement elements capable of absorbing the energy
released by a rockburst through yielding became commercially available in Canadian mines in
the late 1990s with the introduction of the modified conebolt (Simser et al., 2007). Currently,
ground control engineers have access to a relatively wide selection of yielding reinforcement
elements to mitigate the severity of rockbursts. In addition, the selection of reinforcement
elements can be supported by technical information on the performance under static and dynamic
loads. The majority of the available information on the performance of reinforcement elements
has been provided over time by testing-rig facilities from Canada, Australia, and South Africa.
Furthermore, quality control programs have been implemented at mine sites to complement and
cross-validate the information on ground support capacity available in the technical literature.
The collected information can comprise in-situ pull test data, observations on the severity of
corrosion, qualitative observations of support performance during rockbursts, etc.

Rockbursts have become an important challenge over the last 20 to 30 years and proactive
measures have been adopted over time in deep underground mines to face the problem. There
have been significant developments in the fields of seismic monitoring and numerical modelling
 268

and most importantly, there has been a progression in the application of those technologies at the
mine site level. Mines have used information extracted from seismic monitoring and numerical
stress modelling to design and validate mining sequences and develop re-entry protocols to
prevent exposing the workforce to adverse seismic conditions. Over time, an increasing level of
attention has been paid to the development of new rock reinforcement technologies. Currently,
yielding reinforcement elements are available from several bolt manufacturers. The introduction
of new support elements into a ground control standard is now supported by available technical
data on the support performance under controlled static and dynamic conditions.

Despite all the technological improvements to manage mine seismicity and rockbursts, the
methodologies employed for the design of ground support systems have progressed at a much
slower rate. The approach toward ground support design in burst-prone mines is influenced to a
degree by a lack of full understanding of the mechanisms of action and interaction of support
elements under dynamic loads. Despite the fact that there has been important knowledge
developed on the rock mass response to mining-induced stresses, the performance of ground
support systems remains poorly understood and cannot be fully predicted or even quantified. In
the absence of robust support guidelines, ground control engineers rely mostly on empirical
experience in order to select appropriate combinations of reinforcement and surface support
elements. These decisions are often based on limited data.

The primary objective of this research was to develop a ground support design strategy,
supported by high-quality field data, for mine openings subjected to rockbursts. Passive
monitoring of ground support systems under dynamic loading conditions was conducted through
the investigation of rockburst case studies. Rockburst data were collected at three seismically-
active mines in the Sudbury area, Canada, over a continuous time period. Consequently, the
developed database tracked the evolution of the support practice and rockburst frequency and
severity at the three mines. The collected rockburst data were used to assess the performance of
the installed ground support systems under a range of seismic, ground, and stress conditions. In
order to fulfill the primary objective, the following tasks were executed:

• A data collection campaign was undertaken at the three investigated mine sites:
Creighton, Copper Cliff, and Coleman mines. The data covered a 14-year, 10-year, and 8-
year period at each site, respectively.
 269

• The collected data were organized in a retrievable database. As the rockburst data were
extracted from internal and external company reports, the information was reviewed
alongside the mine ground control personnel.

• The collected data were validated through cross-referencing, personnel interviews,

statistical analyses, and numerical stress modelling.

• A multivariate statistical analysis was performed, which led to the classification of

rockburst case studies based on similitudes in the parameters that affected the dynamic-
load demand.

• The ground support systems were classified based on similarities in the conceptual
capacity of the configurations of reinforcement elements.

• The performance of ground support systems during rockbursts was assessed consistently
using the Support Damage Scale (SDS), as proposed by Kaiser et al. (1992) and Mikula
(2012).

• In the support performance analysis, the dynamic-load demand was accounted for by the
size of mine excavations, stress conditions, event magnitude, and distance between the
damage and the epicentre of seismic events. For defined ranges of ground and stress
conditions, the ppv (Hedley, 1992) was used to define performance thresholds of various
categories of ground support systems.
The employed methodology led to the definition of practical guidelines that can assist ground
control personnel in defining the thresholds that initiate a transition from a static to a dynamic
ground support strategy. For various severity of dynamic-loading conditions, the guidelines
provide support recommendations based on the ppv-performance threshold of ground support
systems designed for supporting excavation’s back and walls. These guidelines are based
exclusively on high quality and verified rockburst data collected in deep and high stress mines of
the Sudbury Basin.

9.2 Accomplishments and contributions

The accomplishments and contributions made as part of this thesis are listed in this section.

• A comprehensive database was constructed of high-quality and verified rockburst field

data. Compared to other work (Albrecht, 2005; Heal, 2010; Liang, 2012) the rockburst
 270

database presented in this thesis is the largest and most complete to this date. It is the
only database that provides time-continuity over periods varying from 8 to 14 years per
mine site. The rockburst data were reviewed with the assistance of the ground control
personnel at Creighton, Copper Cliff, and Coleman mines. This level of interaction with
the mine personnel contributed to increasing the level of confidence in the collected data.
The rockburst data were further cross-validated using other sources of information, e.g.
the local seismic monitoring and CAD systems employed at the mine sites, the public
Earthquake Database (National Resources Canada, 2015), external engineering reports,
site inspections, etc. The collection of rockburst data that covered 32 years of continuous
mining operation between three sites (14 years at Creighton, 10 at Copper Cliff, and 8 at
Coleman) was an important contribution to mining rock engineering. The developed
database comprised 209 seismic events and the associated damage to 324 mine locations.

• Limitations in the template used by ground control personnel in Ontario mines for
reporting falls of ground and rockbursts have been identified through the data collection
and revision process. An improved version of the Unusual Occurrence Report for
Groundfalls/Rockbursts has been submitted to Workplace Safety North (WSN), a Health
and Safety Ontario (HSO) partner. The proposed version of the template was adopted by
WSN and is currently used to report on rockburst occurrences and the performance of
ground support systems under dynamic-loading conditions in the underground mines of
Ontario. Before the official adoption of the improved template, ground control personnel
at Vale Sudbury mines had already started incorporating the proposed modification into
their internal reports.

• The magnitude of seismic events is a critical parameter in assessing the performance of

ground support systems under dynamic-loading conditions. As part of this research, a
high level of confidence in the magnitude data was developed. This was possible given
that the study had focused on three mine sites located in the same area. Seismic data
obtained from the seismic monitoring systems at Creighton, Copper Cliff and Coleman
mines have been reviewed and compared to the magnitudes provided by the Geological
Survey of Canada (National Resources Canada, 2015). Using the available seismic data,
magnitudes from the local seismic monitoring systems were cross-validated and further
adjusted in order to ensure the comparability of the magnitude data between the three
 271

sites. The validity of the performed adjustments was verified using statistical tests of
hypotheses. The magnitude-time history analysis and frequency-magnitude relationship
at Creighton Mine were used to further validate the time continuity in the quality of the
adjusted magnitudes as the mine upgraded its seismic monitoring system. The
comparability of the collected event magnitudes from a site to another was not
investigated as part of other studies on the support performance under dynamic-loading
conditions (Albrecht, 2005; Heal, 2010; Liang, 2012). This accomplishment was unique
to this thesis.

• The level of focus of this thesis allowed for very detailed mining-induced stress analyses.
The stress conditions have been assessed for the 324 damage locations using three-
dimensional elastic numerical models provided by the mines. The numerical models were
updated using cavity-monitoring surveys (CMS) and geological mapping of large-scale
structures. The employed models represented the stopes of the 400, 461, 1290, and 402
orebodies at Creighton, 100 and 900 orebodies at Copper Cliff, Main and 153 orebodies
at Coleman. Geological structures such has the numerous shear zones mapped within
Creighton Deep, the Quartz (Trap) and Olivine Diabase Dykes at Copper Cliff, and the
Lunchroom Fault at Coleman have also been included in the models. The mining steps
employed in each numerical model represented stopes (or cuts) that had been mined
between each rockburst occurrence. Such approach testifies of the commitment to
provide accurate estimates of the mining-induced stress field at the location and time of
occurrence of rockbursts. Such level of detail in representing the stress conditions
associated with rockbursts has not been achieved in the previous large-scale back-
analyses of rockbursts (Albrecht, 2005; Heal, 2010; Liang, 2012). In those analyses,
when stresses were considered, they consisted predominantly of an estimate of the far-
field stress conditions, therefore omitting the influence of mine geometries on stress
redistributions. In this thesis, a degree of correlation was observed between the
normalized differential stress (NDS) and the occurrence of rockbursts. The NDS 10th and
90th percentiles consisted of 0.13 and 0.46, indicating that using an elastic modelling
approach, areas that are undergoing stress changes are more susceptible of facing
rockbursts rather than areas of very high stress.
 272

• All the collected rockburst variables that had a possible effect on the dynamic-load
demand on support systems were used to classify rockbursts using the principal
component analysis (PCA) technique. A three-component model was fitted to the
collected data. The model was dominated by the influence of variables related to the
seismic events (magnitude, distance between epicentre and damage, inferred seismic
source mechanism), the depth of mine openings and stress conditions, and the span of
excavation and length of installed reinforcement elements. Those variables were
employed to group the rockbursts based on expected similitudes in the dynamic-load
demand. This classification was further used to assess the performance of ground support
systems under comparable conditions.

