Documente Academic
Documente Profesional
Documente Cultură
Rock Mines
by
Luke Nicholson
Luke Nicholson
Masters of Applied Science
Graduate Department of Civil Engineering
University of Toronto
2016
Rock bolts are the principal reinforcement element of many underground support systems. This thesis
investigates and characterizes the behaviour and performance of rock bolts as measured by a pull test.
A database composed of 985 pull tests from six mines in the Sudbury Basin was assembled. Procedures
and apparatuses used to conduct these tests were compared to ASTM’s standards and ISRM’s suggested
methods. The results from the pull tests were used to compare the behaviour of reinforcement elements
with theoretical models and to quantify performance metrics and their distributions. The influence of
bolt, installation and rock mass parameters on the performance of certain rock bolts was investigated,
and distributions of expected behaviour were constructed. These may be used in the design of hard rock
underground excavations using methodologies that incorporate both the load capacity and displacement
ii
Acknowledgements
I would like to thank my supervisor, Professor John Hadjigeorgiou, for the guidance and educational
experience I have had the pleasure to experience over the last two years.
This project would not have been possible without the funding and active participation by Vale. In
particular, I would like to thank Dr. Mike Yao, Lindsay Moreau-Verlaan, Derek Boucher and the rest
of the ground support staff at Vale’s Sudbury operations for the support they provided me.
I also extend my gratitude to ground support suppliers who provided technical information on test-
ing, including Mansour Mining Technologies Inc, Jennmar Canada, DSI, Normet and Atlas Copco. In
particular, Francois Charette, Lynn Mainville-Beach and Bryan Lamothe.
Thank you to my friends and research associates Marie-Helene Fillion, Philippe Morissette and
Stratos Karampinos for their education, support and generally putting up with me.
I am very grateful to all my friends home and abroad for being there for me when I needed it and
generally improving my life.
Last but certainly not least, I thank my family for their advice and unlimited support on every step
of my journey.
iii
Contents
1 Introduction 1
1.1 Problem Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Significance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.4 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.5 Structure of the Thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
iv
3.2.5 Expandable Bolts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.3 Database . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.3.1 Database Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.3.2 Comparison to Other Pull Test Databases . . . . . . . . . . . . . . . . . . . . . . . 26
3.3.3 Specific and General Limitations of the Pull Test Database . . . . . . . . . . . . . 26
3.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
v
5.6.2 Observed Behaviour of Expandable Bolts . . . . . . . . . . . . . . . . . . . . . . . 69
5.6.3 Characterization of Performance Metrics for Expandable Bolts . . . . . . . . . . . 71
5.7 Other Reinforcement Elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
5.7.1 Yield-Lok . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
5.7.2 Fibreglass Rebar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
5.7.3 DS Bolt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
5.7.4 Other Expandable Bolts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
5.7.5 MD Bolt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
5.8 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
vi
7.4.2 Characterisation of Modified Cone Bolt Behaviour . . . . . . . . . . . . . . . . . . 125
7.5 D-Bolt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
7.5.1 Characterisation of D-Bolt Performance . . . . . . . . . . . . . . . . . . . . . . . . 127
7.5.2 Characterisation of D-Bolt Behaviour . . . . . . . . . . . . . . . . . . . . . . . . . 129
7.6 Expandable bolts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
7.6.1 Characterisation of Expandable Bolt Performance . . . . . . . . . . . . . . . . . . 132
7.6.2 Characterisation of Expandable Bolt Behaviour . . . . . . . . . . . . . . . . . . . . 133
7.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
8 Conclusions 140
8.1 Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
8.2 Load Capacities of Reinforcement Elements . . . . . . . . . . . . . . . . . . . . . . . . . . 141
8.3 Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
8.4 Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
8.5 Implications and Path Forward . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
Appendices 151
vii
List of Figures
3.1 Map of Sudbury area showing locations of relevant mines (after Eckstrand & Hulbert, 2007) 14
3.2 Longitudinal section of Coleman Mine (Morissette et al, 2014) . . . . . . . . . . . . . . . . 16
3.3 Longitudinal section of Copper Cliff Mine (Chinnasane et al, 2014) . . . . . . . . . . . . . 17
3.4 Longitudinal section of Creighton Mine (Snelling et al, 2013) . . . . . . . . . . . . . . . . 17
3.5 Longitudinal section of Stobie Mine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.6 Longitudinal section of Totten Mine (After Sudbury Platinum Corporation, 2015) . . . . 19
3.7 FRS A schematic (Courtesy of Supplier A) . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.8 FRS B schematic (Courtesy of Supplier B) . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.9 Schematics of rebar manufactured by Supplier C (top; courtesy of Supplier C), Supplier
A (middle; courtesy of Supplier A) and Supplier B (bottom; courtesy of Supplier B) . . . 21
3.10 MCB33 (Courtesy of Mansour) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.11 D-Bolt schematic (Normet, 2014) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.12 VersaBolt schematic (Courtesy of Mansour) . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.13 Schematic of a Python bolt, and cross sections before (1) and after (2) inflation (Courtesy
of Jennmar) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
viii
4.4 Determination of working capacity from a pull test . . . . . . . . . . . . . . . . . . . . . . 34
4.5 Measurement of displacement for a pull test on a generic reinforcement element . . . . . . 35
4.6 Generic reinforcement system (Thompson et al, 2012) . . . . . . . . . . . . . . . . . . . . 35
4.7 Measurement of displacement for axial loading of a point anchored rock bolt . . . . . . . . 36
4.8 Design of support systems using the ground reaction curve (Brady & Brown, 2006) . . . . 37
4.9 Calculation of secant and tangent stiffness from a pull test . . . . . . . . . . . . . . . . . . 38
4.10 Pull test campaign results for partially encapsulated Rebar A from November 16th , 2012
at Coleman Mine (Mainville et al, 2012) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
5.1 Shear stress distribution along a frictionally coupled bolt subject to axial load (Li &
Stillborg, 1999) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
5.2 Shear stress and axial load along a Swellex rock bolt (Li & Stillborg, 1999) . . . . . . . . 44
5.3 Pull test performed on an FA39 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
5.4 Ultimate capacity per unit length distributions for FRSs with nominal diameters of 35,
39 and 46 mm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
5.5 Stiffness metrics for an FRS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
5.6 Stiffness distributions for the FA35 and FA39 . . . . . . . . . . . . . . . . . . . . . . . . . 49
5.7 Model of the shear stress profile in a grouted rock bolt (Li & Stillborg, 1999) . . . . . . . 50
5.8 Model of tensile load and shear stress profile for a rebar (Li & Stillborg, 1999) . . . . . . 51
5.9 Pull test performed on a 20 mm Rebar B . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
5.10 Pull test performed on a 20 mm Rebar A . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
5.11 Working capacity distributions for rebar . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
5.12 Secant stiffness for 20 mm rebar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
5.13 Tangent stiffness for 20 mm Rebar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
5.14 Comparison of Rebar A unloading stiffness to secant (a) and tangent (b) stiffness . . . . . 56
5.15 Laboratory pull test of an MCB33 (Simser et al, 2006) . . . . . . . . . . . . . . . . . . . . 57
5.16 Conceptual load–displacement behaviour of a cone bolt subject to quasi-static loading
(Simser et al, 2006) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
5.17 Pull test performed on an MCB33 with plough . . . . . . . . . . . . . . . . . . . . . . . . 58
5.18 Pull test performed on an MCB33 without a linear plough response . . . . . . . . . . . . . 58
5.19 Close-up of the bolt response shown in Figure 5.18 . . . . . . . . . . . . . . . . . . . . . . 59
5.20 Amended conceptual load–displacement behaviour of a cone bolt . . . . . . . . . . . . . . 60
5.21 Performance metrics measured from a cone bolt pull test . . . . . . . . . . . . . . . . . . . 60
5.22 Load metric distributions for the MCB33 . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
5.23 Stiffness metric distributions for the MCB33 . . . . . . . . . . . . . . . . . . . . . . . . . . 62
5.24 Apparatus for a simulated joint laboratory test on a D-Bolt (Li, 2012) . . . . . . . . . . . 63
5.25 Results of simulated joint laboratory tests on 20 mm D-Bolts (Li, 2012) . . . . . . . . . . 63
5.26 Pull tests performed on 22 mm D-Bolts . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
5.27 Pull tests performed on 20 mm D-Bolts . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
5.28 Pull tests performed on 22mm D-Bolts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
5.29 Distributions of stiffness for 20 and 22 mm D-Bolts . . . . . . . . . . . . . . . . . . . . . . 68
5.30 Pull tests performed on Pm12 and Mn12 expandable bolts . . . . . . . . . . . . . . . . . . 69
5.31 Working capacities of Swellex Pm12 and Mn12 . . . . . . . . . . . . . . . . . . . . . . . . 71
5.32 Secant stiffness of Swellex variants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
ix
5.33 Secant stiffness of Swellex sorted by installation medium . . . . . . . . . . . . . . . . . . . 73
6.1 Relationship between ultimate capacity and length for the FA35 . . . . . . . . . . . . . . 79
6.2 Relationship between ultimate capacity and length for the FB35 . . . . . . . . . . . . . . 80
6.3 Relationship between ultimate capacity and length for the FB39 . . . . . . . . . . . . . . 80
6.4 Relationship between ultimate capacity and anchorage length for all FRS bolts . . . . . . 82
6.5 Comparison of ultimate capacities between jackleg and bolter installations of the FA35 . . 83
6.6 Comparison of ultimate capacities between jackleg and bolter installations of all FRS bolts 84
6.7 Relationship between drive time and ultimate capacity for all FRS bolts . . . . . . . . . . 85
6.8 Relationship between drive time and ultimate capacity for FB35, distinguishing between
installation methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
6.9 Relationship between drive time and absolute ultimate capacity for all FRS configurations
installed using a bolter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
6.10 Relationship between installation time and ultimate capacity for all bolter-installed FRS
fit with linear and power functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
6.11 Relationship between drill bit diameter and ultimate capacity for all FRS configurations . 89
6.12 Relationship between drill bit diameter and ultimate capacity for two testing campaigns
performed on the FB46 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
6.13 Relationship between drill bit diameter and ultimate capacity for FA35 and FB35, sepa-
rated by installation method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
6.14 Bolt diameter to drill bit diameter ratios for all FRS variants . . . . . . . . . . . . . . . . 91
6.15 Relationship between bolt diameter to drill bit diameter ratio and ultimate capacity . . . 92
6.16 Relationship between bolt diameter to drill bit diameter ratio and ultimate capacity for
all FRS configurations by testing campaign . . . . . . . . . . . . . . . . . . . . . . . . . . 93
6.17 FRS ultimate capacity by lithology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
6.18 Relationship between UCS and ultimate capacity . . . . . . . . . . . . . . . . . . . . . . . 95
6.19 Distributions of FRS bolts installed in ore and waste rock normalized to the campaign
average ultimate capacity for bolts installed in waste rock . . . . . . . . . . . . . . . . . . 96
6.20 Ultimate capacities recorded for FA46 bolts installed in poor and good quality ground . . 97
6.21 Ultimate capacities recorded for FRSs installed in poor and good quality ground . . . . . 98
6.22 FA39s pulled at Garson, 9/20/2013 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
6.23 Rebar performance metrics by length . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
6.24 Relationships between stiffness and grout length : rebar length for Suppliers A and B . . 102
6.25 Unloading stiffness and grout length : rebar length for Rebar A . . . . . . . . . . . . . . . 103
6.26 Relationship between stiffness and resin spin time for Rebar A . . . . . . . . . . . . . . . 104
6.27 Stiffness comparison of Rebar A installed on the day of testing versus previously . . . . . 104
6.28 Stiffness comparison between lithologies for rebar . . . . . . . . . . . . . . . . . . . . . . . 105
6.29 Performance comparison of MCB33s installed prior to and on the day of testing . . . . . . 107
6.30 Performance comparison of MCB33 bolts installed in ore, igneous/metaigneous and metased-
imentary lithologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
6.31 Relationship between plough stiffness and plough point for the MCB33 . . . . . . . . . . . 110
x
7.2 Distributions of secant stiffness and ultimate capacity for a pull test on an FRS with an
anchorage length of 1.52 m . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
7.3 Ultimate capacity per metre distributions for all FRS configurations . . . . . . . . . . . . 116
7.4 Distribution of secant stiffness and ultimate capacity for all FRS bolts tested . . . . . . . 116
7.5 All FA35 and FA39 pull test load-displacement relationships . . . . . . . . . . . . . . . . . 117
7.6 Distribution of displacement measured during pull testing of FA35 and FA39 with anchor-
age lengths of 1.52 m . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
7.7 Conceptual displacement distribution of FRSs with anchorage lengths of 1.52 m subject
to a pull test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
7.8 Conceptual displacement distribution of pull tests performed on FRSs . . . . . . . . . . . 119
7.9 Distributions of secant stiffness and working capacity for a pull test with a pre-load of
17.8 kN on a rebar rock bolt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
7.10 All rebar pull test load-displacement relationships . . . . . . . . . . . . . . . . . . . . . . . 122
7.11 Distribution of displacement measured during pull testing of 20 mm rebar 1.8 m to 2.4 m
in length with a pre-load of 17.8 kN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
7.12 Distribution of displacement measured during pull testing of 20 mm rebar 1.8 m to 2.4 m
in length without a pre-load . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
7.13 Conceptual distribution of displacement during pull testing of 20 mm rebar without a
pre-load . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
7.14 Distributions of secant stiffness and working capacity for a pull test with a pre-load of
17.8 kN on an MCB33 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
7.15 56 MCB33 pull tests collected from Vale’s Sudbury operations . . . . . . . . . . . . . . . . 125
7.16 Distribution of displacement measured during pull testing of a 2.4 m MCB33 with a
pre-load of 17.8 kN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
7.17 Conceptual distribution of displacements for a pull test of a 2.4 m MCB33 without a
pre-load . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
7.18 Distribution of secant stiffness for a pull test with a pre-load of 17.8 kN on a 20 mm D-Bolt128
7.19 Distribution of secant stiffness for a pull test with a pre-load of 17.8 kN on a 22 mm D-Bolt128
7.20 20 mm D-Bolt pull tests collected from Vale’s Sudbury operations . . . . . . . . . . . . . 129
7.21 22 mm D-Bolt pull tests collected from Vale’s Sudbury operations . . . . . . . . . . . . . 129
7.22 Load–displacement behaviour of a 20 mm D-Bolt with a pre-load of 17.8 kN . . . . . . . . 130
7.23 Load–displacement behaviour of a 22 mm D-Bolt with a pre-load of 17.8 kN . . . . . . . . 130
7.24 Conceptual distribution of displacements for a pull test of a 20 mm D-Bolt without a
pre-load . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
7.25 Conceptual distribution of displacements for a pull test of a 22 mm D-Bolt without a
pre-load . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
7.26 Distributions of secant stiffness and working capacity for a pull test on Swellex Pm12 and
Mn12 without a pre-load . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
7.27 Swellex Pm12 and Mn12 pull tests collected from Vale’s Sudbury operations . . . . . . . . 133
7.28 Swellex Pm24 and Mn24 pull tests collected from Vale’s Sudbury operations . . . . . . . . 134
7.29 Load–displacement behaviour of Pm12 and Mn12 bolts . . . . . . . . . . . . . . . . . . . . 134
7.30 Load-displacement behaviour of Pm24 and Mn24 bolts . . . . . . . . . . . . . . . . . . . . 135
7.31 Conceptual load–displacement behaviour of Pm12 and Mn12 bolts subject to a pull test . 135
xi
7.32 Conceptual load–displacement behaviour of Pm24 and Mn24 bolts subject to a pull test . 136
7.33 Conceptual load–displacement behaviour extrapolated to 0 load for various rock bolts
subject to a pull test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
7.34 Median load–displacement behaviour for all bolts with no pre-load . . . . . . . . . . . . . 138
A.1 Rock bolt pull test sample form (ASTM D4435, 2013) . . . . . . . . . . . . . . . . . . . . 152
A.2 Rock bolt pull test data sheet (After ISRM, 1981) . . . . . . . . . . . . . . . . . . . . . . 153
A.3 Proposed pull test campaign information sheet . . . . . . . . . . . . . . . . . . . . . . . . 155
A.4 Proposed pull test data sheet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
xii
List of Tables
6.1 Statistics on the relationship between ultimate capacity and length for the FA35 . . . . . 79
6.2 Statistics on the relationship between ultimate capacity and length for the FB35 . . . . . 80
6.3 Statistics on the relationship between ultimate capacity and length for the FB39 . . . . . 81
xiii
6.4 Single factor ANOVA performed on the relationship between ultimate capacity and length
for the FB39 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
6.5 Breakdown of bolts contributing to major sets of anchorage length . . . . . . . . . . . . . 81
6.6 Comparison of 1.52 m and 1.83 m of anchorage length for all FRS bolts . . . . . . . . . . 82
6.7 Ultimate capacity statistics for jackleg and bolter installations of the FA35 . . . . . . . . 83
6.8 Ultimate capacity statistics for jackleg and bolter installation of all FRS bolts . . . . . . . 84
6.9 Description of relationships between drive time and ultimate capacity for FRSs installed
with a MacLean Bolter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
6.10 Summary statistics for bolt diameter to drill bit diameter ratios for all FRS variants . . . 91
6.11 Comparison of average ultimate capacities for pull tests performed in ore and waste rock
in the same campaign . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
6.12 Comparison of FRS bolts installed in ore and waste rock . . . . . . . . . . . . . . . . . . . 96
6.13 Comparison of FA46 bolts installed in poor and good quality ground . . . . . . . . . . . . 97
6.14 Comparison of all FRSs installed in poor and good quality ground . . . . . . . . . . . . . 99
6.15 Comparison of partial and full encapsulation test statistics for Rebar A and B . . . . . . . 102
6.16 Statistics regarding residence time for Rebar A . . . . . . . . . . . . . . . . . . . . . . . . 104
6.17 Comparison of stiffness across different lithologies . . . . . . . . . . . . . . . . . . . . . . . 106
6.18 Performance comparison of MCB33s installed prior to and on the day of testing . . . . . . 107
6.19 Performance comparison of MCB33 bolts installed previously and on the day of testing . 108
6.20 Summary of observed relationships of between various factors and performance indicators
for each rock bolt type . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
xiv
Chapter 1
Introduction
1
Chapter 1. Introduction 2
1.2 Significance
The appropriate use of rock bolts in underground mines represents a cost-benefit problem. Over–design
of ground support systems may lead to inefficiencies in labour and equipment usage, inflating costs and
cycle times. Under–design may result in serious safety concerns and production stoppages. An important
step in the design and optimization of a ground support system is the selection of appropriate elements
and systems for a set of conditions. The performance that may be expected from a particular element
given a set of installation parameters and its associated variability should be recognised by those who
design the support system. With improvements in the understanding, measurement, and quantification
of performance, safer and more cost–effective ground support systems may be developed.
1.3 Objectives
This thesis aims to provide information that can be used for design purposes based on in situ pull
tests rather than laboratory results. There are three principal objectives of this thesis. The first is
to develop an understanding of the load–displacement behaviour of various rock bolts, and use it to
interpret pull test data. The second is to identify relationships between recorded pull test parameters
and the performance of the rock bolt. The third is to develop input parameters in terms of bolt load
capacity and displacement that can be used for design purposes in hard rock mines.
1.4 Methodology
The progression of this thesis can be broadly summarized in the following sequence:
1. Data Collection: Pull test campaign reports from several rock bolt suppliers at six Vale mine sites
in the Greater Sudbury area were collected, and their results and all recorded testing parameters
input into a database. The majority of the pull tests used for this analysis were performed between
2011 and 2014.
2. Data Analysis: Statistical methods including linear regression and analysis of variance (ANOVA)
amongst others were used to quantify the performance of the rock bolts tested, as well as investigate
the influence of various factors on bolt performance.
3. Development of Performance Guidelines: Load–displacement data from individual pull tests was
analysed and combined to create performance envelopes for various rock bolts.
Chapter 1: Introduction —- The need for data regarding rock bolt capacity is addressed. The method-
ology used to undertake the thesis is laid out, and the thesis structure stated.
Chapter 2: Testing of rock bolts — A review of rock bolts used in underground hard rock mines is pre-
sented. Standardized testing methods used to characterize rock bolt behaviour are identified, including
Chapter 1. Introduction 3
Chapter 3: Composition of the database — The mines from which the pull tests were collected, the rock
bolts tested, and the database used for analysis are described.
Chapter 4: Review of implemented pull test methods — Typical pull test methodologies and apparatuses
used in the constructed database are described. Deviations from standards and suggested methods are
identified. The information obtained from a pull test is discussed.
Chapter 5: Summary statistics and interpretation of pull test data — Methods of data analysis are
described. A theoretical understanding of rock bolt behaviour is used to interpret pull tests, and per-
formance metrics are quantified.
Chapter 6: Factors influencing pull test performance —- The influence of a variety of individual factors
on rock bolt performance are examined and discussed.
Chapter 8: Conclusions — The main conclusions of this thesis are presented. Contributions and limita-
tions of the current work are identified, and recommendations are made for future work.
Chapter 2
In this chapter, rock bolts are categorized, and various laboratory and field testing methods applicable
to rock bolts are described. While the focus of this thesis is on in situ pull tests (described in Section
2.3), a comparison between testing methods targeting a well-defined set of material properties and a test
more representative of in situ performance is of value.
Mechanical
Bolts
Resin Bolts
Paddle bolt
Posimix
Solid deformed
Wriggle bolt
Chemical
anchor
Friction bolts
Split set
Swellex
4
Chapter 2. Testing of Rock Bolts 5
Mechanical bolts are steel tendons fitted with an expansion shell that anchors the bolt at the toe of
the hole, pressing against the sides of the hole as the bolt tendon is torqued and tensioned. Grouted bolts,
such as rebar, may use either a cement or resin grout to bond the bolt to the rock mass continuously
along its length. Changes in the pattern on the bolt, bolt shape (e.g. paddle bolts) and fixtures on the
bolt (e.g. Posimix) are used to more effectively mix the grout and/or increase the strength of the bolt–
resin interface. A friction bolt also transmits load from the bolt to the rock mass continuously along its
length, but does so using frictional resistance. An example is the friction rock stabilizer (FRS), composed
of a steel tube with a gap in the circumference running along its length, tapered on one end and with a
steel ring welded to the other (or alternatively the bolt’s head is crimped). An FRS is installed in a drill
hole with a diameter slightly smaller than that of the bolt, compressing the bolt radially and generating
forces that result in frictional resistance to pull out. The Swellex bolt is another example of a friction
bolt. In recent years since the expiration of Atlas Copco’s patent, several similar bolts from different
manufacturers have become available. These are commonly referred to as expandable or inflatable bolts.
Expandable bolts are rock bolts that are expanded using pressure exerted by a water pump once the
bolt is placed loosely in a hole. Resistance to pull out is not only frictional, but is also attributed to
mechanical interlock between any undulations in the sides of the drill hole and the resulting shape of
the inflated bolt (Hadjigeorgiou & Charette, 2001).
More recently, a distinction is made for yielding or energy-absorbing rock bolts. While these are often
grouted bolts, they are specifically designed to address stability problems caused by high stress, such as
rock bursting or squeezing (Li et al, 2014). Yielding bolts generally have larger deformation and energy
capacities than bolts only intended for use in static loading scenarios. This is usually achieved through
one of two mechanisms. The first is an increase in uncoupled tendon length between two or more anchor
points. An example of a bolt that uses this mechanism is the “D-Bolt” (available from Normet), where
a smooth bar deforms between anchors set in grout. The second mechanism is movement of the bolt or
a section of the bolt relative to an internal fixture or the drill hole; for example a “Modified Cone Bolt”
(available from Mansour) ploughing through resin (Thompson et al, 2012).
The yield point and the yield strength of the test sample may be determined using a variety of methods
during the test or post-processing. Subsequent to the yield of the bolt, its tensile strength is determined,
as well as elongation, elongation at fracture and reduction of area.
Figure 2.2: Tensile test specimen with a reduced section (ASTM A370, 2012)
This particular test is a recent addition and is applicable to what ASTM refers to as “friction stabilizers”,
synonymous with FRS. An FRS bolt is installed in a test plate made flush against the head of the bolt
while maintaining clearance between the circumference of the plate hole and the FRS (i.e. the bolt is
loose in the plate). The other end of the bolt is gripped and plugged so as to be held in place (Figure 2.3).
The bolt is then tensioned to the minimum ultimate load specified by the standard for the applicable
nominal bolt diameter without failure of the plate or any visual destruction of the bolt head (ASTM
F432, 2013).
Figure 2.3: Apparatus for a tensile test on an FRS (ASTM F432, 2013)
yield strength is obtained using the “drop of the beam” method (ASTM D4435, 2013), which designates
the yield strength as the point where either a drop or a plateau in load is registered while deformation of
the bolt continues. Tensioning progresses in order to determine the wedge tensile strength of the sample,
while also demonstrating head quality and ductility of the product (ASTM F606, 2013).
Figure 2.4: Bolt head configuration for a wedge tension test (ASTM F606, 2013)
Expansion Test
The expansion test entails fully expanding the product at the manufacturer’s recommended pressure.
Ferrule Test
The tensile strength of the ferrule-to-bolt weld must be found to meet or exceed the bolt profile breaking
load. This is verified by gripping the ferrule on one end of the bolt and the other, plugged end of the
Chapter 2. Testing of Rock Bolts 8
bolt itself (Figure 2.5), and applying tension. The specific test methodology must be supplied to the
customer.
Figure 2.5: Two alternative configurations for the ferrule test (ASTM F432, 2013)
A fully inflated bolt (it is permissible to use the bolt subject to the expansion test) is tensioned in
order to find the yield and ultimate tensile loads. These must reach the specifications laid out in ASTM
Standard A370-12.
Figure 2.6: Apparatus for the Laboratory Rock Anchor Capacity Pull Test (ASTM D7401, 2008)
Chapter 2. Testing of Rock Bolts 9
Figure 2.7: Apparatus for the Laboratory Drop Test (ASTM D7401, 2008)
ASTM D4435-13: Standard Test Method for Rock Bolt Anchor Pull Test
ASTM D4435-13 dictates that when developing a rock bolt pull test program, the program should reflect
all possible installations of bolt, such as rock unit, orientation relative to any anisotropy present in the
rock mass, anchor configurations, etc. 10 to 12 tests are recommended for each set of variables.
The apparatus described in the standard (Figure 2.8) includes a loading system with sufficient ca-
pacity to fail the bolt and induce at least 50 mm of displacement. Additionally, it should be compatible
with any rock surface condition and not deviate more than 5◦ from the bolt axis. Either a load cell on
the bolt or a pressure gauge on the ram is to be used to measure applied load, and displacement is to
Chapter 2. Testing of Rock Bolts 10
be measured to an accuracy of at least 0.025 mm and resolution of 0.013 mm. The displacement trans-
ducer must be supported from a point no closer than 0.9 m from the loading system if attached to the
same rock face, or alternatively from the floor. A borehole diameter measuring gauge with a resolution
of at least 0.25 mm is to be used to measure the drilled hole diameter at the location of the anchor,
and a thermometer used to record temperature at a resolution of 0.5◦ C. The bolt, anchor and drilling
equipment are to be used per typical operational procedure, although bolts are not to be tensioned.
