This document is downloaded from DR‑NTU (https://dr.ntu.edu.

sg)
Nanyang Technological University, Singapore.

Reinforcement mechanism of rockbolt system for

underground excavation

Nie, Wen

2019

Nie, W. (2019). Reinforcement mechanism of rockbolt system for underground excavation.

Doctoral thesis, Nanyang Technological University, Singapore.

https://hdl.handle.net/10356/83549

https://doi.org/10.32657/10220/49777

Downloaded on 09 Feb 2024 03:58:45 SGT

Reinforcement Mechanism of Rockbolt System for
Underground Excavation

NIE WEN

SCHOOL OF CIVIL AND ENVIRONMENTAL ENGINEERING

2019
Reinforcement Mechanism of Rockbolt System
for Underground Excavation

NIE WEN

School of Civil and Environmental Engineering

A thesis submitted to the Nanyang Technological University

in partial fulfilment of the requirements for the degree of
Doctor of Philosophy
 Statement of Originality

I hereby certify that the work embodied in this thesis is the result of original research,

is free of plagiarised materials, and has not been submitted for a higher degree to any

other University or Institution.

 Supervisor Declaration Statement

I have reviewed the content and presentation style of this thesis and declare it is free

of plagiarism and of sufficient grammatical clarity to be examined. To the best of my

knowledge, the research and writing are those of the candidate except as

acknowledged in the Author Attribution Statement. I confirm that the investigations

were conducted in accord with the ethics policies and integrity standards of Nanyang

Technological University and that the research data are presented honestly and

without prejudice.
 Authorship Attribution Statement

This thesis contains material from 2 papers published in the following peer-reviewed

journals where I was the first author.

Chapter 3 is published as Nie, W., Zhao, Z.Y., Guo, W., Shang, J., Wu, C. Bond-slip
modeling of a CMC bolt element using 2D-DDA method. Tunnelling and
Underground Space Technology, 2019, 85: 340-353. DOI:
https://doi.org/10.1016/j.tust.2018.12.025

The contributions of the co-authors are as followings:

• A/Prof Zhao provided guidance in the numerical models.

• I initiated the project direction and wrote the drafts of the manuscript.

• Dr Guo assisted in the data analyzing and edited the manuscript drafts.

• Dr Shang and Dr Wu suggested the data analyzing methods.

Chapter 5 is published as Nie, W., Zhao, Z.Y., Ma, S.Q., and Guo W. Efforts of joints
on the reinforced rock units of fully-grouted rockbolts. Tunnelling and Underground
Space Technology, 2018, 71, 15-26. DOI: https://doi.org/10.1016/j.tust.2017.07.005

The contributions of the co-authors are as followings:

• A/Prof Zhao provided guidance in the numerical models and edited the
manuscript drafts.

• I initiated the project direction and wrote the drafts of the manuscript.
 • Dr Ma assisted in the data analyzing.

• Dr Guo assisted in the data analyzing and revised the manuscript.

Chapter 7 is published as Nie, W., Zhao, Z.Y., Goh A.T.C., Song, M.K., Guo W., and
Zhu X. Performance based support design for horseshoe-shaped rock caverns using
2D numerical analysis. Engineering Geology, 2018, 245, 266-279. DOI:
https://doi.org/10.1016/j.enggeo.2018.09.007.

The contributions of the co-authors are as followings:

• A/Prof Zhao provided the initial project direction and edited the manuscript
drafts.

• I performed all the numerical simulation and collected the data.

• I conducted data evaluation together with Dr Zhu and Dr Guo.

• I wrote the drafts of the manuscript. The manuscript was revised by Dr Guo.

• A/Prof Goh and Dr Song suggested the data analyzing methods.

 Acknowledgement

ACKNOWLEDGEMENT

I would like to express the sincerest gratitude to my supervisor, Professor Zhao Zhiye,
for his never-ending inspiration, patience and guidance. His endlessly support and
invaluable suggestions came handy throughout the last three years. I have greatly
benefited from his wealth of knowledge and insight. I could not ask for a better
supervisor.

I also would like to thank Prof Goh Anthony, Prof Chu Jian, Prof Zhou Yingxin, Dr Lee
Hong Song, Dr Song Myungkyu, Dr Ma Shuqi, Dr Shang Junlong and Dr Zhang
Wengang for their technical supports and their sharing research experiences. I would
like to give my appreciation to my colleagues and friends including Dr Ning Youjun,
Dr Chen Huimei, Dr Wu Shifan, Ms Liu Qian, Mr Xiao Fei, Mr Yokota Yasuhiro, Dr
He Lei and Dr Sun Jianping for their encouragement and friendship.

And finally, I wish to give my appreciation to my husband, colleague and best friend,
Dr Guo Wei, for his technical sharing, unbiased understanding and endless support. I
want to thank my little boy, Mr. Guo Youjia, for his lovely encouragement. His smiles
are the therapy for me in troubles and the energy to keep me going. I also could not have
done this work without the support and love from my mother Ms. Wang Lianyun, my
farther Mr. Nie Peixun and my parents-in-law. I could not thank them enough.

I
Acknowledgement

II
 Table of contents

TABLE OF CONTENTS

1.1 Background ................................................................................................... 1

1.2 Key objectives............................................................................................... 3

1.3 Outline of thesis ............................................................................................ 3

2.1 Rockbolt ........................................................................................................ 5

2.1.1. Types of rockbolts ............................................................................. 5

2.1.2. Load transfer in rockbolt.................................................................... 8

2.1.3. Load carrying capacity of a rockbolt ............................................... 16

2.2 Rockbolt design principles.......................................................................... 20

2.2.1. Analytical methods .......................................................................... 21

2.2.2. Empirical methods ........................................................................... 25

2.2.3. Numerical methods .......................................................................... 29

III
 Table of contents

2.3 Rock-rockbolt interaction in underground excavation ............................... 36

2.3.1. GRC ................................................................................................. 38

2.3.2. SCC .................................................................................................. 43

2.3.3. LDP .................................................................................................. 44

2.3.4. Application of rock-rockbolt interaction diagram ........................... 45

2.4 Basic discontinuous deformation analysis (DDA) theory .......................... 47

2.4.1. Submatrices of elastic strains ........................................................... 49

2.4.2. Submatrices of point loading ........................................................... 50

2.4.3. Submatrices of inertia forces ........................................................... 50

2.4.4. Submatrices of bolting connection................................................... 51

2.4.5. Submatrices of contacts ................................................................... 53

2.4.6. DDA coding framework .................................................................. 56

2.5 Summaries and conclusions ........................................................................ 57

3.1 Introduction ................................................................................................. 59

3.2 Trilinear bond-slip model ........................................................................... 60

3.3 Crack propagation in 2D-DDA ................................................................... 62

3.4 Numerical Modelling .................................................................................. 64

3.4.1. 2D-DDA models .............................................................................. 64

3.4.2. Determination of the properties of artificial joint in mortar ............ 65

3.4.3. Determination of the frictional properties at the M-B interface ...... 67

3.5 Numerical Results ....................................................................................... 69

3.6 Parametric studies ....................................................................................... 71

3.6.1. Effects of the rib face angle ............................................................. 71

3.6.2. Effects of the rib spacing ................................................................. 74

3.6.3. Effects of the normal stress .............................................................. 77

IV
 Table of contents

3.7 Conclusions ................................................................................................. 82

4.1 Introduction ................................................................................................. 83

4.2 Development of the rockbolt model ........................................................... 84

4.2.1. Implementation of the rockbolt model in 2D-DDA......................... 84

4.2.2. Analytical solutions of a rockbolt model ......................................... 85

4.2.3. Axial behaviour................................................................................ 87

4.2.4. Bond behaviour ................................................................................ 87

4.2.5. Shear behaviour ............................................................................... 89

4.2.6. Programming ................................................................................... 93

4.3 Verification of the rockbolt model by experimental results ....................... 97

4.3.1. Verifying bond behaviour by pull out test ....................................... 97

4.3.2. Verifying shear behaviour by shear test .......................................... 99

4.3.3. Verifying axial behaviour by pull out tests .................................... 103

4.4 Parameters for rockbolt design ................................................................. 107

4.4.1. 2D-DDA models ............................................................................ 107

4.4.2. Effects of the types of rockbolt ...................................................... 109

4.4.3. Effects of the embedded length ..................................................... 113

4.4.4. Effects of the shear angle ............................................................... 116

4.4.5. Effects of the rock deformability ................................................... 117

4.5 Conclusions ............................................................................................... 119

5.1 Introduction ............................................................................................... 121

5.2 2D-DDA model......................................................................................... 123

V
 Table of contents

5.3 RRU of single rockbolt ............................................................................. 124

5.3.1. Influence of the joint orientation ................................................... 124

5.3.2. Influence of the friction angle and Young’s modulus of rock mass ....
....................................................................................................... 128

5.3.3. Influence of the rockbolt length ..................................................... 130

5.4 Grouped rockbolts ..................................................................................... 130

5.5 Discussion ................................................................................................. 135

5.6 Pressure arch estimation using RRUs ....................................................... 137

5.6.1. 2D-DDA model ............................................................................. 137

5.6.2. Influence of the horizontally persistent joints ............................... 139

5.6.3. Influence of the vertically persistent joints.................................... 142

5.6.4. Discussion and limitation .............................................................. 144

5.7 Conclusions ............................................................................................... 146

6.1 Introduction ............................................................................................... 147

6.2 2D rock wedges stability in tunnel............................................................ 149

6.2.1. 2D-DDA model ............................................................................. 150

6.2.2. Model verification ......................................................................... 150

6.2.3. Parameter studies ........................................................................... 156

6.3 Case study ................................................................................................. 158

6.4 Conclusions ............................................................................................... 166

7.1 Introduction ............................................................................................... 167

7.2 Theoretical background ............................................................................ 169

7.2.1. Characteristic curves ...................................................................... 169

VI
 Table of contents

7.2.2. Performance functions ................................................................... 171

7.3 Analysis of horseshoe cavern using numerical method ............................ 173

7.3.1. Numerical models .......................................................................... 173

7.3.2. Parametric study and characteristic curve ..................................... 176

7.4 Prediction of cavern performance using ANN ......................................... 179

7.5 Development of an evaluation chart based on ANN models .................... 183

7.6 A case study .............................................................................................. 187

7.7 Conclusions ............................................................................................... 191

8.1 Summary ................................................................................................... 193

8.2 Future work plan ....................................................................................... 196

VII
Table of contents

VIII
 Summary

SUMMARY

A rockbolt is a reinforcing element installed untensioned or tensioned into the

ground to restrict the deformation of the surrounding rock mass in rock engineering.
The rockbolt is usually a solid bar or tube element which is installed within the
boreholes drilled into the rock. To improve the rockbolt design in complex rock
conditions, it is necessary to have a good understanding of the reinforcement
mechanism of the rockbolts. In this thesis, the reinforcement mechanism of a
rockbolt system under excavation conditions is studied using the two-dimensional
discontinuous deformation analysis (2D-DDA) method and other numerical tools.

The load transfer capacity of a rockbolt element is controlled by the components of

bonds at the rock and rockbolt, such as the rock mass, the mortar and the profile
configuration of the rockbolt. 2D-DDA method is used to investigate the pullout
performance of a continuously mechanically coupled (CMC) rockbolt element. The
flat joint contact model is introduced into the 2D-DDA code to simulate the force
versus displacement behaviour of an artificial joint. Parametric studies show that the
confining pressure plays an important role in the bond-slip modeling. The obtained
key parameters of the trilinear bond-slip model can be used to simulate the bond
behaviour between the rockbolt and rock under similar conditions.

A rockbolt model is developed and integrated into the 2D-DDA code based on the
load transfer theory. Three mechanical behaviours are considered in the model, i.e.,
the bond, the axial tension and the shear at joint. The proposed rockbolt model
could be used to simulate four major failure modes of rockbolt, such as the de-
bonding along the interface, tensile failure, faceplate failure and shear failure. The
numerical results show that the numerical models have reasonable agreements with
the experiments results. The parameter studies show that the load transfer and load
capacity should be reconsidered when the ground conditions are changed.

IX
 Summary

The rock/rockbolt interactions are examined by simulating the pullout test in rock
mass with systematic joint sets. The effects of the rock properties, joint orientations,
rockbolt lengths and in-plane spacing are studied. The reinforcement rock unit
(RRU) induced by single rockbolt is normally in a cone shape and restrained by the
orientation of the discontinuities. The RRU induced by grouped rockbolts is heavily
influenced by the joint orientations and the rockbolt spacing. Further applications of
the RRUs on the estimation of the artificial pressure arch in the roof are illustrated.

The rockbolt to stabilize roof wedges in underground opening is studied using the
joint relaxation method to consider the rock wedge in deformable rock conditions.
Parametric studies are carried out to investigate the key parameters that influence
the effects of rockbolts when they are used to stabilize the rock wedge. The results
show that 2D-DDA could be used to find the critical horizontal pressure to sustain
the rock wedges. A cross section in storage gallery in Jurong Rock Caverns (JRC),
Singapore, is conducted to show the application of the 2D-DDA models for
rockbolt design to support roof wedges. The stress versus vertical displacement
curves showed clearly the effect of rockbolt on sustaining the movement of rock
wedge. The way to create characteristic diagram could be used by the site engineers
to evaluate the rockbolting design to sustain the possible roof wedges.