• The performance of ground support systems over 324 rockburst case studies was
consistently assessed using the Support Damage Scale (SDS) (Kaiser et al., 1992;
Mikula, 2012). A grouping of ground support systems was proposed based on similitudes
in the conceptual capacity of reinforcement configurations and the employed surface
support elements. The overall performance of the identified categories of ground support
systems was quantified using the Rate of Tolerance (RoT), a performance indicator
developed in this thesis based on the interpreted SDS. A more specific characterisation of
the support performance was achieved by considering each category of rockburst
individually. This allowed the characterisation of the support performance in
consideration of the stress conditions, span of mine excavations, magnitude, and distance
from the epicentre of seismic events. The performance threshold of each category of
support systems was defined using the peak particle velocity (ppv) obtained from a
scaling law applicable to Ontario mines (Hedley, 1992). The ppv-performance thresholds
of ground support systems can be used as an indicator of the performance under given
dynamic-loading conditions and can assist in the selection of ground support systems.

• A ground support strategy for deep and high stress mines was proposed. There were four
steps to this strategy.
o Determining whether the level of seismic risk necessitates transitioning from a
static to a dynamic ground support strategy. The seismic risk threshold is
determined by the event magnitude range of 1.6 to 2.0 MN. The largest anticipated
 273

magnitude can be obtained as the x-axis intercept of the frequency-magnitude

relationship.
o Determining the radius of influence or critical distance from identified
seismically-active structures. The distance over which dynamic-loading
conditions can be anticipated is influenced by the magnitude of the seismic event.
In analysing seismic events associated with specific structures, a design
magnitude can be selected. Figure 9–1 is a practical tool that can assist ground
control engineers in determining the distance around seismically-active structures
over which the ground support should be enhanced in order to face dynamic-
loading conditions.

Figure 9–1 Radius of influence of seismic events determined based on the collected and verified rockburst data

o Determining the location and timing of implementation of a dynamic ground

support strategy. Using a 3D elastic numerical modelling approach, verified using
seismic and rockburst data, the implementation of a dynamic ground support
strategy was recommended in areas where the normalized differential stress
(NDS) is expected to reach or exceed 0.25. In areas where large-scale seismically-
active structures are reported within 15 m from mine excavations, the NDS
threshold is lowered to 0.20. The installation of enhanced ground support is
recommended prior to the NDS reaching 0.20.
 274

o Determining the appropriate ground support system to face the anticipated

dynamic-loading conditions. The ground support systems are selected by
comparing the ppv-performance thresholds to the anticipated dynamic-loading
conditions for the specified range of excavation span and stress conditions. This
approach, supported by the collected rockburst field data, determines the
performance limit of stiff support systems and the ppv range over which yielding
support must be implemented.

• Based on trends observed in the performance of ground support systems and the achieved
performance under controlled conditions of the recently developed D-Bolt, a ground
support strategy for burst-prone ground conditions was proposed. The D-Bolt was
selected as an example of dynamic reinforcement that has been recently introduced with
success in deep and high stress mining environments (Counter, 2012; Counter, 2014).
The strategy, which should be incrementally implemented when the requirement for
yielding support is identified, is presented in a four-step format.
1) D-Bolts, or bolts similar in concept, and welded-wire mesh
2) D-Bolts, or bolts similar in concept, and welded-wire mesh with a second pass of
cable bolts
3) D-Bolts, or bolts similar in concept, and welded-wire mesh installed over a
shotcrete membrane
4) D-Bolts, or bolts similar in concept, and welded-wire mesh installed over a
shotcrete membrane and a second pass of cable bolts
It is possible that comparable performance may be achievable using support systems that
employ yielding bolts that are similar in concept to the D-Bolt. This type of bolt,
characterized by a stretching mechanism, as opposed to plowing or sliding, can achieve
both higher loads and energy absorption capacities, based on dynamic tests performed in
drop test facilities. It is foreseeable that other bolts of similar behaviour may be
developed in the near future.
The transition from a step to another in the recommended ground support strategy can be
initiated when a system achieves its performance limit. This can only be recorded by
committing to monitor the performance of ground support systems. The design of ground
 275

support systems is a continuous learning process and monitoring the support performance
in situ is essential to a pro-active ground support strategy.

9.3 Limitations and future work

This thesis has contributed to an improved understanding of the performance of ground support
systems in seismically active mines. It is still not possible at the present time to fully quantify the
dynamic-load demand and the in-situ capacity of ground support systems to mitigate dynamic
loads. Through exploring the collected data, it was realized that the paradigm had to shift from a
conventional factor-of-safety design approach to an approach based on the performance and
passive monitoring. Fully quantifying the demand and the capacity is not practical in the context
of rockbursts. It is possible, however, to better understand the interaction between capacity and
demand through the investigation of support performance.

In the collected rockburst data, the ground support systems subjected to dynamic loading
reflected the evolution of support practice at Creighton, Copper Cliff, and Coleman mines, from
2000 to 2013. As mining keeps progressing to greater depths and through higher stress
conditions, the required performance threshold of ground support systems is likely going to
increase over time. The introduction of new ground support technologies, not available at the
time in the collected rockburst data, may eventually provide a solution to match higher levels of
dynamic-load demand in the future. As the ground support practice keeps evolving, it will be
important to continue monitoring of support performance and to report on the variables outlined
in this thesis. As part of future work, the ppv-performance thresholds of new ground support
systems, determined and validated using high-quality field data, could be added to the ground
support-selection charts in order to provide continuous improvement to the proposed guidelines.
The improved version of the Unusual Occurrence Report for Groundfall/Rockbursts (Appendix
A) is already used for reporting on the performance of support systems under dynamic-loading
conditions.

The collected rockburst data treated in this thesis were specific to three mines, and perhaps more
importantly to one mining district. Although the specific recommendations of this thesis can be
potentially applied in other mines, it will require site-specific calibration to gain confidence in
the recommendations for assessing the location and timing of installation of enhanced ground
support. The 3D stress analysis elastic model employed in this thesis was calibrated with
 276

rockburst incidents and provided confidence going forward in the conditions used. Nevertheless,
it was acknowledged that there could be variations in the absolute stress thresholds when
applying the proposed methodology at other mine sites. Further variations in the absolute stress
values can result from the use of numerical models that employ different assumptions in model
behaviour such as non-linear constitutive laws. Irrespective of the model use, it is necessary to
undertake a comprehensive site-specific calibration whereby the model predictions are compared
to field observations. Recommendations based on the results of site-specific calibrated numerical
models will probably result in variations in the recommended threshold design values. This
should be taken into consideration when establishing site-specific guidelines using different
numerical tools.

This thesis demonstrated that the performance of ground support systems is sensitive to changes
in the excavation dimensions and stress conditions. It is possible that variations in rock mass
quality, strength of intact rock, and rockburst mechanisms could also impact the performance of
ground support systems. The rockbursts reported in the developed database often occurred in
good quality rock masses, with the intact rock typically characterized by a relatively high
strength (> 200 MPa). Furthermore, rockburst mechanisms were typically associated with either
strain bursting or seismic shaking triggered by remote fault slip events. The recommended
thresholds of support performance could vary under different ground conditions, e.g. softer
rocks, or damage mechanisms, e.g. buckling as opposed to strain bursting. Variations from the
conditions represented in the developed rockburst database should be taken into consideration
when applying the proposed ground support design methodology to other mine sites. Until site-
specific databases of support performance under rockburst conditions are constructed, the
recommended threshold values can provide for a robust ground support design strategy for
assessing the support requirements based on the anticipated stress, excavation, and seismic
conditions. These recommendations should be revised or fine-tuned based on local experience.

The proposed ground support strategy is a validated tool to assist ground control engineers
during the design phase. Employed in conjunction with good quality control and assurance, and
monitoring practice, this strategy can contribute to an extension of the service life of mine
excavations subjected to dynamic-loading conditions, reduction in rehabilitation costs and
production delays due to rockbursts, and increased workplace safety.
 277

References
Abdi, H., & Williams, L. J. (2010). Principal component analysis. Wiley Interdisciplinary Reviews:
Computational Statistics, 2(4), 433-459.

Albrecht, J. (2005). Delineating rockburst damage to underground development subjected to seismic

loading. (Doctoral dissertation). University of Western Australia, Perth, WA.

Andrews, D., & Courchesne, T. (2009). Creighton Mine mineral resource and mineral reserve estimate.
(Technical report). Sudbury, ON: Vale Inco Ltd.

Andrieux, P., Brummer, R. K., Li, H., & O'Connor, C. P. (2007). Elastic versus inelastic numerical
modelling of deep and highly stressed mining fronts. In Y. Potvin (Ed.), Proceedings of the fourth
international seminar on deep and high stress mining (pp. 51-64). Perth, WA: Australian Centre for
Geomechanics.

Atlas Copco. (2013). Swellex rock bolts – Product catalogue.

Barton, N. (2002). Some new Q-value correlations to assist in site characterisation and tunnel design.
International Journal of Rock Mechanics and Mining Sciences, 39(2), 185-216.

Barton, N., Lien, R., & Lunde, J. (1974). Engineering classification of rock masses for the design of
tunnel support. Rock Mechanics, 6(4), 189-236.

Beck, D., Reusch, F., & Arndt, S. (2007). Estimating the probability of mining-induced seismic events
using mine-scale, inelastic numerical models. In Y. Potvin (Ed.), Proceedings of the fourth
international seminar on deep and high stress mining (pp. 31-42). Perth, WA: Australian Centre for
Geomechanics.

Berruar, O. N. S., Yao, M., & Sampson-Forsythe, A. (2012). Challenges in multiple sill pillar mining at
Vale's Coleman Mine. In Y. Potvin (Ed.), Proceedings of the sixth international seminar on deep
and high stress mining (pp. 297-310). Perth, WA: Australian Centre for Geomechanics.

Bieniawski, Z. T. (1989). Engineering rock mass classifications: A complete manual for engineers and
geologists in mining, civil, and petroleum engineering. New York, NY: John Wiley & Sons.