Figure 2.8: Apparatus for a rock bolt anchor pull test (ASTM D4435, 2013)
At least half of the bolts tested are to have three loading/unloading cycles performed to 1/4, 1/2 and
3/4 of the estimated failure load. For these cycles, 10 equal load increments are to be used, and load is to
be applied smoothly and rapidly. These bolts are then pulled until failure in either the same increments
as the previous cycle, or increments of 2.2 kN, whichever is less. Non-cycled bolts are tested until failure
in either 20 equal load increments, or increments of 2.2 kN, whichever is less. Displacement is recorded
to the nearest 0.013 mm after each loading increment. Bolts are to be pulled 12.7 mm beyond the failure
displacement, defined either by the peak load or 12.7 mm of recorded displacement. After failure occurs,
load is to be recorded every 1 mm of displacement, as opposed to recording at load intervals (ASTM
D4435, 2013).
As per ASTM D4435-13, the stress on the bolt is to be calculated, as well as the corrected bolt head
displacement (total displacement discounting elastic deformation of the bolt). Working and ultimate
capacities (Figure 2.9) are to be interpreted from the resulting load–displacement graph. Working
capacity is defined as “the load on the anchor system at which significantly increasing displacement
begins”, and ultimate capacity as “the maximum load sustained by the anchor system” (ASTM D4435,
2013).
Chapter 2. Testing of Rock Bolts 11
Figure 2.9: Conceptual load versus bolt head deflection curve for a rock bolt pull test (ASTM D4435,
2013)
Suggested Methods for Determining the Strength of a Rockbolt Anchor (Pull Test)
Published by the International Society for Rock Mechanics (ISRM), the scope of this suggested method is
“to measure the short-term strength of a rockbolt anchor installed under field conditions” (ISRM, 1981).
It is suggested that at least 5 tests are performed for a given set of rock and installation conditions.
The standard rock bolt assembly is to be used for the pull test, along with a hydraulic jack and
handpump with a travel of at least 50 mm and a method of measuring load with an accuracy of 2%
of the maximum load reached in the test. Displacement should be measured by a device (a dial gauge
is suggested) accurate to 0.05 mm, to be mounted on “firm rock” (ISRM, 1981). Figure 2.10 shows a
schematic of the suggested apparatus. The bolt should be installed in conditions representative of those
commonly encountered on site with standard installation, and if possible the bolt should not be tensioned.
Dimensions of the drill hole, bolt, and anchor should be measured, and the type and condition of the
rock should be noted. It is suggested that the pre-load (the load at which 0 displacement is recorded)
should not exceed 5 kN. Further load should be applied in increments of either 5 kN or 5 mm, whichever
is less, at a rate of 5-10 kN/min. Load and displacement readings should be allowed to stabilize before
recording, and the time taken for this stabilization noted. The bolts should be pulled until 40 mm of
displacement, yield, or failure, whichever occurs first.
Total displacement should calculated, along with the anchor strength (defined as the maximum load
before yield or failure of the anchor). If the tendon yields or fails before the anchor, the failure/yield load
is recorded as the minimum anchor strength. Elastic elongation should be calculated and contrasted
with bolt behaviour, and the effect of cement cure time examined (ISRM, 1981).
Chapter 2. Testing of Rock Bolts 12
similarly oriented and positioned forces. Shear and rotational forces may occur at one or multiple points
along the bolt. The working environment is a difficult one, limiting instrumentation due to delicacy
or set–up time. With this lack of complete instrumentation, there also comes uncertainty with regard
to how the bolt physically fails or yields, or the displacement mechanisms recorded – ASTM D4435-13
concedes that “interpretation of the [load-deflection] curve often requires some engineering judgement.”
2.5 Summary
Several rock bolt types are used in underground mines. Each of these bolts may be subject to variations
of several tests described by ASTM, designed to quantify various aspects of behaviour and performance.
The in situ pull test best incorporates certain aspects of operational bolt performance, namely the bolt
installation process and the rock mass in which the bolt will be installed. Despite the shortcomings of
the pull test, as a site-specific evaluation of the performance of a rock bolt, it is a widely applicable and
relatively inexpensive test influenced by a greater number of relevant factors than any of the laboratory
tests discussed.
This thesis focused on the behaviour of rock bolts subjected to pull tests. Chapter 3 provides a
description of the mines that provided the the results of the pull tests, and the specific bolts tested.
This information was used to construct a database of pull tests performed at these mine sites within a
specific time frame.
Chapter 3
This chapter provides a brief description of the regional geology of the Sudbury Basin, and of the mines
participating in the development of the pull test database. The latter part of this chapter provides
a description of the specific bolts tested in the database, and compares the database assembled with
previous work.
Milnet N
Strathcona
Coleman Capreol
Longvac
Fecunis
Fecunis Lake Fraser
Levack
Levack West McCreedy
Boundary
Hardy Norduna
Windy
Chelmsford
Lw Lw
East Falconbridge
Falconbridge
Little StobieGarson
Lw Stobie
Frood
Murray
SUDBURY
Trillabelle
Lw
Ramsey
Collins Lw Lake
Lw
Copper Cliff North
Sultana Clarabelle Copper Cliff South
Chicago
Creighton
Victoria
Worthington
Totten
Proterozoic LEGEND
Quartzite
Sudbury Igneaous Complex
Figure 3.1: Map of Sudbury area showing locations of relevant mines (after Eckstrand & Hulbert, 2007)
14
Chapter 3. Composition of the Database 15
MOB
153 OB
170 OB
Figure 3.3: Longitudinal section of Copper Cliff Mine (Chinnasane et al, 2014)
1000LL 305Lm
3 Shaft
Ramp
3800LL 1158Lm
Footwall Orebodies
Orebodies
Hangingwall Orebodies
Granite I Gabbro Complex
Sudbury Igneous Complex
6 Shaft 5800LL 1768Lm
Shear Zone
7000LL 2134Lm
8 Shaft
7680LL 2341Lm
461Orebody
8200LL 2500Lm
0 2000 m
2000 N
8000 N
10000LL 3048Lm
quartz diorite dike dipping at 80◦ , the geological model of which is often compared to that of the Copper
Cliff Offset (Lightfoot & Farrow, 2002). 113 tests from between April 2010 and December 2014 were
obtained.
Figure 3.6: Longitudinal section of Totten Mine (After Sudbury Platinum Corporation, 2015)
FRS A
The FRS A (Figure 3.7) is manufactured in 4 nominal diameters: 33, 35, 39 and 46 mm (FA33, FA35,
FA39 and FA46 respectively), although the 33 mm variant was neither used nor tested at Vale’s Sudbury
operations during the time frame for which pull test data was collected. After the tube is electrolytically
galvanized, a ring is welded to the head of the bolt, and the weld and ring are painted over with zinc-
based paint to prevent corrosion. Bolts manufactured in this sequence are referred to as “pre-galvanized”
(i.e. the bolt is galvanized before assembly; Lynn Mainville-Beach, June 2014). Length-normalized load
capacities are calculated with an anchorage length equal to the bolt length minus 6” (15 cm). This is
the specified anchorage length used for such calculations in Supplier A’s pull test reports, and accounts
for the taper length and a portion of the bolt that is not fully inserted into the rock mass so that the
pull testing apparatus may be attached.
FRS B
The FRS B (Figure 3.8) is manufactured in the same 4 nominal diameters as FRS A bolts (FB33,
FB35, FB39 and FB46), although as with the FRS A, the 33 mm variant was not used or tested by
Vale’s Sudbury operations. The bolts were manufactured with a crimped head until 2012, and are now
manufactured with a welded ring head. Bolts are post-galvanized; the galvanization is performed after
the ring head is welded on to the bolt (Lamothe, June 2014). Length-normalized load capacities are
once again calculated using a length 6” (15 cm) less than the total bolt length.
Table 3.2 compares what is referred to as the ‘minimum breaking capacity’ by Supplier A and the
‘minimum ultimate tensile strength’ by Supplier B of the FRS A and FRS B, as well as the initial
capacity claimed by the suppliers, and the bit sizes suggested for installation.
The FRS Bs have slightly larger minimum ultimate tensile strengths than their FRS A equivalents,
although the recommended initial load ranges are identical. There is also variation in the bit size
recommendations; Supplier B recommends a larger minimum diameter for the 33 mm bolt, smaller
minimum and maximum diameters for the 35 mm bolt, and a larger maximum diameter for the 46 mm
bolt.
Chapter 3. Composition of the Database 21
Minimum Ultimate Tensile Strength Initial Load Capacity Nominal Bit Size
FA33 71 kN 27 to 54 kN 30 to 33 mm
FB33 89 kN 27 to 54 kN 31 to 33 mm
FA35 71 kN 27 to 54 kN 32 to 35 mm
FB35 89 kN 27 to 54 kN 31.8 to 33.3 mm
FA39 89 kN 27 to 54 kN 35 to 38 mm
FB39 102 kN 27 to 54 kN 35 to 38 mm
FA46 133 kN 54 to 89 kN 41 to 44 mm
FB46 145 kN 54 to 89 kN 41 to 45 mm
Figure 3.9: Schematics of rebar manufactured by Supplier C (top; courtesy of Supplier C), Supplier A
(middle; courtesy of Supplier A) and Supplier B (bottom; courtesy of Supplier B)
Table 3.3: Steel properties for a 20 mm threaded rebar (Courtesy of Suppliers A, B and C)
As shown in Table 3.3, the Grade 60 rebar (Suppliers A and C) has a marginally higher yield strength
as well as higher tensile strength than Supplier B’s Grade 400 W rebar. It should also be noted that
each bolt type has a different surface pattern, which may influence the interaction between the bolt and
grout during and after mixing. All three of the suppliers’ rebar are manufactured with 10 UNC threads.
Chapter 3. Composition of the Database 22
Minimum Typical
Yield Strength 98.5 kN 113.6 kN
Ultimate Tensile Strength 151.5 kN 163.6 kN
3.2.4 D-Bolt
The D-Bolt (Figure 3.11) is a yielding reinforcement element supplied by Normet. It consists of segments
of smooth bar punctuated by anchor sections that are composed of perpendicularly oriented flattened
paddles. The D-Bolt is designed such that load is distributed evenly along the smooth steel section,
allowing for relatively large deformations and thus capacity to absorb energy. Due to the strength of the
steel and relatively stiff bolt response, the D-Bolt may also hypothetically be used as a static support
element (Li, 2011). The D-Bolt is manufactured with a smooth bar diameter of either 20 mm or 22 mm,
with pull tests on both diameters of bolt present in the database.
Mansour manufacture a similar, 20.5 mm diameter bolt: the VersaBolt (or DS-Bolt). Figure 5.19
shows one possible configuration of the Versabolt. Table 3.5 shows the yield and ultimate tensile loads
of the D-Bolt and VersaBolt.
Table 3.5: VersaBolt and D-Bolt mechanical properties (Courtesy of Mansour, Normet)
1 2
Figure 3.13: Schematic of a Python bolt, and cross sections before (1) and after (2) inflation (Courtesy
of Jennmar)
Atlas Copco manufacture three different types of Swellex; the Premium (Pm) line for standard
applications, the Manganese (Mn) line for large deformation circumstances, and the Spartan (Sp) line
for low convergence, low energy scenarios (Atlas Copco, 2012). DSI supply a “Standard” and a “Plus”
line, which are equivalent to the Pm and Mn lines respectively, and Jennmar only have one line. Table
Chapter 3. Composition of the Database 24
3.6 compares different bolt variants. Swellex Spartan bolts are omitted as they were not tested at Vale’s
operations, and there were no DSI or Jennmar equivalent at the time of writing. Note that breaking load
and elongation values are expressed as minimums for Swellex and Omega bolts, but as typical values for
Python bolts.
Table 3.6: Summary of Swellex, Omega and Python bolt mechanical properties (Atlas Copco, 2012;
Courtesy of Jennmar, DSI)
In addition to the different bolt variants supplied by Atlas Copco, the bolts may also be manufactured
with a plastic coating, denoted by adding a prefix of “Pc” to the bolt name (e.g. PcPm12). Two types
of plastic coating are used; a polyvinyl (PVC) coating is used exclusively on the Pm12 bolt, while a
polyethylene (PE) powdered coating is used on other variants (Leung, 2014).
3.3 Database
In total, pull test data for at least 26 bolt configurations from 7 suppliers across 5 classes of bolt was
collected.
tested. The results of testing take the form of a load–displacement curve, a “yield” load (i.e. working
capacity) and/or a maximum recorded load. The pull test reports were evaluated on an individual
basis to ensure consistency in the data input into the database. For example, the working capacity was
evaluated directly from the load–displacement curve of a test; the “yield” value recorded by the author
of the pull test report was not accepted without verification. Table 3.7 summarizes the number of pull
tests performed on each configuration.
Some of the bolt types shown in Table 3.7 could have been further subdivided, however inconsistent
recording of precise bolt configuration made it difficult to distinguish between these subdivisions. For
example, many FRSs were noted to be galvanized. However, in a large number of cases, it was not
noted whether bolts were galvanized, and there was a small sample size of bolt explicitly note as non-
galvanized. Given the size of the non-galvanized sample, it proved to be impossible to separate the
Chapter 3. Composition of the Database 26
influence of galvanization from other factors. A similar situation exists with Swellex bolts being uncoated,
versus having various plastic coatings. In these cases, bolts are grouped and these distinctions are not
incorporated into the analysis.
3.4 Summary
The six mines for which pull test data was compiled are all situated in or near the Sudbury Igneous
Complex. A wide range of rock units are present as a result of the complex geology of the area, a
challenge compounded by inconsistent geotechnical data logging between the mines.
While the bolt types discussed have different operational principles behind them, the provided spec-
ifications between suppliers of the same bolt configuration generally agree with one another. Although
suppliers do not provide an explicit value of strength to be used for the design of an excavation, they do
often provide minimum and typical material strengths in terms of both yield and ultimate load.
Databases of pull tests have been previously assembled, however the data collect for this thesis
originates from 6 mines in relative proximity to one another. More types of rock bolt are included than
in other databases, and bolt behaviour as well as load capacity is examined.
Chapter 4 details the pull testing process in the campaigns that constitute the database and compares
the methods used in practice with those suggested by ISRM or prescribed by ASTM.
Chapter 4
Chapter 2 provided a description of the ASTM Standards for in situ pull testing of rock bolts, as well
as the ISRM suggested methods for pull testing. This chapter provides a commentary on pull tests as
implemented in the reported database. It is based on all pull test reports in the database, and witnessing
a number of pull test campaigns performed by different suppliers at different mine sites. This chapter will
also discuss the significance of any deviation from the ASTM standards and ISRM suggested methods.
28
Chapter 4. Review of Implemented Pull Test Methods 29
Load is applied by pressurizing the hydraulic ram with a hand pump, and measured from a cali-
brated pressure gauge. Measurements are made by reading displacement off of the Vernier calliper at
predetermined load intervals (for the 22 mm D-Bolt, these intervals were generally one reading every
two tons of load up to 10 tons, and one per ton above that, although this may vary with the type of
bolt being tested). Testing is generally performed until the bolt reaches its working capacity, although
tests may be stopped before this is reached. In contrast, an FRS is tested until it slips (i.e. its ultimate
capacity), although displacement is not measured for the majority of pull tests on this type of bolt.
While suppliers employ different apparatuses, this procedure was the one most widely used to con-
tribute to the database, with manual application of load and data recording. In general, Supplier A, and
on occasion other suppliers, used a more digitized apparatus which logs data electronically. A displace-
ment sensor measures displacement between the pulling rod and the lower part of the hydraulic ram
(essentially the same displacement measurement as that shown in Figure 4.1), and load is measured using
a load cell. Recordings are made at a certain frequency, as opposed to intervals of load or displacement.
Load may be applied with either a manual or an electric pump.
methods are not as rigorous, it is still a thorough methodology. Differences between the pull test as
practised at the mine sites discussed and the standard and suggested methods are described herein.
Only ISRM’s suggested methods refers to the implementation of a pre-load. A pre-load is the limited
loading of the bolt before recording displacement data in order to tighten the apparatus mounted on
the bolt. A 5 kN maximum is suggested, although the typical pre-load used in practice was 2 tons
(17.8 kN) and could be as high as 4 tons (35.6 kN). While ASTM does not make reference to pre-load,
it does describe quite a detailed and extensive loading procedure. For at least half of the bolts, three
loading/unloading cycles in 10 equal increments are specified, loading the bolt up to 1/4, 1/2 and 3/4 of the
estimated failure load. This guideline did not appear to have been followed in any pull test campaign,
and bolts were fully loaded in the first and only loading cycle. Unloading and reloading only occurred
when significant movement of the apparatus resulted in a loss of load on the bolt. ASTM specifies that
non-cycled bolts are to be loaded to failure in 20 equal increments or increments of 2.2 kN, whichever is
less (ISRM suggests increments no greater than 5 kN), but in general increments of 1 ton (8.9 kN) were
used when increments were recorded manually. There are also guidelines regarding failure of the bolt;
ASTM specifies the bolt must be pulled to failure (defined by ASTM as the peak load, or a deflection
of 12.7 mm), and then pulled an additional 12.7 mm, with load recorded every 1 mm. ISRM specifies
testing to be performed until either 40 mm of displacement, yield or failure of the anchor. In general,
pull tests were performed until the working capacity (i.e. the point at which a significant increase in
the displacement observed per load increment; ASTM, 2013c) of the element was surpassed. Peak load
(ultimate capacity) or failure load was not generally found for bolts besides the FRS, as there are safety
concerns associated with energy release when failing a bolt.
There are several calculations mandated by ASTM which are either not performed, or are not included
in the pull test report. These include stress, elastic deformation and corrected bolt head displacement.
It should be recognised that these calculations are not straightforward for all bolt types; for example,
elastic deformation in a grouted bolt is difficult to calculate unless one makes the assumption that all
displacement before the bolt yields may be attributed to elastic deformation.
control investigation. ISRM’s suggested methods are more forgiving, especially in terms of procedure,
however are somewhat dated. They pre-date the introduction of FRSs, expandable bolts and yielding
bolts to the industry, and as a result are focused on cement-grouted rebar and mechanical bolts, which
must be seen as a significant shortcoming of the procedures described.
250 250
Load (kN)
200 200
Load (kN)
150 150
100 100
Plate load
50 50
0 0
0 20 40 60 80 100 120 140 160 180 0 50 100 150 200 250
Plate End Displacement (mm) Displacement (mm)
Figure 4.2: Static laboratory pull test on a 22 mm D-Bolt 2.1 m in length (left; Doucet & Voyzelle,
2012) and dynamic impact test on a 22 mm D-Bolt 1.5 m in length (right; Li & Doucet, 2012)
Chapter 4. Review of Implemented Pull Test Methods 33
The results of static and dynamic tests cannot be directly compared, as the difference in loading rate
results in the reinforcement element assuming different properties. For example, in Figure 4.2, impact
testing results in a higher yield load and larger ultimate displacement than a static laboratory pull test on
a longer 22 mm D-Bolt. However, by indirectly comparing the tests it can be demonstrated why working
capacity may be used as a design value for static scenarios. The energy capacity of a rock bolt is equal to
the area under the load–displacement curve (Li et al, 2014). The D-Bolt absorbs energy by distributing
strain across a length between two anchors, allowing for larger deformations at equivalent loads than
reinforcement elements continuously bonded to the rock mass, such as rebar, where strain is concentrated
on a relatively short section of the element. If a D-Bolt bears a load less than its working capacity, most
of the D-Bolt’s energy capacity is conserved as little deformation has been incurred. However, if the
working capacity is surpassed, the energy capacity of the bolt diminishes much more significantly as the
magnitude of plastic deformation per unit load is greater than that of elastic deformation. As a result,
if a D-Bolt (or other reinforcement elements that physically yield without slipping) is subject to static
loads greater than its working capacity, it will have significantly less capacity to absorb energy under
dynamic loading.
In most pull test reports, the working capacity is referred to as “yield” for all bolt types. Yield
strength is defined as “the engineering stress at which, by convention, it is considered that plastic elon-
gation of the material has commenced” (ASTM E6, 2009). Figure 4.3 shows two methods of determining
yield strength according to ASTM Standard E6-09b.
Figure 4.3: Methods of determining yield strength: halt of the pointer method (left) and offset method
(right; ASTM E6, 2009)
In Figure 4.3, the yield strength (Sy ; UYS and LYS are upper and lower yield strength respectively)
are determined at either a perfectly plastic response to load (i.e. an increase in displacement without
a corresponding increase in load), or at a predefined offset in strain from linear elastic behaviour. The
resolution of recorded load values for the data collected is variable and in some cases quite poor. If load
is recorded manually, increments of 0.5, 1 or even 2 tons (4.45, 8.9 or 17.8 kN) are used in different
tests. Digital recording is frequency-based, resulting in variable load resolution which is a function of
loading rate and measurement frequency. Neither of the methods of determining yield strength presented
in Figure 4.3 are practical when applied to the data collected. Perfectly plastic behaviour is seldom
observed since displacement is recorded in increments of load and defining a strain offset to determine
yield when bolt behaviours may be non-linear and highly variable is not feasible.
The working capacity was determined for the purposes of this thesis as the greatest load measured
Chapter 4. Review of Implemented Pull Test Methods 34
before a softening of the bolt response attributed to plastic deformation of the reinforcement element
material or movement of the entire element is observed, as demonstrated in Figure 4.3.
Working Capacity
Load
Displacement
The working capacity may be indicative of a lower bound of yield strength for some bolts, however
some bolt types may reach their working or ultimate capacity before the bolt material itself actually
yields. For example, an FRS will generally slip rather than yield or fail if pull testing is performed
soon after installation, and its ultimate capacity will be defined by the maximum load sustained by the
bolt without slipping. As working and ultimate capacity are generally applicable measures of bolt load
capacity, they are the primary load metrics discussed in this thesis.
ds
da
db
Figure 4.5: Measurement of displacement for a pull test on a generic reinforcement element
The measurement of da may not accurately reflect the performance of the bolt alone if the objective
of testing is solely the investigation of the bolt response to load, however it could still be representative
of certain loading scenarios. Figure 4.6 shows a representation of a generic rock reinforcement system
(Thompson et al, 2012).
A pull test using an independently mounted dial gauge measures two of the interactions shown in
Figure 4.6: element–internal fixture, and internal fixture–rock. By recording the displacement of the
loading system as was practised in all cases used to build the pull test database, a third interaction
is simulated. The pull test apparatus acts as an external fixture, and its interaction with the rock is
recorded as well as the element-internal fixture and internal fixture-rock interactions. This results in a
loading scenario analogous to the one presented in Figure 4.7.
Figure 4.7 presents a simplified scenario in which a point-anchored bolt is loaded axially by a wedge.
The excavation surface around the face plate of the bolt compresses as it would during a pull test,
displacing by an amount ds relative to the rest of the wedge’s surface. The bolt deforms, resulting in a
displacement db at the bolt head. The summation of these two displacements, da , is the displacement
undergone by the surface of the wedge relative to the surface of the stable excavation. In the case of a
point–anchored reinforcement element, the da of a pull test may reflect the da of such an axial loading
scenario, depending on factors such as plate size relative to the footprint of the pull test apparatus on
Chapter 4. Review of Implemented Pull Test Methods 36
the surface of the excavation, use of surface support, wedge size, etc. For other types of reinforcement
elements, the agreement depends on where along its length a bolt is loaded, how load attenuates to the
head of the bolt (influencing ds ), and the length of bolt that will deform (influencing db ).
da = d b + d s
da
ds
db
Figure 4.7: Measurement of displacement for axial loading of a point anchored rock bolt
The rock mass is an integral part of the system, but its complexity makes its role relatively difficult
to monitor during a pull test. Shearing of a bolt may occur across one or a series of joints, especially if
the bolt is not tested immediately after installation, which may affect its behaviour. Loss of resin into
fractures in the rock mass may impact the volume of effective resin and its mixing, thereby potentially
affecting the performance of any resin-grouted bolt. The drillability of the rock influences the diameter of
a drill hole relative to the size of a drill bit and could impact the performance of frictionally coupled bolts.
The strength and degree of fracturing of the rock mass and surface support may affect the measured
displacements, as shown in Figure 4.5. It follows that the behaviour of a rock bolt during a pull test is
influenced by the condition and properties of the rock mass in which it is installed.
support pressure relative to either a more ductile system (System 4), or systems installed once more
radial displacement has occurred (i.e. the face has advanced further; Systems 1, 2 and 5).
Figure 4.8: Design of support systems using the ground reaction curve (Brady & Brown, 2006)
Stiffness is measured in terms of load per unit displacement, kN/mm. The calculation of stiffness
is a way of investigating how stress attenuates down the bolt. Continuously coupled bolts will not
have an even load distribution along their lengths, and their stiffness may describe how well they are
bonded or coupled to the rock mass. It would appear that “stiffness” is reported in several forms in
technical literature. The two principal measures discussed in this thesis are illustrated in Figure 4.9.
Secant stiffness is the average relationship between displacement and load from initial loading up until
the working capacity (i.e. a secant stiffness at the working capacity). Stiffness calculated along a linear
portion of the load–displacement relationship is the tangent stiffness (Bieniawski et al, 1978). The former
will incorporate displacement effects not necessarily directly related to the behaviour of the rock bolt
such as fracturing of the rock mass, compression of pre-existing fractures and movement of the loading
rig among others. The tangent stiffness was used in an attempt to minimize the influence of these sources
of displacement, and may potentially provide a better description of bolt performance.
Chapter 4. Review of Implemented Pull Test Methods 38
160
140
120 SecantcStiffness
100
Load (kN)
80
60
TangentcStiffness
40
20
0
0 1 2 3 4 5 6 7 8 9 10
Displacement (mm)
Figure 4.9: Calculation of secant and tangent stiffness from a pull test
This thesis also uses the metric “unloading stiffness”, which is calculated by examining the recovery of
displacement (recorded as negative) when the bolt is unloaded, representing elastic recovery of the bolt.
This should further reduce the effect of elements of the testing system extraneous to the rock bolt being
tested. In some cases, the pull test report would explicitly state that unloading was examined, qualifying
the data. For the tests in which unloading was recorded but was not an explicit objective of testing, two
clearly visible recordings of load and displacement in the unloading phase must be visible (see Figure
4.10); as logging was frequency based, loading may have continued beyond the maximum displacement
recorded, affecting the slope of the unloading line. Similarly, when a load of 0 kN is reached, the pull
test apparatus and rock mass are no longer held together in compression, and large displacements not
associated with elastic recovery of the bolt may occur as the system loosens. Figure 4.10 shows a testing
campaign in which only Test 6 provides acceptable data as measurements of load and displacement are
recorded twice between the beginning and the end of unloading (circled in red).
Figure 4.10: Pull test campaign results for partially encapsulated Rebar A from November 16th , 2012
at Coleman Mine (Mainville et al, 2012)
Chapter 4. Review of Implemented Pull Test Methods 39
In addition to the metrics discussed, load and stiffness metrics applicable only to the Modified Cone
Bolt are introduced in Chapter 5. As the MCB33 is designed to plough through the resin column (unlike
any other bolt presented in this thesis), it has a unique behaviour for which the metrics of performance
discussed are not satisfactory.
4.4 Summary
Pull tests as performed at Vale’s Sudbury operations deviated from the standard laid out by ASTM and
the methods suggested by ISRM. Perhaps the most significant difference in implemented apparatus and
procedure is the method of measuring displacement. While both ASTM and ISRM suggest independently
mounted displacement measurement systems, all suppliers use an instrument mounted on to the loading
system itself. This results in measurement of the response of the rock mass surface immediately adjacent
to the bolt tested, and does not solely represent how the bolt and its anchoring system perform.
This thesis uses metrics of load capacity and stiffness to quantify bolt performance. Data was carefully
interpreted in order to attribute observed displacements to certain effects or mechanisms. Despite the
identified limitations, there is great value gained from pull tests conducted by experienced personnel
that employ engineering judgement in collecting and interpreting the pull test data.