A support design method for horseshoe-shaped rock caverns is proposed with

considerations of the progressive damage of the rock mass using the 2D finite
element method (FEM) and the artificial neural network (ANN). The performances
of the rock cavern during excavation are investigated based on the convergence-
confinement method (CCM). The ANN models are built using the numerical results
to find the complex relationships among the rock mass condition, the sequential
excavation parameters and the cavern performances. An evaluation chart is
proposed by integrating the ANN models into the EXCEL software. The proposed
evaluation chart provides an effective method to evaluate the support safety, the
functions of the patterned rockbolt and the optimization of subdivisions of the
excavation cross section. More advanced function is still required to present the
relationship between the parameters of sequential excavation method (SEM) and the
cavern performance.

X
 List of Publications

LIST OF PUBLICATIONS

1. Nie, W., Zhao, Z.Y., Guo, W., Shang, J. and Wu, C.Z. Bond-slip modeling of a
CMC rockbolt element using 2D-DDA method. Tunnelling and Underground
Space Technology, 2019, 85: 340-353.

2. Nie, W., Zhao, Z.Y., Ma, S.Q. and Guo, W. Efforts of joints on the reinforced rock
units of fully-grouted rockbolts. Tunnelling and Underground Space Technology,
2018, 71: 15-26.

3. Nie, W., Zhao, Z.Y., Goh, A.T.C., Song, M.K., Guo, W. and Zhu, X. Performance
based support design for horseshoe-shaped rock caverns using 2D numerical
analysis. Engineering Geology, 2018, 245: 266-279

4. Ma, S.Q., Zhao, Z.Y., Nie, W.* and Zhu, X. An analytical model for fully grouted
rockbolts with consideration of the pre- and post-yielding behaviour. Rock
Mechanics and Rock Engineering, 2017, 50(11): 3019-3028

5. Nie, W., Zhao, Z.Y. and Ma, S.Q. Numerical evaluation of rockbolt reinforcement
unit in jointed rock mass by DDA method. In Reşat Ulusay, Ömer Aydan, Hasan
Gerçek, Ali Mehmet Hindistan and Ergün Tuncay(Eds), Rock Mechanics and
Rock Engineering: From the past to the future, July 2016, Taylor & Francis Group,
pp. 487-492. (Conference paper)

XI
 List of Figures

LIST OF FIGURES

Figure 2-1 Illustration of three types of the rockbolt elements (modified after Windsor
and Thompson, 1993) ....................................................................................... 6
Figure 2-2 Sketch of the cross-section of a CMC rockbolt element .............................. 9
Figure 2-3 Diametrical slice of a grouted rockbolt (after Farmer, 1975) .................... 11
Figure 2-4 Axial load and bond stress distributions along the rockbolt based on
illustrated bond models ................................................................................... 13
Figure 2-5 Examples of the bond load versus slip displacement relationships (after
Mertoğlu et al., 2016) ..................................................................................... 14
Figure 2-6 Illustration of the bond slip model ............................................................. 15
Figure 2-7 Sketch of the mechanical system at joint (Pellet and Egger, 1996) ........... 15
Figure 2-8 Load-deformation results by Stillborg (after Li, 2010) .............................. 16
Figure 2-9 Sketches of the stress/strain distributions along rockbolts when subjected
to pullout loads (Li et al., 2014) ..................................................................... 17
Figure 2-10 Stresses along the rockbolt at joint opening (Li and Stillborg, 1999) ...... 17
Figure 2-11 Shear load versus displacement of various rockbolts (Li et al., 2014) .... 18
Figure 2-12 Comparison of shear load and shear displacement relationship with
different bolt orientations ............................................................................... 19
Figure 2-13 Forces on the rockbolt at the shearing joint (Pellet and Egger, 1996) ..... 19
Figure 2-14 Strain distributions along the rebars under double shear tests (Forbes et al.,
2017) ............................................................................................................... 19
Figure 2-15 Sketches to illustrate the limited stability analysis (after Lang and
Bischoff, 1982) ............................................................................................... 21
Figure 2-16 Illustration of the reinforced rock unit (after Lang and Bischoff, 1983) . 23
Figure 2-17 Tom Lang’s explanation of the systematic rockbolts (Hoek, 2007) ........ 24
Figure 2-18 Conceptual natural arch in underground excavation (Hoek, 2007) ......... 24
Figure 2-19 The Q-chart (NGI, 2015).......................................................................... 27
Figure 2-20 Material models of cable element adopted in FLAC (Itasca Consulting
Group, 2012) ................................................................................................... 31

XII
 List of Figures

Figure 2-21 A spring-slider system to represent the bond behaviour of the grout
annulus (Itasca Consulting Group, 2012) ....................................................... 32
Figure 2-22 Springs representing the assumed reinforcement due to shear movement
(Itasca Consulting Group, Inc. 2011) ............................................................. 34
Figure 2-23 Constitutive model of the local reinforced systems (Itasca Consulting
Group, Inc. 2011) ............................................................................................ 35
Figure 2-24 Ground reaction curve (Vlachopoulos and Diederichs, 2009) ................. 38
Figure 2-25 Illustration of the influence of support stiffness and timing of installation
on support performance (modified after Brady and Brown, 2006) ................ 43
Figure 2-26 Illustration of longitudinal displacement profile (LDP) (Vlachopoulos and
Diederichs, 2009) ............................................................................................ 44
Figure 2-27 Schematic representation of LDP, GRC and SCC (Lüet al., 2012) ........ 46
Figure 2-28 Schematic illustration of global stiffness matrix for a 3-block DDA model
(Bao, 2010) ..................................................................................................... 48
Figure 2-29 Bolting connection in original DDA by Shi (1998) ................................. 52
Figure 2-30 Framework of the multi-time step calculation adopted in DDA .............. 57
Figure 3-1 Sketch of the rib profile of a CMC rockbolt element.................................. 60
Figure 3-2 Sketch of a CMC element for (a) one element, (b) axisymmetric view, and
(c) a trilinear bond-slip model to simulate the bond stress versus slip
displacement curve ......................................................................................... 62
Figure 3-3 Contact at an artificial joint in a 2D-DDA model ...................................... 63
Figure 3-4 Numerical model of the CMC rockbolt element in 2D-DDA (unit: mm).. 64
Figure 3-5 Simulation of the UC test of the motor specimen to determine the
properties of motor (a) numerical model (unit: mm), (b) vertical stress
contours of the specimen at failure, and (c) stress-strain curve and the relative
percentage of broken bonds ............................................................................ 66
Figure 3-6 Simulation of the direct shear test using 2D-DDA (a) schematic diagram of
the lab test (unit in: mm), (b) numerical model (unit in: mm) and (c)
comparisons of the bond stress versus slip displacement curves ................... 68
Figure 3-7 Simulated bond stress-slip displacement curve of the CMC element with
rib face angle α = 45°and rib spacing s = 17.8 mm under p = 2 MPa ........... 70
Figure 3-8 Crack propagations in mortar of the CMC rockbolt element with rib face
angle α = 45° and rib spacing s = 17.8 mm under p = 2 MPa at the slip

XIII
 List of Figures

displacements of (a) Δδ = 0.06 mm, (b) Δδ = 0.3 mm, (c) Δδ = 0.7 mm and (d)
Δδ = 0.9 mm .................................................................................................... 70
Figure 3-9 Effects of the rib face angles on the bond stress-slip displacement curves
when (a) α = 30°, (b) α = 60°and (c) α = 90°................................................ 72
Figure 3-10 Effects of the rib face angles on the crack propagations when (a) α = 30°,
Δδ = 0.3 mm, (b) α = 30°, Δδ = 0.8 mm, (c) α = 60°, Δδ = 0.3 mm, (d) α = 60°,
Δδ = 0.8 mm, (e) α = 90°, Δδ = 0.3 mm and (f) α = 90°, Δδ = 0.8 mm .......... 73
Figure 3-11 Effects of the rib spacings on the bond stress versus slip displacement
curves when (a)  = 30°, (b)  = 45°, (c)  = 60°and (d)  = 90°................ 76
Figure 3-12 Effects of the rib spacing s on the fracturing in mortar block at slip
displacement Δδ ≈ 1.0 mm when (a) α = 30°, (b) α = 45°, (c) α = 60°and (d)
α = 90°............................................................................................................ 77
Figure 3-13 Effects of the normal stress p in different rib faces angles (a) α = 30°, (b)
α = 45°, (c) α = 60°and (d) α = 90°............................................................... 79
Figure 4-1 Sketch of the components of the proposed rockbolt model (Nie et al.,
2014a)......................................................................................................... 83
Figure 4-2 Rock block and rockbolt model in 2D-DDA program (Nie et al, 2014a) .. 84
Figure 4-3 Transformation from 1D bar element to 2D space..................................... 86
Figure 4-4 Stress-strain curves for the elastic, linear hardening rockbolt material
(modified after Owen and Hinton, 1980) ................................................... 87
Figure 4-5 Trilinear model used to present bonding behaviour ................................... 88
Figure 4-6 Numerical models for four types of rockbolts (Nie et al., 2014a) ............. 88
Figure 4-7 Simplified rockbolt element subjected shear movement ........................... 89
Figure 4-8 A free body diagram of a rockbolt installed through a joint ...................... 91
Figure 4-9 Loading state judgement using shear stress versus axial load curves ........ 92
Figure 4-10 An example of a rockbolt element in 2D-DDA ....................................... 93
Figure 4-11 Flowchart of the calculation of rockbolt forces ....................................... 95
Figure 4-12 Revised flowchart of 2D-DDA codes by integrating with rockbolt models
.................................................................................................................... 96
Figure 4-13 The parameters of rockbolt model used in 2D-DDA (a) Dimensions and
boundaries of the model, (b) Bond-slip model and (c) Axial model of the
rebar ........................................................................................................... 97
Figure 4-14 Comparisons of the axial load versus displacement curves from present
study and model test conducted by Rong et al.(2004) ............................... 98

XIV
 List of Figures

Figure 4-15 Comparisons of (a) the simulated axial force distribution with the results
of model test conducted by Rong et al. (2004) and (b) the simulated bond
stress distribution with the analytical solutions by Ma et al. (2010) .......... 99
Figure 4-16 DDA model setting of the shearing tests (a) DDA model, (b) bond–slip
model and (c) material model .................................................................. 100
Figure 4-17 Failures of the rockbolts in numerical simulations (a) Pull-out test and
shear tests at angles of (b) β = 90°, (c) β = 120°and (d) β = 140°........... 101
Figure 4-18 Comparisons of load-displacement curves between experiments and
numerical simulations .............................................................................. 101
Figure 4-19 Simulated shear behaviour of rockbolt elements at joints ..................... 103
Figure 4-20 Sketches of the pull-out test and the various rockbolt models ............... 104
Figure 4-21 Comparisons of load-deformation curves between experiments and
numerical simulations of pullout tests ..................................................... 105
Figure 4-22 Stress distributions along the length of rockbolt models ....................... 107
Figure 4-23 2D-DDA model of a rockbolt pullout/shear test ..................................... 108
Figure 4-24 Comparisons of the simulated load-displacement curves ...................... 109
Figure 4-25 Simplified bond-slip models .................................................................. 110
Figure 4-26 (a) Nodal bond force and (b) nodal axial force along the rockbolts at the
pull-out loading of 108 kN ....................................................................... 112
Figure 4-27 (a) Nodal bond force and (b) nodal axial force along the rockbolts at the
extreme loading ........................................................................................ 113
Figure 4-28 Simulated nodal axial loads along the rockbolt models when the tensile
loads at joint are (a) 60% of Fy and (b) Fult............................................. 115
Figure 4-29 Simulated nodal bond forces along the rockbolt models when the tensile
loads at joint are (a) 60% of Fy and (b) Fult............................................. 116
Figure 4-30 Simulated nodal axial loads in the shear tests of rockbolt with shear angle
β = 90° when the loads at joint are (a) 60% of Fy and (b) Fult ................. 118
Figure 4-31 Axial loads variations in case of deformable rock blocks ...................... 118
Figure 5-1 Sketches of RRUs of rockbolts in homogeneous rock mass (after Hobst
and Zajic, 1983) ..................................................................................... 122
Figure 5-2 DDA models of pulling rockbolt.............................................................. 124
Figure 5-3 Effects of the rock joint orientations on the pullout resistance of rockbolt
(for Er = 0.3 GPa, φ = 20°, L = 1.0 m and α = 0˚, 30˚, 45˚, 60˚ and 90˚,
respectively) ........................................................................................... 125