Bieniawski, Z. T. (1992). Invited paper: Principles of engineering design for rock mechanics. 33rd US
Symposium on Rock Mechanics (USRMS), American Rock Mechanics Association.

Brady, B. H. G., & Brown, E. T. (2004). Rock mechanics for underground mining (3rd ed.). Dordrecht:
Springer Science + Business Media.

British Geological Survey. (2013). World mineral production 2007-2011. Keyworth, Nottingham: British
Geological Survey.

Bruneau, G., Hudyma, M. R., Hadjigeorgiou, J., & Potvin, Y. (2003). Influence of faulting on a mine
shaft—a case study: part II—numerical modelling. International Journal of Rock Mechanics and
Mining Sciences, 40, 113-125.

Bureau, M., & Doucet, C. (2013). Performance of Atlas Copco Swellex Mn line bolts in simulated
dynamic environment. CIM 2013 Convention, Toronto (ON).
 278

Charette, F. (2012). Applicability of D-bolt in high stress conditions. Dynamic Ground Support
Applications Symposium, Sudbury, ON.

Chinnasane, D. R., Yao, M., Landry, D., & Paradis-Sokoloski, P. (2012). Performance of dynamic
support system in highly burst-prone ground conditions at Vale's Copper Cliff Mine – a case study.
In Y. Potvin (Ed.), Proceedings of the sixth international seminar on deep and high stress mining
(pp. 57-70). Perth, WA: Australian Centre for Geomechanics.

Cochrane, L. B. (1984). Ore deposits of the Copper Cliff Offset. In E. G. Pye, A. J. Naldrett & P. J. Giblin
(Eds.), The geology and ore deposits of the Sudbury structure (pp. 347-359). Toronto, ON: Ontario
Ministry of Natural Resources.

Cochrane, L. B. (1991). Analysis of the structural and tectonic environments associated with rock mass
failures in the mines of the Sudbury district. (Doctoral dissertation). Queen's University, Kingston,
ON.

Coulson, A. L. (1996). Mine induced seismicity in highly stressed ground: A case study – Creighton Mine
INCO ltd., Sudbury. (Master's thesis). Queen's University, Kingston, ON.

Counter, D. B. (2012). Support system evolution at Kidd Mine. Dynamic Ground Support Applications
Symposium, Sudbury, ON.

Counter, D. B. (2014). Kidd Mine – dealing with the issues of deep and high stress mining – past, present,
and future. In M. R. Hudyma, & Y. Potvin (Eds.), Proceedings of the seventh international
conference on deep and high stress mining (pp. 3-22). Perth, WA: Australian Centre for
Geomechanics.

Diederichs, M. S. (2003). Rock fracture and collapse under low confinement conditions. Rock Mechanics
and Rock Engineering, 36(5), 339-381.

Dietz, R. S. (1964). Sudbury structure as an astrobleme. The Journal of Geology, 72(4), 412-434.

Doucet, C., & Voyzelle, B. (2012). Technical information data sheets. (Technical report). Ottawa (ON):
CanmetMINING.

Dowding, C. H. (1996). Construction vibrations. Upper Saddle River, NJ: Prentice Hall.

Durrheim, R. J. (2012). Functional specifications for in-stope support based on seismic and rockburst
observations in South African mines. In Y. Potvin (Ed.), Proceedings of the sixth international
seminar on deep and high stress mining (pp. 41-55). Perth, WA: Australian Centre for
Geomechanics.

Eriksson, L., Johansson, E., Kettaneh-Wold, N., Trygg, J., Wikström, C., & Wold, S. (2006). Multi-and
megavariate data analysis: Part I: Basic principles and applications (2nd ed.). Umeå, SE: Umetrics
AB.

Everitt, B. S., Landau, S., Leese, M., & Stahl, D. (2011). Hierarchical clustering. Cluster analysis (5th ed.,
pp. 71-110). Chichester, West Sussex, U.K.: Wiley.

Fang, Z., & Harrison, J. (2002). Development of a local degradation approach to the modelling of brittle
fracture in heterogeneous rocks. International Journal of Rock Mechanics and Mining Sciences,
39(4), 443-457.
 279

Gibowicz, S. J. (1990). The mechanism of seismic events induced by mining – A review. In C. Fairhurst
(Ed.), Proceedings of the second international symposium on rockbursts and seismicity in mines (pp.
3-27). Rotterdam: Balkema.

Hadjigeorgiou, J., & Charette, F. (2001). Rock bolting for underground excavations. In W. A. Hustrulid,
& R. L. Bullock (Eds.), Underground mining methods (pp. 547-554). Littleton, CO: Society for
Mining, Metallurgy, and Exploration.

Hadjigeorgiou, J., & Potvin, Y. (2007). Overview of dynamic testing of ground support. In Y. Potvin
(Ed.), Proceedings of the fourth international seminar on deep and high stress mining (pp. 349-371).
Perth, WA: Australian Centre for Geomechanics.

Hadjigeorgiou, J., & Potvin, Y. (2011). A critical assessment of dynamic rock reinforcement and support
testing facilities. Rock Mechanics and Rock Engineering, 44(5), 565-578.

Hadjigeorgiou, J., & Charette, F. (2009). Guide pratique du soutènement minier (2nd ed.). Québec, QC:
Association minière du Québec.

Hammah, R. E., & Curran, J. H. (2009). It is better to be approximately right than precisely wrong: Why
simple models work in mining geomechanics. 43rd US Rock Mechanics Symposium and 4th U.S.-
Canada Rock Mechanics Symposium, Asheville, NC. 55-61.

Heal, D. P. (2010). Observations and analysis of incidences of rockburst damage in underground mines.
(Doctoral dissertation). University of Western Australia, Perth, WA.

Hedley, D. G. F. (1992). Rockburst handbook for Ontario hardrock mines. (No. SP92-1E). Ottawa, ON:
Canmet.

Hoek, E., & Brown, E. T. (1980). Underground excavation in rock. London: Institution of Mining and
Metallurgy.

Hudson, J. A., & Harrison, J. P. (1997). Engineering rock mechanics: An introduction to the principles.
Oxford: Pergamon.

Hudyma, M. R. (2008). Analysis and interpretation of clusters of seismic events in mines. (Doctoral
dissertation). University of Western Australia, Perth, WA.

Hudyma, M. R., Heal, D. P., & Mikula, P. (2003). Seismic monitoring in mines – old technology – new
applications. In B. K. Hebblewhite (Ed.), Proceedings of the first australasian ground control in
mining conference (pp. 201-218). Sydney, NSW: UNSW School of Mining Engineering.

Hudyma, M. R., & Potvin, Y. H. (2010). An engineering approach to seismic risk management in
hardrock mines. Rock Mechanics and Rock Engineering, 43(6), 891-906.

Hudyma, M. R., & Brummer, R. (2007). Copper Cliff North Mine seismicity review: 2004-2006.
(Technical report). Sudbury, ON: Itasca Consulting Canada.

Hutchinson, D. J., & Diederichs, M. S. (1996). Cablebolting in underground mines. Richmond, BC:
BiTech Publishers.

Kaiser, P. K., McCreath, D. R., & Tannant, D. D. (1996). Canadian rockburst support handbook.
Canadian rockburst research program 1990-1995. CAMIRO.
 280

Kaiser, P. K., Tannant, D. D., McCreath, D. R., & Jesenak, P. (1992). Rockburst damage assessment
procedure. In P. K. Kaiser, & D. R. McCreath (Eds.), Rock support in mining and underground
construction (pp. 639-647). Rotterdam: Balkema.

Kaiser, P. K., & Cai, M. (2012). Design of rock support system under rockburst condition. Journal of
Rock Mechanics and Geotechnical Engineering, 4(3), 215-227.

Kaiser, P. K., & Cai, M. (2013). Critical review of design principles for rock support in burst-prone
ground – time to rethink! In Y. Potvin, & B. Brady (Eds.), Proceedings of the seventh international
symposium on ground support in mining and underground construction (pp. 3-38). Perth, WA:
Australian Centre for Geomechanics.

Kaiser, P. K., Vasak, P., Suorineni, F. T., & Thibodeau, D. (2005). New dimensions in seismic data
interpretation with 3-D virtual reality visualization for burst-prone mines. In Y. Potvin, & M.
Hudyma (Eds.), Proceedings of the sixth international symposium on rockburst and seismicity in
mines (pp. 33-45). Perth, WA: Australian Centre for Geomechanics.

Karampinos, E., Simser, B., & Hadjigeorgiou, J. (2012). Determination of rock mass failure and rock
mass damage criteria for the felsic norites at Nickel Rim South Mine. In Y. Potvin (Ed.),
Proceedings of the sixth international seminar on deep and high stress mining (pp. 401-414). Perth,
WA: Australian Centre for Geomechanics.

Li, C. C. (2012). Performance of D-bolts under static loading. Rock Mech Rock Eng, 45, 183-192.
doi:10.1007/s00603-011-0198-6

Li, C. C., & Doucet, C. (2012). Performance of D-bolts under dynamic loading. Rock Mech Rock Eng, 45,
193-204. doi:10.1007/s00603-011-0202-1

Liang, G. (2012). Performance of support systems subjected to dynamic loads at two underground nickel
mines. (Master's thesis). University of Toronto, Toronto, ON.

Malek, F., Suorineni, F. T., & Vasak, P. (2009). Geomechanics strategies for rockburst management at
Vale Inco Creighton Mine. 3rd CANUS Rock Mechanics Symposium, Toronto, ON.