In Chapter 5, the behaviour of rock bolts subject to a pull test is interpreted, and statistics on the
performance metrics described in this chapter are presented for a series of bolts.
Chapter 5
This chapter provides a description of the statistics and analyses used to characterise the behaviour of
rock bolts when subjected to a pull test. It also explores the variation in performance of rock bolts as
predicted by field data and a series of conceptual assumptions.
where n is the number of samples. The sample standard deviation (s) is also routinely quantified to
describe dispersion, calculated as
r Pn
− x̄)2
i=1 (xi
sx =
n−1
Both of these metrics are used to calculate the coefficient of variation for the sample (ĉv ) as a way
of describing the size of the sample standard deviation relative to the mean of the data (Baecher &
Christian, 2003). This allows for the direct comparison of dispersion between variables with different
magnitudes of mean.
40
Chapter 5. Summary Statistics and Interpretation of Pull Test Data 41
sx
cˆv =
x̄
The shape of a distribution may be described by skewness and kurtosis. The measure of skewness
elected for use is the adjusted Fisher-Pearson standardized moment coefficient (G1 ).
n n x − x̄
i
)3
P
G1 = (
(n − 1)(n − 2) i=1 s
Skewness is a measure of the tendency of the mean of a dataset to be on one side of the median as
the result of the distribution exhibiting a tail. A positively (or right) skewed distribution has a mean
greater than the median (and a tail pointing right), and will have a positive G1 . A normal distribution is
described by G1 = 0 (Doane & Seward, 2011). The second statistic used to describe distribution shape
is a measure of excess kurtosis, referred to as kurtosis herein, calculated as
4 !
3(n − 1)2
n
n(n + 1) P xi − x̄
Kurtosis = −
(n − 1)(n − 2)(n − 3) i=1 s (n − 2)(n − 3)
p(N + 1) = k + d
where k is an integer value and d the remaining decimal. There are three ways Yp is subsequently
calculated:
Note that the 50th percentile is the median of the data (NIST-SEMATECH, 2003).
yi = β0 + β1 xi + i
where
Pn
Sxy (x − x̄)(yi − ȳ)
β1 = = Pn i
i=1
2
Sxx i=1 (xi − x̄)
and
β0 = ȳ − β1 x̄
Chapter 5. Summary Statistics and Interpretation of Pull Test Data 42
and i is a random error term. The output for a bivariate analysis is a line for which the mean
distance between the regression line and all data points is minimized. The coefficient of determination
(R2 ) may be calculated for a linear regression, as
Pn
RSS (yˆi − y¯i )2
2
R = = Pi=1
n 2
T SS i=1 (yi − ȳ)
where yˆi is the fitted value for the term β0 + β1 xi . The coefficient of determination assesses the level
of correlation between y and x; R2 = 1 describes a perfectly linear relationship, and R2 = 0 indicates
no linear relationship is present (Fox, 2008).
Two methods are used to determine if a statistically significant difference in means of two or more
sets of data exists, given the null hypothesis that the means of two populations are equal (µ1 = µ2 ).
The first method, used to compare only two datasets, is the studentized t-test. Various configurations of
this test exist, all of which assume distribution normality. The two used for the purposes of this thesis
are the t-test for data sets Y1 and Y2 with equal variance, and the t-test for populations with assumed
unequal variance. For the former, the t-statistic is calculated as
Ȳ1 − Ȳ2
t= q
sY1 Y2 n11 + 1
n2
where
s
(n1 − 1)s2Y1 + (n2 − 1)s2Y2
sY1 Y2 =
n1 + n2 − 2
v = n1 + n2 − 2
the t-statistic is compared to the t-distribution. A level of significance is chosen (5% is used in this
thesis). If the t-statistic calculated is greater than the t-value that corresponds to the degrees of freedom
and significance level chosen (tcrit ), the null hypothesis is rejected, suggesting a significant difference
between population means. In addition, a p-value is estimated. If t > tcrit , the p-value will be less than
the significance level chosen (NIST-SEMATECH, 2003).
If samples are assumed to have unequal variance, the same procedure is followed but the calculation
of t and v are as follows:
Ȳ1 − Ȳ2
t= p
s1 /n1 + s22 /n2
2
Single factor, one-way Analysis of Variance (ANOVA) is used to determine whether a significant
difference in mean exists between three or more sets of data (although it may be used for two datasets
as well). Table 5.1 shows the general format of an ANOVA table, comparing J groups, each containing
I entries.
Chapter 5. Summary Statistics and Interpretation of Pull Test Data 43
where
1 PJ
ȳi. = yij
J j=1
and
1 PI P J
ȳ.. = yij
IJ i=1 j=1
Similar to the t-test, the F0 value calculated is compared to the F -distribution, accounting for degrees
of freedom (DoF) and significance level. If F0 > Fcrit , the null hypothesis is rejected and a difference
between means is suggested at a predetermined level of significance (NIST-SEMATECH, 2003).
Although many more sophisticated methods of data analysis exist, their application was limited due
to the nature of the database. The database was assembled from independent pull test campaign reports
that inconsistently record a variety of parameters. As a result, large gaps in data exist; for example, of
545 pull tests performed on various configurations of FRS, rock mass quality was recorded for only 126
of those tests. Such a data set is incompatible with some methods of analysis, such as the assembly of a
generalized linear model. Other statistical analyses are suitable for an incomplete dataset, for example
principal component analysis (PCA). However, due to the often limited frequency and widely varying
combinations of parameters recorded, PCA was found to be an ineffective method of analysis for this
database. As such, analysis is limited to the methods that have been discussed in this section.
Figure 5.1: Shear stress distribution along a frictionally coupled bolt subject to axial load (Li &
Stillborg, 1999)
The interface between the rock mass and the rock bolt is assumed to have constant shear strength
s. Once the strength of the interface is overcome, the bolt is said to have “decoupled” at that point
and cannot resist further application of load. A decoupling front develops (x2 ), beyond which the shear
stress on the interface decreases following function τb (x). Once the decoupling front has reached the end
of the element, the entire bolt is free to slip (Li & Stillborg, 1999). Figure 5.2 shows how load attenuates
down the bolt. Although the figure presented is modelled off of a Swellex bolt, the concept is similar for
an FRS. The shear strength of the interface between a Swellex bolt and the rock mass is greater than
that of an FRS due to the mechanical interlock of the bolt and rock, otherwise they are presented as
the same.
Figure 5.2: Shear stress and axial load along a Swellex rock bolt (Li & Stillborg, 1999)
Load decreases linearly in the decoupled length of bolt, and then continues to decrease at a rate
proportional to the shear stress (i.e. the load in the bolt is proportional to the integral of the shear
stress on the interface). As the decoupling front proceeds along the length of the bolt, a progressively
longer portion of the bolt is subject to higher loads. This results in an apparent softening of the bolt
behaviour, as total deformation is proportional to the integral of function P (x), representing load in the
bolt.
Chapter 5. Summary Statistics and Interpretation of Pull Test Data 45
90
80 A B C D
70
E F
60
Loadm(kN)
50
40
30
20
10
0
0 1 2 3 4 5 6 7 8 9 10
Displacementm(mm)
In Phase A load is built on the bolt, although significant displacement occurs. Most displacement
may be attributed to the loading rig adjusting position as it becomes fully flushed with the surface of
the rock mass, as well as the result of the initial rock mass and surface support response to pressure
from the pull test apparatus. To eliminate this phase, a pre-load may be applied to rock bolts during
a pull test before displacement is recorded. Phase B shows a response with decreasing stiffness as the
decoupling front progresses and a larger length of bolt is subject to higher axial load, until limited slip
of the bolt appears to occur in Phase C (i.e. the decoupling front has progressed along the entire length
of the bolt). Once slip stops, the decoupling front regresses which results in a stiffer reaction observed
in Phase D before slip occurs again entering Phase E. Load is then repeatedly built before slip occurs
again and load is released. The test is stopped once pull testing personnel are satisfied that higher loads
will not be reached, and Phase F shows unloading of the bolt as it re-couples with the rock mass and
some elastic deformation is reversed (although there will still be elastic energy stored in the bolt due to
frictional resistance acting against elastic recovery).
Displacement data from the test in Figure 5.3 was collected by a data logger; load intervals are
dictated by a frequency at which data is collected, and are inconsistent due to a variable loading rate.
This results in a load-displacement graph that may appear different from one for which data is manu-
ally collected. In a manually-recorded pull test, testing personnel will generally wait for the load and
displacement readings to stabilize before recording them and will seldom record two displacements for
a single load, which makes perfectly plastic behaviour difficult to observe. Manual recordings of load
and displacement also generally do not incorporate the pre-loading (Phase A) and unloading (Phase F)
portions of a pull test for any bolt type.
Chapter 5. Summary Statistics and Interpretation of Pull Test Data 46
Most types of rock bolt in the database were only tested until their working capacity is observed.
However, FRS pull tests are performed to determine ultimate capacity. As the failure mechanism of a
recently installed FRS is slip, there are fewer safety concerns associated with loading to ultimate capacity
than exist for a bolt for which ultimate capacity is dictated by bolt failure. Additionally, most FRS pull
test reports do not fully record the load–displacement behaviours of the tested bolts and only note a
maximum load. As such, ultimate capacity is investigated as opposed to working capacity for this bolt
type.
Figure 5.4 shows the distributions of the ultimate capacities for FRS A and B bolts. Bolts are
separated by supplier and nominal diameter, and loads are recorded per unit length of anchorage (kN/m).
The length of bolt providing anchorage is assumed to be 6” (0.152 m) less than the total length of the
bolt, as the tapered end and the section of bolt to which the pull test apparatus is attached are not in
contact with the rock mass.
Table 5.2 shows summary statistics for the maximum loads for the six bolt variants. The average
measured diameter of the bolts is also noted. Diameter measurements were not performed using a
consistent method between campaigns (although most reports mention the use of Vernier callipers);
some diameters are an average of 3 or 5 measurements along the length of the bolt, some are a single
measurement at the midpoint, and some are not explained.
There appears to be no statistically significant difference between the ultimate capacities of bolts
with different nominal diameters, or between bolts from either supplier. ANOVA was performed on
the data sets composed of FRS A bolts, FRS B bolts, and across all FRS configurations. Failure to
reject the null hypothesis occurred for both the FRS A and B (p = 0.745 and 0.405 respectively), as
well as for the ANOVA of all configurations (p = 0.695). This indicates that there is not a statistically
significant difference in the ultimate capacity between configurations within or across supplier. While a
larger diameter FRS will have greater surface area in contact with the rock mass on which to generate
friction, it would appear that the larger diameter bolts do not generate the equivalent stress normal to
the bolt-rock mass interface as the smaller bolts. This is demonstrated as resistance to pull is essentially
the same between the different diameters of FRS. In fact, the 35 mm nominal diameter FRSs had the
highest average ultimate capacities (although the difference is marginal and not statistically significant).
This implies that if immediate resistance to axial loading is the main objective of bolt installation,
there is apparently no advantage gained in the selection of a larger diameter FRS. Amalgamating all
FRS configurations into one dataset results in a distribution of ultimate capacities that is very close to
Chapter 5. Summary Statistics and Interpretation of Pull Test Data 47
20 100% 16 100%
18 14
16 Frequency 80% 80%
12
Cumulative
Cumulative5Frequency
Cumulative5Frequency
14
12 60% 10 60%
Frequency
Frequency
10 8
8 40% 6 40%
6
4
4 20% 20%
2 2
0 0% 0 0%
0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80
Ultimate5Capacity5(kN/m) Ultimate5Capacity5(kN/m)
Cumulative5Frequency
Cumulative5Frequency
8 12
60% 60%
Frequency
Frequency
10
6
8
40% 40%
4 6
20% 4
20%
2
2
0 0% 0 0%
0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80
Ultimate5Capacity5(kN/m) Ultimate5Capacity5(kN/m)
10
80% 20 80%
Cumulative5Frequency
Cumulative3Frequency
8
Frequency
60% 15 60%
Frequency
6
40% 10 40%
4
20% 5 20%
2
0 0% 0 0%
0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80
Ultimate5Capacity5(kN/m) Ultimate3Capacity3(kN/m)
Figure 5.4: Ultimate capacity per unit length distributions for FRSs with nominal diameters of 35, 39
and 46 mm
and thus were potentially exposed to a wider range of conditions, resulting in larger standard deviations.
This may also be the cause of the relatively high skew, as it is possible that the FA39 may have been
tested more often in conditions that would result in marginally lower ultimate capacities.
Kurtosis values are generally consistent. The exception is the FA46, with a kurtosis of 0.88, indicating
a peakedness of the distribution. In Figure 5.4e, two outliers are observed: one in the 5-10 kN/m bin,
and one in the 70-75 kN/m bin. This is the only distribution in Figure 5.4 that has single outliers this
obvious. As a result, the shoulders have a lower weight than the centre of the distribution, elevating the
value of kurtosis. With a greater sample size, this may be expected to reduce as the distribution fills in.
Tomory et al. (1998) found that the 33 mm nominal diameter Split Set (SS33) and the 39 mm Split
Set (SS39) also performed very similarly. The average ultimate capacity of their dataset of over 900
pull tests was 1.09 tons/ft, or 31.9 kN/m, with ĉv = 0.42. Although the mean ultimate capacity is
significantly lower than those observed in Figure 5.2, much of their database was composed of Split Sets
installed using jacklegs, while the emergence of bolters since 1998 and their use at Vale’s operations in
Sudbury may explain the higher capacities observed in more recent times. It should also be noted that
Tomory et al. collected data from over 50 mine sites. This means it is likely the pull tests contained
therein were performed across a wider range of installation and ground conditions, and possibly testing
methods/equipment. As a result, the higher coefficient of variation is expected.
Overall, it does not appear as though one FRS configuration significantly outperforms any other in
terms of either ultimate capacity or consistency of performance. Although there are irregularities in
values of the coefficient of variance, skewness and kurtosis, these would likely be addressed by expanding
the database.
Stiffness
Only campaigns of FA35 and FA39 bolts recorded the bolts’ load–displacement behaviour in its entirety.
As such, an analysis of stiffness is limited to these two bolt configurations. Figure 5.5 shows how two
measures of stiffness are calculated for an FRS.
70
60
FirstmDropmStiffness
50
Loadm(kN)
40
30
20
SecantmStiffness
10
0
0 2 4 6 8 10 12 14
Displacementm(mm)
Measuring the stiffness of an FRS is made difficult by the constantly changing coupling length and
the occurrence of slip before maximum load is achieved. The “first drop stiffness” outlined in Figure
5.5 represents the stiffness of the bolt at the displacement at which the first drop in load is observed.
Whenever the bolt slips, a drop in load is expected. This drop is not consistently captured in the data
recording due to the logging frequency. Slip distance appears to often be relatively short (less than 1
mm) before the bolt reaches a position with a higher frictional state at loads below the ultimate capacity.
Where the first drop in load is observed is where it takes the bolt a greater period of time than the
data logging frequency to reach this position and rebuild an equivalent load, and can thus be assumed
to be a greater slip distance than any previously observed slip. The secant stiffness is representative
of the displacement at which the maximum load is achieved. Both measures have their drawbacks,
however they are satisfactory as broad descriptors of behaviour; the resolution of the data recorded is
not sufficient to build a more robust model of bolt response. Figure 5.6 shows the distribution of stiffness
for the FA35 and FA39. Table 5.3 summarizes the data.
6 100/ 7 100/
5 6
80/ 80/
CumulativeSFrequency
CumulativeSFrequency
Frequency 5
4 Cumulative
Frequency
Frequency
60/ 60/
4
3
3
40/ 40/
2
2
20/ 20/
1 1
0 0/ 0 0/
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
FirstSDropSStiffnessS(kN/mm) FirstSDropSStiffnessS(kN/mm)
(a) FA35 first drop stiffness (b) FA39 first drop stiffness
6 100% 10 100%
9
5
80% 8 80%
CumulativekFrequency
CumulativekFrequency
7
4
Frequency
Frequency
60% 6 60%
3 5
40% 4 40%
2
3
20% 2 20%
1
1
0 0% 0 0%
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
SecantkStiffnessk(kN/mm) SecantkStiffnessk(kN/mm)
The difficulties of assessing FRS performance emerge; standard deviations calculated for the distri-
butions are very large relative to the means of the data (ĉv > 1 in the case of first drop stiffness of the
FA39). The heavy skew of the data calls into question the use of statistics such as standard deviation
and kurtosis in these circumstances, as the distribution does not appear to be normal. The FA35 seems
to have a higher secant stiffness than the FA39, however given the similarity of the first drop stiffnesses
Chapter 5. Summary Statistics and Interpretation of Pull Test Data 50
between the bolts and the high coefficient of variations, it is unclear whether this is indicative of a
consistent difference in responses. It is reiterated that these metrics are used as broad observational
descriptors of behaviour. They are heavily influenced by several parameters unrelated to the bolt or how
it interacts with the rock mass, such as data logging frequency and how much displacement the personnel
conducting the test allow before stopping the pull test. As such, no further analysis was performed on
the stiffness of FRSs although the information gained will be used to describe the expected behaviour
of the bolts in a pull test.
Figure 5.7: Model of the shear stress profile in a grouted rock bolt (Li & Stillborg, 1999)
In this model, a length of bolt (x0 ) is fully decoupled from the rock mass. Between x0 and x1 , the
bolt is partially decoupled with the shear stress along the interface between the bolt and the grout (τb )
equal to the residual strength of the interface (sr ). Beyond point x1 , the grout is less damaged and has
higher strength, with the peak strength (sp ) occurring at point x2 . Beyond this point, the full strength of
the grout is not mobilized, and shear stress attenuates down the remainder of the bolt following function
τb (x). Figure 5.8 shows a similar shear stress profile, but shows function P (x), representing tensile load
in the bolt (Li & Stillborg, 1999).
Chapter 5. Summary Statistics and Interpretation of Pull Test Data 51
Figure 5.8: Model of tensile load and shear stress profile for a rebar (Li & Stillborg, 1999)
As in the case of the FRS model, load (and thus stress) in the bolt decreases at a rate proportional
to the shear stress. As x2 (the location of sp , the maximum strength) progresses down the bolt, an
increasing length of the element is subject to higher stresses, and a decrease in measured stiffness is
expected. Should x2 reach the end of the bolt, the rebar would be pulled out through the grout. However,
all pull tests in the database displayed behaviour suggesting P (x) exceeds the working capacity of the
bolt at x = 0, i.e. at the bolt’s threads.
160
140 A B
120
100
Load (kN)
80
60
40
20
0
0 1 2 3 4 5 6 7 8 9 10
Displacement (mm)
The stiffness of a 1.8 m 400W steel bar with a diameter of 20 mm and an elastic modulus of 200
Chapter 5. Summary Statistics and Interpretation of Pull Test Data 52
GPa (as is used to manufacture Rebar B; Lamothe, November 2014) is 34.9 kN/mm. The stiffness for
the 1.8 m rebar shown in Figure 5.9 with an approximately linear response between 26.7 and 115.6
kN (R2 = 0.998) is 34.2 kN/mm. If displacement is interpreted as equal to the deformation of the
rebar, this would suggest that stress is distributed uniformly along its length. However, this may in
fact demonstrate the degree to which sources of displacement beyond elastic deformation of the bolt
interfere with the measurement displacement during a pull tests. As very small displacements are being
measured, displacements on the scale of one millimetre attributed to the rock mass or surface support
will strongly affect the stiffness calculation. This may also explain why no progressive softening of the
bolt response is observed.
Figure 5.10 shows the same relationship for a 20 mm Rebar A, and perhaps a more typical pull test,
where adjustments and rock fracturing are observable on the load–displacement graph.
160
140
120
100
Load (kN)
80
60
40
20
0
0 1 2 3 4 5 6 7 8 9 10
Displacement (mm)
A tangent stiffness of 59.3 kN/mm is calculated for the linear portion of the graph, circled in red
2
(R = 0.999). The minimum elastic modulus of Rebar A, like Rebar B, is 200 GPa (Mainville-Beach,
October 2014), and thus also has a stiffness of 34.9 kN/mm if subject to constant axial stress along its
length. This stiffness calculation agrees more closely with the analytical model of fully grouted rock bolts
(Li & Stillborg, 1999; Martı́n et al, 2010), suggesting that stress and strain is not evenly distributed
along the length of the bolt. In any case, it is apparent by contrasting these two examples that the
properties of the rebar material are not the only factors influencing the stiffness of the bolt system.
It is noted that if the load were to be evenly distributed along a bolt’s length, a longer bolt would
appear less stiff. Although different lengths of rebar are present in the database, length–normalized
stiffness metrics are not used as the degree of deformation the bolt undergoes is much better described
by the location of maximum shear stress on the bolt (x2 in Figure 5.7) than by the overall length of
bolt.
Chapter 5. Summary Statistics and Interpretation of Pull Test Data 53
Figure 5.11 shows the distribution of working capacities for rebar supplied by Suppliers A and B. The
rock bolts were typically installed using the proprietary resins of each manufacturer.
12 100% 14 100%
Frequency 12
10
Cumulative 80% 80%
Cumulative5Frequency
Cumulative5Frequency
10
8
Frequency
Frequency
60% 60%
8
6
6
40% 40%
4
4
20% 20%
2 2
0 0% 0 0%
100 105 110 115 120 125 130 135 140 145 100 105 110 115 120 125 130 135 140 145
Working5Capacity5(kN) Working5Capacity5(kN)
Working capacity appears to be distributed normally, although data resolution poses a problem for
pull tests conducted by Supplier B. Although the Rebar B working capacities are not as well distributed
as those of Rebar A, this may be the result of differing data recording methods. Tests performed by
Supplier B typically are performed in loading increments, with manual recording of displacement, while
pull tests conducted by Supplier A generally log data digitally. As loading increments in a manual test
are usually one ton, working capacity may only be calculated to the nearest ton (8.9 kN). As such, no
working capacities may be recorded for Rebar B between 14 and 15 tons (124.6 kN and 133.5 kN), thus
the lack of entries in the 125 kN to 130 kN bin and the resulting appearance of the distribution. On
the other hand, Supplier A’s load recordings are distributed according to loading rate and measurement
frequency. As such, the variance of the distribution observed is dependent not only on the variability in
rebar material properties, but also the variability in loads recorded by the apparatus. This introduces
a type of error that is also distributed normally as a random variable. Table 5.4 provides the summary
statistics for the working capacities of the rebar.
The average working capacities of the two rebar are very similar. The small discrepancy between the
two averages may be a result of the difference in the method of data recording rather than performance
of the elements themselves. A 1 kN difference in the medians of the two datasets also suggests that little
difference would exist between the two, if the same method of data recording were used for both rebar.
It may also be concluded that the negative skew and high kurtosis observed for the Rebar B distribution
is the result of the rounding of working capacity down to the nearest ton.
Chapter 5. Summary Statistics and Interpretation of Pull Test Data 54
Stiffness
Tangent and secant stiffness were calculated for the results of the pull tests on rebar. A large portion of
the pull tests performed were partial encapsulation tests. In these tests, a limited length of active resin
(usually one 12” or 18” cartridge of fast setting resin) is used in tandem with two or three cartridges
of “dummy” or inert resin used to simulate typical mixing conditions for the active cartridge. These
tests are performed to verify the competence of a limited length of resin. While this test configuration
should not affect the working capacity of the rebar, it may influence the load–displacement behaviour
of the bolt. As such, a distinction is made between the fully and partially encapsulated tests in Figures
5.12 and 5.13, which show the secant stiffness and tangent stiffness respectively of both rebar. Table 5.5
shows summary statistics for the stiffness of the two brands of rebar.
12 100p 7 100%
10 6
80p 80%
CumulativeNFrequency
CumulativeNFrequency
PartiallyNEncapsulated
5
8 FullyNEncapsulated
60%
Frequency
60p
Frequency
Cumulative 4
6
3 40%
40p
4
2
20p 20%
2 1
0 0p 0 0%
0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80
SecantNStiffnessN(kN/mm) SecantNStiffnessN(kN/mm)
9 10) 7 1009
8
6
80) 809
CumulativesFrequency
Cumulative(Frequency
PartiallysEncapsulated 5
6 FullysEncapsulated
Frequency
Frequency
4 3
40) 409
3
2
2 20) 209
1
1
0 0) 0 09
0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80 90
TangentsStiffnessspkN/mmd Tangent(Stiffness((kN/mm)
70% of the rebar pull tests in the data base were performed on 1.8 m, 20 mm diameter rebar with
an elastic modulus of 200 GPa, and thus a stiffness of 34.9 kN/mm when non-grouted. This is a very
conservative lower bound of stiffness in which recorded displacement can only be attributed to the
deformation of the rebar, as it assumes an even distribution of stress along the bolt. Figures 5.12a
and 5.12b show that a large number of secant stiffness calculations fall below this value. This is likely
due to the incorporation of sources of displacement besides deformation in the measurements. It is
interesting to see that the tangent stiffness also frequently falls short of 34.9 kN/mm, suggesting that
these alternate displacement mechanisms may in some cases have linear load–displacement responses at
the load resolutions in question, making it difficult to isolate the bolt/resin response from that of the
rock mass. High values of skewness and kurtosis are calculated for Rebar A in particular. Although
these are calculated for a dataset that incorporates both partially and fully encapsulated bolts, it can
be observed in Figures 5.12a and 5.13a that high stiffness outliers exist which are responsible for these
values. Table 5.6 separates stiffness statistics by encapsulation length.
Table 5.6: Comparison of stiffness between partially and fully encapsulated rebar
Interestingly, the secant stiffness of fully encapsulated rebar from both suppliers are quite similar
(averages of 29.2 kN/mm and 27.4 kN/mm respectively), with similar dispersions (coefficients of variation
equal to 0.59 and 0.52). However, Rebar A have greater values of tangent stiffness (an average of 50.6
kN/mm compared to 30.9 kN/mm). Comparing the distributions presented in Figures 5.12 and 5.13, the
fully encapsulated Rebar B appear to have a much more defined distribution shape for both tangent and
secant stiffness, suggesting the sample size of Rebar A is too small to adequately define these statistics,
despite being the same size as the Rebar B dataset. This may be the result of the difference in data
recording methods. While the displacements measured by Supplier A are taken instantaneously at loads
defined by loading rate and data logging frequency, it is common practice in manual data recording
of pull tests to wait for displacement measurements at a certain load to stabilize before reading them,
potentially resulting in what appears to be lower stiffness. The reason the secant stiffness measurements
are similar for the two brands of rebar despite the disparate distributions of tangent stiffness may lie in
the pre-loads used. Relatively large displacements may be observed at low loads while the rock mass
Chapter 5. Summary Statistics and Interpretation of Pull Test Data 56
and pull test apparatus tighten up. 80% of the pull tests on fully encapsulated Rebar A were performed
without a pre-load, compared to only 11% of pull tests performed on Rebar B. This can explain both
the ragged distribution shape of the stiffness metrics’ distributions (as the amount of displacement at
low loads is highly variable), and the high tangent stiffness relative to the secant stiffness calculated for
Rebar A.
In addition to tangent and secant stiffness calculations, a total of 10 unloading stiffness measurement
were calculated from the available data, including both partially and fully encapsulated Rebar A. Figure
5.14 compares the stiffness of the unloading phase with the corresponding tangent and secant stiffness
calculated for that rock bolt.