XV
 List of Figures

Figure 5-4 Contours of normalized major principal stresses σ1/σ0 in rock mass for Ua
= 10 mm, Er = 0.3 GPa, φ = 20°, L = 1.0 m and (a) α = 0°, (b) α = 30°, (c)
α = 45°, (d) α = 60°, and (e) α = 90°...................................................... 127
Figure 5-5 Deformation at the rock surface during the pullout process of rockbolts for
Er = 0.3 GPa, φ = 20°, L = 1.0 m and (a) α = 0°, (b) α = 30°, (c) α = 45°,
(d) α = 60°, and (e) α = 90°................................................................... 128
Figure 5-6 Parametric studies on the key influence parameters ................................ 129
Figure 5-7 Axial loads vs. displacement curves for the rockbolt in jointed rock mass
with dip angle of 0°............................................................................... 133
Figure 5-8 Axial loads of the middle rockbolt in cases of primary joints with dip
angles of (a) 30°, (b) 45°, (c) 60°and (d) 90°....................................... 134
Figure 5-9 Percentages of the axial loads of middle rockbolts .................................. 134
Figure 5-10 Overlap ratio of the RRUs...................................................................... 136
Figure 5-11 Illustration of the relationship between RRU and axial load ................. 136
Figure 5-12 Comparison of the RRU areas with the maximum axial loads of rockbolt
at UA = 10 mm ....................................................................................... 137
Figure 5-13 DDA models for pressure arches in layered rock with inclined angles of
(a) α = 0ºand (b) α = 90º....................................................................... 138
Figure 5-14 Effect of the in-situ stress ratios on (a) the vertical displacement of M1,
and (b) the contour maps of the transverse stresses ............................... 139
Figure 5-15 Effect of the number of rockbolts on (a) the vertical displacement of M1
and (b) contour maps of the transverse stresses..................................... 140
Figure 5-16 Distributions of stresses and displacements in rock blocks along vertical
cross sections (a) x-stress, (b) y-stress and (c) y-displacement .............. 141
Figure 5-17 Effect of the stress ratios on the vertical displacement of M2 in verticlly
jointed rock mass (a) Displacement histories of measuring point M2 and
(b) Contour maps of the stress ratio k = 6.0, 4.0 and 2.0 ....................... 142
Figure 5-18 Vertical displacement of M2 in cases of artificial roof arches formed in
vertically jointed rock mass (a) Displacement histories of measuring
point M2; (b) Contour maps of the transverse stresses in cases of 5
rockbolts, 1 rockbolt and 2 rockbolts are installed ................................ 143
Figure 5-19 Changes in the stresses and vertical displacements in various rockbolts
reinforced conditions (a) x-stress, (b) y-stress and (c) y-displacement .. 144

XVI
 List of Figures

Figure 5-20 Axial load of the rockbolt and its sketched RRU in the horizontally
laminated rock blocks (a) two rockbolts and (b) five rockbolts ............ 145
Figure 5-21 Axial load of the rockbolt and its sketched RRU in the vertically
laminated rock blocks (a) two rockbolts and (b) five rockbolts ............ 145
Figure 6-1 Examples of the definitions of rockbolting design parameters in wedge
problems (a) sketch of unstable wedges formed in an underground opening,
(b) changes of embedded lengths and (c) changes of response modes of
rockbolts ................................................................................................... 148
Figure 6-2 2D-DDA model for a symmetric wedge with straight wedge free face (a)
geometry model, (b) contacts at the wedge block and (c) parameters of
rockbolts ................................................................................................... 151
Figure 6-3 Free-body diagrams of a 2D roof wedge in (a) rigid joint condition and (b)
relaxed joint condition ............................................................................. 152
Figure 6-4 Comparisons between the simulated critical horizontal pressures and the
analytical solutions at different friction angles ........................................ 154
Figure 6-5 Comparisons of the vertical displacements versus time curves obtained
from numerical analysis with different contact stiffness ......................... 155
Figure 6-6 Comparisons of the normal contact forces versus time step curves when
friction angle φ = 40°(a) the vertical displacements and the normal contact
forces in the cases of (b) kn = 0.1E, (c) kn = 1.0E and (d) kn = 10E ......... 156
Figure 6-7 Effects of the horizontal pressure p on the reinforced efforts of rockbolts to
stabilize the 2D wedge ............................................................................. 157
Figure 6-8 Effects of the bolt spacing sb on the reinforced efforts of rockbolts to
stabilize the 2D wedge ............................................................................. 157
Figure 6-9 Effects of the installation angle θ on the reinforced efforts of rockbolts to
stabilize the 2D wedge ............................................................................. 158
Figure 6-10 Site information after top heading (a) geological and (b)support
conditions ................................................................................................. 159
Figure 6-11 Method to estimate the removable key blocks in the cavern surface using
key block theory ....................................................................................... 160
Figure 6-12 An approximated blocky model of (a) the cavern cross-section, (b)
rockbolt arrangement of Case A and (c) rockbolt arrangement of Case B
.................................................................................................................. 161

XVII
 List of Figures

Figure 6-13 Simulated reactions of (a) vertical displacement and (b) the variations of
stress versus displacement at block A under different joint relaxation
situations .................................................................................................. 162
Figure 6-14 The total support force acting on block A in Cases 5 to 9 ..................... 164
Figure 6-15 Simulated axial forces along the rockbolts through block A ................. 165
Figure 6-16 Simulated reactions of block A with varies rockbolting time ................ 165
Figure 7-1 Influence zones due to excavation in rock (modified after Siren et al., 2015)
.................................................................................................................. 168
Figure 7-2 Characteristic curves for excavation in rock (modified after Lüet al., 2012)
.................................................................................................................. 170
Figure 7-3 Numerical models using RS2 program for (a) horseshoe cavern, (b) circular
tunnel, and (c) horseshoe cavern under subdivision ................................ 173
Figure 7-4 Illustration of the progressive core replacement method (span = 20 m
under fair rock condition) ........................................................................ 175
Figure 7-5 Effects of Q-value on the cavern performance (a) LDP, (b) damage depths
at the roof from the excavation face, and (c) GRC .................................. 178
Figure 7-6 Effects of the SEM parameters on cavern performance (a) the round length,
(b) the cavern size, and (c) the cavern size for 3-section excavation in case
of Q = 1.0 ................................................................................................. 180
Figure 7-7 Architecture of the ANN models ............................................................. 181
Figure 7-8 Performances of the ANN models to train the numerical results to predict
(a) the damage depth and (b) the normal stress ....................................... 182
Figure 7-9 Comparisons between the targets and the predicted values of (a) the
damage depth and (b) the normal stress ................................................... 182
Figure 7-10 Evaluation chart for the preliminary support design .............................. 184
Figure 7-11 Calculated GRCs and SCCs using evaluation chart to assess the support
designs proposed by Sari & Pasamehmetoglu (2004) for (a) Section 2, (b)
Section 7, and (c) Section 5. .................................................................... 188
Figure 7-12 GRCs and SCCs to evaluate the revised support designs for (a) Section 2,
(b) Section 7, and (c) Section 5. ............................................................... 189
Figure 7-13 Comparisons between the results from numerical analysis and evaluation
chart in terms of (a) displacement and (b) damage depth of EDZ from the
excavation surface .................................................................................... 190

XVIII
 List of Tables

LIST OF TABLES

Table 2-1 Features of the reinforcement (after Nie et al., 2014a).................................. 7

Table 2-2 Parameters to control the bond strength of fully grouted rockbolt/cables... 10
Table 2-3 Applications of rockbolt (modified after Hoek et al., 2000) ....................... 20
Table 2-4 Empirical design recommendations (after Stillborg (1986) and Li (2017b))
..................................................................................................................... 25
Table 2-5 Examples of 10 m span rock tunnels in accordance with the RMR system
(after Hoek, 2007) ....................................................................................... 26
Table 2-6 Support category illustrated based on Q-chart ............................................ 28
Table 2-7 Examples of the numerical models of a rockbolt ........................................ 36
Table 2-8 Useful methods to represent 3D problems using 2D models (modified after
Karakus, 2007) ............................................................................................ 37
Table 2-9 Analytical solutions of typical axisymmetric circular tunnel problem
(modified after Brady and Brown, 2006; Carranza-Torres and Fairhurst,
2000) ........................................................................................................... 39
Table 2-10 Analytical solution for the circular tunnel excavation without rock
reinforcement (after Bobet and Einstein, 2011) .......................................... 40
Table 2-11 Analytical solution for the circular tunnel excavation with DMFC element
reinforcement (after Bobet and Einstein, 2011) .......................................... 41
Table 2-12 Analytical solution for the circular tunnel excavation with CMC/CFC
element reinforcement (after Bobet and Einstein, 2011) ............................ 42
Table 2-13 Limitations and caveats when applying CCM in engineering................... 46
Table 2-14 Contact status in DDA ............................................................................... 56
Table 3-1 Parameter setting of 2D-DDA calculation .................................................. 65
Table 3-2 Physical parameter setting of the rockbolt element model .......................... 67
Table 3-3 Parameters defining the trilinear bond-slip models under p = 2 MPa ......... 73

XIX
 List of Tables

Table 3-4 Effects of the normal stress on crack propagations in the mortar of CMC
rockbolt element with rib spacing s = 17.8 mm and different rib face angles
at the slip distances ∆δ = 1.0 mm ............................................................... 80
Table 3-5 Results of case studies ................................................................................. 81
Table 4-1 Rockbolt model properties .......................................................................... 94
Table 4-2 Parameters used in 2D-DDA to simulate the pullout test............................ 98
Table 4-3 Parameters setting in the simulation of shear tests .................................... 100
Table 4-4 Comparisons between numerical simulations and the experiment results 102
Table 4-5 Parameter settings in pullout models ......................................................... 104
Table 4-6 Parameters of three bond-slip models and the fitted exponential models . 114
Table 5-1 The maximum axial stress of the rockbolt installed in different rock mass
(modified after Hobst and Zajìc, 1983) ...................................................... 122
Table 5-2 Parameters used in 2D-DDA to simulate the pullout test ......................... 123
Table 5-3 Displacement contours in rock blocks during rockbolt pullout process ... 126
Table 5-4 Displacement contour of rock blocks when UA = 10 mm ......................... 131
Table 5-5 Deformation of rock surface...................................................................... 132
Table 6-1 Parameters setting in the numerical simulation ......................................... 152
Table 6-2 Roof wedge stability based on the analytical solutions ............................. 154
Table 6-3 Characteristic of the discontinuities according to the site geological
mapping after top heading........................................................................ 159
Table 6-4 Contour maps of the horizontal stress σxx at different time step ................ 163
Table 7-1 Support capacities defined by Hoek (2007) .............................................. 172
Table 7-2 Geotechnical parameters for numerical models ........................................ 174
Table 7-3 Magnitudes of Pi and Oj for normalization based on numerical results .... 183
Table 7-4 Predictions of the support safety function g2(x) for shotcrete linings with
thickness tc = 0.1 m .................................................................................. 186
Table 7-5 Predictions of the rockbolt length criterion function g3(x) for rockbolt
support...................................................................................................... 186
Table 7-6 Support design parameters (B – bolting, SL – shotcrete lining) ............... 189
Table 7-7 Displacement contours around the tunnel with/without supports ............. 191

XX
 List of Symbols

LIST OF SYMBOLS

Symbols Description

rb : Radius of a rockbolt

Db : Diameter of a rockbolt

Ab : Cross-section area of rockbolt

DH : Diameter of borehole

Eb : Elastic modulus of a rockbolt

Lb : Length of a rockbolt

le : Assumed hinge length

l0 : Embedded length of a rockbolt

sb : In-plane spacing of pattern rockbolts

Fs : Unit bond force

τ : Bond stress

δ : Slip displacement at bond

τmax : Maximum bond strength at the rock and rockbolt interface

τres : Residual bond strength at the rock and rockbolt interface

k : Bond stiffness of a bond-slip model

Ks : Shear stiffness of a rockbolt element

σy : Yield strength of a rockbolt element

εext : Extreme plastic strain of a rockbolt element

EG : Elastic modulus of mortar

Er : Elastic modulus of rock

XXI
 List of Symbols

φ : Friction angle of a joint

c : Cohesion of a joint

σt : Tensile strength of a joint

ρ : Density of a block

γ : Unit weight of a block

v : Poisson’s ratio of a block

kn : Normal stiffness of joint/contact

ks : Shear stiffness of joint/contact

Rn : Normal contact force at a joint

Rs : Shear contact force at a joint

g1 : Pre-set maximum time increment in 2D-DDA calculation

g2 : Maximum allowed displacement ratio in 2D-DDA calculation

B : Cavern span

Dp : Damage depth of excavation damaged zone (EDZ)

sf : Safety factor

pi : Support pressure

ui : Displacement at the excavation surface

pcr : Critical horizontal confining pressure

XXII
 Chapter 1 Introduction

INTRODUCTION

1.1 Background

Underground space has been increasingly exploited in recent years. The potential for
instability in the rock mass surrounding the underground excavation is a threat to the
lives and safety of construction workers. To mitigate safety hazards, it is necessary to
understand the causes of instability and design supports to eliminate or minimise the
risks (Hoek et al., 2000).

The support is the procedures and materials used to improve the stability and maintain
the load-carrying capability of rock mass surrounding the underground excavation
(Brandy and Brown, 2006). Based on the application of reactive forces to the
surrounding rock mass, the supports could be divided into support and reinforcement
(Windsor and Thompson, 1993). Support (i.e., timber, shotcrete, steel frame, etc.) will
apply reactive forces to the excavation surface. Reinforcement is targeting to improve
the overall rock mass properties. Rockbolting is one of the important techniques to
provide reinforcement in the rock mass.

The rockbolt has been the primary reinforcement structure in underground tunnelling
and mining for quite a long period. It is usually simple to apply and can be modified to
reinforce the rock mass during excavation. To mobilize the reinforcement efforts and
improve the design of the rockbolt, it is necessary to have a good understanding of its
reinforcement mechanism. However, the reinforcement mechanism of rockbolt system
has not been fully understood as the rockbolt system design remains an art rather than a
science since most designs are based on historical experiences (Hoek et al., 2000). To
ensure the stability of underground excavations, the rockbolt system is usually
overdesigned, leading to extra time and cost.

1
Chapter 1 Introduction

The load transfer capacity between rockbolt and rock is one of the most important
factors controlling the reinforcement efforts of the rockbolt. When the rock moves
towards the excavation surface, the rockbolt elongates. Tension is created in the rockbolt
and transferred to the rock as bond stress, thus helps the rock to support itself by
increasing confinement (Bobet and Einstein, 2011). To increase the efficiency of the
load transfer capacity, many types of rockbolt elements have been developed. The bond-
slip models are introduced to describe the relationship between the bond stress and the
slip at the rock and rockbolt interface for different types of rockbolt elements
(Benmokrane et al., 1995; Moosavi et al., 2005; Ren et al., 2010; BlancoMartín et al.,
2011; Deb and Das, 2011; Ma et al., 2013; Ma et al., 2014; Zhandarov and Mäder, 2016).
However, there is a lack of studies on the application of the bond-slip models to assess
the reinforcement mechanism of the rockbolt.