Malek, F., Trifu, C., Suorineni, F. T., Espley, S., & Yao, M. (2008). Management of high stress and
seismicity at Vale Inco Creighton Mine. 42nd US Rock Mechanics Symposium, San Francisco, CA.

Mansour Mining. (2015). Product catalogue. Retrieved Apr. 17, 2015, from
http://mansourmining.com/Downloads/MMTI%20Product%20Catalogue%20V2.pdf

Martin, C. D. (1997). Seventeenth Canadian Geotechnical Colloquium: The effect of cohesion loss and
stress path on brittle rock strength. Canadian Geotechnical Journal, 34(5), 698-725.

Martin, C. D., Kaiser, P. K., & McCreath, D. R. (1999). Hoek-Brown parameters for predicting the depth
of brittle failure around tunnels. Canadian Geotechnical Journal, 36(1), 136-151.

McGarr, A., Green, R., & Spottiswoode, S. M. (1981). Strong ground motion of mine tremors: Some
implications for near-source ground motion parameters. Bulletin of the Seismological Society of
America, 71(1), 295-319.

Mendecki, A. J. (2008). Forecasting seismic hazard in mines. In Y. Potvin, J. Carter, A. Dyskin & R.
Jeffrey (Eds.), Proceedings of the first southern hemisphere international rock mechanics
symposium (pp. 55-69). Perth, WA: Australian Centre for Geomechanics.
 281

Merriam-Webster. (2008). Collegiate dictionary (11th ed.). Springfield, MA: Merriam-Webster, Inc.

Michailidis, G. (2007). Principal component analysis. In N. J. Salkind (Ed.), Encyclopedia of

measurement and statistics (pp. 783-786). Thousand Oaks, CA: SAGE Reference.

Mikula, P. (2012). Progress with empirical charting for confident selection of ground support in seismic
conditions. In Y. Potvin (Ed.), Proceedings of the sixth international seminar on deep and high
stress mining (pp. 71-90). Perth, WA: Australian Centre for Geomechanics.

Milev, A. M., Spottiswoode, S. M., Noble, B. R., Linzer, L. M., van Zyl, M., Daehnke, A., &
Acheampong, E. (2002). The meaningful use of peak particle velocities at excavation surfaces for
the optimisation of the rockburst criteria for tunnels and stopes. (SIMRAC Final Project Report
GAP 709 No. 2002 - 0305). CSIR: Division of Mining and Technology.

Montgomery, D. C., & Runger, G. C. (2007). Applied statistics and probability for engineers (4th ed.).
Hoboken, NJ: John Wiley & Sons.

Morissette, P., Hadjigeorgiou, J., Punkkinen, A. R., & Chinnasane, D. R. (2014). The influence of change
in mining and ground support practice on the frequency and severity of rockbursts. In M. R.
Hudyma, & Y. Potvin (Eds.), Proceedings of the seventh international conference on deep and high
stress mining (pp. 165-178). Perth, WA: Australian Centre for Geomechanics.

Morissette, P., Hadjigeorgiou, J., & Thibodeau, D. (2012). Validating a support performance database
based on passive monitoring data. In Y. Potvin (Ed.), Proceedings of the sixth international seminar
on deep and high stress mining (pp. 27-40). Perth, WA: Australian Centre for Geomechanics.

Naldrett, A. J. (1984). Ni-Cu ores of the Sudbury Igneous Complex – Introduction. In E. G. Pye, A. J.
Naldrett & P. J. Giblin (Eds.), The geology and ore deposits of the Sudbury structure (pp. 302-307).
Toronto, ON: Ontario Ministry of Natural Resources.

Naldrett, A. J. (1989). Magmatic sulphide deposits. New York: Clarendon Press.

National Resources Canada. (2015). National earthquake database. Retrieved 18 March, 2015, from
http://www.earthquakescanada.nrcan.gc.ca/stndon/NEDB-BNDS/index-eng.php

Normet. (2014). D-bolt technical data sheet. Retrieved May 22, 2015, from
http://www.normet.com/inet/normet/flow.nsf/(AttachmentsByFileName)/D-
Bolt%20TDS%2020140129.pdf/$File/D-Bolt%20TDS%2020140129.pdf;

O’Connor, C., Cotesta, L., Brummer, R., & Thibodeau, D. (2010). Non-linear modelling calibration
process for Vale nickel mines Ontario division. In M. Van Sint Jan, & Y. Potvin (Eds.), Proceedings
of the fifth international seminar on deep and high stress mining (pp. 521-535). Perth, WA:
Australian Centre for Geomechanics.

Ortlepp, W. D. (1997). Rock fracture and rockbursts: An illustrative study. Johannesburg, SA: South
African Institute of Mining and Metallurgy.

Ortlepp, W. D. (1992). The design of support for the containment of rockburst damage in tunnels – An
engineering approach. In P. K. Kaiser, & D. R. McCreath (Eds.), Rock support in mining and
underground construction (pp. 593-609). Rotterdam, NL: Balkema.

Ortlepp, W. D., & Stacey, T. R. (1998). Testing of tunnel support: Dynamic load testing of rockbolt
elements to provide data for safer support design. (No. GAP 423). SIMRAC.
 282

Ortlepp, W. D. (2001). The behaviour of tunnels at great depth under large static and dynamic pressures.
Tunnelling and Underground Space Technology, 16(1), 41-48.

Peck, R. B. (1969). Advantages and limitations of the observational method in applied soil mechanics.
Geotechnique, 19(2), 171-187.

Petty, S. E. (2013). Forensic engineering: Damage assessments for residential and commercial structures
CRC Press.

Player, J. R., Thompson, A. G., & Villaescusa, E. (2008). Dynamic testing of reinforcement systems. In
T. R. Stacey, & D. F. Malan (Eds.), Proceedings of the sixth international symposium on ground
support in mining and civil engineering construction (pp. 597-622). Cape Town (SA): Southern
African Institute of Mining and Metallurgy.

Player, J. R., Villaescusa, E., & Thompson, A. G. (2009). Dynamic testing of friction rock stabilisers. 3rd
CANUS Rock Mechanics Symposium, Toronto (ON).

Potvin, Y., & Wesseloo, J. (2013). Towards an understanding of the dynamic demand on ground support.
In Y. Potvin (Ed.), Proceedings of the seventh international symposium on ground support in mining
and underground construction (pp. 287-304). Perth, WA: Australian Centre for Geomechanics.

Potvin, Y., Wesseloo, J., & Heal, D. (2010). An interpretation of ground support capacity submitted to
dynamic loading. Mining Technology, 119(4), 233-245.

Punkkinen, A. R., & Mamidi, N. R. (2010). Effective ground support system design to manage seismic
hazard in a high stress diminishing pillar at a Vale mine. In Y. Potvin, & M. Van Sint Jan (Eds.),
Proceedings of the fifth international seminar on deep and high stress mining (pp. 367-381). Perth,
WA: Australian Centre for Geomechanics.

Punkkinen, A. R., & Yao, M. (2007). Change of ground support system and mining practice in the Deep
at Creighton Mine. CIM Conference and Exhibition 2007, Montreal, QC.

Rickard, J., & Watkinson, D. (2001). Cu-Ni-PGE mineralization within the Copper Cliff Offset dike,
Copper Cliff North Mine, Sudbury, Ontario: Evidence for multiple stages of emplacement.
Exploration and Mining Geology, 10(1-2), 111-124.

Roth, A., Cala, M., Brändle, R., & Rorem, E. (2014). Analysis and numerical modelling of dynamic
ground support based on instrumented field-scale tests. In M. R. Hudyma, & Y. Potvin (Eds.),
Proceedings of the seventh international seminar on deep and high stress mining (pp. 151-164).
Perth, WA: Australian Centre for Geomechanics.

Roussell, D. H., & Card, K. D. (2009). Geological setting. In D. H. Roussell, & G. H. Brown (Eds.), A
field guide to the geology of Sudbury, Ontario (pp. 1-6) Ontario Geological Survey.

Sampson-Forsythe, A. (2011). Unusual occurrence report for rockburst ME-156. (Technical report).
Levack, ON: Vale.

Simser, B. (2007). The weakest link – ground support observations at some Canadian Shield hard rock
mines. In Y. Potvin (Ed.), Proceedings of the fourth international seminar on deep and high stress
mining (pp. 335-348). Perth, WA: Australian Centre for Geomechanics.

Simser, B., Andrieux, P., Langevin, F., Parrott, T., & Turcotte, P. (2007). Field behaviour and failure
modes of modified conebolts at the Craig, LaRonde and Brunswick mines in Canada. In Y. Potvin,
 283

Hadjigeorgiou, J., & Stacey, D. (Eds.), Challenges in deep and high stress mining (pp. 347-354).
Perth, WA: Australian Centre for Geomechanics.

Simser, B., & Andrieux, P. (1999). Seismic source mechanisms at the Brunswick Mine. 14th Mine
Operators Conference of the CIM,

Simser, B. (2012). Ground support in "deep" underground mines. Dynamic Ground Support Applications
Symposium, Sudbury, ON.

Snelling, P. E., Godin, L., & McKinnon, S. D. (2013). The role of geologic structure and stress in
triggering remote seismicity in Creighton Mine, Sudbury, Canada. International Journal of Rock
Mechanics and Mining Sciences, 58, 166-179.

South African Mining and Engineering Supplies. (2006). Tunnel guard. Retrieved 11 June, 2015, from
http://www.samining.com/prod_tg_tg.htm

Stacey, T. R. (2009). Design—a strategic issue. Journal of the South African Institute of Mining &
Metallurgy, 109(3), 157.