70 70
Fullmencapsulation
Partialmencapsulation
60 60
Fullmencapsulationmregression y/=/0.5356x/+/1.543
R²/=/0.7528
Tangent/Stiffness/(kN/mm)
Partialmencapsulationmregression
SecantmStiffnessmukN/mmp
50 50
All
ym=m0.199xmrm13.34 y/=/0.511x/+/1.4277
40 40
R²m=m0.2884 R²/=/0.7503
30 30
y/=/0.3272x/+/7.2054
20 20 R²/=/0.9949
10 10
0 0
0 20 40 60 80 100 120 0 20 40 60 80 100 120
UnloadingmStiffnessmukN/mmp Unloading/Stiffness/(kN/mm)
Figure 5.14: Comparison of Rebar A unloading stiffness to secant (a) and tangent (b) stiffness
As shown in Figure 5.14, a positive correlation exists between unloading stiffness and both secant
and tangent stiffness. However, the relationship has a much higher coefficient of determination for the
unloading stiffness–tangent stiffness relation. Note that this relationship appears to show that tangent
stiffness overestimates the amount of displacement attributable to the elastic deformation of the bolt
by a factor of 2 for partially encapsulated pull tests, and a factor of 3 for the fully encapsulated tests.
This demonstrates that tangent stiffness may be used as an indicator of the quality of the bond between
rebar and rock mass, although not necessarily a direct quantifier of elastic deformation of a rebar rock
bolt subject to a pull test.
20
18 17 mm
cone
16 plow
14
12
(tonnes)
6 mm cone plow
10
8
6
4
2
0
0 50 100 150 200 250
(mm)
Around 50 mm of displacement occurred before strain hardening is induced. Only 17 mm of this was
attributed to cone plough. Of the remaining 170 mm of displacement recorded, just 6 mm is attributed
to plough. This suggests that there is no constant load in a quasi-static loading scenario at which plough
occurs consistently before the bolt fails, although plough does occur as load increases. A conceptual
depiction of the behaviour is shown in Figure 5.16.
Steel stretching
Load
Eventual failure
of the bolt
Limited plow of the
cone (~ 20-40 mm)
Deformation
Figure 5.16: Conceptual load–displacement behaviour of a cone bolt subject to quasi-static loading
(Simser et al, 2006)
magnitude occurring between yield and strain hardening may be observed for a carbon steel, depending
on composition and manufacturing process (ASM International, 2002). It should be noted that testing
on the bolt shown in Figure 5.17 was stopped before failure occurred.
160 2
140
0
120
A B C D0
100
Load (kN)
80
60
40
20
0
0 10 20 30 40 50 60 70 80
Displacement (mm)
The linear plough–elastic deformation phase of the bolt was not observed in all pull test on the
MCB33. 17% of the tests exhibited a behaviour that suggested that the bolts responded to load with
relatively little plough, although they may exhibit a progressive softening of the bolt/grout system.
Figure 5.18 illustrates an example of such behaviour.
160
140
120
100
Load (kN)
80
60
40
20
0
0 10 20 30 40 50 60 70 80
Displacement (mm)
Figure 5.18: Pull test performed on an MCB33 without a linear plough response
The dashed red line in Figure 5.18 represents the linear elastic behaviour of a bar of equivalent length,
diameter and elastic modulus (200 GPa) as the tested bolt. Figure 5.19 focuses on the initial response
recorded for the same bolt, which seems to closely follow a second order polynomial trend between the
start of the test and the yield of the tendon (R2 = 0.9992). This gradual deviation from the elastic
response of the tendon suggests that as load increases, there is movement of the cone and the resin is
Chapter 5. Summary Statistics and Interpretation of Pull Test Data 59
becoming more damaged at an accelerating rate, but is competent enough to resist a linear plouging
response until the bolt tendon yields.
160
140
120
R2t=t0.9992
100
Load (kN)
80
60
40
20
0
0 1 2 3 4 5 6 7 8
Displacement (mm)
Examining both of these divergent displacement behaviours highlights the difficulties associated with
assessing the performance of a bolt. Cone bolts are designed to plough through resin under dynamic
loads, so from one perspective it may be considered proof of concept and a successful pull test if the
bolts plough under static conditions (as in Figure 5.17). However, if the cone bolt is loaded in static
conditions it is possible that as little deformation as possible is desirable (as in Figure 5.18), and signif-
icant displacements should ideally only occur during dynamic events if a bolt is to be said to perform
well in both loading conditions (although there is no widespread methodology used to test the dynamic
capabilities of a rock bolt in situ).
lt
Strain hardening of bo
Yield of bolt tendon ilure
To fa
Plough begins
Displacement
of the bolt will be a controlling factor on any stiffness metric dependent on the deformation of the bolt
tendon, all MCB33s in the database were the same length. As a result, stiffness is not normalized to
length, and is expressed in kN/mm.
160
140 SteelkYieldkStrength
120
PloughkStiffness
100
Load (kN)
80
SecantkStiffness
60
PloughkPoint
40
InitialkStiffness
20
0
0 10 20 30 40 50 60 70 80
Displacement (mm)
Figure 5.21: Performance metrics measured from a cone bolt pull test
The initial response to loading is measured by the initial stiffness. The plough point is the load at
which behaviour deviates from the initial stiffness, and plough begins. The plough stiffness measures the
gradient of the combined plough/elastic deformation response, until the steel yield strength. The secant
stiffness is the average stiffness between the first measurement of load and displacement and yield of the
bolt. The term “working capacity” is not used to describe the capacity of a cone bolt as its determination
in this context is somewhat ambiguous. Unlike the other grouted bolts discussed, the response to load
consistently deviates from linearity before the tendon yields. As such, the working capacity could be
defined as the plough point, but this neglects the fact that substantially more load may be borne by the
element at larger displacements. This also results in a potential complication if the bolt ploughs from
the beginning of a pull test, as the plough response becomes the linear response from which deviation
Chapter 5. Summary Statistics and Interpretation of Pull Test Data 61
defines working capacity. Values of cone bolt capacity used in the design of a support system should
depend on the design methodology, chiefly whether or not displacement is taken into account. As such,
the term “working capacity” will not be used in the context of cone bolts, and either the plough point
or the yield strength will be specified.
Figure 5.22 shows the distributions of plough point and yield strength. Table 5.7 shows the summary
statistics calculated for these two metrics.
14 100% 100%
14
12
80% 12 80%
Cumulative Frequency
Cumulative)Frequency
10
10
Frequency
Frequency
60% 60%
8
8
6 40% 6 40%
4 4
20% 20%
2 2
0 0% 0 0%
0 20 40 60 80 100 120 140 0 20 40 60 80 100 120 140
Plough Point (kN) Yield)Strength)(kN)
The observed distributions can be explained with a careful review of the pull tests compiled. The
initial peak in Figure 5.22a and the resulting heavy skew of the MCB33 yield strength distribution is
attributed to one testing campaign from 2006 which contributed 5 of the 7 values in the 80 - 90 kN bin
for the yield strength, and as a result is excluded from the calculation of statistics. The yield strength
is the variable with the lowest coefficient of variation, which is to be expected as it should be heavily
dependant on the properties of the steel as opposed to the resin. The plough point appears to be
normally distributed about an average of 53.2 kN, less than half of the yield strength of the bolt. There
are a large number of bolts that plough between 20 and 30 kN, which may be representative of bolts
where ploughing began during pre–loading.
Stiffness Metrics
Distributions for initial stiffness, plough stiffness and secant stiffness are shown in Figure 5.23. Summary
statistics are presented in Table 5.8.
Chapter 5. Summary Statistics and Interpretation of Pull Test Data 62
10 100% 7 100a
9 6
8 80% 80a
CumulativemFrequency
5
Cumulative(Frequency
7
Frequency
6 60% 4 60a
Frequency
5
3
40a
4 40%
3 2
20a
2 20% 1
1
0 0a
0 0% 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
0 2 4 6 8 10 12 14 16 18 20 22 24 26
Initial(Stiffness((kN/mm) PloughmStiffnessm(kN/mm)
12
80%
Cumulative Frequency
10
Frequency
60%
8
6 40%
4
20%
2
0 0%
0 2 4 6 8 10 12 14 16
Secant Stiffness (kN/mm)
The stiffness of a 200 GPa, 2.4 m bar with a diameter of 17.2 mm is 19.3 kN/mm. As the initial
reaction of the MCB33 to load is though to be a result of elastic deformation of the tendon, one would
expect the initial stiffness to be approximately equal to 19.3 kN/mm. However, very few bolts exhibit a
response this stiff. As seen in Section 5.3.3, a linear measure of stiffness may overestimate displacement
attributable to bolt deformation by a factor of 2 to 3, which could explain why so few bolts exhibit a
stiffness of 19.3 kN/mm. The secant stiffness of the MCB33 has a strong right skew and high coefficient of
variation; as it is a function of the four other parameters, non-normality is not unexpected. The plough
stiffness is quite evenly distributed between 1 and 3 kN/mm as indicated by the negative kurtosis,
showing that the cone–resin interaction appears to be highly variable. This could suggest that the in
situ dynamic performance of the cone bolt may be subject to a similarly variable energy capacity. It
must be noted that the pull test is strictly analogous to static or quasi-static loading of a bolt. The
dynamic performance of a reinforcement element may be assessed using impact testing (as reported in
Chapter 5. Summary Statistics and Interpretation of Pull Test Data 63
Hadjigeorgiou & Potvin, 2011), or using passive monitoring in situ (as in Morissette et al, 2014). The
MCB33 has been demonstrated to perform significantly differently when subject to impact loading versus
quasi-static (Doucet & Voyzell, 2012), as such the results of one loading rate may not indicate a similar
effect for another.
5.5 D-Bolts
5.5.1 Theoretical Behaviour of a D-Bolt
D-Bolt performance is difficult to evaluate from a conventional pull test. The principle behind the
design of the bolt is to evenly distribute load across the length of the smooth bar between two anchors,
potentially resulting in significant differences in load between two adjacent smooth sections. Published
laboratory static testing was performed across a simulated joint, where load is applied at the midpoint
between two anchors (Li, 2012). Figure 5.24 shows the apparatus and test set-up used for this test, and
Figure 5.25 shows the results.
Figure 5.24: Apparatus for a simulated joint laboratory test on a D-Bolt (Li, 2012)
Figure 5.25: Results of simulated joint laboratory tests on 20 mm D-Bolts (Li, 2012)
The test shown in Figure 5.25 is performed on two bolts grouted in cement: OP1 and OP2. OP2 has
a shrink sleeve on the test section while OP1 does not, otherwise the bolts are identical (Li, 2012. The
D-Bolts used by Vale do not have sleeves, and are grouted in resin). Bolt load is very similar for both
tests, and shows typical steel stress-strain behaviour. For test OP1, the plate experiences an increase
in load around when the “test section” or tendon starts to yield, while OP2’s plate is loaded slightly
Chapter 5. Summary Statistics and Interpretation of Pull Test Data 64
before. With such a small sample size, it is difficult to determine whether this difference in behaviour
may be attributed to the difference in bolt surfaces. In any case, it is apparent that load does appear
to propagate past the anchors into adjacent smooth sections, insinuating that there is some limited
movement of the anchors through the grout.
The D-Bolts in the database are 2.4 m in length with three 0.12 m long anchor sections, at 0.34
m, 1.14 m and 2.14 m along the length of the bolt. Although an in situ pull test is different from the
procedure shown in Figure 5.24 as load is applied to the head of the bolt, possible behaviour may be
hypothesised. Load could be limited to the first, 0.34 m segment of the bolt between the head and the
first anchor assuming good resin encapsulation, resulting in a very stiff response (the test sections for the
bolts shown in Figure 5.25 were 0.9 m in length; Li, 2012). If load does propagate past the first anchor,
a drop in stiffness would be observed as the adjacent smooth section will also be strained. Table 5.9
shows the stiffness that would be expected of both the 20 and 22 mm D-Bolts used at Vale’s Sudbury
operations, depending on the length of bolt exposed to load. This is defined by the anchor at which no
movement occurs, and thus prevents load from further propagating down the bolt. The steel used in the
manufacture of D-Bolts has an elastic modulus of 200-210 GPa (Charette, 2014).
The calculations shown in Table 5.9 should be used only as a rough guideline. An assumption is
made that the anchor sections are of equivalent diameter to the smooth sections. It also assumes that
if, for example, the second anchor is stable, the first bears no load and the same magnitude of stress
is present in both smooth sections on either side of it. If the first anchor were to be load-bearing, the
stiffness of the element as a whole would be dependent on the magnitude of that borne load.
250
200
150 Creighton,C2014
LoadC(kN) CopperCCliff,C2009
BeforeCfirstCanchorC(0.34Cm)
100
BeforeCsecondCanchorC(1.14Cm)
BeforeCthirdCanchorC(2.24Cm)
50
0
0 2 4 6 8 10 12
DisplacementC(mm)
An insight into how load is distributed along the bolt is provided by the displacement attributed to
plastic deformation of the bolt before strain hardening occurs. Laboratory testing performed on D-Bolts
by Li (2012; Figure 5.25) exhibited slightly more than 10 mm of perfectly plastic deformation between
yield and strain hardening on a smooth section of 0.8 m, corresponding to a strain of around 1.3%. The
bolt pulled at Creighton in 2014 shown in Figure 5.26 exhibits about 2.5 mm of near perfectly plastic
deformation before strain–hardening. Assuming similar steel was used for the manufacture of this bolt
as was used for the bolts tested by Li (2012), 2.5 mm of deformation representing a strain of 1.3%
corresponds to a deformation length of about 190 mm. The distance between the end of the threaded
section of the D-Bolt head and the first anchor is 215 mm. This calculation is complicated by the fact that
the 22 mm D-Bolt uses an M24 x 3.0 thread (Normet, 2014). This thread has a major (i.e. maximum)
diameter of 24 mm and the thread crests are 3 mm apart. The minor (i.e. minimum) diameter for
an M24 x 3.0 thread is 20.7 mm for flat form threads (ASME, 2005). Assuming the threading process
does not significantly alter the yield strength of the bolt material, the lowest diameter part of the bolt
(the 20.7 mm minor thread diameter) will yield first, about 20 kN before the bar as cross-sectional area
is proportional to yield load. The thread is by definition of varying diameter, so yield will be a more
protracted response (as observed in Figure 5.26) in comparison to the yield of the smooth bar section
alone in Figure 5.25. As the major (maximum) diameter is 24 mm, the 22 mm smooth section will
yield before the entire threaded section does. As a result, it can be concluded that the near perfectly
plastic response observed in Figure 5.26 is attributed to the first smooth section between the thread and
the first anchor, implying that although the bolt acts in a manner soft enough to suggest little to no
anchorage provided by the first two anchors, the first section bears the most load. This does not agree
with the findings of an analysis solely of secant and tangent stiffness, thus demonstrating the value of
measuring displacement beyond the working capacity and analysing bolt behaviour.
However, the measured stiffness of the bolts should not be ignored. D-Bolts are often pulled until
just after, or even before, their working capacity. Figure 5.27 shows the results of three 20 mm D-Bolts
installed in the same rock mass at the same location during one testing campaign. These bolts were
installed at angles between 15◦ and 30◦ off of perpendicular from the wall face. It is not clear how this
Chapter 5. Summary Statistics and Interpretation of Pull Test Data 66
affects the results, although the appearance of a working capacity of about 120 kN was attributed to
movement of the testing rig (the minimum yield load of a 20 mm D-Bolt is 140 kN).
160
140
120
100
Loadl(kN)
Boltl1
80 Boltl2
Boltl3
60
Beforelfirstlanchorl(0.34lm)
40 Beforelsecondlanchorl(1.14lm)
Beforelthirdlanchorl(2.24lm)
20
0
0 1 2 3 4 5 6 7
Displacementl(mm)
As in Figure 5.26, the expected stiffness of varying length of smooth bar in tension is displayed, in
this case the diameter of which is 20 mm. Bolt 1 has the least stiff response, on average performing
very similar to what would be expected if only the third anchor was stable. Bolt 2 is the most stiff,
its behaviour indicating that strain in the bolt is largely concentrated before the first anchor. After an
initially softer response, Bolt 3 develops a stiffness that almost exactly matches the expected deformation
of a 1.14 m bar. The initial response could be explained by rock mass compression, or perhaps by the
second anchor fully seating after limited displacement and preventing further development of load on
the third anchor. With these three pull tests, assuming the installation and ground conditions are near-
identical between them, it is clear that a variety of behaviours can be expected from the D-Bolt in a
pull test.
Limited unloading data for D-Bolts was also present in the database. The results of a single campaign
on 22 mm D-Bolts are shown in Figure 5.28. Ground conditions at the location of the pull tests (which
were performed in the back) are described in the pull test report as “broken”, and the large displacements
observed during loading are likely a result of this. As was the case with rebar, the unloading phase is
used as a way to measure elastic deformation of the bolt. Unlike the rebar pull tests, the unloading
phase was explicitly targeted by the personnel conducting the pull test, and data was recorded manually.
As such, the stiffness may be calculated with the maximum and minimum load values recorded. Table
5.10 shows the unloading stiffness calculated for the three pull tests.
Table 5.10: Unloading stiffness calculated for pull tests performed on 22 mm D-Bolts
Test Stiffness
1 635 kN/mm
2 54 kN/mm
3 36 kN/mm
Chapter 5. Summary Statistics and Interpretation of Pull Test Data 67
160
140
120
100
Load (kN)
80
Test 1
60 Test 2
Test 3
40
20
0
0 10 20 30 40 50 60
Displacement (mm)
Each of the three tests exhibits different degrees of elastic recovery during the unloading phase,
implying variable load distributions between the three bolts. Comparing the stiffness calculated for these
bolts with the different values of stiffness expected of a 22 mm D-Bolt (Table 5.9), certain similarities are
observable. Test 1 exhibits very little displacement during unloading, resulting in a very high stiffness
value. It must be acknowledged that for such high values of stiffness, the measured displacements will be
correspondingly very small, and any error in measurement will be magnified. Test 3 unloads in a manner
that would be similar to that of a bolt with an even load distribution along its length until the third
anchor, and Test 2 as if the bolt were subject to a reduced load between the second and third anchors.
The fact that different bolt behaviours are characterized over limited testing agrees with the conclusions
drawn from Figure 5.27; the stiffness of the D-Bolt is dependent on the mobility of its anchors, which is
apparently subject to change between bolts even if installed in very similar conditions.
An in-depth analysis of working capacity is omitted due to an insufficient amount of data, with a total
of 6 reliably observed working capacity values (to the nearest ton) between the two D-Bolt diameters.
As too few measurements of working capacity exist in the database to reliably characterise dispersion,
all values of working capacity for the D-Bolt present in the database are shown in Table 5.11.
Table 5.11: Working capacities obtained from all D-Bolt pull tests
Stiffness
Figure 5.29 shows the distribution of stiffness for the two diameters of D-Bolt. Refer to Table 5.9 for
the anticipated stiffness of each bolt for different distributions of load along its length.
CumulativekFrequency
CumulativesFrequency
80.00%
6 60.00% 22smm
6
60) 60%
Frequency
Frequency
5 20smmFrequency 5
40.00%
Cumulatives)
Cumulative %
4 20.00% 4
40) 40%
3 0.00% 3
40
45
50
55
60
65
70
75
80
85
More
30
35
2 20) 2 20%
Bin
1 1
0 0) 0 0%
0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80
TangentsStiffnesss(kN/mm) SecantkStiffnessk(kN/mm)
These figures show that in most of the pull tests performed, either all bolts acted in a manner that
would suggest that the load distribution extended beyond the first anchor (and the second anchor for
the majority of bolts), or rock mass compression played a large role in the tests. Two 20 mm bolts have
tangent stiffnesses that strongly suggest the second anchor is firmly held in the resin, and although there
is one test with higher stiffness, none of the tests approach the stiffness that would be observed if there
were no movement in the first anchor. Table 5.12 shows summary statistics for the D-Bolts.
The undertaken analysis is to be interpreted carefully. A very small volume of data is available, and
in reality there are three possible clusters of bolt stiffness, one for each anchor. Normal distributions
are not to be expected with such variable behaviour. Insufficient data was obtained to examine each
of these clusters on an individual basis. The data appears to portray the 20 mm D-Bolts as generally
acting stiffer than the 22 mm equivalent. This is likely misleading; two thirds of the 20 mm D-Bolt pull
tests were performed at Copper Cliff Mine, on 3550 and 3710 Levels (1,080 and 1130 m), while over half
of the 22 mm D-Bolt pull tests were performed at Creighton Mine at twice that depth, between 7680
and 7940 Levels (2340 and 2420 m) in rock masses noted to be fractured. While the quality of the rock
was not noted for the shallower Copper Cliff tests, the lower far field stresses may correlate to a less
intensely damaged rock mass in which the bolts were installed.
Chapter 5. Summary Statistics and Interpretation of Pull Test Data 69
120
WorkingWCapacity
100
Pm12W(2.4Wm)
80
PcPm12W(1.8Wm)
Load (kN)
60 PcMn12W(2.4Wm)
1.8WmWStiffness
40 2.4WmWStiffness
20
0
0 5 10 15 20 25
Displacement (mm)
Figure 5.30: Pull tests performed on Pm12 and Mn12 expandable bolts
Some expandable bolts exhibit an extremely stiff response before gradually softening as load increases,
implying that a limited section of the bolt is initially deforming and the decoupling front mobilizes as
load increases (Li & Stillborg, 1999). Conversely, some bolts have a very soft initial response before
stiffening. This may be due to the response of the rock mass, but may potentially also indicate bolt
movement. The third behaviour is simply a linear response of uniform stiffness, and the fourth behaviour
is highly variable, with no discernible pattern in response to load.
A number of partial embedment tests are included in the database. In these tests, the bolt is inserted
Chapter 5. Summary Statistics and Interpretation of Pull Test Data 70
into a steel tube with a diameter less than that of the installation hole, and a limited length left exposed
(in the cases present in the database, this is usually 1’, or .3048 m). When the bolt is inflated, only the
exposed part of the bolt comes in contact with the rock mass and provides anchorage. Table 5.13 shows
bolt behaviour by bolt type, omitting partial embedment tests. Acknowledging very limited data, it is
interesting to note that the Pm12 bolts appear to soften or displace in a uniform manner more often
than the other bolt types, although it must be noted that all pull tests on the Pm12 were performed in
sandfill or paste.
If the objective of a pull test on an expandable bolt was to determine the strength of the bolt–rock
interface (as it is for an FRS), then a length–normalized value of capacity would be adopted as capacity
would be dependent on the surface area (and thus length) of the interface. However, Swellex bolts in the
database generally yield before slipping extensively. The working capacity of the bolt is thus dependent
on the cross-sectional area of the steel, and independent of the bolt’s length. As such, load is expressed
in absolute terms (i.e. kN). In these cases it is only possible to calculate a minimum strength of the
bolt–rock interface.
Bolts subject to a partial embedment pull test usually fail by slip, in which case the shear strength
of the interface can be calculated in terms of load per length. While a limited number of fully embedded
pull tests on both Pm12 and Pm24 did fail by slipping, all tests were performed in sandfill. Table 5.14
summarizes all slipped tests. Insufficient data is available to distinguish between bolt configurations.
Table 5.14: Coupling strength of partially embedded and slipped Swellex pull tests
It is worth noting that 8 of the 12 slipped fully embedded Swellex bolts came from only 3 testing
campaigns, suggesting that slip in fill generally occurs when there is an underlying issue with either bolt
installation or the fill itself. It is acknowledged that for the fully embedded tests, there does exist an
upper bound on the value of couple strength calculable, as it is not possible to determine for tests that
do not slip. As such this should not be seen as representative of all bolts, while the partial embedment
tests may be a better approximation. With an average coupling strength of 228 kN/m, the mechanical
interlocking of an inflatable bolt with the rockmass seems to greatly contribute to the strength of this
interface; as seen in Section 5.2, an FRS tends to reach its ultimate capacity (i.e. fully decouple) at 40
kN/m.
Contrasting the number of tests where Swellex bolts slipped in the current database with the number
of slipped tests in a pull test database previously assembled by Soni (2000) shows a large difference in the
proportion of slipped tests relative to what Soni terms as “destructive” or “non-destructive” tests. 12 out
of 111 tests (excluding partial embedment tests and tests performed with malfunctioning equipment) in
Chapter 5. Summary Statistics and Interpretation of Pull Test Data 71
the database assembled for this thesis are known to have slipped. Soni’s database included 173 “slipped”
tests out of 304 entries in the data set (Soni, 2000). This is a result of a fundamental difference in the
definition and interpretation of a “slipped” test. Tests in which 15 mm of displacement or more was
observed were classified as “slipped” by Soni (2000). For the purposes of this thesis, a “slipped” test
occurs if the pull tester is unable (and acknowledges their inability) to build further load on the bolt
as it continues to displace at loads that do not suggest plastic deformation of the bolt. Soni is not
necessarily incorrect; it is possible that slip is the displacement mechanism that contributes to the low
stiffness calculated for the bolts in the database, but it would be slip comparable in observed effect to
cone plough, only occurring as load increases rather than the sustained slip at constant load observed
in the case of an FRS.
Figure 5.31 shows the distributions of working capacity for the Pm12 and Mn12, with further information
in Table 5.15. Working capacity is measured to the nearest 1 ton (8.9 kN), as this is generally the
resolution used for pull tests performed by Atlas Copco. In general, pull tests on the Pm24 and Mn24
bolts were verification tests, likely due to the large loads involved, and very rarely were the bolts
loaded until they yielded. Load resolution for these tests was 2 tons (17.8 kN), further complicating the
interpretation of the data. As such, an analysis on the working capacity of the Pm24 and Mn24 bolts
was not performed.
13 100P
12
11
80P
10
CumulativepFrequency
9
8 Mn12
60P
Frequency
Pm12
7
Cumulative
6
5 40P
4
3
20P
2
1
0 0P
0 20 40 60 80 100 120 140 160 180
WorkingpCapacityp(kN)
Table 5.15: Summary statistics of Swellex Pm12 and Mn12 working capacity
Bolt n x̄ s ĉv
Pm12 15 88.4 kN 5.5 kN 0.06
Mn12 5 96.1 kN 7.4 kN 0.08
Mn12 bolts appear to have a higher working capacity than the Pm12 bolts. This is attributed to one
outlier and the low number of tests on the Mn12. As expected, the coefficients of variation are relatively
Chapter 5. Summary Statistics and Interpretation of Pull Test Data 72
low, as the working capacity is dependent on the material properties of the bolt.
Stiffness
Figure 5.32 shows the distributions of bolt secant stiffness for all variants, and Table 5.16 shows summary
statistics of secant stiffness.
14 100% 7 100%
12 6
80% 80%
Cumulative3Frequency
CumulativeNFrequency
10 5
Mn12 Mn24
Frequency
60% 60%
Frequency
8 Pm12 4 Pm24
Cumulative Cumulative
6 40% 3 40%
4 2
20% 20%
2 1
0 0% 0 0%
0 4 8 12 16 20 24 28 32 0 4 8 12 16 20 24 28 32
Secant3Stiffness3(kN/mm) SecantNStiffnessN(kN/mm)
The average secant stiffness for the Pm24 and Mn24 bolts is significantly greater that of the Pm12
and Mn12 bolts. This corresponds to the larger cross-sectional area of the Pm/Mn24 configurations.
However, the values of secant stiffness are well below what would be expected if the only displacement
mechanism were to be axial deformation. Using a maximum bolt length of 2.44 m for the Pm12 and
Mn12, a stiffness of 20.1 kN/mm would be observed for pure axial deformation spread evenly across
the length of the bolt, and for the Pm24 and Mn24 with a maximum length of 3.6 m, a stiffness of
26.7 kN/mm would be observed. If a deformation model similar to that of an FRS is considered with a
decoupling front marking the onset of steel deformation as postulated by Li & Stillborg (1999), higher
stiffness should be observed considering the strength of the couple appears to be in the region of 200
kN/m (Table 5.14). Of the Pm12 and Mn12 bolts, none approached this stiffness, and only one Pm24
and one Mn24 bolt surpassed their respective minimum anticipated stiffness. Of the few values of tangent
stiffness recorded (16 across all bolts), none reached their minimum anticipated values. This strongly
suggests a displacement mechanism beyond axial elastic deformation is being mobilised.