The reinforcement mechanism of rockbolt is complex in field due to the existing of

joints, faults and other discontinuities in rock mass. Three theories have been proposed
in recent years to study the reinforcement mechanism of rockbolt in complex ground
conditions (Lang and Bischoff, 1982; Hoek et al, 2000; Brandy and Brown, 2006). For
massive rock blocks, the limited rock block stability analysis is used to analyse the
stability of the rock wedges. For stratified rock mass, the rockbolts are designed to
transfer the weight from weak strata to the strong strata. In jointed rock mass, the loose
zone could be stabilized by the systematic rockbolts in the formation of artificial arch.
Numerical studies have been carried out to simulate the interaction between rockbolt
and rock mass (Sari and Pasmehmetoglu, 2004; Bobet and Einstein, 2011; Deb and Das,
2014; Ma et al., 2016; Nie et al., 2014a; Feng et al., 2018). Among these studies, the
discontinuous element method becomes a primary choice when dealing with the study
of reinforcement mechanism of rockbolt as it can simulate the the mechanical behaviour
of an assembled rock blocks. Discontinuous Deformation Analysis (DDA), introduced
by Shi in 1980s (Shi, 1988), is one of the discontinuum methods which could conduct
the numerical simulation on rock engineering problems, such as the ground reaction
after excavation. However, the rockbolt models have not been developed in the original
2D-DDA program, which makes it hard to be applied in support design for underground
excavation projects.

Although the mentioned researches have made some achievements in the understanding
of the reinforcement mechanism of rockbolt, there are still some limitations to assess

2
 Chapter 1 Introduction

the interactions between rockbolt and rock mass. Attempts are made in this thesis to
investigate the reinforcement mechanism and the design principles of the rockbolt.

1.2 Key objectives

The thesis aims to investigate the reinforcement mechanism of the rockbolt and improve
its design efficiency for underground excavation. To achieve the objectives, the study
has focus on the following items:
(1) To study the bonding behavior of the rockbolt element by considering the
rockbolt rib configurations;
(2) To develop a rockbolt model in 2D-DDA based on the load transfer mechanism;
(3) To investigate the reinforced rock units of the rockbolt and rockbolt system in
jointed rock mass;
(4) To construct numerical models to evaluate the stability of rockbolt reinforce rock
wedge in the roof of underground opening, and
(5) To propose the rockbolt design method for horse-shoe shaped rock caverns
based on the cavern performance.

1.3 Outline of thesis

This thesis investigates the reinforcement mechanism of the rockbolt for underground
excavation using 2D-DDA method. It is divided into eight chapters.

In Chapter 2, literature reviews are conducted to understand the rockbolt theory and the
load transfer mechanism of the rockbolt. In addition, the basic 2D-DDA method is
introduced.

In Chapter 3, numerical simulations are carried out to study the load transfer between
rock and rockbolt element. The influences of the rib profiles on the bond stress versus
slip displacement curves are investigated, followed by the discussion of the trilinear
bond model along the rock and rockbolt interface.

In Chapter 4, the rockbolt model in 2D-DDA method is improved by integrating the

non-linear axial behaviour, the shear behaviour at joint and the bond behaviour at the
rock and rockbolt interface. The models are verified by experimental results.

3
Chapter 1 Introduction

In Chapter 5, the pullout tests are simulated using the developed rockbolt model to
investigate the reinforced rock unit of rockbolt in jointed rock mass. Meanwhile, the
interactions between rock and rockbolt are evaluated based on the reinforce rock unit of
single rockbolt and grouped rockbolts.

In Chapter 6, numerical simulations are carried out using 2D-DDA to investigate the
stability of rockbolt reinforced rock wedge in the roof. The effects of the spacing and
the installation angle of rockbolt on the reinforcement efforts are presented with
consideration of the changes of stress state in the surrounding rock blocks.

In Chapter 7, a rockbolt design method is proposed using artificial neural network

models which are built based on the numerical evaluations of ground reaction curves.

In Chapter 8, conclusions are drawn based on the current studies. Future studies on the
support design/management for underground excavation are presented.

4
 Chapter 2 Literature review

LITERATURE REVIEW

The research works on the rockbolt reinforcement mechanism are reviewed to

understand the rockbolt behaviour and the interaction between rockbolt and rock mass.
After a brief introduction of rockbolt reinforcement system, the load transfer in
rock/rockbolt is explained. The reinforcement effects are then discussed with aids of
ground-support diagrams using Convergence-Confinement method (CCM). The Two-
dimensional Discontinuous Deformation Analysis (2D-DDA) is introduced for further
development purpose.

2.1 Rockbolt

The rockbolt is a reinforcing element, normally made of solid or tube formed steel and
installed un-tensioned or tensioned into the rock mass (Stillborg, 1986). The installation
of rockbolts in rock mass is to reinforce and improve the inherent strength of the rock
mass, so that the rock mass becomes self-supported. In general, when the rock moves
towards the excavation surface, the rockbolt elongates. Tension is created in the rockbolt
and transferred to the surrounding rock mass as compression, thus helps the rock to
support itself by increasing confinement (Bobet and Einstein, 2011). With rockbolts,
local reinforcement is used to secure and prevent detachments of individual blocks.
General or systematic reinforcement is required to control the deformation occurring in
the entire tunnels. Since the rockbolt is usually simple to apply and can be modified
during construction, it becomes a key component in technologies used for designing
tunnels, such as New Austrian Tunneling Method (NATM).

2.1.1. Types of rockbolts

Many types of devices have been developed to be used as reinforcement elements. In

general, these devices are solid bars or tube elements and installed within boreholes

5
Chapter 2 Literature review

drilled into the rock. Based on the source of reaction forces, the rockbolt could be
classified as three categories (Windsor and Thompson, 1993).

(1) Continuously Mechanically Coupled (CMC) element has continuously load

transfer through the length. The source of reaction forces relies on the securing agent
(such as grout or resin) which fills the annulus between the element and the borehole
wall. Figure 2-1(a) presents an example of the cross-section of a fully grouted rebar.
(2) Continuously Frictionally Coupled (CFC) element also has continuously
load transfer through the length. However, as the element is in contact with rock
directly, the source of reaction forces relies on the friction between the element and the
borehole, as well as the radial pressure set up. Figure 2-1(b) presents an example of the
cross-section of a split set.
(3) Discretely Mechanically or Frictionally Coupled (DMFC) element provides
either mechanical or frictional reaction forces over a short interval of its total element
length. The interval is known as anchorage length which provides either mechanical or
frictional load transfer over this relatively short interval. This anchorage must be
sufficiently strong to sustain the full material strength of the reinforcing element. Figure
2-1(c) presents an example of the cross-section of the D-bolt (Li, 2010).

(a) CMC element

(i.e., fully grouted rebar)

(b) CFC element

(i.e., the split set)

(c) DFMC element

(i.e., the D-bolt○
R (Li,

2010))

Figure 2-1 Illustration of three types of the rockbolt elements (modified after Windsor
and Thompson, 1993)

Some examples of the reinforcement devices with consideration of their advantages and
disadvantages are summarized in Table 2-1.

6
 Chapter 2 Literature review

Table 2-1 Features of the reinforcement (Nie et al., 2014a)

Types DMFC CMC CFC

Examples Expansion shell or slot and Fully grouted/resin rebar Split set/tube
wedge Fully grouted/resin threaded Expanded tube
Mechanical anchors bar (Swellex○ R , Omega○
R,

The D-bolt○R Fully grouted/resin cable bolt etc.)

Cement or resin is used as
The frictional
Two ends of the rockbolt grouting agents to integrate
resistance to sliding is
are mechanically bolt with the rock via ribs on
generated by a radial
Load transfer

anchored in borehole. the bolt surface. The load

force against the
Once the fracture opens, transfer through the bond
borehole wall over the
the bolt shank between between the element and the
length of the bolt.
two ends is tensioned rock cause tension in the
Mechanical interlock
according with the bolt. The bearing capacity is
between the bolt and
anchoring capacity of two provided primarily by the
the rock then prevents
anchoring points. shear resistance at the
the bolt from sliding.
interface.
Rock type
Preferred

Moderately hard to hard

All rock conditions All rock conditions
rock

Provide immediate
The most commonly used
support action after
rockbolt; Rapid and simple
installation;
Predominantly used in installation;
Serve as permanent
Advantages

temporary support systems Provide immediate

reinforcement by post-
grouting; Provide high bolt loads in support action after
various rock conditions; installation;
High bolt loads can be
achieved; Provide immediate support Can be used in a
if a “fast-setting” resin is variety of ground
A versatile system for conditions.
used for bottom anchoring of
rock reinforcement in
the bar.
hard rock conditions.
Difficult to install
reliably; Difficult to check the resin Relatively expensive;
Must be monitored and cartridges/cement; Corrosion protection
Disadvantages

checked for proper Relatively expensive; required if used in

tensioning; Use of standard cement in long term
Lose bearing capacity as the grout requires several installations;
a result of blast days curing before the bolt Borehole diameter is
vibrations; can take load; crucial in the
Not recommended for Affected by corrosion. prevention of failure.
very hard rock.
The void between the The bolt needs to be
Notes

bolt and the borehole protected against

need to be post grouted. corrosion.

7
Chapter 2 Literature review

Another classification of rockbolt based on the view of loading and deformation

performance is also widely used in design which classifies the rockbolts as stiff, ductile
and energy-absorbing rockbolt (Li et al., 2014). A stiff rockbolt has high load-carrying
capacity, but a small amount of displacement prior to failure, such as the fully
encapsulated rebar bolts. A ductile rockbolt can tolerate a large deformation, but
relatively low load-carrying capacity, such as the split set. An energy-absorbing rockbolt
can carry the load equal or close to the strength of the bolt material and deform largely,
so that it can absorb a large amount of energy prior to failure, for example, the D-bolt
(Li, 2010) and the CRLD bolts (He et al., 2014).

The rockbolt can also be classified based on the types of grout and the auxiliary fittings
it used. Grouting materials are broadly classified as two materials, i.e., cementations and
resinous materials (Windsor and Thompson, 1993). Cement grout is suited to the longer
reinforcing elements due to its availability, ease of mixing and placement. Some
additives can be added to aid the placement and improve the quality of cement grout
and the set time. Resinous material is generally confined to the shorter elements due to
the higher costs of the resin materials and placement difficulties in longer holes.
Auxiliary fittings are either attached to the reinforcing element at the rock mass
boundary (external fittings) or used to modify the internal response of continuously
coupled devices (internal fittings) (Bawden, 2011). External fittings can provide varying
degrees of surface restraint to the rock mass surface. Internal fittings are used
predominantly with continuous mechanically coupled devices and comprise intermittent
anchors and decoupling sleeves. The combination auxiliary fittings modify the load
transfer mechanism between the rock and the reinforcing element.

2.1.2. Load transfer in rockbolt

The load transfer concept is fundamental to understand the reinforcement mechanisms

of different rockbolt elements. This concept is composed of three basic mechanisms
(Windsor and Thompson, 1993),
(1) Rock movement with requires load transfer from the unstable rock to the
reinforcing element;
(2) Transfer load via the reinforcing element from the unstable surface region to a
stable interior region, and
(3) Transfer the load of reinforcing element to the stable rock mass.

8
 Chapter 2 Literature review

The load transfer capacity of the rockbolt is controlled by the components of bonds at
the rock and rockbolt, such as the surrounding rock mass, the mortar, the profile
configuration and the boundary conditions of the rockbolt. For different kinds of
rockbolt element, the bonding behaviours are different. The bonding behaviour of the
CMC element relies on a securing agent that fills the annulus between the element and
the borehole wall. The dominant reinforcement is mainly created as a result of friction
or mechanical interlocking between steel and grout (Zhandarov and Mäder, 2004;
Moosavi et al., 2005; Singer, 1990). The bonding of the CFC element depends on the
friction between the element and the rock. The DMFC element provides either
mechanical or frictional interlocking bonding.

Take the CMC element as an example, Figure 2-2 shows the anchorage models in a view
of cross-section of the element. In the field, two interfaces are the major concerns of
rockbolt failure. One interface is between rockbolt and grout, whilst the other is between
grout and rock mass. In laboratory test, PVC tubes or steel tubes may be used to confine
the rockbolt (Gurung, 2001; Hyett et al., 1995; Kaiser et al., 1992). In such boundary
condition, the major concern is the interface between rockbolt and grout.

(a) Example of CMC element (b) Model of section A-A

Figure 2-2 Sketch of the cross-section of a CMC rockbolt element

The fully grouted rockbolt is the CMC rockbolt element. Experimental results show the
bond of fully grouted rockbolt is created mainly because of the mechanical interlocking
between steel and grout (Moosavi et al., 2005; Singer, 1990; He et al., 2015; Feng et al.,
2018). This implies that the bond strength depends on the mechanical properties of both
the rock and the mortar, the geometry of interlock devices (such as the profile of a
rockbolt rib), the boundary conditions, and so on. Some parametric studies have been
conducted as summarized in Table 2-2.