Stacey, T. R. (2012). Support of excavations subjected to dynamic (rockburst) loading. In Q. Qian, & Y.
Zhou (Eds.), Proceedings of the 12th international congress of the international society of rock
mechanics (pp. 137-145). London: Tayor & Francis Group.

Stacey, T. R., & Ortlepp, W. D. (2004). The capabilities of various types of wire mesh under dynamic
loading and the contribution of wire rope lacing in surface support. In Y. Potvin, T. R. Stacey & J.
Hadjigeorgiou (Eds.), Surface support in mining (pp. 403-408). Perth, WA: Australian Centre for
Geomechanics.

Suorineni, F. T., & Vasak, P. (2008). Expert opinion report on North Mine seismic events of September
11th 2008. (Technical report). Sudbury, ON: Mirarco.

Tan, R. (2003). Numerical modelling of a new “V” mining sequence for Creighton Deep. (Technical
report). Sudbury, ON: Inco Ltd.

Tannant, D. D. (2004). Load capacity and stiffness of welded wire, chain link, and expanded metal mesh.
In Y. Potvin, T. R. Stacey & J. Hadjigeorgiou (Eds.), Surface support in mining (pp. 399-402). Perth,
WA: Australian Centre for Geomechanics.

Tannant, D. D., & Kaiser, P. K. (1997). Evaluation of shotcrete and mesh behaviour under large imposed
deformations. International Symposium on Rock Support, Lillhammer.

Tarasov, B. G., & Potvin, Y. (2012). Absolute, relative, and intrinsic rock brittleness at compression. In
Y. Potvin (Ed.), Proceedings of the sixth international seminar on deep and high stress mining (pp.
119-133). Perth, WA: Australian Centre for Geomechanics.

Tarasov, B. G., & Randolph, M. F. (2011). Superbrittleness of rocks and earthquake activity.
International Journal of Rock Mechanics and Mining Sciences, 48(6), 888-898.

Turcotte, P. (2014). Practical applications of a rockburst database to ground support design at LaRonde
mine. In M. R. Hudyma, & Y. Potvin (Eds.), Proceedings of the seventh international conference on
deep and high stress mining (pp. 79-92). Perth (WA): Australian Centre for Geomechanics.

Umetrics. (2008). User guide to simca-P+ (version 12). Umeå, SE: Umetrics AB.
 284

Umetrics. (2009). Simca-P+ (version 12.0.1.0) [software]. Umeå, SE: Umetrics AB.

Vale. (2010a). Geology. In D. R. Chinnasane (Ed.), Copper Cliff Mine – 2010 mine design package (pp.
12-27). Sudbury, ON: Vale.

Vale. (2010b). Mining methods. In D. R. Chinnasane (Ed.), Copper Cliff Mine – 2010 mine design
package (pp. 33-49). Sudbury, ON: Vale.

Vale. (2011). Geology and geotechnical data. In A. Sampson-Forsythe (Ed.), Plant 10 Coleman Mine –
mine design package 2011 update (pp. 8-19). Sudbury, ON: Vale.

Vale. (2012a). Geology and geotechnical data. In A. R. Punkkinen (Ed.), Plant 17 Creighton Mine - mine
design package 2012 update (pp. 18-63). Sudbury, ON: Vale.

Vale. (2012b). Ground support. In A. R. Punkkinen (Ed.), Plant 17 Creighton Mine - mine design
package 2012 update (pp. 81-93). Sudbury, ON: Vale.

Villaescusa, E., Player, J. R., & Thompson, A. G. (2014). A reinforcement design methodology for highly
stressed rock masses. 8th Asian Rock Mechanics Symposium, Sapporo, JP.

Ward, J. H. (1963). Hierarchical grouping to optimize an objective function. Journal of the American
Statistical Association, 58(301), 236-244.

Wawersik, W. R., & Brace, W. F. (1971). Post-failure behavior of a granite and diabase. Rock Mechanics,
3(2), 61-85.

Wawersik, W. R., & Fairhurst, C. (1970). A study of brittle rock fracture in laboratory compression
experiments. International Journal of Rock Mechanics and Mining Sciences, 7(5), 561-575.

Wiles, T. D. (1998). Correlation between energy released rate and observed bursting conditions at
Creighton Mine. (Technical report). Mt. Eliza, VIC: Mine Modelling Pty. Ltd.

Wiles, T. D. (2005). Rockburst prediction using numerical modelling: Realistic limits for failure
prediction accuracy. In Y. Potvin, & M. R. Hudyma (Eds.), Proceedings of the sixth international
symposium on rockbursts and seismicity in mines (pp. 57-63). Perth, WA: Australian Centre for
Geomechanics.

Wiles, T. D. (2007). Loading system stiffness – A parameter to evaluate rockburst potential. In Y. Potvin,
J. Hadjigeorgiou & T. R. Stacey (Eds.), Challenges in deep and high stress mining (pp. 57-63).
Perth, WA: Australian Centre for Geomechanics.

Wiles, T. D., & MacDonald, P. (1988). Correlation of stress analysis results with visual and microseismic
monitoring at Creighton Mine. Computers and Geotechnics, 5(2), 105-121.

Woldemedhin, B. Y., & Mwagalanyi, H. (2011). Investigation of rock-fall and support damage induced
by seismic motion at Kiirunavaara Mine. (Master's thesis). Luleå University of Technology, Luleå,
SE.

WSN. (2010). Unusual occurrence report for groundfall/rockburst. (Technical Report). North Bay, ON:
WSN.
 285

Yao, M., Chinnasane, D. R., & Harding, D. (2009). Mitigation plans for mining in highly burst-prone
ground conditions at Vale Inco Copper Cliff North Mine. 3rd CANUS Rock Mechanics Symposium,
Toronto, ON.
 286

Appendix A:
Unusual Occurrence Report for Groundfall/Rockbursts
 287
UNUSUAL OCCURRENCE REPORT
FOR GROUNDFALL/ROCKBURST
(UNDERGROUND MINE)

THIS REPORT IS FOR: FALL OF GROUND ROCKBURST

GENERAL
Company Internal Report Reportable Incident
incident code: (see Sect. 21 of Ontario Regulation 854))
Company: Mine: Address:

Date: Unknown Time of occurrence: AM PM Unknown

Damage sustained by mine
None Single location Multiple locations
openings:
General description of occurrence:

WORKERS
At time of incident workers were: Underground Surface No one Working Unknown

Were workers normally required in area: Yes No Was access to the area restricted? Yes No
Were workers in immediate area of damage: Yes To within what distance of the
No incident were workers present: m ft
Were there any injuries: Yes No Nature of Injuries:

SEISMICITY (FOR ROCKBURSTS ONLY)

Seismic event that Magnitude: Coordinates: North East Depth m ft
most likely triggered
Apparent seismic
damage: Undetermined Strain burst Pillar burst Fault slip
source mechanism:
Magnitude scale: Nuttli Richter
Magnitude of first event: Magnitude of largest event:
Other:
Event magnitudes: <1 1- 2 2– 3 >3

Number of events: Unknown Unknown Unknown Unknown

Period of time over which events occurred (if more than one): Unknown Seconds Minutes Hours

Location of major events: Hanging wall Footwall Ore Zone Not Located
Location determined by: Visual Inspection Seismic Monitoring Equipment Other Monitoring Equipment
Estimated Not Located
The Rockburst: Triggered a fall of ground Displaced material violently Was contained by ground support

Revised November 28, 2013

 288
DAMAGE LOCATION #1

DESCRIPTION OF OCCURRENCE
Mine level: Location:
This area was:
Coordinates: North East Depth m ft
Active Inactive Abandoned
Geological zone: H/W F/W Ore Rock type:

The incident occurred in: Raise Drift/XC Pillar Shaft Ore/waste pass Stope Other:

Opening dimensions: Height: Width: Length: Span: ft m

Damage sustained to: Excavation Ground Support Equipment Unknown

Associated mining activity: Nothing apparent Backfilling Blasting Bolting Drilling Mucking Scaling

Ore Recovery in Immediate Area: None Primary Recovery Pillar or Secondary Recovery
Mining Method: None Shrinkage Cut & Fill Post Pillar Cut & Fill Undercut & Fill Blasthole
VRM Slot & Slash Uppers Retreat Sublevel Caving Block Caving Other:
If pillar sustained damage: Type: Rib Post Sill Crown Other:

Pillar dimensions Height: Width: Length: m ft

Material displaced from: Face Back Wall Floor Shoulder Brow Unknown Other:
From behind support From unsupported
Material displaced Contained by support: Total:
(uncontained): ground:
tonnes tons

Damage dimensions Length: Width: Max. depth: m ft

Displaced material Wedge Tabular Blocky Thin/slabbing Irregular Shotcrete Unknown

Rockburst damage Rock bulking due to fracturing Rock ejection due to seismic energy transfer
mechanism
Rock fall due to seismic shaking Unknown Not applicable

Comments:

Rock mass characteristics

Massive Bedded Blocky/Chunks Fractured Slabbing Unknown
(choose one only)
Dyke Fault/shear Contacts Steeply dipping joints Flat lying joints
Structural geology and water
Joint alteration/infilling Water
Orientation: trend/plunge
Fault/dyke description Thickness: m ft
dip/dip direction
Comments:

Revised November 28, 2013

 289
DAMAGE LOCATION #1

ROCK SUPPORT SYSTEM

Binder Type and
Backfill Type Location or Opening Backfilled Percentage Filled
Content

Location Pattern Performance

Reinforcement Type Length
Back Walls Wide Long Failed Beyond
Mechanical
bolts
Resin rebars
Friction
stabilizers
Expandable
bolts
Dynamic bolts