Within bolt sizes, the Pm and Mn variants seem to perform quite similarly to one another. It is
also noted that the skew value of the Pm12 is significantly less than that of the other 3 bolts; the other
distributions appear more log-normal in nature, although the sample size seems to be too small to be
definitive. The Pm12 also has the lowest coefficient of variation. This may be the result of installation
Chapter 5. Summary Statistics and Interpretation of Pull Test Data 73
only in backfill, a relatively controlled substance when compared with the variety of rock masses that
may be encountered in the Sudbury Basin. Figure 5.33 shows the secant stiffness for the different bolt
types sorted by the medium in which they were installed, and Table 5.17 shows summary statistics.
t-tests are performed assuming unequal variances (the properties of sandfill and of various lithologies
are not assumed to be equally variable), and the p-value is calculated for a two-tailed test.
35 35
Secant Stiffness (kN/mm)
Table 5.17: Secant stiffness summary statistics on Swellex sorted by installation medium
As the Mn variants of Swellex are designed as yielding bolts, they are generally not installed in
backfill, and very few of the Pm bolts tested were installed in rock or ore. The Pm12 bolts installed
in sandfill appear to react with similar stiffness as the Mn12s installed in rock or ore; with a p-value of
0.524, the null hypothesis is not rejected. However, both the Pm24s and Mn24s installed in rock or ore
seem to react significantly stiffer than the Pm24s installed in sandfill (from the t-test, p=0.027). This
discrepancy in observed effect of the installation medium may be due to a lack of data, be it volume or
the testing of Pm12s exclusively in sandfill, and Mn12s only in rock and ore. Having observed that in
the assembled database slipping failure of the entire length of a Swellex bolt only appears to occur in
sandfill, it does appear that inflatable bolt performance in rock and in backfill is not equivalent. This is
attributed to the difference in material properties between the two installation media.
5.7.1 Yield-Lok
Two pull test campaigns were performed on Yield-Lok bolts by Vale’s Sudbury operations between 2011
and 2014, both at Totten Mine. A total of 9 bolts were tested, one of which is discarded as the data
is incongruous with the other tests. One campaign was a verification test, and bolts were not pulled
to their working capacity (in the case of the Yield-Lok defined by the yield strength of the tendon),
resulting in very limited data. A summary of the working capacity and secant stiffness of the bolts is
shown in Table 5.18.
Although only three working capacities were recorded, they were all around 150 kN. The 2013 testing
campaign stopped testing at about 16 tons (142 kN) without observing yield. The expected stiffness of
a 3/4”, 8’ (19 mm, 2.44 m) bar is 23.4 kN/mm. One test acts in a stiffer manner, although the bolt was
unloaded midway through the test and then reloaded, which may have affected the result. The lower
stiffness of the other tests suggests that there is some degree of movement of the upset head through the
polymer in static conditions, although it is possible that this is the rock mass response being measured.
19 tons (169 kN), with one test reaching the 25 ton (222 kN) capacity for the pump. This falls well short
of the FRP rebar failure loads anticipated, so resin performance was not verified for higher loads.
5.7.3 DS Bolt
One testing campaign was performed on 4 DS Bolts (now known as the VersaBolt; Lamothe, September
2014). Comparable to the D-Bolt, it is a smooth bar punctuated by oval anchors along its length. 20.5
mm bolts were pulled at Totten on April 19th , 2013. Load and displacement data was recorded for 3
of these tests. Working capacity and tangent and secant stiffness are shown in Table 5.19. It should be
noted that all bolts exhibited a very gradual yielding behaviour in comparison to the D-Bolt.
The typical thread and bar yield strengths of the 20.5 mm DS Bolt are 125 kN and 138 kN respectively
(courtesy of Mansour), indicating yield of the thread.
Table 5.20 shows that the data collected for these bolts is of insufficient quality for the purposes of
this thesis. As such, no further analysis was performed.
5.7.5 MD Bolt
A campaign testing 14 MD Bolts, supplied by Sandvik, was conducted at Copper Cliff mine. Only “peak
load” and displacement at that peak load were recorded. It was not apparent how peak load was defined
in the report for this trial, potentially denoting slip, yield, failure or an arbitrary load at which the test
was stopped. Additionally, without intermediate displacements recorded during loading, the behaviour
of the bolt cannot be fully assessed from this dataset. As a result, the MD bolt was not investigated
further.
Chapter 5. Summary Statistics and Interpretation of Pull Test Data 76
5.8 Summary
Performance metrics have different degrees of variability for the reinforcement elements discussed. Met-
rics that are dependant primarily on the material properties of the bolt itself (i.e. working capacity as
defined by the yield strength of the steel) tend to have relatively low variance. Larger variance observed
for other metrics (such as measures of stiffness, or ultimate capacities defined by frictional as opposed
to mechanical properties) may be attributed to variable installation conditions as a result of a number
of potentially influential factors, as will be discussed in Chapter 6.
The ultimate capacity of an FRS appears to be independent of the nominal diameter of the bolt.
FRS A and FRS B bolts perform similarly, with average ultimate capacities of about 40 kN/m. The
stiffness of FA35 and FA39 bolts was found to be similar, although displacement is comprised of various
degrees of limited slip and elastic deformation of the bolt. Consequently, stiffness as calculated in this
thesis is a somewhat arbitrary descriptor of how an FRS behaves.
The rebar rock bolts tested yielded in a relatively consistent manner, although stiffness is much more
variable. In most cases, stiffness values are below the minimum that would be expected of an equiva-
lent non-grouted steel bar. Investigating the unloading of the rebar shows that while the stiffness (in
particular tangent stiffness) of the bolt may be proportional to the elastic deformation, it overestimates
deformation by a factor of at least 2. Partial encapsulation tests show that even a limited bond length
between the rebar and the rockmass has sufficient strength to surpass the yield strength of the rebar.
For the cone bolts, five performance metrics were identified to describe behaviour. The bolt head is
displaced with an initial stiffness while the cone remains anchored in the resin. The cone then begins to
displace at the plough load, and a new linear relationship between load and displacement is developed.
This was dubbed “plough stiffness,” and accounts for simultaneous bolt movement and deformation.
This linear displacement behaviour occurs until the bolt tendon yields, at which point the test is usually
stopped as large plastic deformation of the bolt tendon is induced. This sequence of events is not
universal; a fraction of the pull tests show that the MCB33 may potentially yield before any significant
plough occurs.
Divergent behaviours appear to be characteristic of D-Bolt pull tests for the collected data set.
Although relatively little data was available, it could be observed that stiffness of the bolt may be a
function of the length between the point of load application and the first, second or third anchor on
the bolt, or a value in between or even beyond any of the anchors. The anchors have the potential to
prevent load transmission down the entire bolt length. However, movement of one or more anchors on
the bolt results in stress transmission and a less stiff response.
A large variety of expandable bolt test data was collected; bolts manufactured with different steels,
coatings and capacities were tested at Vale’s Sudbury operations. Limited data volume resulted in the
analysis of four types of Swellex: the Pm12, Mn12, Pm24 and Mn24. These bolts also had several
possible behaviours: the bolt may displace linearly with load, but may also stiffen, soften, or displace in
a non-uniform manner. This resulted in variable stiffness, although the stiffness generally reflected the
capacity of the bolt used; higher capacity bolts are made with thicker, larger tubes resulting in stiffer
behaviour. As there exist four bolt configurations split between installation in backfill and rock, and
such divergent behaviours are consistently observed, meaningful analysis is difficult.
Bolts for which a small amount of data is available include the Yield-Lok, FRP rebar, the DS Bolt,
one Python bolt configuration, two Omega Bolt configurations and the MD Bolt. Further discussion of
these bolts in this thesis is limited.
Chapter 5. Summary Statistics and Interpretation of Pull Test Data 77
The interpretation of the pull tests was consistently hindered by displacements attributed to rock
mass or surface support compression, which did not represent a reaction of the bolt or its anchoring
mechanism. This led to stiffness calculations resulting in lower-than-anticipated values. This could be
addressed by implementing an alternative method of measuring displacement, such as that described in
ASTM D4435-13. However, stiffness as measured in the creation of the database may still be used as a
relative metric, albeit requiring slightly more interpretation.
Table 5.21 summarizes the findings of this chapter. Note that in the case of the FRS, ultimate
capacity as opposed to working capacity is shown, with capacity given per metre of anchorage length.
In the case of the MCB33, the yield load is shown.
Table 5.21: Summary of working capacities for all bolts pull tested
Recognizing the limitations of the testing methods, the statistical analysis provides useful indicators
of the working capacity of several bolt types in underground hard rock conditions. Chapter 5 focuses
on how factors related to the installation of the bolt, the rock mass in which the bolt was installed in
and the characteristics of the bolt itself affect the performance of reinforcement elements in underground
hard rock mines.
Chapter 6
The behaviour of rock bolts as recorded by a pull test is influenced by a number of factors, some of
which may be specific to a particular type of bolt. These can be grouped into three categories: those
pertaining to the bolt itself, those that are specific to the installation of the bolt, and factors associated
with the rock mass in which the bolt is installed. As the behaviour and performance of different bolt
types is dictated by different mechanisms, the influence of various factors are examined for each type of
bolt individually.
The nature, quantity, and quality of information recorded varied between rock bolt suppliers and
was also dependent on the type of bolt being tested. Sufficient data was recorded only for an analysis
on factors that affect the performance of FRS bolts, rebar rock bolts, and MCB33s. This chapter will
describe these analyses.
78
Chapter 6. Factors Influencing Pull Test Performance 79
120
100
60
40
20
0
0.0 0.5 1.0 1.5 2.0 2.5
Anchorage Length (m)
Figure 6.1: Relationship between ultimate capacity and length for the FA35
Table 6.1: Statistics on the relationship between ultimate capacity and length for the FA35
As only two treatments (lengths) are present in the FA35 data, a t-test was performed to compare
their means. The critical t- and the p-values shown in Table 6.1 are for a one-tailed t-test assuming
unequal variance (α = 0.05). As the p-value is 0.0111, the null hypothesis is rejected, implying a
significant difference between the two means. It is acknowledged that there was very little data for the
bolt with 1.37 m of anchorage (10 data points from 2 testing campaigns), especially considering the wide
range of installation conditions that may be encountered. This is addressed to some extent by assuming
unequal variances when performing the t-test, but it should also be taken into account that there is
only a 0.15 m difference in length. It seems unlikely that this relatively small additional length of the
bolt/rock mass interface has a strength of 79 kN/m, while the entire interface length of the 1.37 m bolt
has an average strength of 36 kN/m.
The same statistical test (one-tailed, unequal variance, α = 0.05) is performed for the FB35. A third,
intermediate length of bolt was pull tested, but with only four tests performed it was omitted from the
analysis, and the t-test was performed on the 1.52 m and 1.83 m bolts (presented in Figure 6.2 and
Table 6.2).
Chapter 6. Factors Influencing Pull Test Performance 80
120
100
60
40
20
0
0.0 0.5 1.0 1.5 2.0 2.5
Anchorage Length (m)
Figure 6.2: Relationship between ultimate capacity and length for the FB35
Table 6.2: Statistics on the relationship between ultimate capacity and length for the FB35
The null hypothesis was rejected, albeit marginally, which once again suggests a length–dependency
of performance. Although the data set was larger for this test, it was still limited and the results of the
t-test are difficult to interpret. In Figure 6.3 and Table 6.3, the equivalent graph and table for the FB39,
the opposite trend seems to be present for the larger bolt diameter.
120
100
Ultimate Capacity (kN)
80
60
40
20
0
0.0 0.5 1.0 1.5 2.0 2.5
Anchorage Length (m)
Figure 6.3: Relationship between ultimate capacity and length for the FB39
Chapter 6. Factors Influencing Pull Test Performance 81
Table 6.3: Statistics on the relationship between ultimate capacity and length for the FB39
Three lengths of FB39 were tested, so the t-test must be abandoned in favour of ANOVA. For the
purposes of this analysis, single factor ANOVA using a fixed effect model was used. Single factor ANOVA
on a fixed effect model is to be applied when one factor has been isolated. As the database is composed
of different testing campaigns which have variable installation, rock mass and bolt parameters, it is very
difficult to isolate a single factor when comparing two or more campaigns. As such, the results of the
analysis presented in the Table 6.4 are to be interpreted cautiously, as the degree of confidence calculated
is calculated under the assumption that a single factor is varied between the datasets.
Table 6.4: Single factor ANOVA performed on the relationship between ultimate capacity and length
for the FB39
Total 35313 84
A p-value of 0.0032 was calculated, thus the null hypothesis was rejected. This analysis implies that
the capacity of an FB39 in fact decreases with length, in direct contradiction to the findings of the
analyses on the FA35 and FB35 datasets, the model proposed by Li & Stillborg, and a conventional
understanding of friction. This shows the shortcomings of this type of analysis on data sets with high
potential variability not necessarily captured by the low volume of data. As a result, the confidence in
the findings of similar analyses on the length–dependency of the FA35 and FB35 is affected.
In order to address the problem of a lack of data volume, datasets composed of each bolt configuration
were combined. Table 6.5 shows that all 1.52 m bolts were either 35 or 39 mm nominal diameter,
and almost all 1.83 m bolts were 46 mm nominal diameter. As such, this analysis is only valid if the
assumption is made that there is no difference in performance between different diameters and suppliers of
FRS. Although from Section 5.2.3 this does appear to be the case, it must be a considered a shortcoming
of the analysis. The overall dataset is shown in Figure 6.4.
1.52 m 1.83 m
FA35 30.5% 0%
FA39 12.9% 0%
FA46 0% 31%
FB35 29.2% 12.7%
FB39 27.5% 0%
FB46 0% 56.3%
Chapter 6. Factors Influencing Pull Test Performance 82
Distribution/p10/kN/m/binsc
0a 10a 20a 30a 40a 50a
160
L=1.52m
140
L=1.83m
Ultimate/Capacity/pkNc
120 y/=/14.947x/h/39.486
R²/=/0.0207
100
80
60
40
20
0
0.0 0.5 1.0 1.5 2.0 2.5
Anchorage/Length/pmc
Figure 6.4: Relationship between ultimate capacity and anchorage length for all FRS bolts
As seen in Figure 6.4, the coefficient of determination for the linear regression performed was very
low. However, as the majority of the data is for bolt lengths of 1.52 m and 1.83 m, these two subsets were
further analysed. The distributions for both are shown in Figure 6.4. The longer bolt does appear to
generally have a larger load capacity as it not only has a larger mean, but the overall distribution occurs
at higher loads. Table 6.6 directly compares the two data sets, including the results of a one-tailed t-test
assuming equal variances (α = 0.05).
Table 6.6: Comparison of 1.52 m and 1.83 m of anchorage length for all FRS bolts
The results of the t-test performed in Table 6.6 strongly suggest a difference between the mean
ultimate capacities of the two FRS lengths. The low coefficient of determination calculated as part of
the linear regression may be explained by the high variability in ultimate capacity. This limitation is
overcome by the t-test through sheer volume of pull test results, allowing a thorough characterisation of
distributions for two lengths of bolt.
Tomory et al. (1998) do not discuss the relationship between capacity and FRS length. However,
Tomory (1997) found no observable trends in an analysis that spanned a larger variety of bolt lengths
than is presented here. This analysis was performed on 475 39 mm Split Sets, over a much wider range of
installation conditions (over 50 mines participated in the study), and it was hypothesized that a variety
of factors obscured any potential relationship between bolt length and ultimate capacity (Tomory, 1997).
Similar conclusions must be drawn for the unsuccessful linear regression presented in Figure 6.4. However,
on both the bases of the analysis presented in Table 6.6, as well as existent theoretical justification (Li
& Stillborg, 1999), a load per unit length basis of performance evaluation will be used for the remainder
of the FRS analysis in this chapter.
Chapter 6. Factors Influencing Pull Test Performance 83
70
60
UltimatedCapacityd(kN/m)
50
Bolter-installed
10
0
Jackleg Bolter
Figure 6.5: Comparison of ultimate capacities between jackleg and bolter installations of the FA35
Table 6.7: Ultimate capacity statistics for jackleg and bolter installations of the FA35
Jackleg Bolter
n 31 44
x̄ 32.60 kN/m 46.25 kN/m
s 8.49 kN/m 7.47 kN/m
ĉv 0.26 0.16
t0 −7.2
tcrit 2.001
p 6 ∗ 10−10
There is a significant difference between the means of the two data sets, with bolter-installed FRSs
exhibiting almost 50% more capacity than a jackleg-installed bolt. The jackleg bolts represent four
testing campaigns in different conditions, and 90% of the bolter-installed bolts outperform 65% of the
jackleg-installed bolts. The t-test strongly suggests that installation method is a critical factor in regards
to FRS performance. In Figure 6.6 and Table 6.8, the dataset is expanded to compare installations of
all FRS configurations.
Chapter 6. Factors Influencing Pull Test Performance 84
30%
25%
MacLean
20% Jackleg
Frequency
15%
10%
5%
0%
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75
Ultimate Capacity (kN/m)
Figure 6.6: Comparison of ultimate capacities between jackleg and bolter installations of all FRS bolts
Table 6.8: Ultimate capacity statistics for jackleg and bolter installation of all FRS bolts
Jackleg Bolter
n 39 289
x̄ 31.8 kN/m 39.8 kN/m
s 7.83 kN/m 11.59 kN/m
ĉv 0.25 0.29
t0 5.56
tcrit 1.67
p 0.0000
While relatively few bolts are added to the jackleg-installed dataset, many more bolter-installed bolts
are incorporated into the analysis. The conclusions are similar, although Table 6.7 seems to overstate
the difference in bolt performance. A bolter-installed FRS appears to have approximately 25% greater
capacity than an FRS installed with a jackleg. This may be due to the stability that a bolter provides
during drilling. A steadier drill would result in a smaller, more accurately drilled hole. This would in
turn result in a tighter fitting FRS with higher radial stresses, and greater frictional resistance to pull.
In addition to the productivity and safety benefits of mechanized bolting, it also appears to result in a
higher quality FRS installations.
In Section 5.2.3, it was found that the average FRS ultimate capacity was greater than that found by
Tomory et al. (1998) for the Split Set (38.9 kN/m versus 31.9 kN/m). Tomory et al. (1998) state that
the jackleg was “commonly used” for bolt installation, while for the database assembled for this thesis,
bolter installation is much more common. In fact, comparing the average ultimate capacity of Split Sets
pull tests collected by Tomory et al. (1998) and the average capacity of pull tests on jackleg-installed
FRS in this thesis, very similar capacities are observed (31.8 kN/m in the current database versus 31.9
kN/m found by Tomory et al; 1998). As such, the discrepancy in average load values is attributed to
Chapter 6. Factors Influencing Pull Test Performance 85
160 160
140 yI=I1.4847xI+I38.407 140
UltimatenCapacityn(kN)
UltimateICapacityI(kN)
R²I=I0.5252
120 120 yn=n0.4436xn+n56.484
100 100 R²n=n0.03
80 80
60 60
40 40
20 20
0 0
0 20 40 60 80 100 0 20 40 60 80 100
InstallationITimeI(s) InstallationnTimen(s)
UltimateICapacityI(kN)
R²I=I0.3946
120 yI=I1.119xI+I46.415 120
R²I=I0.1513
100 100
80 80
60 60
40 40
20 20
0 0
0 20 40 60 80 100 0 20 40 60 80 100
InstallationITimeI(s) InstallationITimeI(s)
UltimatenCapacityn(kN)
R²n=n0.2298
R²I=I0.5517
120 120
100 100
80 80
60 60
40 40
20 20
0 0
0 20 40 60 80 100 0 20 40 60 80 100
InstallationITimeI(s) InstallationnTimen(s)
Figure 6.7: Relationship between drive time and ultimate capacity for all FRS bolts
Most configurations of FRS appear to show a trend between installation time and ultimate capacity.
The two exceptions are the FB35 and the FA39. In the case of the FA39, there are relatively few data
points, and those present are concentrated in a narrow band of installation times. In the case of the
Chapter 6. Factors Influencing Pull Test Performance 86
FB35, the low coefficient of determination appears to be due to different installation methods. Figure
6.8 distinguishes between the bolts known to have been installed with a MacLean bolter (black), and
those installed with either a jackleg or with equipment not noted in their respective pull test report
(red).
160
140 yn=n2.2248xn+n44.69
R²n=n0.4931
UltimatenCapacitynukNw
120
100
80 Bolter
Jacklegnornunknown
60
40
20
0
0 20 40 60 80 100
InstallationnTimenusw
Figure 6.8: Relationship between drive time and ultimate capacity for FB35, distinguishing between
installation methods
Separating the MacLean-installed FB35 bolts results in a clearer relationship (R2 = 0.493, compared
to 0.030 for the full data set). This shows that although jackleg–installed FRS bolts have lower load
capacities, they require a similar amount of time for installation. Figure 6.9 and Table 6.9 show the
results of analysing data only from pull tests known to have been installed with a MacLean Bolter.
Table 6.9: Description of relationships between drive time and ultimate capacity for FRSs installed
with a MacLean Bolter
160 160
140 140
UltimateICapacityI(kN)
UltimatenCapacityn(kN)
yI=I0.8646xI+I56.744 yn=n2.2248xn+n44.69
120 R²I=I0.6984 120 R²n=n0.4931
100 100
80 80
60 60
40 40
20 20
0 0
0 20 40 60 80 100 0 20 40 60 80 100
InstallationITimeI(s) InstallationnTimen(s)
UltimateICapacityI(kN)
120 yI=I1.119xI+I46.415 120
R²I=I0.1513
100 100
80 80
60 60
40 40
20 20
0 0
0 20 40 60 80 100 0 20 40 60 80 100
InstallationITimeI(s) InstallationITimeI(s)
R²)=)0.5383
120 yI=I1.5291xI+I40.25 120
100 R²I=I0.4127 100
80 80
60 60
40 40
20 20
0 0
0 20 40 60 80 100 0 20 40 60 80 100
InstallationITimeI(s) Installation)Time)(s)
Figure 6.9: Relationship between drive time and absolute ultimate capacity for all FRS configurations
installed using a bolter
From Table 6.9, it appears as though different bolt configurations have different degrees of sensitivity
to the relationship between drive time and ultimate capacity; regression coefficients range from 0.86
kN/s for the FA35 to 3.05 kN/s for the FB39. This may be the result of an over-simplification of the
relationship illustrated by performing a linear regression. Both the FB35 and FB39 have a large number
of high leverage data points with installations less than 10 seconds. If the relationship were logarithmic
or a power function as opposed to linear in nature (as is perhaps hinted for the FB35 and FA46 in Figure
6.9 and the fact that such a relationship should pass through the origin), data sets with concentrations
of low values would have higher gradients than those without. Datasets of individual FRS configurations
do not clearly capture this behaviour, and it is unclear whether these datasets may be combined. Figure
6.10 shows a power function fit to the data from all FRS configurations.
Chapter 6. Factors Influencing Pull Test Performance 88
160
140
y = 1.4801x + 45.695
UltimatenCapacityn+kN)
120 R² = 0.3641
FA35
100 y = 23.661x0.3904 FA39
R² = 0.4217
80 FA46
FB35
60
FB39
40
FB46
20
0
0 20 40 60 80 100
InstallationnTimen+s)
Figure 6.10: Relationship between installation time and ultimate capacity for all bolter-installed FRS
fit with linear and power functions
Figure 6.10 shows that a linear regression has a lower coefficient of determination than a power
function fit to the same dataset, although neither correlation is very strong. The FB35 appears to develop
high ultimate capacities with relatively low installation times. The FB39 has longer installation times for
low ultimate capacity bolts, but the high regression coefficient results in relatively low installation times
for high ultimate capacity bolts. The FB46 and FA46 display similar relationships between ultimate
capacity and installation time. The FA35 develops high capacities at relatively low installation times, but
has a low regression coefficient so only relatively long installation times exhibit higher capacities. Overall,
it would appear that as trends between ultimate capacity and installation time are quite different between
FRS diameters as well as suppliers, this relationship should be defined on the basis of an individual FRS
configuration. A larger number of tests that cover a wider range of installation times for individual
configurations are necessary in order to determine whether these trends may be characterised as linear
or power functions. Tomory et al. (1998) fit linear functions to their dataset which, when extrapolated,
intercepted the origin. This is not the case of the data presented in Figure 6.10, possibly due to the fact
that Tomory et al. were primarily investigating bolts installed using a jackleg, while the data presented
in Figure 6.10 is for bolts installed using a bolter.
It is clear that there is a relationship between drive time and the ultimate capacity of an FRS.
However, drive time is not an independent factor, but is just as dependent on installation conditions as
the performance of the FRS itself. As such, its relationship with ultimate capacity is not particularly
elucidating in terms of explaining how the performance of an FRS relates to its installation conditions.
Having said this, tracking installation time could be a useful quality control tool. If the operator of
the equipment installing the bolts consistently achieves drive times of less than 10 seconds, it may be
prudent to implement a denser bolting pattern. Conversely, a 30 second drive time for any bolt suggests
the bolt will perform well in a loading scenario analogous to a pull test.
smaller drill bit diameters due to higher radial stresses induced in the bolt as a result of a tighter fit.
Figure 6.11 shows the relationship between ultimate capacity and drill bit diameter for the different
configurations of FRS.
80 80
UltimateDCapacityD(kN/m)
UltimateDCapacityD(kN/m)
60 60
R²D=D0.093
40 40 R²D=D0.0053
20 20
0 0
29.0 31.0 33.0 35.0 37.0 29.0 31.0 33.0 35.0 37.0
DrillDBitDDiameterD(mm) DrillDBitDDiameterD(mm)
UltimateDCapacityD(kN/m)
60 60
40 R²D=D0.1082 40 R²D=D0.0009
20 20
0 0
31.0 33.0 35.0 37.0 39.0 31.0 33.0 35.0 37.0 39.0
DrillDBitDDiameterD(mm) DrillDBitDDiameterD(mm)
UltimateDCapacityD(kN/m)
60 60
R²D=D0.0809 R²D=D0.3651
40 40
20 20
0 0
39.0 41.0 43.0 45.0 47.0 39.0 41.0 43.0 45.0 47.0
DrillDBitDDiameterD(mm) DrillDBitDDiameterD(mm)
Figure 6.11: Relationship between drill bit diameter and ultimate capacity for all FRS configurations
From Figure 6.11 there initially does not appear to be a substantive relationship between drill bit
diameter and ultimate capacity, and the weak relationships that do exist for the smaller diameter bolts
appear to be positive, defying the expectation of a negative relationship. The exception to this is the
FB46, the data for which has the largest span in drill bit diameters tested. In order to further investigate
this, two of the largest campaigns on pull tests in which the drill bit diameter was varied were examined,
both on the FB46. These are shown in Figure 6.12.
Chapter 6. Factors Influencing Pull Test Performance 90
80
70
UltimateDCapacityD(kN/m)
CreightonD9/12/2005
60
CreightonD4/28/2008
50
40
R²D=D0.2482
30
20
R²D=D0.8044
10
0
39.0 40.0 41.0 42.0 43.0 44.0 45.0 46.0 47.0
DrillDBitDDiameterD(mm)
Figure 6.12: Relationship between drill bit diameter and ultimate capacity for two testing campaigns
performed on the FB46
Both of the campaigns presented in Figure 6.12 show negative relationships between drill bit diameter
and ultimate capacity. This implies that while individual campaigns may reflect differences in bolt
performance when a wide range of bit diameters are used, there are other factors that strongly influence
the way bit diameter translates to hole size, and thus tightness of bolt fit and ultimate capacity. Results
were filtered by installation method for further examination. There were no further findings for the
FA39, FA46, FB39 or FB46 bolts. FA35 and FB35 pull tests (Figure 6.13) further clarify the relationship
between ultimate capacity, drill bit diameter and installation method.