9
Chapter 2 Literature review

Table 2-2 Parameters to control the bond strength of fully grouted rockbolt/cables
Influenced
Parameters Relationship vs. Bond strength References
mediums
(Yazici and Kaiser, 1992;
Increase with the stiffness ratio or
Rock Stiffness Salcher and Bertuzzi,
rock to grout
2018)
(Kılıc et al., 2003; Kılıc
Increase with the shear strength of et al., 2002; Moosavi et
Strength the grouting material al., 2005; Yazici and
logarithmically Kaiser, 1992; Zhu et al.,
2015)
(1) Optimum w:c ratio = 0.34~0.4,
Water but groutibility (pumpability)
Mortar decreases at w:c ratio = 0.34 (Hyett et al., 1995; Kılıc
cement
et al., 2002)
ratio (2) Higher bond strengths are
results of lower w:c ratio grout
Increases rapidly in first 7 days,
Curing time (Kılıc et al., 2002)
but increases slowly after
Increase with the Young’s
Stiffness (Kılıc et al., 2002)
modulus
(Ghadimi et al., 2014;
Diameter of
Inversely proportional Yazici and Kaiser, 1992)
borehole
(Aziz et al., 2005)
A conical lugged bolt has larger
Shape (Kılıc et al., 2003)
bond strength than a ribbed bar
Geometry of The optimum spacing width for (Aziz et al., 2008; Cao et
interlock Rib spacing particular type of bolt is 37.5 al., 2013; Ghadimi et al.,
devices mm 2014)
Rib face (Cao et al., 2013; Kılıc et
Inversely proportional
angle al., 2003)
Rib height
to length Inversely proportional (Cao et al., 2013)
ratio
(Blanco Martín et al.,
Increase with the presure,
Pressure at 2013; Hyett et al., 1992;
especially for confining
Boundary interface Kaiser et al., 1992;
pressure below 10 MPa
conditions Yazici and Kaiser, 1992)
Installation The field results showed more
(Singer, 1990)
processes variability.

Considering the features of the interactions among rockbolt, grout and rock mass, a bond
model is developed to illustrate the load transfer in conceptual view. In the following,
the bond-slip model of a CMC element is introduced to discuss the load transfer in the
fully grouted rockbolt. Yazici and Kaiser (1992) have defined the bond stress as the

10
 Chapter 2 Literature review

shear resistance along contact surface. Analytical studies and laboratory tests have been
widely used to study the bond strength. Farmer (1975) has proposed a fundamental bond
model which is widely used now. As shown in Figure 2-3, a diametrical unit of the CMC
element between x and x+dx is studied. The force equilibrium along the horizontal
direction gives (Farmer, 1975),

 rb2 d x = −2 rb x dx (2.1)

d x 2
or： = − x (2.2)
dx rb

where: σx – axial stress of rockbolt;

rb – radius of rockbolt;
τx – shear stress at the rockbolt/resin interface.

Figure 2-3 Diametrical slice of a grouted rockbolt (after Farmer, 1975)

Assuming the element is elastic, the axial stress is given as,

d x
 x = − Eb (2.3)
dx
where: εx – axial strain of rockbolt;
Eb – elastic modulus of rockbolt.

Combining Eqs (2.2) and (2.3), gives,

d 2 x 2  x
= (2.4)
dx 2 rb Eb

The bond stress τx could be assumed as:

11
Chapter 2 Literature review

 x = k x (2.5)

Gg
for R − rb  rb : k= (2.5a)
R − rb
Gg
for R − rb  rb : k= (2.5b)
rb ln R rb

where: k – bond stiffness which is assumed to be proportional to the shear modulus

of grouting material, unit in Pa/m;
Gg – rigidity modulus of grout;
R – borehole radius.

Submitting Eq.(2.5) to Eq.(2.4) gives,

d 2 x
2
−  2 x = 0 (2.6)
dx
2Gg
for R − rb  rb 2 = (2.6a)
Eb rb ( R − rb )

2Gg R
for R − rb  rb 2 = 2
ln (2.6b)
Eb rb rb

Integration of Eq.(2.6) gives,

 x = A exp  x + B exp  x (2.7)
where: A and B are the integration constants.

An important assumption in Farmer’s model is that the bond stress is proportional to the
relative bond slip. Li and Stillborg (1999) have enhanced this model and added
progressive decoupling in it if the slip displacement is too large. By adopting the above
bond model, the axial load and bond stress distribution along the rockbolt are shown in
Figure 2-4.

The pullout tests on short encapsulated bolts are suggested to derive the bond-slip model.
Some experiments to test the bond behaviour at reinforced bar and concrete are also
reviewed to verify the model as shown in Figure 2-5 (Ko et al., 2014; Lee et al., 2016;
Mertoğlu et al., 2016; Mousavi et al., 2017). According to the experimental results, a
trilinear bond-slip model is proposed and widely used in recent year (Benmokrane et al.,

12
 Chapter 2 Literature review

1995; Moosavi et al., 2005; Ren et al., 2010; Blanco Martín et al., 2011; Deb and Das,
2011; Ma et al., 2013; Ma et al., 2014; Zhandarov and Mäder, 2016).

(a) Farmer’s model (1975)

Debonding

Bond stress, MPa

Axial load, kN

(b) Li and Stillborg’s model (1999)

Figure 2-4 Axial load and bond stress distributions along the rockbolt based on
illustrated bond models

Figure 2-6(a) shows a sketch of the trilinear bond-slip model. Totally three stages can
be found from the plot. The first stage is a linear ascending line before the bond stress
reaches the peak bond strength τmax. The slope is defined as k1. A descending bond
strength with slip distance is following. The slope k2 is a negative value which describes
the decoupling process at the rock/grout interface. The third stage begins when the bond

13
Chapter 2 Literature review

stress reaches the residual bond strength τres and the slip distance reaches δ2. This model
can be extended to define the load transfer for other rockbolt elements, such as CFC and
DFMC elements (Li and Stillborg, 1999). For example, the bond of CFC element may
follow the model as shown in Figure 2-6(b). The DFMC element will have a
combination of CFC and CMC elements.

Figure 2-5 Examples of the bond load versus slip displacement relationships (after
Mertoğlu et al., 2016)

14
 Chapter 2 Literature review

(a) Mechanical bond model (b) Frictional bond model

Figure 2-6 Illustration of the bond slip model

The mechanical behaviour of rockbolt elements subjecting to shear is important as the

shear force may reduce the load carrying capacity of the rockbolt. Analytical models are
developed to describe the mechanical behaviour of CMC element subjected to shear
loadings. The resultant force along the bolt can be divided into a parallel component
force N0 and a normal component force Q0. Based on the beam theory (as shown in
Figure 2-7), the yield and failure shear force on rockbolt is given as (Li et al., 2015;
Pellet and Egger, 1996):

1   Db2 el 
Yield shear force: Q0e = pu Db  − N oe  (2.8a)
2  4 
2
 Db2  N0 f 
Failure shear force: Q0 f =  ec 1 − 16   (2.8b)
  Db  ec
2
8 

where: pu – yielding pressure of the grout;

Db – rockbolt diameter;
σel – yield stress of the bolt material;
σec – failure stress of the rockbolt material;

Figure 2-7 Sketch of the mechanical system at joint (Pellet and Egger, 1996)

15
Chapter 2 Literature review

The reaction force to the rock is (Pellet and Egger, 1996):

pu = f r c Db (2.9)

where: fr – load factor (fr > 1);

σc – Strength of the host material (such as grout or rock).

2.1.3. Load carrying capacity of a rockbolt

(1) Pullout test

The pullout test results are the most used method to evaluate the load-carrying capacity
of a rockbolt. Typical experimental results of the load-deformation curves of various
types of rockbolt are shown in Figure 2-8. According to the model developed by Farmer
(1975), if tension occurs on rod, it will be transferred to the grout, causing differential
rod extension and grout shear along the length of rockbolt. The stress distribution along
the rockbolt has been evaluated by experiments (Li and Stillborg, 1999; Li et al., 2014).
Some typical models are shown in Figure 2-9.

Figure 2-8 Load-deformation results by Stillborg (after Li, 2010)

It is also necessary to appreciate the fact that any pull-out test results are depends on the
ground conditions. Researches have proposed analytical solutions of the model for bolts
subjected to the joint opening (Deb and Das, 2011; Deb et al., 2018; Li and Stillborg,
1999; Liu et al., 2017). As shown in Figure 2-10, decoupling of the bolt interface during
a joint opening is activated firstly at the joint and then propagates along the interface
with an increase in the opening displacement. It also indicated when the opening
displacement of the joint is small, both the bond stress and the axial stress decrease

16
 Chapter 2 Literature review

exponentially with increasing distance from the joint. When the opening is large enough,
decoupling will be activated at the bolt interface, while the bond and axial stresses will
be changed their distribution as shown in Figure 2-10(b).

(a) DMFC, two-point anchored bolt (b) CMC, a fully grouted bolt

(c) CFC, frictional bolt (d) DMFC, multi-point anchored

Figure 2-9 Sketches of the stress/strain distributions along rockbolts when subjected to
pullout loads (Li et al., 2014)

(a) Small joint opening (b) Large joint opening

Figure 2-10 Stresses along the rockbolt at joint opening (Li and Stillborg, 1999)

17
Chapter 2 Literature review

(2) Shear tests

The load-carrying features of the rockbolt under shear test is also examined using the
load-displacement curves (see Figure 2-11). The incline angle of the rockbolt with
respected to the shear face is one of the most important factors. Comparisons of shear
load and shear displacement relationship with different bolt orientations are shown in
Figure 2-12. When the angle between the bolt and the joint surface is less than 40°, the
failure of the rockbolt is in tension. If the angle is larger than 40°, the failure is in
combined tension and shear. As the angle approaches 90°, the rockbolt shows a shear
type of failure. The strain/force along rockbolt induced by shearing is illustrated in
Figure 2-13. The shear bearing capacity influence by cyclic shear loading is also
examined by Wu et al. (2018). Forbes et al. (2017) have proposed a new technique to
examine the strain along the rockbolt under shearing test. Demonstrated tests are
showing the progressively failure at the joint places (see Figure 2-14).

(a) DFMC, Expansion shell (b) CMC, Cement rebar

(c) CFC, Expansion shell d) DFMC, Twin strand cable

Figure 2-11 Shear load versus displacement of various rockbolts (Li et al., 2014)

18
 Chapter 2 Literature review

(a) CMC, fully grouted rebar (Li et al., 2016) (b) DMFC, D-bolt (Chen and Li, 2015)

Figure 2-12 Comparison of shear load and shear displacement relationship with
different bolt orientations

Figure 2-13 Forces on the rockbolt at the shearing joint (Pellet and Egger, 1996)

Figure 2-14 Strain distributions along the rebars under double shear tests (Forbes et
al., 2017)

19
Chapter 2 Literature review

2.2 Rockbolt design principles

The rockbolt system is designed to minimize the movement induced by the dead weight
of loosened rock, as well as those induced by stress redistribution in the rock
surrounding an excavation. In principle, the rockbolt must be strong enough to support
the deadweight of the loosened rocks. Thus, the strength of the rockbolt is the most
important parameters in the design. Analytical, empirical and observational methods
have been proposed to design the rockbolts for underground excavations (Choquet and
Hadjigeorgiou, 1993). Hoek et al. (2000) have summarized the applications of rockbolts
in underground excavations as shown in Table 2-3. The stability of a shallow
underground excavation under low stress environment depends on the structural
conditions. However, at a great depth, the relationship between the stress of rock mass
and the strength of rock determines the stability of the excavation. For deep level rock
excavation, where rock stresses locally exceed the inherent strength of the rock, design
methods for rockbolting are more complex.

Table 2-3 Applications of rockbolt (modified after Hoek et al., 2000)

Geological strength of rock mass
Massive rock Blocky rock Heavily jointed rock
Low
Stress levels (= in-situ stress/UCS)

Light pattern rockbolts

‘Pre-tensioned’
No support or with mesh and/or
rockbolts to prevent
‘safety bolts’ or shotcrete will control
failure of individual
dowels and mesh raveling of near surface
blocks and wedges.
rock pieces
High

Heavy rockbolt or
Pattern rockbolts or Heavy rockbolts or
dowel pattern with steel
dowels with mesh or dowels, inclined to
fiber reinforced shotcrete.
shotcrete to inhibit cross rock structure,
In extreme cases, steel
fracturing and to keep with mesh or steel fiber
sets with sliding joints
broken rock in place reinforced shotcrete
may be required.

20
 Chapter 2 Literature review

2.2.1. Analytical methods

There are three theories used for the design of the rockbolt system in underground
excavations.

(1) Limited stability analysis of rock block

The sketches to illustrate the method of using the rockbolt to reinforce rock wedge in an
underground excavation driven by gravity is shown in Figure 2-15. The gravity of the
rock wedge is denoted as W, while the shear friction between the boundary of rock
wedge and surrounding rock is R. For the rock wedge formed in the roof, the friction is
not considered. By introducing the safety factors, the number of rockbolt to reinforce
the rock wedges is designed based on the limited stability analysis as follows (Lang and
Bischoff, 1982).

L
α

R
W sb W

β
w
(a) Reinforcement of a wedge free (b) Reinforcement of a wedge free
to slide on the sidewall to fall on the roof

Figure 2-15 Sketches to illustrate the limited stability analysis (after Lang and Bischoff,
1982)

The number of rockbolt to reinforce the rock wedge in side wall (see Figure 2-15(a)) is
calculated as:
W ( f sin  − cos  tan  ) − cA (2.10)
n=
B ( cos  tan  + f sin  )

where: R is the shear friction between the boundary of rock wedge and is (2.10a)
calculated as R = cA + W cos  tan  .