Cable bolts

Location Dimension Performance

Surface
Type or
support Cracked or
Back Walls thickness Broken Failed
bulged
Wire-mesh

Shotcrete

Straps

Used to support Performance

Other system
Back Walls Deformed Broken Failed

Comments Regarding Effectiveness of Support Systems:

Follow-up Action:

Revised November 28, 2013

 290
DAMAGE LOCATION #2

DESCRIPTION OF OCCURRENCE
Mine level: Location:
This area was:
Coordinates: North East Depth m ft
Active Inactive Abandoned
Geological zone: H/W F/W Ore Rock type:

The incident occurred in: Raise Drift/XC Pillar Shaft Ore/waste pass Stope Other:

Opening dimensions: Height: Width: Length: Span: ft m

Damage sustained to: Excavation Ground Support Equipment Unknown

Associated mining activity: Nothing apparent Backfilling Blasting Bolting Drilling Mucking Scaling

Ore Recovery in Immediate Area: None Primary Recovery Pillar or Secondary Recovery
Mining Method: None Shrinkage Cut & Fill Post Pillar Cut & Fill Undercut & Fill Blasthole
VRM Slot & Slash Uppers Retreat Sublevel Caving Block Caving Other:
If pillar sustained damage: Type: Rib Post Sill Crown Other:

Pillar dimensions Height: Width: Length: m ft

Material displaced from: Face Back Wall Floor Shoulder Brow Unknown Other:
From behind support From unsupported
Material displaced Contained by support: Total:
(uncontained): ground:
tonnes tons

Damage dimensions Length: Width: Max. depth: m ft

Displaced material Wedge Tabular Blocky Thin/slabbing Irregular Shotcrete Unknown

Rockburst damage Rock bulking due to fracturing Rock ejection due to seismic energy transfer
mechanism
Rock fall due to seismic shaking Unknown Not applicable

Comments:

Rock mass characteristics

Massive Bedded Blocky/Chunks Fractured Slabbing Unknown
(choose one only)
Dyke Fault/shear Contacts Steeply dipping joints Flat lying joints
Structural geology and water
Joint alteration/infilling Water
Orientation: trend/plunge
Fault/dyke description Thickness: m ft
dip/dip direction
Comments:

Revised November 28, 2013

 291
DAMAGE LOCATION #2

ROCK SUPPORT SYSTEM

Binder Type and
Backfill Type Location or Opening Backfilled Percentage Filled
Content

Location Pattern Performance

Reinforcement Type Length
Back Walls Wide Long Failed Beyond
Mechanical
bolts
Resin rebars
Friction
stabilizers
Expandable
bolts
Dynamic bolts

Cable bolts

Location Dimension Performance

Surface
Type or
support Cracked or
Back Walls thickness Broken Failed
bulged
Wire-mesh

Shotcrete

Straps

Used to support Performance

Other system
Back Walls Deformed Broken Failed

Comments Regarding Effectiveness of Support Systems:

Follow-up Action:

Revised November 28, 2013

 292
DAMAGE LOCATION #3

DESCRIPTION OF OCCURRENCE
Mine level: Location:
This area was:
Coordinates: North East Depth m ft
Active Inactive Abandoned
Geological zone: H/W F/W Ore Rock type:

The incident occurred in: Raise Drift/XC Pillar Shaft Ore/waste pass Stope Other:

Opening dimensions: Height: Width: Length: Span: ft m

Damage sustained to: Excavation Ground Support Equipment Unknown

Associated mining activity: Nothing apparent Backfilling Blasting Bolting Drilling Mucking Scaling

Ore Recovery in Immediate Area: None Primary Recovery Pillar or Secondary Recovery
Mining Method: None Shrinkage Cut & Fill Post Pillar Cut & Fill Undercut & Fill Blasthole
VRM Slot & Slash Uppers Retreat Sublevel Caving Block Caving Other:
If pillar sustained damage: Type: Rib Post Sill Crown Other:

Pillar dimensions Height: Width: Length: m ft

Material displaced from: Face Back Wall Floor Shoulder Brow Unknown Other:
From behind support From unsupported
Material displaced Contained by support: Total:
(uncontained): ground:
tonnes tons

Damage dimensions Length: Width: Max. depth: m ft

Displaced material Wedge Tabular Blocky Thin/slabbing Irregular Shotcrete Unknown

Rockburst damage Rock bulking due to fracturing Rock ejection due to seismic energy transfer
mechanism
Rock fall due to seismic shaking Unknown Not applicable

Comments:

Rock mass characteristics

Massive Bedded Blocky/Chunks Fractured Slabbing Unknown
(choose one only)
Dyke Fault/shear Contacts Steeply dipping joints Flat lying joints
Structural geology and water
Joint alteration/infilling Water
Orientation: trend/plunge
Fault/dyke description Thickness: m ft
dip/dip direction
Comments:

Revised November 28, 2013

 293
DAMAGE LOCATION #3

ROCK SUPPORT SYSTEM

Binder Type and
Backfill Type Location or Opening Backfilled Percentage Filled
Content

Location Pattern Performance

Reinforcement Type Length
Back Walls Wide Long Failed Beyond
Mechanical
bolts
Resin rebars
Friction
stabilizers
Expandable
bolts
Dynamic bolts

Cable bolts

Location Dimension Performance

Surface
Type or
support Cracked or
Back Walls thickness Broken Failed
bulged
Wire-mesh

Shotcrete

Straps

Used to support Performance

Other system
Back Walls Deformed Broken Failed

Comments Regarding Effectiveness of Support Systems:

Follow-up Action:

Revised November 28, 2013

 294
ATTACHEMENTS
Please, provide a list of attached documents (e.g. photos, mine plans, etc.) if applicable.

SIGN-OFF
Date Report Completed Name of Person Completing Report Title

Phone: ( ) Fax: ( ) E-Mail:

Title Name Signature Date

If this is a reportable incident, please send report to:

• District Office, Mining Health and Safety Program, Ontario Ministry of Labour [email protected]
• Ground Control Specialist, Workplace Safety North, 690 McKeown Avenue, PO Box 2050, North Bay, Ont. P1B 9P1
[email protected]

To obtain a copy of the Guidelines for completing the Unusual Occurrence Report for Groundfall/Rockburst, or for additional information,
please contact WSN’s Ground Control Specialist, (705) 474-7233 [email protected]

Revised November 28, 2013

 295

Appendix B:
A Review of Dynamic Testing Rig Facilities

This appendix provides a brief review of dynamic testing rig facilities that have generated a
considerable amount of results on reinforcement and surface support elements employed in
mining, namely the SRK, CanmetMINING, and WASM rigs. For more detailed information on
these rigs, the reader is invited to consult the work of Hadjigeorgiou and Potvin (2007, 2011).
The appendix further covers the Geobrugg testing rig, which was commissioned more recently
with the purpose of testing integrated ground support system. For more information about this
rig, the reader is referred to the work of Roth et al. (2014).

B.1 SRK drop weight test facility

The SRK drop weight test facility (Figure B–1) was located in Johannesburg, South Africa. The
rig was originally designed for testing the performance of surface support elements used in South
African mines but was further modified for testing reinforcement elements as well. The rig
configuration was described by Stacey and Ortlepp (2001, 2004). The dimensions of the tested
pannels were 1.6 m x 1.6 m. Those pannels were hung from the support beams by four 22 mm
diameter cone bolts installed 1 m appart from each other. Yielding reinforcement elements were
used so they could survive extended series of tests, i.e. over 100 drops, without having to be
replaced. The cone bolts were deliberately over-designed so that they would not contribute, by
yielding, towards the energy-absorption capacity of the surface support tested (Stacey & Ortlepp,
2007). Stay ropes anchored into the ground provided extension of the boundary conditions of the
test in an attempt to account for the continuous nature of surface support installed in
underground mines. The stay ropes further prevented the bar frame sides from bending inwards
towards the point of impact (Ortlepp, Swart, & Erasmus, 2004).

The dynamic load was provided by dropping weights onto a ‘simulated rock mass’. Masses of
1,048 kg and 2,706 kg were dropped from a maximum drop height of 3.3 m. The ‘simulated rock
mass’ consisted of steel-encased concrete blocks laid onto, and in direct contact with, the tested
pannel. This setup aimed at distributing the load onto the surface support. The maximum
achieved impact velocity and kinetic energy inputs were 8.1 m/s and 70 kJ/m2.
 296

Figure B–1 Sketch of the SRK drop weight test facility (Stacey & Ortlepp, 2001)

Based on dynamic testing of several surface support elements or systems at the SRK drop weight
facility, Stacey and Ortlepp (2001) have determined the following energy absorbing capacities
(Table B–1). Stacey and Ortlepp (2004) estimated the additional contribution of yielding
reinforcement elements based on good engineering intuition. These observations were
representative of a total support system with rock bolts spaced at 1 m centres.