80 80
UltimateDCapacityD(kN/m)
UltimateDCapacityDgkN/m)
MacLeanDBolter MacLeanDBolter
Jackleg R²D=D0.2354 Unknown
60 60
R²D=D0.0342
40 40
20 R²D=D0.4411 20 R²D=D0.0169
0 0
29.0 31.0 33.0 35.0 37.0 29.0 31.0 33.0 35.0 37.0
DrillDBitDDiameterDgmm) DrillDBitDDiameterD(mm)
Figure 6.13: Relationship between drill bit diameter and ultimate capacity for FA35 and FB35,
separated by installation method
Although no correlation between ultimate capacity and drill bit diameter is observed for bolts installed
by a bolter in Figure 6.13a, bolts installed using a jackleg show a negative relationship between ultimate
capacity and drill bit diameter (R2 = 0.4411). In Figure 6.13b, no relationship is observed for bolts
installed using an unknown method, but a weak relationship seen for bolts installed with a bolter
(R2 = 0.2354). The relatively low coefficient of determination calculated for either regression suggests
that while ultimate capacity does appear to be influenced by the diameter of the drill bit, it is heavily
influenced by other factors that obscure the relationship. These factors are likely associated with the
translation of drill bit size to hole size, and could include the equipment operator’s skill, or how the
strength and quality of the rock mass influences the size of the hole drilled. It is thus difficult to describe
Chapter 6. Factors Influencing Pull Test Performance 91
a universally applicable model of FRS performance dependant solely on the diameter of the drill bit
used.
1.18
1.16
Bolt Diameter/Bit Diameter
1.14
1.12
1.10
1.08
1.06
1.04
1.02
1.00
0.98
FA35 FB35 FA39 FB39 FA46 FB46
Bolt Configuration
Figure 6.14: Bolt diameter to drill bit diameter ratios for all FRS variants
Table 6.10: Summary statistics for bolt diameter to drill bit diameter ratios for all FRS variants
Figure 6.14 and Table 6.10 suggest that for the given data set, FRS A bolts are larger relative to the
drill bit than the equivalent FRS B. However, it must be recognized that this is not representative of the
entire database (only 228 of 545 FRS pull tests), and as seen in Section 6.1.4 this does not necessarily
translate directly into performance. Table 6.10 reinforces both of these points; although the FA35 and
FB35 have quite different bolt to drill bit diameter ratios, the average ultimate capacities for the two
data sets are similar. Conversely, the bolt to drill bit diameter ratios of the FA46 and FB46 are similar,
but the FA46 dataset has a much higher average ultimate capacity. Only a comparison of the 39 mm
nominal diameter bolts shows the expected effect. The FA39 dataset has a larger bolt diameter to
drill bit diameter ratio, and a larger ultimate capacity than the FB39 dataset. Figure 6.15 shows the
Chapter 6. Factors Influencing Pull Test Performance 92
relationship between bolt to drill bit diameter ratio and ultimate capacity.
80 80
UltimateoCapacityo(kN/m)
UltimateDCapacityD(kN/m)
60 60
R²o=o0.0945 R²D=D0.0306
40 40
20 20
0 0
0.98 1.03 1.08 1.13 1.18 0.98 1.03 1.08 1.13
Bolto/oDrilloBitoDiameter BoltD/DDrillDBitDDiameter
UltimateDCapacityD(kN/m)
60 60
R²D=D0.0091
40 R²D=D0.0019 40
20 20
0 0
0.98 1.03 1.08 1.13 1.18 0.98 1.03 1.08 1.13
BoltD/DDrillDBitDDiameter BoltD/DDrillDBitDDiameter
UltimateDCapacityD(kN/m)
60 60
R²o=o0.0253
40 40 R²D=D0.2289
20 20
0 0
0.98 1.03 1.08 1.13 1.18 0.98 1.03 1.08 1.13
Bolto/oDrilloBitoDiameter BoltD/DDrillDBitDDiameter
Figure 6.15: Relationship between bolt diameter to drill bit diameter ratio and ultimate capacity
Figure 6.15 appears to indicate that the relationship between the bolt diameter/drill bit diameter
ratio and ultimate capacity is weak; coefficients of determination are generally low, and gradients are both
positive and negative for different bolt configurations, when positive relationships would be expected. In
an attempt to address this, as in Section 6.1.4, Figure 6.16 trims the data, sorting by test campaign in
which multiple drill bit sizes are used (to increase the range of bolt diameter to drill bit diameter ratios
covered). Linear regressions are shown for data sets with 10 or more data points.
Chapter 6. Factors Influencing Pull Test Performance 93
80 80
UltimateBCapacityBhkN/mO
UltimatenCapacityn(kN/m)
60 60
40 40
Totten,B04/22/2010
Totten,B4/4/2014
20 Creighton,B4/29/2014BhOreO 20 Totten,n2014/11/19
Creighton,B4/29/2014BhNoriteO
Totten,B9/23/2010 R²B=B0.4567
0 0
0.98 1.03 1.08 1.13 1.18 0.98 1.03 1.08 1.13 1.18
BoltB/BDrillBBitBDiameter Boltn/nDrillnBitnDiameter
UltimatenCapacityn(kN/m)
UltimatesCapacitys(kN/m)
60 60 Coleman,n3/7/2014
40 40
Garson,s9/30/2013
20 Coleman,s4/2/2014s(Ore) 20
Coleman,s4/2/2014s(Granite)
0 0
0.98 1.03 1.08 1.13 1.18 0.98 1.03 1.08 1.13 1.18
Bolts/sDrillsBitsDiameter BoltnDiametern:nDrillnBitnDiameter
Creighton,g6/10/2014gXOre) CopperfCliff,f2014/1/31f(QuartzfDiorite)
Creighton,g6/10/2014gXSubX) CopperfCliff,f2014/1/31f(Ore)
60 60
Totten,f2014/12/3
40 40
20 20
0 0
0.98 1.03 1.08 1.13 1.18 0.98 1.03 1.08 1.13 1.18
BoltgDiameterg:gDrillgBitgDiameter BoltfDiameterf:fDrillfBitfDiameter
Figure 6.16: Relationship between bolt diameter to drill bit diameter ratio and ultimate capacity for
all FRS configurations by testing campaign
Figure 6.16 shows that there is no clear relationship for the data sets with 5 or fewer entries, and
only one data set (Totten, 4/22/2010) shows a positive relationship between bolt to bit diameter ratio
and ultimate capacity. This data set is the largest and also has the widest ratio range, so while it does
make a case for the existence of a relationship when testing campaigns are evaluated individually, this
is not validated with any other data in the database.
6.1.6 Geology
Rock types in the pull test database broadly fall into five categories: ore, igneous rocks (norite, quartz
diorite, granite, etc.), metamorphosed igneous rocks (greenstone, amphibolite and mafic gneiss), brec-
cias (Sudbury and granite breccias), and metasedimentary rocks. Additionally, some pull tests were
performed on bolts installed in sand fill. Figure 6.17 separates test results by these lithological cate-
gories. Drill bit diameter is indicated in this figure, with green indicating a smaller drill bit diameter, and
red larger. Note that all bolts for which the data could be found are included, regardless of installation
method.
Chapter 6. Factors Influencing Pull Test Performance 94
80 80
UltimatekCapacityk(kN/m)
UltimateUCapacityU(kN/m)
60 60
40 40
LargeUdrillbitUdiameterU(relative)U
20 20
SmallUdrillbitUdiameterU(relative)
0 0
Ore Igneous neous Breccia etased andfill Ore Igneous neous Breccia etased andfill
ig S ig S
Meta M Meta M
UltimatekCapacityk(kN/m)
60 60
40 40
20 20
0 0
Ore Igneous neous Breccia etased andfill Ore Igneous neous Breccia etased andfill
ig S ig S
Meta M
Meta M
60 60
40 40
20 20
0 0
Ore Igneous neous Breccia etased andfill Ore Igneous neous Breccia etased andfill
ig S ig S
Meta M Meta M
Despite ore in Sudbury being a generally weaker and softer material than the host rock (lower
UCS and modulus of elasticity), it is difficult to see any category of lithology consistently over- or under-
perform relative to the others, potentially due to other factors or measures of rock mass characterization.
Figure 6.18 shows ultimate capacity plotted against the UCS of the rock in which the bolt is installed.
Performing linear regression on the UCS–ultimate capacity data results in no obvious relationship
for any bolt, except the FA35. The likely cause of this relationship, however, is the fact that the bolts
installed with a jackleg prior to pull testing were all in the relatively strong host rock and not the weaker
ore, resulting in a misleading relationship with rock strength. As part of Vale’s quality control program,
pull test campaigns covered more than one rock unit in some cases (usually ore and host rock). These
serve to isolate the effect of the properties of the intact rock as much as possible. Usually, the same
equipment, operator and drill bit (or set of drill bits) were used to install the bolts, and testing in one
area reduces the variability of rock mass conditions between the two rock units. A direct comparison of
the results from ore and waste rock is presented in Table 6.11.
Chapter 6. Factors Influencing Pull Test Performance 95
80 80
20 20
0 0
0 100 200 300 0 100 200 300
UCSP(MPa) UCS (MPa)
40 40
R²P=P0.0356 R² = 0.0262
20 20
0 0
0 100 200 300 0 100 200 300
UCSP(MPa) UCS (MPa)
60 60
40 R²P=P0.0027 40
20 20 R² = 0.0611
0 0
0 100 200 300 0 100 200 300
UCSP(MPa) UCS (MPa)
Table 6.11: Comparison of average ultimate capacities for pull tests performed in ore and waste rock in
the same campaign
In general, very little difference in ultimate capacity is observed between installations in ore versus
waste, as 7 of the 13 campaigns listed show a difference of less than 2 kN/m between the average ultimate
capacities installed in the two rock types. In order to further investigate, the data from each campaign
is normalized with respect to the average ultimate capacity for the pull tests performed in waste, set
to equal 1. This reduces variability across different testing campaigns, while preserving the contrast
between the ultimate capacities measured in ore and waste rock within individual campaigns. Figure
6.19 and Table 6.12 show the results. Table 6.12 also shows the results of the reverse analysis, where the
average result of the bolts installed in ore in a campaign are equal to 1. The t-tests performed assume
unequal variances, and the tcrit and t values are two-tailed (α = 0.05).
Distribution
0% 10% 20% 30% 40% 50% 60%
1.8
Normalized Ultimate Capacity
1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
Ore Waste
Figure 6.19: Distributions of FRS bolts installed in ore and waste rock normalized to the campaign
average ultimate capacity for bolts installed in waste rock
Table 6.12: Comparison of FRS bolts installed in ore and waste rock
Table 6.12 shows that while pull test results for bolts installed in ore do appear to be more variable
than those installed in the waste rock, there is on average no significant difference between their ultimate
capacities immediately after installation. It should be noted that while the ore is generally weaker than
the waste rock in Sudbury as was shown in Table 3.1, it is still quite strong compared to some rock types
with UCS values between 91 and 170 MPa. The same investigation performed in rock weaker than the
ore found in Sudbury may not yield the same results. Additionally, there are other factors that may
influence how an FRS performs with time between the ore and the waste; as the ore is a sulphide, an
FRS installed in the ore may corrode more quickly than one installed in other rock types. The nature of
structures present in waste rock versus those in ore is another aspect of the rock masses to be considered,
as well as difference in rock quality.
Chapter 6. Factors Influencing Pull Test Performance 97
80
70
Ultimate Capacity (kN/m)
60
50
40
30
20
10
0
Poor Good
Rock Mass Quality
Figure 6.20: Ultimate capacities recorded for FA46 bolts installed in poor and good quality ground
Table 6.13: Comparison of FA46 bolts installed in poor and good quality ground
Performing the t-test results in the rejection of the null hypothesis, thus there appears to be a
significant difference between the means of the data sets; FA46 bolts installed in good ground averaged
30% higher ultimate capacity than those installed in poor ground. Data, however, is limited, especially
for those bolts installed in poor ground. To address this, the analysis is expanded to include all bolt
Chapter 6. Factors Influencing Pull Test Performance 98
configurations. Figure 6.21 shows the results. Bolts installed using a jackleg are omitted on the basis of
the findings in Section 6.1.2, where it was shown that an FRS installed using a bolter has 25% greater
capacity than one installed with a jackleg.
Poor
70 Good
60
50
40
30
20
10
0
RockCMassCQuality: Poor Good
Figure 6.21: Ultimate capacities recorded for FRSs installed in poor and good quality ground
The highlighted pull tests performed in poor rock are from a campaign of FA39 bolts installed at
Garson, 9/20/2013, and are the obvious outliers in the data set. Further investigation showed that the
report included a photograph of the bolts as installed, shown in Figure 6.22. Although the bolts appear
to be installed along a fracture, only Bolts 1 and 3 appear to be on the fracture on the surface of the
excavation, and registered ultimate capacities of 49.8 kN/m and 29.9 kN/m respectively, two of the three
lowest results of the campaign (Bolt 3 by a large margin). The surrounding rock mass appears to be
intact, and the assessment of this particular testing area as having “fractured” rock, per the report, is
questionable. As a result, Table 6.14 compares the dataset from good quality rock, and poor quality
rock with and without this campaign included.
Table 6.14: Comparison of all FRSs installed in poor and good quality ground
If the Garson FA39 test is included, the p-value of 0.061 is calculated, resulting in a marginal t-test
result; strictly speaking, the null hypothesis is not rejected if α = 0.05, and no difference is observed
between the means at 5% significance. However if this campaign is omitted, very similar results are
obtained as when only the FA46 is examined, indicating that the average ultimate capacity of an FRS
installed in good quality rock is 30% larger than that of one installed in poor rock. A larger number
of tests and less variance in the results of the bolts installed in good rock result in a very low p-value,
suggesting a significant difference. As such, it is concluded a difference does exist in the performance of
FRSs installed in poor versus good quality rock.
was to investigate parameters that influence the performance of Split Sets, although there is relatively
little overlap in findings. In Section 6.1.1, a significant difference in performance between FRS bolts of
different lengths was found, which was not touched on by Tomory et al. (1998). While the emergence of
the use of bolters allows for a comparison between installation methods, Tomory et al. (1998) did not
perform an equivalent analysis, presumably due to lack of variation in installation equipment. While both
this thesis and Tomory et al. (1998) examine drive time, this thesis found that it appears as though the
relationship between bolt capacity and drive time is either non-linear, or inconsistent between different
diameters and suppliers of FRS. Tomory et al. (1998) appears to have had greater success analysing
the effect of drill bit diameter on performance, likely due to the larger dataset on fewer configurations.
The findings presented in this thesis only show broad trends in the cases of some individual campaigns,
and relatively little on a consistent basis. Similarly, results comparing different geologies (aside from
ore versus waste rock) were inconclusive for this thesis, while the database assembled by Tomory et al.
(1998) covered a wider range of geologies, and presented more conclusive results. However, no direct
comparison between ore and waste installations of Split Sets was performed, for which there does not
appear to be a difference in FRS ultimate capacity. Rock mass quality is another parameter investigated
that seems to influence ultimate capacity which Tomory et al. (1998) did not discuss extensively. The
factor that was thoroughly investigated by Tomory et al. (1998) but not covered in this thesis was
capacity development with time. It was found that the capacity of the SS33 increased with time after
installation. All pull tests in the database assembled for this thesis were conducted immediately after
bolt installation, so this parameter could not be examined.
200 200
100 100
50 50
0 0
0.0 1.0 2.0 3.0 0.0 1.0 2.0 3.0
BoltbLengthb(m) Bolt Length (m)
150 150
100 100
50 50
0 0
0.0 1.0 2.0 3.0 0.0 1.0 2.0 3.0
Bolt Length (m) Bolt Length (m)
120 120
80 80
40 40
0 0
0.0 1.0 2.0 3.0 0.0 1.0 2.0 3.0
Bolt Length (m) Bolt Length (m)
It can be seen that for both suppliers’ rebar, high tangent and secant stiffness occur with the longer
(2.4 m) rebar length. Rebar A data (unlike the Rebar B data) shows two data clusters for the greater
length: one low stiffness, and one high stiffness. It is assumed that the longer Rebar B acts in a stiffer
manner than its shorter counterpart is a result of the low number of tests collected, and is not an effect
that would be systematically observed with further testing. The 2.4 m Rebar A had a lower working
capacity than the 1.8 m rebar. This may be a result of a different steel and/or manufacturing plant or
process used for the production of the longer bolt - these results come from 3 different testing campaigns
at 2 mines, so it is not attributed to bolt or installation quality. 2.4 m Rebar B does have one low
working capacity test, but the others are equivalent to results seen for the 1.8 m bolts.
60 60
TangenthStiffnessh(kN/mm)
TangenthStiffnessh(kN/mm)
50 50
40 40
30 30
20 20
10 10
0 0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
GrouthLengthh:hBolthLength GrouthLengthh:hBolthLength
SecanthStiffnessh(kN/mm)
SecanthStiffnessh(kN/mm)
50 50
40 40
30 30
20 20
10 10
0 0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
GrouthLengthh:hBolthLength GrouthLengthh:hBolthLength
Figure 6.24: Relationships between stiffness and grout length : rebar length for Suppliers A and B
Large variances and a lack of data between fully encapsulated bolts and bolts with less than one
third of their lengths encapsulated makes an analysis of two data clusters more prudent. Table 6.15
compares stiffness calculated for full and partial encapsulation tests.
Table 6.15: Comparison of partial and full encapsulation test statistics for Rebar A and B
There is a clear difference in tangent and secant stiffness between the partial and full encapsulation
tests, presumably due to a smaller length of rebar coupled to the rock mass in the partial encapsulation
tests. Interestingly, the partial encapsulation tests consistently achieve about 70% of the stiffness of the
fully encapsulated tests, although they employ only 15% to 35% of the resin. To further investigate, the
unloading stiffness of partially and fully encapsulated test are shown in Figure 6.25.
Chapter 6. Factors Influencing Pull Test Performance 103
200
Figure 6.25: Unloading stiffness and grout length : rebar length for Rebar A
Figure 6.25 shows the maximum stiffness that may be expected for 1.8 m and 2.4 m bolts, corre-
sponding to grout coverage. This was determined by assuming uniform elastic deformation along the
length of bolt not fully bonded to the rock mass, with no deformation occurring within the bonded
section. At 0% grout coverage, the maximum stiffness is equivalent to the stiffness of the entire length
of rebar, and at 100% grout coverage maximum stiffness is considered infinite. This also assumes that
the length of resin in the hole is equivalent to the length of cartridge, as the cross sectional area of a
20 mm bolt (314 mm2 ) plus a 28 mm resin cartridge (616 mm2 ; combined total of 930 mm2 ) is roughly
equivalent to a 34 mm hole (908 mm2 ) that would be drilled by a 32 to 33 mm drill bit. It is apparent
that for the fully grouted bolts, significant load is distributed down the bolt and/or alternate sources
of displacement greatly influence the results. It is also apparent that the partially encapsulated tests
perform in a relatively stiff manner, even though these tests should also be subject to displacements
extraneous to elastic deformation. This implies that resin may migrate up the bolt into the inert resin
during spinning, resulting in bond length between the rock and the rebar significantly greater than the
length of the cartridge.
70 70
TangentEStiffnessE(kN/mm)
Figure 6.26: Relationship between stiffness and resin spin time for Rebar A
As the majority of the data is concentrated between 8 and 10 seconds (per typical installation
procedures), no clear trends are observed. There was no description of how the bolt was spun, i.e. how
quickly the bolt was spun or inserted into the hole, or how long spinning continued after the bolt was
fully inserted. A wider range of data would be expected to show a trend in stiffness, however this data
set does demonstrate that there is some limited allowable deviation in spin time for which performance
does not seem to be significantly altered.
200 200
TangentlStiffnessl(kN/mm)l
175 175
150 1.8lm 150
2.4lm
125 125
100 100
75 75
50 50
25 25
0 0
Daylofltest Previous Day of test Previous
Installation Installation
Figure 6.27: Stiffness comparison of Rebar A installed on the day of testing versus previously
Recognizing that the dataset is limited, it appears that in some cases rebar stiffness may decay with
time; bearing load and being subject to vibrations from blasting and seismicity may weaken the resin–
rebar bond, allowing a greater degree of stress propagation down the bolt. However, this should not be
regarded as a universally applicable model. Only a very limited sample size was available, and the low
stiffness of the previously installed rebar makes it tempting to assume that no significant shearing has
affected the bolt’s performance. Should shearing occur, higher stiffness could be observed due to locking
of the rebar in place.
6.2.5 Geology
Rebar bolts were pull tested in a variety of lithologies. Due to the relatively low number of tests, the
only lithological distinction made is between ore and host rock. A comparison of stiffness is shown in
Figure 6.28 and Table 6.17.
160 160
Tangent Stiffness (kN/mm)
140 140
120 120
100 100
80 80
60 60
40 40
20 20
0 0
Ore Waste Ore Waste
It appears that Rebar A react in a more stiff manner in ore versus a stronger and stiffer host rock,
although fully encapsulated 1.8 m bolts are conspicuously absent from the data set. Conversely, Rebar B
react with a higher tangent stiffness in the host rock, although the difference in stiffness is less pronounced
and is quite small in the case of the secant stiffness (based on a smaller sample size). This suggests that
for Rebar A, stress is being transmitted further down the bolt if it is installed in host rock, resulting in
more strain and lower stiffness. The results must be kept in perspective, with a limited number of tests
performed and many potentially influential factors.
Chapter 6. Factors Influencing Pull Test Performance 106
30 120
Initial Stiffness (kN/mm)
25 100
6 12
5 10
4 8
3 6
2 4
1 2
0 0
Dayyofytest Previous Day of test Previous
Installation Installation
Figure 6.29: Performance comparison of MCB33s installed prior to and on the day of testing
Table 6.18: Performance comparison of MCB33s installed prior to and on the day of testing
MCB33 behaviour appears to evolve with time. The bolts seem to plough at lower loads, although
with higher plough stiffness as indicated by the two relevant t-tests indicating a significant difference in
means (p = 0.032 and p = 0.028 respectively). Note that of the 14 plough point measurements for the
bolts installed before the test date, 5 of them have noted plough points of 26.7 kN. This is an upper
bound, as it represents the pre-load, and it is possible that plough may have already occurred at loads
below this. It should also be noted that the average initial stiffness of the two installation periods is
very similar, implying that the difference in plough stiffness is not due to factors such as the rock mass
compressing.
This difference in behaviour could be a result of exposure to seismicity or blasting that would have
occurred between installation and testing of the bolts. Resin already damaged by vibrations in the rock
mass would offer less resistance to the onset of a consistent plough response. Resin damage may also be
the result of shearing of the bolt. When sheared, cone bolts convert a portion of the shearing load into
axial (Gaudreau et al, 2004), potentially initialising the ploughing process. Simser et al. (2006) found
that shearing of the rock mass can pinch and lock cone bolts in place, resulting in a much stiffer form
of reinforcement. This may be reflected in the higher initial and plough stiffnesses. Of the 9 previously
installed bolts, only 1 did not plough, indicating that the locking in these circumstances was generally
not significant enough to prevent any movement of the bolt, but did appear to affect the displacement
mechanism. Alternatively, it is also possible that the properties of the resin enveloping the bolt evolved
with time, modifying the behaviour of the bolt as it ploughs. A significant shortcoming of this analysis
is that the length of time that the previously installed bolts had been in the ground relative to one
another is unknown. Further investigation could quantify how performance changes with time, as well
as clarify what the underlying cause in the change of behaviour is.
6.3.2 Geology
The Modified Cone Bolts were installed and tested in various lithologies. Figure 6.30 and Table 6.19
show how properties vary between bolts installed in ore, igneous/metaigneous and metasedimentary
rocks. Previously installed bolts are highlighted in red, and omitted from statistical calculations.
Table 6.19: Performance comparison of MCB33 bolts installed previously and on the day of testing
30 120
DayMofMInstallation
InitialMStiffnessM(kN/mm)
25 PreviouslyMInstalled 100
15 60
10 40
5 20
0 0
Ore (Meta)Igneous Metaseds Ore (Meta)Igneous Metaseds
6
12
5
10
4
8
3
6
2 4
1 2
0 0
Ore (Meta)Igneous Metaseds Ore (Meta)Igneous Metaseds
Figure 6.30: Performance comparison of MCB33 bolts installed in ore, igneous/metaigneous and
metasedimentary lithologies
While all comparisons between ore and host rock must be approached cautiously as only one MCB33
test campaign was undertaken in ore, it would appear that these bolts acted in a less stiff manner and
with a higher plough point. It is interesting to note that this is the precise opposite effect of what was
observed for the bolts installed on a date before the pull test, characterised by a low plough point and
high stiffness. Without further data, it is difficult to posit an explanation of this behaviour, or to assess
its reproducibility.
6 Day of/installation
Plough/Stiffness/rkN/mmv
Previously installed
5
R² = 0.3153
4
2 R² = 0.0009
0
0 20 40 60 80 100 120
Plough Load/rkNv
Figure 6.31: Relationship between plough stiffness and plough point for the MCB33
6.4 Summary
The findings of this Chapter are summarized in Table 6.20. The reinforcement elements are shown
with the parameter that influences performance, as measured by a specified performance indicator. The
nature of the relationship is described, along with its strength and confidence in the findings. Generally, a
weak relationship and low confidence is due to a lack or narrow range of data, while a strong relationship
and high confidence is indicative of a large amount of data clearly showing a certain effect.
Table 6.20: Summary of observed relationships of between various factors and performance indicators
for each rock bolt type
This chapter has discussed performance metrics in terms of working capacity and various measures of
stiffness, although these are single values representing potentially complex responses to load. In Chapter
7, the displacement response of rock bolts subject to a pull test is described in greater detail.
Chapter 7
Having investigated rock bolt response to load and the influence of various parameters, this chapter
describes bolt behaviour observed during a pull test, defining a distribution of displacement development
with load. Bolt performance is compared to manufacturer specifications and recommendations, and to
that of other bolt types.
112
Chapter 7. Characterization of Rock Bolt Behaviour 113
blocks. Hydraulic jacks exerted a load which pushed the two blocks apart as displacement was measured
(Li et al, 2014). Doucet & Voyzelle (2012) present the results of laboratory static pull tests performed
at the CANMET-MMSL facility using the procedures and apparatus described in ASTM D7401-08,
discussed in Section 2.2.5. Instead of concrete blocks, rock bolts were installed in a steel tube, and load
applied to the bolt head, as in an in situ pull test. The results presented by Doucet & Voyzelle (2012)
are from tests performed between 2006 and 2011, and include the behaviours of a mechanical bolt, resin
grouted rebar, the MCB33, the discontinued MCB38, the Durabar Yieldable Bolt, the discontinued
Roofex Rx8D and Rx20D, 20 mm and 22 mm D-Bolts, the Yield Lok Dynamic and the Yield Lok Static.
The use of a database of in situ pull tests has significant advantages over these laboratory campaigns
as tools for design. The bolts in a database of in situ pull tests are installed in conditions found on a
mine site. Instead of installing bolts using lab-scale equipment in blocks of high strength concrete or
steel tubes, they are installed in holes drilled in rock by equipment and procedures used daily on site. In
addition, the database assembled is a much larger dataset than the laboratory tests discussed. Figure
7.1 was made using 3 tests for each bolt type (Stillborg, 1993). Static test results presented by Doucet
& Voyzelle (2012) appear to be results from single tests. With more tests performed, a distribution of
behaviour may be assembled and can be used for probabilistic design and analysis.