21
Chapter 2 Literature review

The number of rockbolt to reinforce the rock wedge in roof (see Figure 2-15(b)) is
calculated as:

Wf (2.11a)
n=
and: B
sb  3e, w  Lb + 1.0 (2.11b)

where: f – Safety factor, in Figure 2-15(a), f = 1.5 for grouted bolts, f = 2.0 for
nongrouted bolts; in Figure 2-15(b), 2  f  5 ;
n – Number of rockbolts;
W – Weight of wedge;
B – Bolt load carrying capacity;
β – Dip of the sliding surface;
c – cohesive strength of the sliding surface;
A – Base area of sliding surface;
R – Resistance to sliding;
α – Angle between the plunge of the bolt and the normal to the sliding
surface;
w – Width of excavation;
sb – Bolt spacing;
e – Joint spacing;
Lb – Bolt length;
φ – Joint surface angle of friction.

It should be noted that the Eqs (2.10) and (2.11) cannot consider the effects of the in-
situ stresses. Generally, the higher horizontal stress may increase the stability of rock
wedge in roof, but its effectiveness is reduced by loosening (Crawford and Bray, 1983).

(2) The beam or slab concept for bedded rock

Assuming the stratified rock mass behaves like slabs or beams, the design is applicable
to the cases where reinforcement is by suspension and the bolt transfers the weight from
weak strata to the strong strata. The rockbolt must be long enough to get through the
potentially unstable zone. As shown in Figure 2-16, the minimum rockbolt tension to
ensure stability is calculated as (Stillborg, 1986):

22
 Chapter 2 Literature review

T = f  sb  sb'  h    g (2.12)
where: f – Safety factor, recommended 1.5 ≤ f ≤ 3;
sb'– Bolt spacing along the axis of excavation;
h – Thickness of unstable layer of rock;
H – Thickness to reach stable rock layer;
ρ – Rock density.

Figure 2-16 Illustration of the reinforced rock unit (after Lang and Bischoff, 1983)

Lang and Bischoff (1983) extended this analysis by incorporating the shear strength
developed by the rock on the vertical boundaries of the reinforced rock unit. The axial
tensiton load in rockbolt, T, should be calculated :

T   c  1 − exp ( −  kH Rs ) 
=  1 −   (2.13)
ARs  k   Rs   1 − exp ( −  kL Rs ) 

where: A – area of roof carrying one bolt, A = s2;

Rs – shear radius of the reinforced rock unit, Rs = A/P;
P – shear perimeter, P = 4s;
α – a factor depending on the time of installation of the rockbolt, α = 0.5
for active support; α = 1.0 for passive reinforcement;
T – axial tension load in rockbolt;
μ – coefficient of internal friction of rock;
γ – unit weight of rock;
c – apparent cohesion of rock mass;
k – ratio of existing horizontal stress to vertical stress;

23
Chapter 2 Literature review

H – total length of rock zone participating in the formation of reinforced

rock unit and the rockbolt length (L).

(3) The rock arch concept

A natural arch may be formed at some depth into the jointed rock mass. Illustrating
experiments conducted by Hoek (2007) have presented there is a zone of compression
might be induced in the reinforced area, resulting in effective reinforcement as shown
in Figure 2-17. As shown in Figure 2-18, the loose zone is stabilized by systematic
rockbolts. Formulas are developed to estimate the distribution of the rockbolt as:

The length: Lb = 1.40 + 0.184w (2.14a)

The spacing: sb  3e (2.14b)

The tension force: 0.5B  T  0.8B (2.14c)

Figure 2-17 Tom Lang’s explanation of the systematic rockbolts (Hoek, 2007)

Assumed deep
boundary for
natural arch

Loose
zone

Span, w

Figure 2-18 Conceptual natural arch in underground excavation (Hoek, 2007)

24
 Chapter 2 Literature review

2.2.2. Empirical methods

Empirical design proposes spot bolting or sparsely spaced pattern bolting in jointed rock
mass which may be good enough to stabilize the loosen wedges. The design parameters
are mainly bolt length, diameter, spacing, pattern of rockbolt, etc.. The empirical design
recommendations proposed by Stillborg (1986) and Li (2017b) are listed in Table 2-4.

Table 2-4 Empirical design recommendations (after Stillborg (1986) and Li (2017b))

Parameters Description Recommendations and suggestions

Lb ≤ 0.5h (Roof);
Rockbolt length, Lb

For operation
Lb ≤ 0.5w (Wall)
For small depth failure zone Lb ≥ df + l0 (m)
For moderate depth failure zone Lb = 1.40 + 0.184 B
For highly fracture zone Lb = 2 ~ 3 (m)
In general case sb = 1.0 ~ 2.5 (m)
Spacing*, sb

For slightly jointed rock mass (e = 0.3

sb = (3~4)e
~ 1m)
For moderated jointed zone (e < 0.3) sb ≤ 3e
For highly fracture zone sb ≤ Lb/2
For gravitational rock falls Spot bolting
Patterns

For less deformable rock Systematic bolting

For squeezing rock Yield support system
Diameters: 16 ~ 20 (mm)
Other bolt parameters

Bond stiffness Mechanical / Frictional / others

e.g., 339 kN for Dydidag○R ;
Ultimate strength 500 kN for Flexirope○R ;
110 kN for Split set○R .
e.g., Ultimate axial strain,
Creep properties Strain hardening index,
etc.
Note: h – tunnel height; w – tunnel width; df – the depth of the failure zone;

l0 – Embedded length, empirically = 1.2 m; e – mean joint spacing.

*In practice, the spacing of a systematic bolting design includes the in-row
spacing and the spacing between rows. Here, both spacings are assumed equal.

25
Chapter 2 Literature review

Empirical assessment of rock reinforcement based on RMR is widely adopted in mining

development. Table 2-5 shows an example of 10 m span rock tunnel using RMR
methods. The proposed method is more suitable for rockbolt design when the field
vertical stress is less than 25 MPa.

Table 2-5 Examples of 10 m span rock tunnels in accordance with the RMR system
(Hoek, 2007)

Ground mass Support design

Excavation
class Rockbolts Others
Very good Generally no support required except for
Full face: 3 m advance
RMR:81-100 occasional spot bolting

Locally bolts in
Good Full face:1.0-1.5 m crown, 3 m long,
Shotcrete: 50 mm in
advance; complete spaced 2.5 m with
RMR:61-80 crown where required
support 20 m from face occasional wire
mesh
Top heading and bench:
Systematic bolts 4 m
1.5-3 m advance in top
Fair long, spaced 1.5-2 m Shotcrete: 50-100 mm in
heading; commence
in crown and walls crown and 30 mm in
RMR: 41-60 support after each blast;
with wire mesh in sides
commence support 10
crown
m from face
Shotcrete: 100-150 mm
Top heading and bench: in crown and 100 mm in
Systematic bolts 4-5
1.0-1.5 m advance in sides
Poor m long, spaced 1-1.5
top heading; install
m in crown and
RMR: 21-40 support concurrently
walls with wire Steel sets: Light ribs
with excavation- 10 m
mesh spaced 1.5 m where
from face
required
Shotcrete: 150-200 mm
in crown, 150 mm in
Multiple drifts: 0.5-1.5 sides, and 50 mm on
Systematic bolts 5-6
m advance in top face
Very poor m long, spaced 1-1.5
heading; install support
m in crown and
RMR < 21 concurrently with
walls with wire Steel sets: Medium to
excavation; shotcrete as
mesh. Bolt invert heavy ribs spaced 0.75
soon as possible
m with steel lagging and
forepoling; close invert

Empirical assessment of rock reinforcement based on the Q-system are more commonly
used in the tunnel design. The Q-system developed by NGI is a classification system for
rock masses. It provides the support and reinforcement designs according to historical

26
 Chapter 2 Literature review

cases (Barton and Choubey, 1978). Using Q-values, the rock mass conditions, the cavern
geometry and the support classes are linked. As shown in Figure 2-19, the Q-values are
plotted along the horizontal axis, whilst the support classes are divided into 9 categories,
such as:

1 Unsupported or spot bolting;

○
2 SB, Spot bolting
○
3 B+Sfr, Systematic bolting, fibre reinforced sprayed concrete, 5 – 6 cm
○
4 Sfr(E500)+B, Fibre reinforced sprayed concrete and bolting, 6 – 9 cm
○
5 Sfr(E700)+B, Fibre reinforced sprayed concrete and bolting, 9 – 12 cm
○
6 Sfr(E700)+RRS I+B, Fibre reinforced sprayed concrete and bolting, 12-15 cm
○
+ reinforced ribs of sprayed concrete and bolting
7 Sfr(E1000)+RRS II+B, Fibre reinforced sprayed concrete > 15 cm + reinforced
○
ribs of sprayed concrete and bolting
8 CCA or Sfr(E1000)+RRS III+B, Cast concrete lining
○
9 Special evaluation
○

Recommended bolt spacing is an expression of quantity of bolts necessary. The position

and direction of bolt should be based on an evaluation of the joint geometry. The length
of the bolts depends on the span or wall height of the underground opening and to some
degree on the rock mass quality. In unfavorable joint geometry, longer bolts are
necessary. Illustrations of using Q-Chart to define the support categories are shown in
Table 2-6.

Figure 2-19 The Q-chart (NGI, 2015)

27
 Table 2-6 Support category illustrated based on Q-chart
Chapter 2 Literature review

28
 Chapter 2 Literature review

2.2.3. Numerical methods

Different numerical methods are developed to assist the support design for underground
structures. As the rockbolt is a popular structure to provide reinforcement for
underground structure, special elements are included in most of the numerical methods
to simulate the rockbolt, such as RS2, FLAC, UDEC, etc..

(1) Bolt model in RS2

RS2 is a 2D finite element program and offers a wide range of support modeling options
(Rocscience Inc., 2011). The fully grouted rockbolt model is developed based on the
studies of Farmer (1975) and Hyett et al. (1995). It is based on a bar element and only
considers the axial behaviour.

The equilibrium along the axial direction of rockbolt unit gives,

d 2u x (2.15)
Ab Eb + Fs = 0
dx 2

where: Fs – the shear force per unit length;

A – the cross-section area of the bolt;
Eb – the modulus of elasticity of the bolt.

The shear force per unit length could be calculated as:

Fs = k (ur − ux ) (2.16)

where: ur – rock movement;

ux – rockbolt element movement;
k – the shear stiffness of the bolt-grout interface.

Substitute Eq.(2.16) to Eq.(2.15) gives the weak form associated with the total
potential energy which is shown as:
 d 2u x 
  =   AEb 2
− ku x + kur  udx
 dx 
  d  du  du d  u  
=   AEb   x  u  − x  − ( ku x − kur )  u dx (2.17)
  dx  dx  dx dx  
du d u
L
du  
= AEb u x −   AEb x + ku x u dx +  ( kur u )dx
dx 0  dx dx 

Eq. (2.17) can be re-written as:

29
Chapter 2 Literature review

 du d  u 
-  AEb x + ku x u dx +  ( kur u )dx
 dx dx 
 u x1 
  (2.18a)
 K b 0  u x 2 
= − u x1 u x 2 ur1 ur 2     
 0 − K r   ur1 
ur 2 
where [Kb] and [Kr] are calculated as:

AEb  1 −1 kL  1 0.5

 Kb  = +
L  −1 1  3 0.5 1 
and (2.18b)
N N N1 N 2  kL  1 0.5
 K r  = k  N1 N1 =
N 2 N 2  3 0.5 1 
 2 1

The rockbolt element are not necessarily connected to the element vertices in RS2
program. Thus, a mapping procedure is carried out to transfer the effect to the element
vertices. This procedure is done for each bolt segment by mapping the stiffness and the
shape functions depending on the intersected side of the elements.

(2) Cable element in FLAC

Cable element is developed in Fast Lagrangian Analysis of Continua (FLAC) when the
bending effects of rockbolt elements are not necessary to be considered in the simulation
(Itasca Consulting Group, Inc. 2012). The cable is divided into many segments with
nodal points at the segment ends. It allows the modeling of bonding resistance along the
length.

The reinforcing element is treated as bar element with capacity to sustain uniaxial
tension. The increment of the axial force is calculated as:
Eb Ab
F t = − u t (2.19)
L

where: u t = ui ti = ( u xi − u xi +1 ) t1 + ( viy − viy+1 ) t2

u xi , v iy – the movement of node i;

t1 , t2 – the tangential (axial) direction of the cable.

30
 Chapter 2 Literature review

In the software, a tensile yield-force limit and a compressive yield-force limit are
assigned to the cable as shown in Figure 2-20. In computing the axial force,
displacements are computed at nodal points along the axis of the reinforcement.

(a) Axial behaviour (b) Bond-slip model

Figure 2-20 Material models of cable element adopted in FLAC (Itasca Consulting
Group, 2012)

The bond behaviour of the grout annulus is represented as a spring-slider system located
the nodal points as shown in Figure 2-21. The bond force per length is calculated as:

Fs
= k ( uc − u m ) (2.20)
L
where k – the grout shear stiffness;
uc – the axial displacement of the cable;
um – the axial displacement of the medium (soil or rock), and
L – the element length.

The bond stiffness k is calculated by Yazici and Kaiser (1992) and shown as:
2 G
k=
10 ln (1 + 2t / Db )
(2.21)

where: G – the shear modulus of grout;

t – the annulus thickness.