Weld mesh was characterized by a low energy absorption capacity. Strands had broken in almost
all of the tests conducted by Stacey and Ortlepp (2001), sometimes due to the action of sharp-
edged steel face plates. Diamond, or chain-link, mesh generally performed better than weld
mesh; however, it tended to unravel once a single wire strand had failed. Wire-rope lacing
proved to increase the energy absorption capacity of surface support significantly. Combined
with lacing, weld mesh performed better than diamond mesh with lacing; the weld mesh could
contain the rock mass even though some of the wire strands had failed.
 297

Table B– 1 Energy absorbing capacities and deformation limits of surface support lining component and system
(Stacey & Ortlepp, 2001, 2004)

Liner component/system Input energy Estimated capacity Characteristic

capacity with yielding rock deformation
(kJ) bolts (kJ) limit (mm)
Unreinforced shotcrete 6 6 50
Weld mesh 10 10 100
Diamond mesh 15 15 220
Special wire mesh Varies, say 45 Varies, say 400
35
Weld mesh reinforced shotcrete 15 15 150
Dramix fibre-reinforced shotcrete 20 20 100
Monofilament propylene fibre- 15 15 120
reinforced shotcrete
Weld mesh and wire rope lacing 50 70 400
Diamond mesh and wire rope lacing 35 45 300
Special wire mesh and wire rope 50 70 450
lacing
Fibre-reinforced shotcrete and wire- 35 45 170
rope lacing
Wire mesh and tendon straps 35 50 280
Fibre-reinforced shotcrete and tendon 40 55 220
straps
Wire mesh and yielding tendon straps 45 65 350
Fibre-reinforced shotcrete and 50 70 300
yielding tendon straps
Weld mesh and yielding wire rope >70 >90 450
lacing
Special mesh and yielding tendon 100 125 450
straps
Fibre-reinforced shotcrete and 150 180 500
yielding tendon straps and yielding
wire rope lacing
Special mesh and yielding tendon 150 180 500
straps and yielding wire rope lacing
 298

The SRK drop weight test facility provided consistent and repeatable tests and was relatively
inexpensive to operate. Nevertheless, the instrumentation was quite limited and consisted of
direct tape measurements, photographs taken before and after the drop, and video recordings
(Hadjigeorgiou & Potvin, 2011). The degree of damage to surface support was assessed visually,
e.g. by counting broken wires on steel mesh.

Energy losses that occurred within the SRK testing rig during drop tests had not been quantified
at the time. Testing results from this rig related the support performance to the input energy
instead of the energy being actually imposed on the tested panel. This has important implications
when comparing results with those obtained from other testing facilities. Hadjigeorgiou and
Potvin (2011) approximated the energy transmitted to the surface support as being half the input
energy. Some of the results from the dynamic testing of surface support performed by (Ortlepp &
Stacey, 1997) have been re-interpreted by Hadjigeorgiou and Potvin (2011) using this
assumption (Figure B–2).

Figure B–2 Interpretation of some of the results from the dynamic testing of surface support performed by Ortlepp
and Stacey (1997) (Hadjigeorgiou & Potvin, 2011)
 299

B.2 CanmetMINING
The CanmetMINING rig was originally designed for Noranda Inc. and was primarily used for
testing modified conebolts. Following the closure of the Noranda Technology Centre, the rig was
donated to Canmet Mining and Mineral Sciences Laboratories, now CanmetMINING, a division
of Natural Resources Canada, in Ottawa. Since its transfer to the Canmet’s Bells Corners
Complex, the dynamic testing rig has been significantly upgraded.

The rig consists of two 4.91 m high columns spaced at 1.18 m. The dynamic load is generated by
weight plates that are suspended by an electromagnet, to which power is cut to initiate the drop.
The weights are dropped directly onto an impact plate, from which the energy is transferred to
the tendon installed into a test tube that simulates a borehole (Figure B–3). The rig has a load
capacity of 3 tons and a maximum drop height of 2 m. The maximum impact velocity and impact
energy are 6.3 m/s and 60 kJ, respectively (Doucet, 2012). Loads and displacements are
measured continuously during testing. The loads are measured at the impact plate by an array of
four piezoelectric force sensors. Linescan cameras are used to measure the displacement of the
bolt end and plate at a rate of 10,000 images per second. The system, however, does not account
for system stiffness and energy losses during testing (Hadjigeorgiou & Potvin, 2011).
Nevertheless, it was demonstrated by comparing levels of potential and kinetic energies that the
energy dissipated by friction of the drop weight along the guiding rails was negligible for most
tests (Plouffe, Anderson, & Judge, 2008).

Two methods are used for testing reinforcement elements at CanmetMINING: direct dynamic
impact tests with a continuous tube or indirect dynamic impact tests with a split tube that
simulates a discontinuity across the borehole (Figure B–3b). The rig allows for comparative
analyses of reinforcement elements. Furthermore, it is possible to investigate the influence of
hole diameter, resin type, installation method, and other factors that may affect the dynamic
performance of the tested reinforcement elements (Plouffe et al., 2008). More recently, the rig
has been adapted for testing friction bolts. The simulation of the frictional interface between a
bolt and the rock mass represented a considerable challenge for impact testing of friction bolts.
Calibration results performed on the Swellex bolt were presented by Doucet (2012). The system
can also accommodate impact testing of shotcrete panels.
 300

Figure B–3 CanmetMINING dynamic testing rig (Doucet, 2012)

A summary of the dynamic capacity of reinforcement elements based on drop tests conducted at
CanmetMINING is presented (Table B–2). Those tests were conducted with an impact velocity
varying between 4.5 and 5.4 m/s. The maximum impact energy, which corresponded to the
maximum energy the bolt could be impacted with before failure, was used to characterize the
bolts’ performance instead of the absorbed energy (Doucet & Voyzelle, 2012) . This is because
reinforcement elements behave differently depending on their yielding mechanism and as a
result, some can absorb more energy than others for the same impact. Consequently, the impact
energy was ultimately recommended for design purposes (Doucet & Voyzelle, 2012) .

In summary, the CanmetMINING rig is equipped with sophisticated instrumentation and can
accommodate testing of mechanically anchored, grouted, and to some extent, friction tendons.
The rig also performs impact tests on shotcrete panels. Perhaps most importantly, the rig as well
as the methodology developed by CanmetMINING, has been adopted by the ASTM organization
as a standard method for laboratory determination of rock anchor capacity by drop tests (ASTM
 301

D 7401–08) (Hadjigeorgiou & Potvin, 2011). The CanmetMINING dynamic testing rig is
currently very active and is regularly used by suppliers to test existing bolts as well as products
under development (Hadjigeorgiou & Potvin, 2011).

Table B–2 Dynamic performance of reinforcement elements based on drop tests conducted at CanmetMINING
(Doucet & Voyzelle, 2012)

Support Bolt type Maximum Displacement Average Test

category impact (mm) load configuration
energy (kN)
(kJ)
Mechanical bolt 2.2 43 ± 6 16 Continuous
“Static” support

(φ 14.1 mm, shell F 32 mm) tube

Resin rebar 5 5 160 Continuous
(Type #6 – 20 mm) tube
Resin rebar 14 58 280 Split tube
(Type #6 – 20 mm)
Modified cone bolt 16 160 ± 72 134 ± 16 Continuous
TM
MCB33 tube
Yielding support – plowing, sliding

Fully debonded cone bolt 30 695 ± 68 55 ± 11 Split tube

TM
MCB33FD
Modified cone bolt 16 89 ± 25 155 ± 46 Continuous
TM
MCB38 tube
Roofex Rx8 Dynamic® 34 914 58 ± 4 Continuous
800 mm sliding length tube
Roofex Rx20 Dynamic® 51 783 ± 45 99 ± 7 Continuous
800 mm sliding length tube
Yield-Lok 43 750 95 ± 7 Continuous
750 mm coating tube
D-Bolt (20 mm) 45 187 256 ± 13 Split tube
stretching
support –
Yielding

1500 mm smooth section

D-Bolt (22 mm) 56 225 279 ± 3 Split tube
1500 mm smooth section
 302

B.3 Western Australian School of Mines (WASM)

The Western Australian School of Mines (WASM) dynamic testing facility is located in
Kalgoorlie, Western Australia, and its construction was initiated in 2002. The WASM facility
applies momentum transfer to test grouted bolts, cables, and friction stabilizers under dynamic-
loading conditions (Player, Thompson, & Villaescusa, 2008). The testing arrangement comprises
the three primary components illustrated (Figure 4–31):

• the reinforcement system which includes the reinforcement tendon and surface hardware;

• the simulated ejected rock which includes the collar zone, lower pipe length, and steel
rings; and

• the simulated rock mass which includes the anchor zone, upper pipe length, and the drop
beam.

Figure B–4 Schematic of the testing arrangement used at WASM (Player, Villaescusa, & Thompson, 2004)

Axial loading is generated by dropping the simulated rock mass onto an engineered impact
surface consisting of hydraulic buffers. The hydraulic buffers rapidly decelerate the drop beam
while protecting the foundations of the testing rig. Deceleration of the simulated ejected rock
 303

(steel rings) is controlled by the stiffness of the reinforcement system and hydraulic buffers. The
energy absorbed by the reinforcement system is determined by the relative velocity of the drop
beam compared to the ejected rock. A maximum weight of 4,500 kg can be dropped from a 6 m
height, hence generating a maximum impact velocity of 10 m/s and a kinetic energy input of 225
kJ (Hadjigeorgiou & Potvin, 2011). The standard tests performed at the WASM Dynamic Test
Facility uses 2,000 kg of simulated ejected rock, an impact velocity of 6 m/s, and a resulting
kinetic energy input of 36 kJ (Player et al., 2008).
The WASM facility has been adapted to conduct dynamic tests on mesh panels (Figure B–5).
The procedure used at WASM requires every square of the mesh to be attached to the frame.
Although this practice is not representative of the boundary conditions encountered underground,
it favours the repeatability of the tests (Hadjigeorgiou & Potvin, 2011). Comparative results have
been provided for welded-wire mesh and chain-link mesh (Player, Morton, Thompson, &
Villaescusa, 2008).