There are two primary advantages to the laboratory tests discussed. The first is that bolts are
Chapter 7. Characterization of Rock Bolt Behaviour 114
pulled until failure. The second is the test configuration. Load may be applied across a simulated
joint (as in Stillborg, 1993, and Li et al, 2014), which is arguably a better analogue of the nature of
loading a reinforcement element may be subject to in practice, rather than loading of the bolt head (as
is performed during an in situ pull test). The test may be instrumented differently to provide more
certainty of the results and interpretation. For example, displacement at the toe of the bolt may be
measured to determine whether the bolt is slipping or if all displacement may be attributed to bolt
deformation. Such a measurement cannot be easily performed in an in situ pull test. Further discussion
of testing methods and how they relate to one another may be found in Chapters 2 and 4.
Previously assembled pull test databases (Tomory et al, 1998; Soni, 2000) have primarily analysed the
performance of rock bolts in terms of load capacity. Tomory et al. (1998) did not include displacement
of the Split Set in their analysis. Soni (2000) used the displacement measured during the pull test of
a Swellex bolt to determine whether a bolt had slipped or not, thus as a binary metric of performance
as opposed to a continuous one. Displacement characteristics of rock reinforcement and ground support
elements in general are critical to enact some methods of design, such as those that incorporate the
ground reaction curve (Section 4.3.3). In this chapter, load-displacement behaviours of certain rock
bolts are compiled across all pull testing campaigns in the database in order to define a distribution of
behaviour for that bolt. The objective is to propose composite diagrams of load-displacement behaviours
of various types of rock bolt as undertaken by Stillborg (1993), using a larger set of in situ pull test data.
The percentiles of performance metrics used for the two bolts are not identical. Although the 10th
and 75th percentiles are quite similar between the two bolts, the FA35 dataset appears to generally have
a higher ultimate capacity. The FA35 also appears to behave in a more stiff manner than the FA39, the
disparity between the two bolts increasing at higher percentiles. However, it has been demonstrated in
Chapter 7. Characterization of Rock Bolt Behaviour 115
Chapter 5 that there is no difference in ultimate capacity between the two bolts when the entire dataset
of ultimate capacities is examined, and that secant stiffness is not necessarily reflective of performance.
As such, the two datasets are combined in Figure 7.2. In this figure, the data points represent the
combination of load and displacement at which the ultimate capacity of a bolt was reached. Boxes are
defined by percentiles of both ultimate capacity and secant stiffness. Median ultimate capacity and
secant stiffness are indicated by solid black lines.
100
90
P90
80
P75
70
P50
60 5 P 10
0
P2
50
Loadt(kN)
P9
P7
P25 FA35
50 P10 FA39
40
30
20
10
0
0 2 4 6 8 10 12 14 16 18 20
Displacementt(mm)
Figure 7.2: Distributions of secant stiffness and ultimate capacity for a pull test on an FRS with an
anchorage length of 1.52 m
In the construction of Figure 7.2, not all data regarding the ultimate capacity of an FRS was used.
Figure 7.3 and Table 7.2 compares the dispersion of ultimate capacities for all configurations of FRS
discussed in this thesis, including the FA35/FA39 dataset for which displacements are measured, and
the FRS dataset as a whole. As multiple lengths of bolt are represented, load is quantified in terms of
kN/m of anchorage length.
80
Median
70 Mean
60
P90
50
P75
LoadV(kN/m)
40 P50
P25
30
P10
20
10
0
FA35 FA39 FA46 FB35 FB39 FB46
FRSVVariant
Figure 7.3: Ultimate capacity per metre distributions for all FRS configurations
Percentiles of ultimate capacity generally agree with each other across bolt type. The main exception
is the FA39, the capacity of which appears to be relatively narrowly distributed between 30 and 40 kN/m.
The FB35 and FB39 have the widest distributions, while those of the FB46, FA35 and FA46 are similar.
The nature of the FA39 dataset carries over into the combined FA35/FA39 dataset. P10 and P25 are
higher than the equivalent percentiles for other bolts, although the median, P75 and P90 better match
the rest of the data. In Figure 7.4, percentiles of secant stiffness and ultimate capacity are generalized
so they may be applied to various lengths of FRS. Load capacity is expressed in kN/m, displacement
in percentage of anchorage length, and percentiles of ultimate capacity shown are for the entire FRS
dataset.
70
60
P90
50 P75
Loadt(kN/m)
40 P50
5 P 10
P90
P2
50
P7
P25
30
P10
20
10
0
0.09 0.29 0.49 0.69 0.89 1.09
Displacementt(9toftanchoragetlength)
Figure 7.4: Distribution of secant stiffness and ultimate capacity for all FRS bolts tested
Chapter 7. Characterization of Rock Bolt Behaviour 117
To use Figure 7.4 for design, a decision must be made regarding what percentiles of performance (i.e.
ultimate capacity and secant stiffness) are to be used. As an example, a designer wishes to use the 25th
percentiles of both ultimate capacity and secant stiffness of an FRS with an anchorage length of 2 m.
The intersect of the 25th percentiles is at 33 kN/m and 0.32 % displacement, so the ultimate capacity
for the 2 m bolt will be 66 kN at 6.4 mm at displacement. 25% of bolts will exhibit lower stiffness, 25%
lower ultimate capacity, but only 6.25% will have a lower ultimate capacity with more displacement.
100
90
FA35
80 FA39
70
60
Load (kN)
50
40
30
20
10
0
0 2 4 6 8 10 12 14 16 18 20
Displacement (mm)
Figure 7.5: All FA35 and FA39 pull test load-displacement relationships
Both bolts show erratic displacement behaviour as load develops, suggesting that in many cases the
bolt slips multiple times before reaching its ultimate capacity. To construct a representative distribution
of behaviour, percentiles of displacement were calculated at 0.5 t (4.45 kN) intervals. As load increases,
the dataset diminishes as the ultimate capacities of bolts are surpassed. To address this, dummy data
entries were used for these bolts so that the number of bolts used to construct the percentiles remained
constant. While the behavioural range is calculated using displacement, this is interpreted as stiffness,
i.e. the 90th percentile of displacement is equal to the 10th percentile of stiffness. Figure 7.6 illustrates
both the percentiles of stiffness (calculated from displacement at load intervals), as well as the percentiles
of ultimate capacity simultaneously. As once the ultimate capacity of a particular test has been exceeded
no displacement data at higher loads may be recorded, the 10th percentile of stiffness must end at the
10th percentile of ultimate capacity, as only 90% of the data remains from which to calculate stiffness
(mathematically, at this point the dummy entry influences the value of the percentile, so the percentile
Chapter 7. Characterization of Rock Bolt Behaviour 118
is no longer shown). This should not be taken to indicate that bolts with low secant stiffness have
proportionally low ultimate capacities; Figure 7.2 shows that this does not appear to be the case. Figure
7.7 shows a smoothed version of the same figure displaying the initial anchorage capacity suggested by
the manufacturer (courtesy of Supplier A).
100
90
P90
80
70 P75
P50
60
Load (kN)
P25
50
P10
40
30
20
10
0
0 2 4 6 8 10 12 14 16 18 20
Displacement (mm)
Figure 7.6: Distribution of displacement measured during pull testing of FA35 and FA39 with
anchorage lengths of 1.52 m
100
90
P90
80
P75
70
P50
60
P25
Loadc(kN)
50 P10
InitialcCapacitycRange
40 (FromcSupplier)
30
20
10
0
0 2 4 6 8 10 12 14 16 18 20
Displacementc(mm)
Figure 7.7: Conceptual displacement distribution of FRSs with anchorage lengths of 1.52 m subject to
a pull test
Figure 7.6 shows that in situ FRS behaviour may be quite variable; it is possible that next to no load
Chapter 7. Characterization of Rock Bolt Behaviour 119
is required to initialize displacement, or that no displacement may be observed until 35 kN of load. Load
then increases with growing displacement before plateauing at a maximum value (ultimate capacity).
In Figure 7.7, behaviour is extrapolated to be perfectly plastic. In reality, after reaching the ultimate
capacity, load is continually built and released as the bolt slips out of the hole incrementally. Note that
the percentiles of ultimate capacity displayed in this figure are for the FA35/FA39 dataset, and not the
entire FRS dataset. Figure 7.8 shows a generalized version of Figure 7.7. Load is expressed in load per
metre of anchorage, and displacement in terms of anchorage length. This characterisation of behaviour
was performed only bolts with an anchorage length of 1.52 m, and its applicability to other bolt lengths
has not been verified.
70
60
P90
50
P75
P50
Loadt(kN/m)
40
P25
P10
30
20
10
0
0.09 0.29 0.49 0.69 0.89 1.09 1.29 1.49
Displacementt(9toftanchoragetlength)
Initial anchorage values of 27 to 54 kN are claimed by Suppliers A and B for both 35 and 39 mm
configurations of FRS. As seen in Figure 7.7, pull testing indicates that these values are generally met or
exceeded. However, both suppliers claim anchorage values of 54 kN to 89 kN for their respective 46 mm
FRSs (courtesy of Suppliers A and B). Table 7.3 compares manufacturer recommendations with observed
pull test results in terms of both absolute and length–normalized load. As both suppliers provide their
load recommendations in terms of absolute load, the typical 35 mm and 39 mm anchorage length of
1.524 m is used to normalize load to length for all bolts.
For 35 and 39 mm bolts, almost all pull tests (97.6%) have ultimate capacities exceeding the lower
bound of the suppliers’ claim, and around 60% exceed the upper bound. Suppliers claim a higher range
of ultimate capacities for their 46 mm bolts – between 54 kN and 89 kN. It has been found in this thesis
that there is no difference in performance across different FRS diameters. 77.7% of the 46 mm FRS
tests exceeded the lower bound of the claimed range, but only 15.5% exceeded the upper bound, a much
lower pass rate than the 59.2% observed for the 35 mm and 39 mm FRSs. When loads are adjusted to
a common anchorage length, the discrepancy in supplier specifications becomes more apparent. 59.4%
of 35/39 mm FRSs pass their upper bound of 35.4 kN/m, compared to only 1.9% of 46 mm FRSs
passing their upper bound of 58.4 kN/m. It is difficult to see how suppliers determine their capacity
Chapter 7. Characterization of Rock Bolt Behaviour 120
specifications, as it has been observed that no difference exists between the length–normalized ultimate
capacities of different FRS diameters. Assuming that the claimed ranges of capacity are determined
consistently (i.e. use of equivalent bolt length, installation procedures, etc.), it is difficult to explain this
variation.
It is recognized that this discussion touches only on recently installed bolts. The 46 mm bolt has
more constituent steel, as it is larger in diameter. This means that the load at which it will physically
fail is greater than the smaller diameters of FRS discussed. If an FRS is sheared an locked in place, a
larger cross-sectional area will result in a higher load required to break the bolt. Similarly, the shear
strengths of the bolts may not be equivalent. However, if these failure modes are not observed at an
operation, the FRS bolts appear to have the same capacity regardless of diameter.
160
140
P7 90
P90 P 10
5
0
P
P5
P75
P50
120
25
P25
P
P10
100
LoadA(kN)
RebarAA
80 RebarAB
60
40
20
0
0 2 4 6 8 10 12 14 16 18 20
DisplacementA(mm)
Figure 7.9: Distributions of secant stiffness and working capacity for a pull test with a pre-load of 17.8
kN on a rebar rock bolt
160
140
120
100
Load (kN)
60
40
20
0
0 2 4 6 8 10 12 14 16 18 20
Displacement (mm)
It can be seen in Figure 7.10 that large displacements at low loads may occur if the bolt is not pre-
loaded. Figure 7.11 shows the load–displacement data neglecting the first 17.8 kN of all pull tests. Figure
7.12 incorporates displacement data below 17.8 kN and correcting the percentiles of the entire data set
to match those of the tests performed without pre-load at 17.8 kN. As pull tests are often performed
only until the rebar yields, if not before, data limitations result in the 75th and 90th percentiles not
clearly showing yield behaviour.
160
P10
140
P25
P50
120 P75 P90
100
Load (kN)
80
60
40
20
0
0 2 4 6 8 10 12 14 16 18 20
Displacement (mm)
Figure 7.11: Distribution of displacement measured during pull testing of 20 mm rebar 1.8 m to 2.4 m
in length with a pre-load of 17.8 kN
A comparison of Figures 7.11 and 7.12 illustrates the objective of using a pre-load during pull testing;
Chapter 7. Characterization of Rock Bolt Behaviour 123
160
P10
140
P25
P50
120 P75 P90
100
Load (kN)
80
60
40
20
0
0 2 4 6 8 10 12 14 16 18 20
Displacement (mm)
Figure 7.12: Distribution of displacement measured during pull testing of 20 mm rebar 1.8 m to 2.4 m
in length without a pre-load
large displacements occur at low loads as the rock mass and testing apparatus tighten up, before a stiffer
linear response to load is developed. As the stiffness and working capacity of rebar subject to a pull
test are independent of one another, they are separated in Figure 7.13. The behaviour shown in Figure
7.12 is extrapolated to P90 of working capacity. Median working capacity is shown in red, and other
percentiles of working capacity are listed. Note that partially encapsulated rebar were incorporated into
the dataset used to calculate percentiles of working capacity.
160
140
P10 P25 P50 P75 P90
120
100
LoadW(kN)
80 PercentilesWofWWorkingWCapacity
P90 131WkN
P75 127WkN
60 P50 124WkN
P25 117WkN
P10 113WkN
40
20
0
0 2 4 6 8 10 12 14 16 18 20
DisplacementW(mm)
Figure 7.13: Conceptual distribution of displacement during pull testing of 20 mm rebar without a
pre-load
Chapter 7. Characterization of Rock Bolt Behaviour 124
Figure 7.13 shows that less than 2 mm of displacement may be observed when pull testing a 20 mm
rebar before working capacity is reached and the rebar begins to yield. About 5 mm of displacement
represents median behaviour if the bolt is not pre-loaded, and in some conditions over 14 mm of displace-
ment may be measured. Table 7.5 compares the rebar strength claimed by the supplier to the observed
working capacities.
All rebar reached their working capacities above the supplier-specified thread yield strength. While
not all rebar were tested until the 134 kN minimum thread tensile strength specified by Supplier A,
thread failure was not reported in any pull test.
Figure 7.14 shows that cone bolt performance is more variable than that of rebar, in terms of both
yield strength and displacement. While the MCB33 yield strengths observed are similar to those of the
rebar, the displacement at which yield occurs is much larger. When both elements are installed in a
support system, this would result in the stiffer rebar bearing more load than the less stiff cone bolt. As
such, the yield strengths presented in Table 7.6 should not be used in a design methodology based solely
on the load demand on and capacity of a support system, but one that also incorporates displacement
into the calculation.
Chapter 7. Characterization of Rock Bolt Behaviour 125
Displacement (mm)
160
140 P90
P75
P50
0 5
120 P25 P 10
75
P5 P2
P9
P
P10
100
Load (kN)
80
60
40
20
0
0 10 20 30 40 50 60 70 80 90 100
Displacement (mm)
Figure 7.14: Distributions of secant stiffness and working capacity for a pull test with a pre-load of
17.8 kN on an MCB33
160
140
120
100
Loadm(kN)
80
60 4mtm(35.6mkN)mPre-load
3mtm(26.7mkN)mPre-load
2mtm(17.8mkN)mPre-load
40
20
0
0 10 20 30 40 50 60 70 80 90 100
Displacementm(mm)
Figure 7.15: 56 MCB33 pull tests collected from Vale’s Sudbury operations
Chapter 7. Characterization of Rock Bolt Behaviour 126
Percentiles of displacement are presented in Figure 7.16. As many pull tests are stopped before
the bolt has yielded, data becomes increasingly limited at higher loads. Figure 7.16 was created by
decreasing the size of the dataset used to calculate percentiles (from n = 45 to n = 32) at a load where
a large number of tests are stopped (12 tons; 107 kN) by eliminating dummy displacements from the
percentile calculations at 107 kN. Above 107 kN, dummy displacements were reintroduced to maintain
the n = 32 dataset as it decreased in size at higher values of load.
160
P10
140 P25 P50
P75
P90
120
100
Load (kN)
80
60
40
20
0
0 10 20 30 40 50 60 70 80 90 100
Displacement (mm)
Figure 7.16: Distribution of displacement measured during pull testing of a 2.4 m MCB33 with a
pre-load of 17.8 kN
Figure 7.16 clearly shows the two distinct behaviours of the cone bolt described in Section 5.4:
the 10th and 25th percentiles are comprised of the bolts that do not develop a linear plough response
before yielding, while a linear plough response is seen in the larger percentiles. The development of this
behaviour is the principal controlling factor over the displacement that may be observed before the bolt
yields. Figure 7.17 shows a conceptual version of 7.16. Using the initial stiffness of the MCB33s in the
database, percentiles are extrapolated to 0 load as if no pre-load were recorded. While it is acknowledged
that larger displacements may be observed at low loads as is the case for the rebar, these displacements
are still relatively small compared to those that occur as the bolt ploughs through the resin.
In Figure 7.17, it can be seen that the median displacement observed before an MCB33 begins to yield
is 35 mm, although can be as much as 100 mm. Extreme cases of low plough fall below P10 , although
a decreasing stiffness is observed. In Table 7.7, Mansour’s specifications regarding yield strength are
compared with the observations made in the database.
160
120
100
Loadf(kN)
80
PercentilesfoffYieldfLoad
P90 137fkN
60 P75 130fkN
P50 125fkN
P25 116fkN
40 P10 107fkN
20
0
0 10 20 30 40 50 60 70 80 90 100
Displacementf(mm)
Figure 7.17: Conceptual distribution of displacements for a pull test of a 2.4 m MCB33 without a
pre-load
The minimum thread yield strength (98.5 kN) appears to accurately reflect a minimum, as all of the 47
bolts with a yield strength recorded begin to yield above these values. Additionally, 81% begin yielding
above the “typical” thread yield strength. As such, pull test findings agree with the specifications
claimed by the supplier. However, it must be reiterated that because of the large displacement the
MCB33 undergoes even in static loading conditions, design methodologies that incorporate the static
reinforcement offered by the Modified Cone Bolt must consider displacement as well as load.
7.5 D-Bolt
Because of the high strength of the D-Bolt, testing personnel usually used pull testing as a verification
tool and did not yield the bolt in every testing campaign due to safety concerns associated with bolt
failure during a pull test. As a result, very few working capacities were recorded in comparison to values
of secant stiffness. As such, stiffness (and thus displacement) are primarily evaluated.
250
200
P10
P25
75
TypicalyYieldyLoad
P5
90
P
P
150
MinimumyYieldyLoad
Loady(kN)
100
50
0
0 5 10 15 20 25 30 35 40 45 50
Displacementy(mm)
Figure 7.18: Distribution of secant stiffness for a pull test with a pre-load of 17.8 kN on a 20 mm
D-Bolt
250
5
P2
50
0
200 5
P1
P 7 TypicalyYieldyLoad P 90
P
MinimumyYieldyLoad
150
Loady(kN)
100
50
0
0 5 10 15 20 25 30 35 40 45 50
Displacementy(mm)
Figure 7.19: Distribution of secant stiffness for a pull test with a pre-load of 17.8 kN on a 22 mm D-Bolt
Figures 7.18 and 7.19 show that the pull tests performed on both D-Bolts resulted in variable dis-
placements and behaviours, with the 22 mm bolt acting in a less stiff manner than a 20 mm bolt. As
discussed in Section 4.3.4, much of the 22 mm D-Bolt testing was performed in very fractured ground
conditions, potentially influencing the degree of displacement observed. Despite this, both D-Bolt config-
urations perform stiffly compared to the Modified Cone Bolt. As the stiffness is so variable, site–specific
pull testing is recommended in order to fully understand displacement development with load, and how
Chapter 7. Characterization of Rock Bolt Behaviour 129
250
200
150
Load (kN)
100
50
20 25 30
0
0 5 10 15 20 25 30 35 40 45 50
Displacement (mm)
Figure 7.20: 20 mm D-Bolt pull tests collected from Vale’s Sudbury operations
250
200
150
Load (kN)
100
50
0
0 5 10 15 20 25 30 35 40 45 50
Displacement (mm)
Figure 7.21: 22 mm D-Bolt pull tests collected from Vale’s Sudbury operations
Figures 7.20 and 7.21 serve to highlight the wide range of behaviour observed for both diameters
of D-Bolt when subject to a pull test. However, a cluster of similarly behaving 22 mm D-Bolts with
Chapter 7. Characterization of Rock Bolt Behaviour 130
relatively high stiffness is observed in Figure 7.21. All of these bolts were tested in one campaign,
demonstrating that the D-Bolt may exhibit consistent behaviour within some testing campaigns, but
not in others. Figures 7.22 and 7.23 show the distributions of behaviour for either diameter of D-Bolt.
250
200
150
Load (kN)
P75
P90
100
50
0
0 5 10 15 20 25 30 35 40 45 50
Displacement (mm)
250
200
P75
150 P90
Load (kN)
100
50
0
0 5 10 15 20 25 30 35 40 45 50
Displacement (mm)
The D-Bolt has the potential to exhibit high displacement at low loads as seen for P75 and P90
of displacement for the 22 mm D-Bolt. This is similar to the behaviour observed for rebar in Figure
7.12, where large initial displacements may be attributed to compression of the rock mass and testing
apparatus. The effect in Figure 7.23 continues to higher loads than the equivalent effect for rebar, but
Chapter 7. Characterization of Rock Bolt Behaviour 131
this is thought to be the result of a large proportion of pull tests on the D-Bolts begin performed in
poor ground. Figures 7.24 and 7.25 show conceptual versions of D-Bolt behaviour with manufacturer
specifications (Normet, 2014). Behaviours are extrapolated to 0 kN.
250
200
100
50
0
0 5 10 15 20 25 30 35 40 45 50
Displacementy(mm)
Figure 7.24: Conceptual distribution of displacements for a pull test of a 20 mm D-Bolt without a
pre-load
250
MinimumyYieldyLoad
150
Loady(kN)
100
50
0
0 5 10 15 20 25 30 35 40 45 50
Displacementy(mm)
Figure 7.25: Conceptual distribution of displacements for a pull test of a 22 mm D-Bolt without a
pre-load
While too little working capacity data was collected to reliably verify that the D-Bolt performs
at or above the supplier’s specifications in a pull test, it should be acknowledged that when working
Chapter 7. Characterization of Rock Bolt Behaviour 132
capacity/yield was captured during the test, it always occurred above the minimum yield load specified.
This includes 3 20 mm D-Bolts yielding above 140 kN, and 3 22 mm D-Bolt yielding above 170 kN.
250
200
150
Load (kN)
0
75
P5 5
P 2P90
100
P
P75
P25/P50
P10
50
0
0 5 10 15 20 25 30 35 40 45 50
Displacement (mm)
Figure 7.26: Distributions of secant stiffness and working capacity for a pull test on Swellex Pm12 and
Mn12 without a pre-load
Chapter 7. Characterization of Rock Bolt Behaviour 133
Little available data only permitted a limited analysis of Swellex performance compared to the
other bolts discussed. Figure 7.26 shows that the Pm12 and Mn12 bolts reach their working capacities
consistently at loads of about 90 kN. In Table 7.9 it can be seen that the Pm24 and Mn24 have the
potential to react in a very stiff, as well as a very soft manner. This variable behaviour is not as apparent
with the Pm12 and Mn12 bolts, although less data is available.
250
200
150
Load (kN)
100
50
0
0 5 10 15 20 25 30 35 40 45 50
Displacement (mm)
Figure 7.27: Swellex Pm12 and Mn12 pull tests collected from Vale’s Sudbury operations
Chapter 7. Characterization of Rock Bolt Behaviour 134
250
200
150
Load (kN)
100
50
0
0 5 10 15 20 25 30 35 40 45 50
Displacement (mm)
Figure 7.28: Swellex Pm24 and Mn24 pull tests collected from Vale’s Sudbury operations
In Figure 7.27 it can be seen that pull tests performed on Pm12/Mn12 may exhibit several behaviours.
In some cases, very little displacement is initially observed, although stiffness decays with load. In other
cases, a linear response to load is observed. The Pm24 and Mn24 bolts act in a more consistent manner,
usually responding linearly to load, although two pull tests exhibit low initial stiffness. The distributions
of these behaviours are constructed in Figures 7.29 and 7.30. Insufficient data was available to calculate
10th and 90th percentiles of displacement for the Pm12 and Mn12 bolts. All pull tests shown for these
configurations had no pre-load. A pre-load of 4 tons (35.6 kN) is depicted for the Pm24 and Mn24.
250
200
150
Load (kN)
50
0
0 5 10 15 20 25 30 35 40 45 50
Displacement (mm)
250
200
150
Load (kN)
100
50
0
0 5 10 15 20 25 30 35 40 45 50
Displacement (mm)
Figures 7.29 and 7.30 show that Swellex expandable bolts appear to react linearly to load, although
may bear low loads with little to no displacement. The Pm24 and Mn24 bolts exhibit a consistently stiff
response to load, however there does appear to exist a relatively small possibility of a much softer response
(note that this may be in part attributable to the reaction of the rock mass and testing apparatus). As
a result of a lack of displacement data at high loads, neither figure clearly shows yielding behaviour.
Figures 7.31 and 7.32 show smoothed relationships and depict the supplier’s specifications regarding
yield load.
250
200
150
Load (kN)
50
0
0 5 10 15 20 25 30 35 40 45 50
Displacement (mm)
Figure 7.31: Conceptual load–displacement behaviour of Pm12 and Mn12 bolts subject to a pull test
Chapter 7. Characterization of Rock Bolt Behaviour 136
250
200
Minimum Yield Load (Pm24)
150
Load (kN)
100
50
0
0 5 10 15 20 25 30 35 40 45 50
Displacement (mm)
Figure 7.32: Conceptual load–displacement behaviour of Pm24 and Mn24 bolts subject to a pull test
Only sufficient working capacity data was collected for the Pm12 and Mn12 to compare with supplier
specifications, shown in Table 7.10. As displacement data was collected in 1 ton (8.9 kN) load increments,
one of the data bins is the 89 kN - 97.9 kN. This means any bolt that yields between 89 kN and 97.9 kN,
it is said to have yielded at 89 kN using procedures enacted throughout this thesis. This is problematic,
as a large portion of the data (47%) falls in this range, and the supplier specified minimum yield load
for the Pm12 and Mn12 is 90 kN. As such, bolts with a recorded working capacity of 89 kN are assumed
to have passed the 90 kN threshold specified by the supplier.
Table 7.10: Swellex Pm12 and Mn12 performance compared to supplier specifications
Two of the Swellex pull tested did not comply with manufacturer specifications (both of them Pm12).
It is possible this is due to a difference in the definition of working capacity or its method of determination
used in this thesis, and the definition of yield strength used by Atlas Copco. Higher load resolution is
desirable to determine with more certainty whether bolts in the 89 kN to 97.9 kN range yielded above
or below 90 kN. Additionally, the supplier’s specifications for the Pm24 and Mn24 should be verified by
pulling these bolts to loads above 200 kN and 180 kN respectively.
7.7 Summary
An important part of designing a support system is the selection of the appropriate reinforcement
elements required to stabilise an excavation (Thompson et al, 2012). Figure 7.33 shows the conceptual
behaviours of all bolts discussed on similar scales. Some bolts (MCB33, both diameters of D-Bolt, and
Swellex Pm24/Mn24) have their behaviours extrapolated to 0 load as a result of a lack of data.