31
Chapter 2 Literature review

Figure 2-21 A spring-slider system to represent the bond behaviour of the grout annulus
(Itasca Consulting Group, 2012)

The maximum bond force per unit length is a function of the cohesive strength of the
grout and the stress-dependent frictional resistance of the grout which can be calculated
as:
Fs ,max (2.22)
= C +  c  tan   Db
L
where: C – intrinsic shear strength or cohesion;
σzz– out-of-plane stress;
φ – friction angle;
Db – perimeter of the cable element.
 c – the mean effective confining stress normal to the element, which can
be calculated as:
  +  zz 
 c = −  nn + p
 2 
where:  nn =  xx n12 +  yy n22 + 2 xy n1n2
ni – the unit vectors;

The advantage of the shear model is that it can consider the confining pressures applying
normally to the cable element. In evaluating the relative displacement at the
grout/medium interface, an interpolation scheme is used to calculate the displacement
of the medium in the cable axial direction at the cable node.

(3) Rockbolt element in FLAC

32
 Chapter 2 Literature review

Rockbolt is simulated using pile element featured in 2D element with 3 degrees of

freedom (two displacements and one rotation) at each end node (Itasca Consulting
Group, Inc. 2012). The rockbolt interacts with the FLAC grid via shear and normal
coupling springs. They are nonlinear connectors that transfer forces and motion between
the elements and the grid at the pile element nodes. The formulation is similar to that
for cable elements. Rockbolt breakage is based on a user-defined tensile failure strain
limit. The total plastic tensile strain is calculated by:

Db  pl
 pl =   plex +  (2.23)
2 L
where: Db – rockbolt diameter;

 pl – total plastic tensile strain;

 plax – axial plastic strain;

L – bolt segment length, and

θpl – average angular rotation over the rockbolt.

A user-defined table can be adopted to give a correction factor for the effective confining
stress, in case of non-isotropic stress, as a function of deviatoric stress ratio. Softening
of the bonding at the rock and rockbolt interface could also considered in the software.

(4) Local reinforcement in UDEC

The cable elements and rockbolt element which are introduced in FLAC are also
adopted in The Universal Distinct Element Code (UDEC) in the deformable blocks.
Besides, local reinforcement can be applied to both rigid and deformable blocks (Itasca
Consulting Group, Inc. 2011).

The local reinforcement formulations consider only the local effect of reinforcement
installed through existing discontinuities. Large displacements are accommodated by
considering the simple geometric changes that develop locally in the reinforcement near
a discontinuity. During shear displacement along a discontinuity, the reinforcement
model deforms as shown in Figure 2-22(a). The short length of reinforcement, which
spans the discontinuity and changes orientation during shear displacement, is referred
to the active length L and determined by:

33
Chapter 2 Literature review

exp ( −  L ) =  max (2.24)

1
 K 4
where  =   in which K = 2 Eg ( d2 d1 − 1)
 4 Eb I 
Eg – Young’s modulus of the grout;
I – second moment of area of the reinforcement element;
d2, d1 – diameter of borehole and reinforcement element respectively, and
ρmax – the proportion of maximum deflection in the reinforcement.

The model consists of two springs located at the discontinuity interface and oriented
parallel and perpendicular to the reinforcement axis (see Figure 2-22(b)). Following
shear displacement, the axial spring is oriented parallel to the active length, while the
shear spring remains perpendicular to the original orientation. The force-displacement
relation that describes the axial response is given by:

Fa = ka ua f ( Fa ) (2.25)

where: ΔFa – incremental changes in axial force;

Δua – incremental change in axial displacement;
ka – the axial stiffness, and
f(Fa) – a function describing the axial loading path.

(a) (b)
Figure 2-22 Springs representing the assumed reinforcement due to shear movement
(Itasca Consulting Group, Inc. 2011)

A reduction factor rf is applied to incremental axial displacements arising from changes

in orientation of the active length to account for the crushing. Note that no reduction (rf

34
 Chapter 2 Literature review

= 1.0) is applied for cases in which there is no change in orientation of the active length.
The reduction factor in the UDEC program is calculated as:
1
rf = uaxial ( u + u
2
s n)
2 −2 (2.26)

where: uaxial – Summation of axial displacement increments;

us – Total discontinuity shear displacement;
ua – Total discontinuity normal displacement.

The shear force-displacement relation is described in incremental form and shown as:

Fs = ks us f ( Fs ) (2.27)

where ΔFs – incremental changes in shear force;

Δus – incremental change in shear displacement;
ks – the shear stiffness, and
f(Fs) – a function describing the shear loading path;

The axial and shear behaviour of this local reinforcement model is shown in Figure 2-23.
Although the local reinforcement model can be used with either rigid blocks or
deformable blocks, the representation is most applicable to cases in which deformation
of individual rock blocks may be neglected in comparison with deformation of the
reinforcing system.

(a) Axial behaviour (b) Shear behaviour

Figure 2-23 Constitutive model of the local reinforced systems (Itasca Consulting
Group, Inc. 2011)

35
Chapter 2 Literature review

The features of the three reinforcement elements are summarized in Table 2-7. It shows
the current models have their advantages and disadvantages. Thus, a more
comprehensive numerical model of the reinforcement element could consider axial,
bond and shear behaviour.

Table 2-7 Examples of the numerical models of a rockbolt

In FEM In FDM In DEM (UDEC)

Illustration

(Rocscience Inc., 2011)

(Itasca Consulting Group, 2012)
(Itasca Consulting Group,
2011)

d 2u x Eb Ab t
Axial

AEb + Fs = 0 F t = − u Fa = ka ua f ( Fa )

dx 2 L
Fs
Fs = k ( ur − u x )
Bond

= k ( uc − u m ) –
L
Shear

– – Ts = ks us f (Ts )

2.3 Rock-rockbolt interaction in underground excavation

A full 3D mechanical analysis of the rockbolt interaction problem is practically difficult

for reinforcement design in tunneling due to time constraints. The rockbolt
reinforcement design often use 2D models to analyze rockbolt and rock interaction
problems. Several numerical methods have been developed to calculate 3D problems
using 2D models as summarized in Table 2-8. Among these methods, the convergence-
confinement method (CCM) is commonly adopted to solve the 3D rock-support
interaction problems associated with the installation of support near a tunnel face in
underground excavations. As shown in Figure 2-24, the CCM consists of three basic
components in the form of graphs: the longitudinal deformation profile (LDP), which
relates tunnel deformation to distance to the tunnel face; the support characteristic curve
(SCC), which represents the stress–strain relationship in the support system; and the
ground reaction or response curve (GRC).

36
 Chapter 2 Literature review

Table 2-8 Useful methods to represent 3D problems using 2D models (modified after
Karakus, 2007)
Methods Illustration Briefing
The radial stress, on the tunnel
confinement method periphery simulates this face
Convergence-

effect and is given by σr = (1-λ)σ0.

An internal pressure, initially
equal to the in-situ stress is applied on the inside of the
excavation boundary. The pressure is incrementally
relaxed until the excavation boundary condition is that
of zero normal stress.
The stiffness reduction method uses a support core
reduction method

with a reduced modulus of elasticity. The excavation

softening
(material
Stiffness

method)

area in the tunnel face is introduced with a modified

elasticity modulus. Modification of the original
modulus of elasticity, E, to the modulus of support
core, Es, is given by: Es = αE.
Tunnel excavation is divided into disks by sections
perpendicular to the tunnel axis. Disk 0 is assigned as
Disk calculation method

ahead of the face, disk 1 represents the excavation of

the tunnel while disk 2 represents the shotcrete
application.
Disk 1 is weakened by the excavation of crown due
to deformations partly reduced by shear stresses
affecting the intersection planes to the more rigid near
disks 0 and 2. At disk 1, excavation is represented by
removing the soil weight at the crown and reducing the
crown rigidity by the value of α.
In Case I, parameter δ stands for the reduction of
lining method
elasticity soft
Hypothetical

the short-term elasticity modulus of the lining.

modulus of

In Case-II, the HME value of the lining takes its

short-term value that varies from 3 to 7 GPa.
In Case III, the lining is fully strengthened with its
long-term elasticity modulus.
A measure of the volume of material that has been
The gap method

excavated in excess of the theoretical volume within

outer diameter of the tunnel lining. Gap parameters
are applied to calculate the associated volume losses
due to the tunnelling process. However, determination
of the gap parameter is very sensitive for types of soil.
Equivalent nodal forces ΔF on the tunnel
Volume loss control method

boundary, which represents the pressure exerted by

soil to be excavated, are calculated and these forces
divided by the number of increments, which simulates
excavation stage. Equal but opposite forces ΔF are
then applied to the tunnel boundary over each n
increments for the excavation stage. After lining
construction, ΔF is still applied to the tunnel boundary
for the remainder of the n increments.

37
Chapter 2 Literature review

Figure 2-24 Ground reaction curve (Vlachopoulos and Diederichs, 2009)

2.3.1. GRC

As the solid line shown in Figure 2-24, the GRC is a plot of the nomralized percentage
of raidal support proeesure, p*, required at a point to limit the normalized radial
boundary dispalcement u*. Analytical solutions have been illustrated by solving a
simple axisymmetric problem (Brown et al., 1983; Wu et al., 2018). It is a circular tunnel
driven in a homogeneous, isotropic, initially elastic rock mass subjected to a hydrostatic
stress field, σ0. The support system is assumed to provide a uniform radial support
pressure, ps. It may happen that the stresses induced in the rock following excavation
will exceed the yield strength of the rock mass and that a plastic zone of radius re will
develop around the tunnel. The rock outside the boundary defined by re is assumed to
remain elastic. There are other assumptions to derive the solutions as following:

(1) Plane strain conditions;

(2) The ground is homogeneous and isotropic, and remains elastic until reaching
failure, which may be brittle and follows the coulomb failure envelop in which post-
peak dilatancy occurs at a constant rate with major principal strain. and
(3) The bolt load is assumed as a uniformly distributed pressure.

38
 Chapter 2 Literature review

Using the β-method to consider the 3D effects, the internal pressure at the excavation
boundary is incrementally relaxed until it is zero (Bobet and Einstein, 2011). The result
is continuous representation of the deformation-internal pressure relationship for a
tunnel. The solution is presented in Table 2-9.

Bobet and Einstein (2011) also derives the analytical soltuions of GRCs for circular
tunnel without reinforcement, circular tunnels with DMFC and CMC/CFC
reinforcement using β-method to consider the 3D effects. The solutions are shown in
Table 2-10, 2-11 and 2-12 respectively.

Table 2-9 Analytical solutions of typical axisymmetric circular tunnel problem

(modified after Brady and Brown, 2006; Carranza-Torres and Fairhurst,
2000)

39
 Table 2-10 Analytical solution for the circular tunnel excavation without rock reinforcement (after Bobet and Einstein, 2011)
Chapter 2 Literature review

40
 Table 2-11 Analytical solution for the circular tunnel excavation with DMFC element reinforcement (after Bobet and Einstein, 2011)

41
Chapter 2 Literature review
 Table 2-12 Analytical solution for the circular tunnel excavation with CMC/CFC element reinforcement (after Bobet and Einstein, 2011)
Chapter 2 Literature review

42
 Chapter 2 Literature review

2.3.2. SCC

Stresses and displacement in rock tunnels depend on the rock mass properties and the
in-situ stress (Alejano et al., 2009). However, the type and stiffness of the support and
the timing of its installation influence the stress/displacement field significantly (Brown
et al., 1983). The interdependence of these factors can be presented in a ground-support
interaction diagram, such as the examples shown in Figure 2-25. Four support catogeries
1 is installed at time t1 and reaches equilibrium
are discussed in this diagram. Support ○
at point B. This support is too stiff to sustain the rock mass deformation. Instead, it is
2 has a lower
easy to fail due to the excessive loading on the support elements. Suport ○
stiffness and it is installed at a later time t2 and reaching equilibrium at C. This support
3 is installed at the same
provides a good time and stiffness to carry the load. Support ○

time t2, but has the lowest stiffness among the four catogeries. It reaches equilibrium at
D where rock mass has initiated falling. This dangeous situation indicates that this
4 which has the same stiffness
support is too flexible to sustain extra loading. Support ○
2 , is installed at time t3. In this case, excessive convergence of the
with support ○

excavation may occur. It means the support elements are installed too late and may not
reach the balance.

Support ○
1 : Too stiff;

Support ○
2 : Most proper

Support ○
3 : Too flexible

Support ○
4 : Too late

Figure 2-25 Illustration of the influence of support stiffness and timing of installation
on support performance (modified after Brady and Brown, 2006)

43
Chapter 2 Literature review

2.3.3. LDP

LDP is developed to work with GRC to determine the appropriate timing for the
installation of support or to optimize the installation of support with specific
displacement capacity. In the LDP, the longitudinal profile of closure or displacement
versus distance from the tunnel face are presented. As shown in Figure 2-26, a portion
of the maximum radial displacements at the tunnel boundary will take place before the
face advances past a specific point. The tunnel boundary will continue to displace
inwards as the tunnel advances further beyond the point in question.

Figure 2-26 Illustration of longitudinal displacement profile (LDP) (Vlachopoulos and

Diederichs, 2009)

To determine the relationship for the LDP, several researchers have suggested
alternative expressions. Vlachopoulos and Diederichs (2009) improved the expressions
and get the best fit relationship of the normalized closure u* as:
u
For x < 0, in the rock mass: u* = = u0*  e x
umax
u0 1 −0.15 R*
For x = 0, at face: u0* = = e (2.28)
umax 3
3x
−
For x > 0, in the tunnel: u* = 1 − (1 − u0* )  e 2 R*

44
 Chapter 2 Literature review

where: R* – the normalized plastic radius, R = RP RT ;

*

RT – tunnel radius;
RP – radius of plastic zone;
umax – maximum radial displacement
u0 – radial displacement at the face location

Graphical LDP templates are also developed by Vlachopoulos and Diederichs (2009) to
provide an alternative way to determine the displacement. It should be noted that, when
adopt this method, umas and RP need to be determined prior to the sequenced analysis
(Alejano et al., 2010; Vlachopoulos and Diederichs, 2014).