Figure B–5 WASM dynamic testing rig adapted for testing mesh panels (Player et al., 2008)
 304

Sophisticated instrumentation is used at the WASM dynamic testing facility to monitor forces,
deformations, accelerations, and strains over different components of the testing arrangement.
Load cells measure forces at the bolt collar and anchor locations. Accelerometers are installed at
strategic locations along the beam and on the simulated ejected rock near the bolt plate to
measure velocities and displacements. Strain gauges attached to the drop beam are used to
determine compressive and tensile strains in the beam as a result of the dynamic load. An
ultrasonic motion sensor measures the compression of the buffers during deceleration of the
beam. Displacements along the centerline of the drop are assessed by processing high-speed
digital video records. Data are recorded electronically to facilitate reviewing, filtering, and
analysis over selected time intervals.

B.4 Geobrugg
The Geobrugg dynamic-testing facility is located in Walenstadt, Switzerland, and was
commissioned in 2012 (Figure B–6). One of the motivations for building this facility was to
provide more data on the interaction between the reinforcement, the surface support, and
connection elements (Roth, Cala, Brändle, & Rorem, 2014). Hence, the Geobrugg rig
accommodates full-scale dynamic tests on integrated ground support systems as opposed to the
individual elements.

Figure B–6 The Geobrugg dynamic testing facility in Walenstadt, Switzerland (Roth et al., 2014)
 305

The rig consists of a metal frame with high tensile chain link mesh horizontally connected to it
by lacing wire rope. An unreinforced concrete slab is poured over the mesh, separated by a
plastic foil in order to keep the two surface support elements separated from each other. Boulders
and gravel are placed over the concrete slab in order to simulate a fractured rock mass. The
surface support setup is held in place by four dynamic rockbolts (D-Bolts) grouted into steel
tubes (Roth et al., 2014). A load distribution plate is mounted over the gravel layer to transfer the
energy impact induced by a free-falling test block. A 6,280 kg test block lifted at a maximum
height of 3.25 m delivers the impact energy to the support system.

The rig is highly instrumented using load cells and high-speed cameras. The load cells are
installed on the toe and collar of one of the dynamic bolts and provide data on the load
transferred from the impact to the bolts. One of the cameras records the impact and provides
information on the impact velocity whereas the second camera records deformations of the mesh
and bolts.

The Walenstadt test setup proved to be suitable to quantify the overall performance of an
integrated dynamic ground support system. The work from Roth et al. (2014) has demonstrated
that the load-distribution between the mesh and the bolts is dependent on the stiffness of the
shotcrete/rock assembly. Consequently, the testing procedure needs to be fine-tuned in order to
simulate different ground conditions. Very few support systems have been tested at the
Geobrugg facility at the current time. Therefore, the application of the results generated at this
facility is still limited.

B.5 References
Doucet, C. (2012). Ground support research at CanmetMINING. Dynamic Ground Support
Applications Symposium, Sudbury, ON.

Doucet, C., & Voyzelle, B. (2012). Technical information data sheets. (technical report). Ottawa
(ON): CanmetMINING.

Hadjigeorgiou, J., & Potvin, Y. (2011). A critical assessment of dynamic rock reinforcement and
support testing facilities. Rock Mechanics and Rock Engineering, 44(5), 565-578.

Ortlepp, W. D., & Stacey, T. R. (1997). Testing of tunnel support: Dynamic load testing of rock
support containment systems. (No. GAP Project 221).SIMRAC.
 306

Ortlepp, W. D., Swart, A. H., & Erasmus, P. N. (2004). The performance of yielding tunnel
support systems under dynamic loading. In Y. Potvin, T. R. Stacey & J. Hadjigeorgiou
(Eds.), Surface support in mining (pp. 409-416). Perth, WA: Australian Centre for
Geomechanics.

Player, J. R., Morton, E. C., Thompson, A. G., & Villaescusa, E. (2008). Static and dynamic
testing of steel wire mesh for mining applications of rock surface support. Proceedings of
the sixth international symposium on ground support in mining and civil engineering
construction (pp. 693-706). Cape Town: South African Institute of Mining and Metallurgy.

Player, J. R., Thompson, A. G., & Villaescusa, E. (2008). Dynamic testing of reinforcement
systems. In T. R. Stacey, & D. F. Malan (Eds.), Proceedings of the sixth international
symposium on ground support in mining and civil engineering construction (pp. 597-622).
Cape Town (SA): Southern African Institute of Mining and Metallurgy.

Player, J. R., Villaescusa, E., & Thompson, A. G. (2004). Dynamic testing of rock reinforcement
using the momentum transfer concept. In E. Villaescusa, & Y. Potvin (Eds.), Proceedings of
the fifth international symposium on ground support (pp. 327-340). Leiden: Balkema.

Plouffe, M., Anderson, T., & Judge, K. (2008). Rock bolts testing under dynamic conditions at
CANMET-MMSL. In T. R. Stacey, & D. F. Malan (Eds.), Proceedings of the sixth
international symposium on ground support in mining and civil engineering construction
(pp. 581-596). Cape Town (SA): Southern African Institute of Mining and Metallurgy.

Roth, A., Cala, M., Brändle, R., & Rorem, E. (2014). Analysis and numerical modelling of
dynamic ground support based on instrumented field-scale tests. In M. R. Hudyma, & Y.
Potvin (Eds.), Proceedings of the 7th international seminar on deep and high stress mining
(pp. 151-164). Perth, WA: Australian Centre for Geomechanics.

Stacey, T. R., & Ortlepp, W. D. (2001). Tunnel surface support – capacities of various types of
wire mesh and shotcrete under dynamic loading. The Journal of the South African Institute
of Mining and Metallurgy, 101(7), 337-342.

Stacey, T. R., & Ortlepp, W. D. (2007). Yielding rock support – the capacities of different types
of support, and matching of support type to seismic demand. In Y. Potvin, J. Hadjigeorgiou
& T. R. Stacey (Eds.), Challenges in deep and high stress mining (pp. 407-411). Perth, WA:
Australian Centre for Geomechanics.
 307

Appendix C:
Evolution of Mining-Induced Stresses within Creighton Mine Deep

The numerical modelling results presented in this appendix include the differential-stress
profiles, seismicity, and rockbursts (damage locations) for each pair of years between 2000 and
2013 at Creighton Mine. Four deep and mature mine levels were represented in this analysis:
7200 L (2,195 m), 7400 L (2,255 m), 7530 L (2,295 m), and 7680 L (2,340 m). The mining-
induced stresses were obtained using Map3D, a linear-elastic boundary element stress analysis
package. The seismicity and rockbursts plotted over the differential stresses consisted of the
events that were localized between the mine levels immediately above and below the studied
level. For example, seismic events that were localized between the 7200 level (2,195 m) and
7530 level (2,295 m) were plotted and compared to the mining-induced stresses on the 7400
level (2,255 m).

Generally, the model was able to capture the high stress zones in the footwall with
relatively good agreement with the observed seismicity, particularly on the east side of the 461
orebody, where most of the mine development is located. By considering both the differential-
stress contours generated for each mine level in 2013 and the seismic events from 2000 to 2013,
it was observed that, seismicity occurred predominantly in areas where the differential stress was
greater than 60 MPa, or 0.25 times the UCS of the footwall granite (240 MPa). Nevertheless, it is
evident that the model did not capture all the mechanisms that lead to seismic activity in the
mine’s footwall, particularly with respect to the shear zones. It is common on every mine level to
observe seismicity diverging from the differential-stress contours and being controlled by
geological structures. The main observations for each mine level are summarized:
• 7200 level: The majority of seismic events plotted in the immediate footwall of the Plum
shear;
• 7400 level: The majority of seismic events occurred between the 461 and Grizzly shear
zones;
• 7530 and 7680 levels: The majority of seismic events occurred between the 461 and Plum
shear zones and along the Grizzly shear zone.
 308

The collected seismic data demonstrated that at Creighton Mine, large-magnitude seismic events
were strongly related to shear zones as events larger than 2.0 MN tended to plot closer to these
structures. Although the presence of seismicity could not be simply explained using the
differential stresses contours, the majority of seismically-active shear zones were those located in
the vicinity of the high differential-stress zone. The model therefore demonstrated that seismic
activity in Creighton Mine’s footwall is driven by the presence of high differential stresses and
further enhanced by the presence of shear zones.

It has been argued that shear zones, especially those of the southwest striking family, contributed
to seismicity indirectly by re-aligning stresses in the footwall of the 7400 level (Snelling et al.,
2013). This perception was not supported by the results of the current analysis. The 3D
numerical stress analysis in the current study resulted in differential-stress contours similar to
those generated by the 2D analysis conducted by Snelling et al. (2013), on the 7400 level. Since
the shear zones in the Map3D model are characterized by the same elastic behaviour as the host
rock, this similitude indicates that stress redistributions in the footwall are primarily driven by
the mine geometry, as opposed to geological structures. This conclusion highlights the
importance of adequately representing the mine geometry in numerical stress analyses and
validates further the selection of a 3D numerical package for this study.
 309

C.1 Mining sequence

 310

C.2 7200 level

2000 – 2001

2002 – 2003

2004 – 2005
 311

2006 – 2007

2008 – 2009

2010 – 2011
 312

2012 – 2013

2000 – 2013
 313

C.3 7400 level

2000 – 2001

2002 – 2003

2004 – 2005
 314

2006 – 2007

2008 – 2009

2010 – 2011
 315

2012 – 2013

2000 – 2013
 316

C.4 7530 level

2000 – 2001

2002 – 2003

2004 – 2005
 317

2006 – 2007

2008 – 2009

2010 – 2011
 318

2012 – 2013

2000 – 2013
 319

C.5 7680 Level

2000 – 2001

2002 – 2003

2004 – 2005
 320

2006 – 2007

2008 – 2009

2010 – 2011
 321

2012 – 2013

2000 – 2013