Chapter 7. Characterization of Rock Bolt Behaviour 137
250 250
200 200
Loady(kN)
Load (kN)
150 150
MedianyWorkingyCapacity
100 100
50 50
0 0
0 20 40 60 80 100 0 20 40 60 80 100
Displacement (mm) Displacementy(mm)
200
Load (kN)
150
100
50
0
0 20 40 60 80 100
Displacement (mm)
(c) MCB33
250 250
200 200
Load (kN)
150 150
100 100
50 50
0 0
0 20 40 60 80 100 0 20 40 60 80 100
Displacement (mm) Displacement (mm)
150 150
100 100
Minimum Yield Load
50 50
0 0
0 20 40 60 80 100 0 20 40 60 80 100
Displacement (mm) Displacement (mm)
(f) Swellex Pm12 and Mn12 (g) Swellex Pm24 and Mn24
Figure 7.33: Conceptual load–displacement behaviour extrapolated to 0 load for various rock bolts
subject to a pull test
Chapter 7. Characterization of Rock Bolt Behaviour 138
Figure 7.34 shows the median behaviours of all bolts discussed in this chapter extrapolated to a load
of 0 kN.
25,
MediangWorkinggCapacity
TypicalgYieldgLoad
MinimumgYieldgLoad
2,, 22gmmgDFBolt
SwellexgPm24
SwellexgMn24
95, 2,gmmgDFBolt
2,gmmgRebar
Loadg8kN(
MCB33
9,,
SwellexgPm923Mn92
FA35sgFA39
5,
,
, 2, 4, 6, 8, 9,,
Displacementg8mm(
Figure 7.34: Median load–displacement behaviour for all bolts with no pre-load
Chapter 7. Characterization of Rock Bolt Behaviour 139
Figure 7.34 shows that the 22 mm D-Bolt appears to have the greatest load capacity, although the
minimum yield loads shown for the Swellex Pm24 and Mn24 are minima specified by Atlas Copco.
The stiffest reinforcement element is the 20 mm rebar, likely due to limited load propagation down
the bolt. The least stiff bolts (in terms of secant stiffness) appear to be the MCB33 and the Swellex
Pm12 and Mn12. While the MCB33 is a yielding bolt, and the ploughing mechanism is the cause
of the large displacement, it is unclear why the Pm/Mn12 acted in so soft a manner as continuously
frictionally coupled bolts. The bolts with the lowest working capacity are the two FRS A diameters
shown, approximately representative of all FRS configurations tested. If the working capacity of the
MCB33 is defined by the onset of plough, it has quite a similar median capacity to that of the FRS
A bolts. However, the load at which plough occurs is much more variable than the ultimate capacities
achieved by FRSs as a whole.
Two bolts in Figure 7.34 were examined by Stillborg (1993), specifically resin-grouted 20 mm rebar
and a 39 mm nominal diameter FRS. The rebar tested by Stillborg (1993) behaves in a slightly stiffer
manner than the median depicted in Figure 7.34. This is to be expected comparing laboratory with
field testing, and the difference is only 3-4 mm. As Stillborg (1993) tested bolts across a simulated joint,
the yield response measured is for the bolt tendon, while the working capacity shown in Figure 7.34 is
the result of the threads yielding; plastic deformation starts around 15.3 tonnes (150 kN) for the rebar
tested by Stillborg (1993), while median rebar working capacity is 124.1 kN.
Stillborg (1993) depicts the SS39 as undergoing no displacement until its capacity is reached, at
which stage it slips out of the hole at a constant load. Pull tests that constituted the database showed
that bolts usually slip multiple times with growing displacement before reaching ultimate capacity; the
median behaviour shows a displacement of 11 mm before ultimate capacity is reached. The SS39 tested
by Stillborg (1993) was 3 m in length, and had a capacity of approximately 5 tonnes (49 kN). However, as
Stillborg (1993) conducted testing across a simulated joint, in reality two 1.5 m lengths of Split Set were
being tested. This means that a length-normalized capacity of 32.6 kN/m was recorded. In comparison,
the median ultimate capacity for 1.68 m FA35/39s was 62 kN (and the overall median was 38.7 kN/m),
however Tomory et al. (1998) recorded an average Split Set capacity of 31.9 kN/m, likely as the result
of jackleg installations.
While the Swellex tested by Stillborg (1993) is no longer marketed, its overall behaviour may still
be compared to modern Swellex bolts. In Stillborg’s (1993) test, the EXL Swellex dowel undergoes a
very stiff response before slipping in a manner comparable to an FRS at 11 tonnes (108 kN). In situ
pull tests on modern Swellex bolts show quite a different behaviour. Although a softer linear response
is observed, the modern Swellex is not intended to slip when installed in hard rock, and was generally
shown to yield in the database, excluding partial embedment tests and paste/sand fill installations.
When using the figures in this section, it must be kept in mind that these results represent pull tests
on bolts, and may not necessarily reflect performance when subject to in situ loading mechanisms. In
Chapter 8, the limitations of this thesis, including the pull tests as a method of assessing reinforcement
element performance, are discussed. Recommendations are also provided to improve the pull testing
procedures observed during this thesis.
Chapter 8
Conclusions
In the final chapter, the contributions made by this thesis are outlined. Limitations are discussed,
recommendations made and the path forward laid out.
8.1 Contributions
In order to conduct this thesis, a large database of pull tests performed on various rock bolts was
assembled. This database includes 985 individual pull tests performed between 2003 and 2015 on friction
rock stabilizers, rebar rock bolts, Modified Cone Bolts, D-Bolts and expandable bolts, amongst others.
The data was collected from 6 Vale operations in and around the Sudbury Basin. Although databases
of pull tests have been assembled previously, this database incorporates a wider range of bolt types. In
addition, collecting pull test data from mines within relatively close proximity to one another under one
company provides consistency in procedures and conditions for bolt installation and testing, as well as
consistency in the ground control products used on and delivered to the 6 mine sites.
A review was performed on methods of testing rock bolts. This included the ASTM standard (ASTM
D4435-13) and ISRM suggested methods for pull testing. These were compared with the methods and
apparatuses used for pull testing in practice, and shortcomings of the standard, the suggested method,
and the practised methods of pull testing outlined. The measurements taken from a pull test, metrics
that may be calculated and the significance of the measurements and metrics was also discussed.
An analysis of the data collected followed. Theoretical bolt behaviour of rock bolts subject to loading
was compared to behaviour observed during the in situ pull tests. Various metrics of bolt performance
describing load and displacement behaviour, primarily ultimate/working capacity and several measures
of stiffness, were quantified and their distributions characterised for different types of bolt. Subsequently,
the influence of factors associated with bolt installation, the rock mass and the bolt itself on rock bolt
performance were investigated. The findings of this analysis built on and was compared to previous
work when possible.
Using the collected load-displacement behaviour of friction rock stabilizers, rebar, the MCB33, D-
Bolts and Swellex, a distribution of behaviour was built for each bolt. Although the characterisation of
rock bolt behaviour has been previously performed in a laboratory setting, the analysis contained herein
uses a larger dataset than these analyses. Additionally, as the pull tests are performed in situ, they
better represent installation conditions found at a mine site and as a result are more applicable to the
140
Chapter 8. Conclusions 141
of many rock bolt types, and although the ASTM pull testing standard (D4435-13) was developed more
recently, it does not distinguish between types of rock bolt beyond grouted versus mechanically anchored
bolts.
8.3 Limitations
The limitations of this thesis can be broadly divided into three classes; the limitations of a pull test as
a method of assessing performance, those associated with the pull test as implemented in practice, and
the limitations of the data present in the database.
The primary conceptual limitation of a pull test is that it is only representative of a very specific
loading scenario as bolts are pulled axially at the head/thread. Loading in an underground environment
may be much more complex; reinforcement elements may have multiple axial, shear or rotational loads
applied at any location on their lengths, and these forces may evolve with time. Additionally, loading is
only applied in a quasi-static manner during a pull test. The response to dynamic load is not measured,
although may be of interest for yielding reinforcement elements.
The most prevalent difficulty encountered when evaluating the performance of the rock bolts tested
in the database was the presence of displacements attributed to the response of the excavation surface
as opposed to that of the bolt alone. While the measurement of this response may be applicable to
certain scenarios, it introduced not only variability, but also bias into the data, as certain elements
(such as the D-Bolt) tended to be tested in poorer quality rock masses than others. ASTM D4435-08
provides a methodology that limits the measurement of the rock mass response by using an apparatus
that measures displacement relative to a stationary datum, however it was not enacted by pull testing
personnel. Additionally, methods of data recording varied in the database between manual and digital
displacement logging. Manual logging is performed in loading increments which may result in poor data
resolution, while digital logging was often performed at a resolution that does not reflect the capabilities
of this method.
Shortcomings existed with the way in which pull tests were reported. Many parameters (for example
rock mass quality or borehole diameter) were inconsistently or seldom noted. Pull tests were almost
always done on rock bolts that had been installed the same day as the pull test, or if not, at some un-
noted previous date. A significant shortcoming of the database is the lack of data describing bolt failure
during a pull test. It is acknowledged that this was to preserve the safety of the personnel conducting
the test, but it is an advantage of laboratory testing that bolts may be failed in a controlled and safe
manner.
8.4 Recommendations
Attempts have previously been made to introduce a standardized pull test data sheet, in which various
parameters are to be recorded. For example, Soni (2000) presents a sheet on which various testing
parameters may be recorded. These include rock strength parameters, rock mass classification(s), the
type and length of bolt, the diameter of the drillbit and drillhole, inflation pressure, inflation or drive
time, installation date, residence time, and grout type, length and collar depth. In the pull test reports
collected, these parameters were noted with varying frequency, although all are pieces of information
that may affect performance. In addition to these, there are further recommendations to be made with
Chapter 8. Conclusions 143
• The objective and type of pull test should be stated. For example: partial embedment/encapsulation,
bolt slip, bolt yield, achieve a load of x kN. If a test is ended prematurely, it should be explained
why.
• Record the bolt name including configuration and presence or lack of possible modifications, for
example a plastic coating. Bolt length (as well as anchorage length) and diameter should be clear.
• Report equipment used for the pull test. This includes the installation equipment, the loading
system, displacement measurement system and the logging equipment. If an electric pump is
used to apply load, record loading rate. If an electronic logger is used, report logging frequency.
It should be noted whether the displacement measurement system is mounted on the pull test
apparatus itself, or if it is stationary relative to the excavation.
• It should be explicitly stated what units are used to record load and displacement. If load is
recorded in tons, then short, long, or metric should be specified.
• If the bolt is pre-loaded before recording displacement, that pre-load should be stated.
• If applicable, a bolt’s spin time should be recorded. For these scenarios, temperature should also
be measured.
Appendix A presents pull test informations sheets from ASTM and ISRM, as well as modified data
sheets, including one for the overall campaign and one to log each pull test individually. In terms of pull
test procedure, modifications could be made without significantly affecting the amount of time required
to perform them. They are as follows.
• If recording data manually, use a maximum load increment of 5 kN or 0.5 tons once load has been
built to within 20 kN or 2 tons of the expected working capacity to obtain a more precise pull test
result.
• If recording data electronically, the additional load resolution possible with this method must be
exercised.
• If load is not built despite an observed increase in displacement, take two or more recordings of
displacement to make it clear slip or perfectly plastic behaviour was observed.
• If a reinforcement element such as an FRS slips, the test should continue until a pre–determined
displacement (e.g. 15 mm) has occurred. This ensures that maximum recorded loads are at least
comparable between tests, and the criterion used to dictate the end of a test is consistent.
• When using a method of measuring displacement that incorporates the rock mass response, un-
loading of the reinforcement element should be recorded to assess elastic deformation, and thus
estimate how much of the displacement response may be attributed to the rock mass reaction.
Displacement should be recorded at the highest load achieved. Pump pressure should then be
released until a load equal to or greater than the pre-load is reached (if no pre-load is used, at
Chapter 8. Conclusions 144
least 15 kN is recommended), and displacement should be recorded again. This is to minimize the
effect of rock mass and pull test apparatus relaxation. The measurement of the stiffness of the
unloading response gives an idea of how load attenuates down the bolt length from the point of
load application, and may be used as a relative measure of performance.
In some cases, the analysis presented was hampered by a lack of data for specific types of bolt
or parameters recorded. These limitations may be overcome by expanding the database. One of the
advantages of the database assembled for this thesis over previously constructed pull test databases is
that all pull tests were performed in and around the Sudbury Basin, reducing variability in installation
conditions and procedures. If possible, this aspect of the database should be preserved, including data
only collected from hard rock mines, and should be tested in a similar or superior manner to the pull
tests presented herein.
Abdellah, W., Raju, G., Mitri, H., and Thibodeau, D. Stability of underground mine development
intersections during the life of mine plan. International Journal of Rock Mechanics and Mining Science,
72:173–181, 2014.
ASM International. Atlas of stress-strain curves. ASM International, Materials Park, OH, 2002.
ASM Standard B1.13M, 2005. Metric screw threads: M Profile, ASME, New York, NY, 2005,
www.asme.org.
ASTM Standard A370, 2012. Standard test methods and definitions for mechanical testing of steel
products, ASTM International, West Conshohoken, PA, 2012, DOI: 10.1520/A0370-12, www.astm.org.
ASTM Standard A615, 2015. Standard specification for deformed and plain carbon-steel bars for con-
crete reinforcement, ASTM International, West Conshohoken, PA, 2015, DOI: 10.1520/A0615-15A,
www.astm.org.
ASTM Standard D4435, 2013. Standard test method for rock bolt anchor pull test, ASTM International,
West Conshohoken, PA, 2013, DOI: 10.1520/D4435-13E01, www.astm.org.
ASTM Standard D7401, 2008. Standard test methods for laboratory determination of rock anchor capac-
ities by pull and drop tests, ASTM International, West Conshohoken, PA, 2008, DOI: 10.1520/D7401-08,
www.astm.org.
ASTM Standard E6, 2009. Standard terminology relating to methods of mechanical testing, ASTM
International, West Conshohoken, PA, 2008, DOI: 10.1520/D0006-09B, www.astm.org.
ASTM Standard F432, 2013. Standard specification for roof and rock bolts and accessories, ASTM
International, West Conshohoken, PA, 2013, DOI: 10.1520/F0432-13, www.astm.org.
ASTM Standard F606, 2013. Standard test methods for determining the mechanical properties of exter-
nally and internally threaded fasteners, Washers, and Rivets, ASTM International, West Conshohoken,
PA, 2013, DOI: 10.1520/F0606M-13, www.astm.org.
Atlas Copco. Swellex rock bolts product catalogue. Nacka, Sweden, 2012.
Baecher, G. and Christian, J. Reliability and statistics in geotechnical engineering. Wiley, Chichester,
England, 2003.
Balanda, K. and MacGillivray, H. Kurtosis: a critical review. The American Statistician, 42:111–119,
1988.
145
List of References 146
Bieniawski, Z. Engineering rock mass classifications: a complete manual for engineers and geologists
in mining, civil and petroleum engineering. Wiley, New York, NY, 1989.
Bieniawski, Z. et al. Suggested method for determining deformability of rock materials in uniaxial
compression. In Rock characterization, testing and monitoring, ISRM suggested methods, pages 111–
116. Permagon Press, Oxford, 1978.
Brady, B. and Brown, E. Rock mechanics for underground mining. Kluwer Academic, Dordrecht,
Netherlands, 3rd edition, 2006.
Cai, M., Champaigne, D., and Kaiser, P. Development of a fully debonded cone bolt for rockburst
support. In Van Sin Jan, M. and Potvin, Y., editors, Deep Mining 2010 - the fifth international
seminar on deep and high stress mining, pages 329–342, Santiago, Chile, 2010. Australian Centre for
Geomechanics.
Card, K., Gupta, V., McGrath, P., and Grant, F. The Sudbury Structure: its regional, geological and
geophysical setting. In The geology and ore deposits of the Sudbury Structure, pages 25–44. Government
of Ontario, Toronto, ON, 1984.
Carmichael, H. Operations suspended. The Sudbury Star, October 19 2012. Retrieved March 5, 2015
from http://www.thesudburystar.com/2012/10/18/vale-shutting-down-frood-mine.
Chinnasane, D., Knutson, M., and Watt, A. Use of cable bolts to reinforce the hanging pillars and
improve the ore recovery when stopes are mined using double top sills at Vale’s Copper Cliff Mine.
In Hudyma, M. and Potvin, Y., editors, Deep Mining 2014 - proceedings of the seventh international
conference on deep and high stress mining, pages 305–314, Nedlands, Australia, 2014. Australian Centre
for Geomechanics.
Cochrane, L. Ore deposits of the Copper Cliff Offset. In Pye, E., Naldrett, A., and Giblin, P., editors,
The geology and ore deposits of the Sudbury Structure, pages 347–359. Government of Ontario, Toronto,
ON, 1984.
CSA Standard G30.18-09. Carbon steel bars for concrete reinforcement, Canadian Standards Associa-
tion, Toroto, ON, 2009.
Doane, D. and Seward, L. Measuring skewness: a forgotten statistic? Journal of Statistics Education,
19:111–119, 2011.
Doucet, C. and Voyzell, B. Technical information data sheets. Technical report, CanmetMINING,
Ottawa, ON, 2012.
Duthin, D. and Starnes, M. Strength and ductility of concrete beams reinforced with carbon FRP and
steel. Technical Report NISTIR 6830, National Institute of Standards and Technology, Gaithersburg,
MD, 2001.
List of References 147
Fox, J. Applied regression analysis and generalized linear models. Sage, Los Angeles, CA, 2nd edition,
2008.
Gaudreau, D., Aubertin, M., and Simon, R. Performance assessment of tendon support systems sub-
mitted to dynamic loading. In Villaescusa, E. and Potvin, Y., editors, Ground support in mining and
underground construction - proceedings of the fifth international symposium on ground support, pages
299–312, Perth, 2004. Taylor and Francis.
Golightly, J. The Ni-Cu-PGE deposits of the Sudbury Igneous Complex. In Rousell, D. and Brown,
G., editors, A field guide to the geology of Sudbury, Ontario, pages 121–131. Ontario Geological Survey,
2009.
Hadjigeorgiou, J. and Charette, F. Rock bolting for underground excavations. In Hustrulid, W. and
Bullock, R., editors, Underground mining methods: engineering fundamentals and international case
studies, pages 547–554. Society for Mining, Metallurgy, and Exploration, Littleton, CO, 2001.
Hadjigeorgiou, J. and Potvin, Y. A critical assessment of dynamic rock reinforcement and support
testing facilities. Rock Mechanics and Rock Engineering, 44:565–578, 2011.
Hawkes, J. and Evans, R. Bond stresses in reinforced concrete columns and beams. Journal of the
Institute of Structural Engineers, 29:323–327, 1951.
Hoek, E., Kaiser, P., and Bawden, W. Support of underground excavations in hard rock. Balkema,
Rotterdam, 1995.
ISRM. Suggested methods for rock bolt testing, Part 1: Suggested method for determining the strength
of a rockbolt anchor (pull test). In Brown, E., editor, Rock characterization, testing and monitoring,
pages 163–166. Pergamon Press, Oxford, NY, 1981.
Li, C. Performance of D-Bolts under static loading. Rock Mechanics and Rock Engineering, 45:183–192,
2012.
Li, C. and Doucet, C. Performance of D-Bolts under dynamic loading. Rock Mechanics and Rock
Engineering, 45:193–204, 2012.
Li, C. and Stillborg, B. Analytical models for rock bolts. International Journal of Rock Mechanics and
Mining Sciences, 36:1013–1029, 1999.
Li, C., Stjern, G., and Myrvang, A. A review on the performance of conventional and energy-absorbing
rockbolts. Journal of Rock Mechanics and Geotechnical Engineering, 6:315–327, 2014.
Lightfoot, P. and Farrow, C. Geology, geochemistry and mineralogy of the Worthington Offset Dike:
a genetic model for offset dike mineralization in the Sudbury Igneous Complex. Economic Geology,
92(7):1419–1446, 2002.
Long, D. The Huronian Supergroup. In Rousell, D. and Brown, G., editors, A field guide to the geology
of Sudbury, Ontario, pages 14–30. Ontario Geological Survey, 2009.
Ma, L., Gagnon, R., and Mainville, L. Report for Yield-Lok bolts qualification tests at Vale. Technical
report, Jennmar Canada, 2011.
Mainville, L., Guenette, A., and Desjardins, M. Pull test report for Vale - Coleman Mine. Technical
report, Jennmar Canada, 2012.
Martin, L., Tijani, M., and Hadj-Hassen, F. A new analytical solution to the mechanical behaviour of
fully grouted rockbolts subjected to pull-out tests. Construction and Building Materials, 25:749–755,
2010.
McCreath, D. and Kaiser, P. Evaluation of current support practices in burst-prone ground and
preliminary guidelines for Canadian hard rock mines. In Villaescusa, E. and Potvin, Y., editors, Rock
support in mining and underground construction: proceedings of the international symposium on rock
support, pages 611–619, Rotterdam, 1992. Balkema.
Morissette, P., Hadjigeorgiou, J., and Thibodeau, D. Investigating the dynamic-load demand on support
systems using passive monitoring data. International Journal of Rock Mechanics and Mining Sciences,
67:115–126, 2014.
Mukwakwami, J., LaFrance, B., Lesher, C., Tinkham, D., Rayner, N., and Ames, D. Deformation,
metamorphism and mobilization of Ni-Cu-PGE sulfide ores at Garson Mine, Sudbury. Miner Deposita,
49:175–198, 2013.
Naldrett, A. Evolution of ideas about the origin of the Sudbury Igneous Complex and its associated
Ni-Cu-PGE mineralization. In Rousell, D. and Brown, G., editors, A field guide to the geology of
Sudbury, Ontario, pages 106–120. Ontario Geological Survey, 2009.
List of References 149
NIST-SEMATECH. e-handbook of statistical methods. 2003. Retrieved June 3, 2015 from http://www.
itl.nist.gov/div898/handbook/.
Potvin, Y. and Hadjigeorgiou, J. Ground support strategies to control large deformations in mining
excavations. In Sixth international symposium on ground support in mining and civil engineering
construction, pages 545–560, Marshalltown, South Africa, 2008. SAIM M, SANIRE and ISRM.
Rousell, D., Fedorowich, J., and Dressler, B. Sudbury Breccia (Canada): a product of the 1850 Ma
Sudbury Event and host to footwall Cu-Ni-PGE deposits. Earth-Science Reviews, 60:147–174, 2003.
Scolari, F. Swellex Manganese offers improved work index. In Smith, M., editor, Rock and soil rein-
forcement, pages 11–13. Atlas Copco, Orebro, Sweden, 3rd edition, 2005.
Scott, J. Friction rock stabilizers - a new reinforcement method. In Brown, W., Green, S., and Hus-
trulid, W., editors, Monograph 1 on Rock Mechanics Applications in Mining, pages 242–249. American
Institute of Mining, Metallurgical, and Petroleum Engineers, 1977.
Sentinel Captial Partners. Sentinel acquires Fasloc from Dupont. (2005, October 3).
Sentinel Captial Partners. Sentinel sells Fasloc to DSI USA. (2007, January 22).
Simser, B., Andrieux, P., Langevin, F., Parrott, T., and Turcotte, P. Field behaviour and failure
modes of Modified Conebolts at the Craig, LaRonde and Brunsick Mines in Canada. In Potvin, Y.,
Hadjigeorgiou, J., and Stacey, D., editors, Deep and High Stress Mining: proceedings of the third
international seminar on deep and high stress mining, page 13p. Australian Centre for Geomechanics,
2006.
Snelling, P., Godin, L., and McKinnon, S. The role of geologic structure and stress in triggering remote
seismicity in Creighton Mine, Sudbury, Canada. International Journal of Rock Mechanics and Mining
Science, 58:166–179, 2013.
Soni, A. Analysis of Swellex bolt performance and a standardized rockbolt pull test datasheet and
database. Master’s thesis, University of Toronto, 2000.
St-Pierre, L., Hassani, F., Radziszewski, P., and Ouellet, J. Development of a dynamic model for a
cone bolt. International Journal of Rock Mechanics and Mining Science, 46:107–114, 2009.
Stillborg, B. Rockbolt tensile loading across a joint. In International Mine Water Association Sympo-
sium Zambia 1993, pages 587–604. IMWA, 1993.
Thompson, A., Villaescusa, E., and Windsor, C. Groud support terminology and classification: an
update. Geotechnical and Geological Engineering, 30:553–580, 2012.
Tomory, P. Analysis of Split Set bolt performance. Master’s thesis, University of Toronto, 1997.
Tomory, P., Grabinsky, M., Curran, J., and Carvalho, J. Factors influencing the effectiveness of Split
Set friction stabilizer bolts. CIM Bulletin, 91:205–214, 1998.
Vale. Copper Cliff mine design package. Technical report, Vale, 2010.
Vale. Coleman Mine mine design package. Technical report, Vale, 2011.
Windsor, C. Rock reinforcement systems. International Journal of Rock Mechanics and Mining Science,
34(6):919–951, 1997.
Windsor, C. and Thompson, A. Reinforcement design for jointed rock masses. In Tillerson, J. and
Wawersik, W., editors, Proceedings of the 33rd US symposium on rock mechanics, pages 521–530,
Rotterdam, 1992. Balkema.
Appendices
151
Appendix A
Figure A.1: Rock bolt pull test sample form (ASTM D4435, 2013)
152
Appendix A. Pull Testing Forms 153
TESTpRESULTS: Maximumppullpforce:
Displacementpatpmaximumppullpforce: Maxbpdisplacementpinptest:
Naturepofpfailureporpyield:
Otherpremarks:
Testedpby: Checkedpby:
Figure A.2: Rock bolt pull test data sheet (After ISRM, 1981)
Appendix A. Pull Testing Forms 154
PullBtestBobjective:
(e.g.wqualitywcontrol,w
productwverification,wetc.)
TestBLocation: BoltBinstallationBdateBEMMNDDNYYG:
o o
TemperatureBduringBinstallation: C NB F Wet(bulb( Dry(bulb
RockwMasswParameters
RockBtype: UCSBEMPaG: EBEGPaG:
RMR: to Q: to RQD: to
RockBmassBquality:
GSI: to OtherBEspecifyGB : to
Notes:B______________________________________________________________________________________
BoltwandwAnchor Parameters
RockBboltBtype: BoltBlength: mN ’ ”B
GalvanizationNcoating: NominalBboltBdiameter: mm N ”
AnchorBtype: InstallationBtorque:
GroutBtype: Resin Cement Measured bolt diameter: mm N ”
GroutBdescription:B
(includingwcartridgew
lengths,ww/cwratio,wetc.)
Equipment andwApparatus
DrillNinstallationBequipment: MeasuredBdrillBbitBdiameter: mm N ”
LoadingBpumpBtype: Manual Electric DataBrecording: Manual Electronic
LoadBunits: kN t((short)( t((long)( LoadBintervalsBorB
t((metric)( measurementBfrequency:
DisplacementBunits: mm( in( Preload: kN N t
DisplacementBmeasured: Ram(travel( Bolt(head(relative(to(stable(point(
AdditionwComments
Pull Test Data Sheet
1 2 3 4 5 6 7 8 9 10
MeasuredSdiameter
DrillholeSdiameter
DriveS/SInflationS/S
SpinStime
PartialSembedmentS
orSencapsulation?
Preload
NOTES Load Disp Load Disp Load Disp Load Disp Load Disp Load Disp Load Disp Load Disp Load Disp Load Disp
StoppingScriteria/S
natureSofSfailure
Notes