2.3.4. Application of rock-rockbolt interaction diagram

The application of rock-rockbolt interaction diagram is evaluated by researchers

(Alejano et al., 2010; Carranza-Torres and Fairhurst, 2000; Dias, 2011; Lüet al., 2012).
The typical approach to use the diagram is summarized as follows.

(1) Calculate the GRC for a tunnel without reinforcement. The relationship
between the pressure applied on the tunnel face and maximum extrusion is calculated.
(2) Simulate the LDP using full 3D analysis (if possible). For 2D analysis, the
methods introduced in Table 2-8 could be adopted to simulate the LDP. The results
could be converted to location along the tunnel using Eq. (2.28).
(3) Plotting GRC and LDP in the same diagram, the intersection point allows
defining the working point of the soil/bolts system. Figure 2-27 illustrates a typical
application process.

However, the application of CCM in engineering practice has some limitations. Some
of them are common in the rock mechanics field, such as the stringent assumptions to
derive equations, the variability of rock mass, complexity of rock structure, etc. The
limitations and caveats when applying the CCM in engineering practice are summarized
in Table 2-13.

45
Chapter 2 Literature review

Figure 2-27 Schematic representation of LDP, GRC and SCC (Lüet al., 2012)

Table 2-13 Limitations and caveats when applying CCM in engineering

Limitations Caveats
(Alejano et al., 2010) (Vlachopoulos and Diederichs, 2014)
The core replacement method is less sensitive to element type
Difficulties in defining (more deviation between roof, floor, wall) in the circular case, and
GRC both are subject to deviations caused by non-ideal geometry (in
the case of the horseshoe).
• Non-isotropic stresses render the standard LDP approach
inaccurate. For moderate values of stress ratio, k0, some
assumptions and adjustments can be made to make the approach
Difficulties in defining
practically viable.
LDP
• Sequenced excavation such as top heading and bench excavation
poses a problem for the LDP approach unless the second
excavation stage is distant from the first.
• For simple tunnel geometries, the 2D LDP and GRC are not
sensitive to the choice of face replacement or pressure reduction
technique but are sensitive to the step size (face too soft or
Complex construction
pressure increment too great).
behaviour
• Tunnel shape is an important factor for the application of 2D
staged modelling although results can be practically acceptable
provided the aspect ratio of tunnel geometry is not extreme
Complex boundary Fixed boundaries should be a minimum of 12 radii from the
condition in engineering tunnel or at least 3 plastic radii away from the plastic zone.
Availability of It is critical to correctly locate the installation step within a
numerical models to staged 2D modelling sequence to prevent overloading or excess
analyze the problem deformation

46
 Chapter 2 Literature review

2.4 Basic discontinuous deformation analysis (DDA) theory

The discontinuous deformation analysis (DDA) method is developed and programmed

by Shi (1988) to perform a complete deformation analysis of a block system. As a
numerical method with a discrete modelling feature, DDA has been widely used in rock
mechanics and rock engineering problems, such as underground structure failures and
reinforcement (Zhao et al., 2007; Kim et al., 1999; Tsesarsky and Hatzor 2006), and
slope-stability analysis (Wu and Chen 2011; Sun et al. 2011). The DDA computes the
static and dynamic behaviours of discrete blocky systems based on block kinematics. It
uses six block deformations as the basic variables in the first-order approximations. In
each time step, an individual block has constant stresses and strains. The displacements
of any point (x, y) inside a block are represented as:

 u0 
 
 v0 
 u   1 0 − ( y − y0 ) ( x − x0 ) 0 ( y − y0 ) 2   r0 
  = 0 1   (2.29)
v  ( x − x0 ) 0 ( y − y0 ) ( x − x0 ) 2    x 
 y 
 
 
 xy 

where (u0, v0) − the rigid body translation at a specific point (x0, y0) within the block;
r0 − the rotation angle of the block with respect to (x0, y0);
εx and εy − the normal strains in the x and y directions, respectively, and
γxy − the shear strain.

Rewriting Eq.(2.29) into a generalized form gives:

 d1i 
d 
 2i 
ui   t11 t12 t13 t14 t15  d 3i 
t16   
 =    (2.29a)
 vi  t21 t22 t23 t24 t25 t26  d 4i 
 d 5i 
 
d 6i 
 

or ui  = Ti Di  (2.29b)

For a system with n blocks, the simultaneous equilibrium equations have the following
form:

47
Chapter 2 Literature review

 K11 K12 K1n   D1   F1 

K K 22 K 2 n   D2   F2 
 21 = (2.30)
    
    
 K n1 Kn2 K nn   Dn   Fn 

or  K ij   Di  =  Fi  (2.30)a

where Kij (i ≠ j) is a 6×6 sub-matrix to represent the contacts between blocks i and j,

Kij (i = j) is the local stiffness matrix, Di and Fi are 6×1 sub-matrices, and Di represents
the deformation variables (u0, v0, r0, εx, εy, γxy) and Fi represents the loading on block i
distributed to the six deformation variables. Figure 2-28 illustrates the structure of the
global stiffness matrix of a 3-block system.

Figure 2-28 Schematic illustration of global stiffness matrix for a 3-block DDA model
(Bao, 2010)

These equilibrium equations are derived by minimizing the total potential energy (Π) of
the block system. The ith row of Eq.(2.30) consists of 6 linear equations:


i = 0, r = 1, 2,...,6
(2.31)
d
ri
where Πi is the potential energy of block i.

Considering the components of dri, the above equation the equilibrium of all the loads
and contact forces acting on block i along the x and y axes respectively, gives:

 i 
= 0, i = 0
u0 v0

48
 Chapter 2 Literature review

The moment equilibrium of all the loads and contact forces on block i. gvies:

 i
=0
r0

The equilibrium of the external forces and stress of block i, gives

 i  i  i
= 0, = 0, =0
 x  y  xy

2.4.1. Submatrices of elastic strains

In the original DDA method, the block is deforming under Hooke’s law. The stress and
strain relationship can be expressed as:
 
 x  1 v 0  x 
  E   
 y  = v 1 0  y  (2.32)
  1 − v
2

 xy 
 1− v   
0 0   xy 
 2 
where E is the elastic modulus of block, v is the Poisson’s ratio. For the plane strain
problem, using E’ and v’ to instead the E and v reprehensively. They are calculated as:
E 
E = ,  = (2.32a)
1 − 2
1 −

Substituting [dri] to replace [εi], the potential strain energy Πe as:

1
 e =   i   i  dxdy
T

2
1
d ri   Ei d ri  dxdy
T
=
2  (2.33)
S
= i d ri   Ei d ri 
T

2
where Si is the area of the ith block.

For block i, the matrix [Ei] is given as:

0 0 0 0 0 0 
0 0 0 0 0 0 

0 0 0 0 0 0 
E  
 Ei  = (2.34)
(1 − v 2 ) 0
0 0 0 1 v 0 
0 0 v 1 0 
 
0 1− v 
0 0 0 0
 2 

49
Chapter 2 Literature review

By taking the derivatives to minimize the strain energy, the contribution of elastic strain
to the stiffness matrix [Kij] can be calculated as:
 2 e
= Si  Ei  →  Kii  (2.35)
di di

2.4.2. Submatrices of point loading

The potential energy Πp of a point load acting at point (x,y) of block i is derived as:
 Fx  T  Fx 
 p = − ( u v )   = −  Di  Ti ( x, y )    (2.36)
 Fy   Fy 

To minimize Πp, the derivatives are computed and formed a 6×1 sub-matrix [Fi] which
is calculated as:
 t11 t21 
t t22 
 12
t13 t23   Fx 
    →  Fi  (2.37)
t14 t24   Fy 
t15 t25 
 
t16 t26 

2.4.3. Submatrices of inertia forces

The time dependent displacements of a point (xi, yi) of block i are denoted as u(t) and
v(t) respectively. The mass per unit area is m. The force of inertia per unit area is given
as:
  2u ( t ) 
 fx   
 t 2 
  = −m 2 (2.38)
 fy    v (t ) 
 
 t 2 
The potential energy Πm is derived as:
 fx 
 m = −  ( u v )   dxdy
 fy 
(2.39)
 2  D ( t )
=  m  Di  Ti  Ti   2  dxdy
T T

t

50
 Chapter 2 Literature review

 2  D ( t )    2 u0 ( t )  2 v0 ( t )  2 r0 ( t )  2 x ( t )  2 y ( t )  2 xy ( t ) 
T

where: = 
t 2  t 2 t 2 t 2 t 2 t 2 t 2 


Using time integration, assume [Di(t)] as the initial block displacement, Δ as the time
interval of this step, we get:
  D ( t ) 
Initial velocity: V0  = (2.40a)
t
  D 2 ( t )  2
Velocity:  i
V =  + V0  =  Di  − V0  (2.40b)
t 2

 2  D ( t )  2 2   D ( t )  2 2
Acceleration: =  D  − = 2  Di  − V0  (2.40c)
t 2   t  
2 i

 2   D ( t )    D ( t ) 
2
Displacement:  Di  = +  (2.40d)
2 t 2
t

By taking the derivatives to minimize the potential energy Πm, the submatrices of inertia
forces are calculated as:

 2
m = 2m T T T  dxdy →  K 
2   i   i 
(2.41)
d d   ii 
i i

−
 m m
=
di  2
(  T  T  dxdy )V  →  F 
i
T
i 0 i (2.42)

2.4.4. Submatrices of bolting connection

The DDA code considers bolting connection between two points. As shown in Figure
2-29, a bolt is connecting point (xi, yi) of block i and point (xj, yj) of block j. The
displacement of the end point is given as:
dxi = ui = u ( xi , yi )
dyi = vi = v( xi , yi )
(2.43)
dx j = u j = u ( x j , y j )
dy j = v j = v( x j , y j )

The increment of bolt length is given as:

  lx   lx    T  lx  T  lx   
dl = ( ui vi )   − ( u j v j )    =  Di  Ti    −  D j  T j    
T T
(2.44)
  ly   l y     ly   l y  

51
Chapter 2 Literature review

The force of bolt due to elongation is given as:

dl
f = − Eb (2.45)
l

Bolting: (xj, yj)

Length: l
Direction: (lx, ly) Block j
Stiffness: Eb
l= ( x1 − x2 ) + ( y1 − y2 )
2 2

Block i 1
lx = ( x1 − x2 )
l
(xi, yi) 1
l y = ( y1 − y2 )
l

Figure 2-29 Bolting connection in original DDA by Shi (1998)

The strain energy of the bolt is calculated as:

1 E
 b = − fdl = b ( dl )
2

2 2l
E
2l
(
= b  Di   Ei   Ei  Di  +  D j   E j   E j   D j 
T T T T
) (2.46)
E
− b  Di   Ei   E j   D j 
T T

l
 lx 
 Ei  = Ti 
T
where:   (2.46a)
 ly 
T  lx 
 E j  = T j    (2.46b)
 ly 

By taking the derivatives to minimize the strain energy Πb, the submatrices of bolting
connection can be calculated as:
 2  b Eb
=  Ei  Ei  →  K ii 
T
(2.47)
di di l

 2 b E
= b  Ei   E j  →  K ij 
T
(2.48)
di d j l

Since the introduction of DDA method in 1980s, rockbolt elements have been developed.
For example, Moosavi and Grayeli (2006) implemented the fully grouted cable element
into the DDA according to the load transfer mechanism. Nie et al. (2014a, b) have

52
 Chapter 2 Literature review

extended the model to simulate different kinds of rockbolt, such as the fully grouted
rockbolt and the D-bolt. He et al. (2018) also have introduced the Euler-Bernoulli beam
elements to simulate the rockbolt element. The results have shown that 2D-DDA could
provide an effective framework to simulate the load transfer between rockbolt and rock
mass.

2.4.5. Submatrices of contacts

In general, there are three kinds of contacts between blocks in DDA: angle to angle,
angle to edge and edge to edge. All the contacts transformed into point-line crossing
inequalities in the calculation. By assuming a point P1 (x1, y1) in block i which has
displacement increment as (u1, v1), the point has potential to pass an inter-penetration
line P2P3 constituted by points (x2, y2) and (x3, y3) in block j. The displacement
increments for the points in the inter-penetration line are assumed as (u2, v2) and (u3, v3)
respectively.

(1) Normal contact

If P1 passes through P2P3, the artificial interpenetration distance dn should be negative

and calculated as:

dn = (2.49)
l
1 x1 + u1 y1 + v1
where: Δ is the area of triangle P1P2P3 and  = 1 x2 + u2 y2 + v2 ,
1 x3 + u3 y3 + v3

l is the length of edge P2P3 and l = ( x2 + u2 − x3 − u3 ) + ( y2 + v2 − y3 − v3 )

2 2
.

Neglect the second order infinitesimal terms, Eq. (2.49) could be revised as:
 u1  
 S0 + ( y2 − y3 x3 − x2 )   
 v1 
d n =  
1
(2.50)
l u   3 
u
 + ( y3 − y1 x1 − x3 )  2  + ( y1 − y2 x2 − x1 )   

  v2   v3  

53
Chapter 2 Literature review

1 x1 y1
where: S0 = 1 x2 y2 .
1 x3 y3

By introducing Eq. (2.19b), Eq. (2.50) could be revised as:

 d1i   d1 j 
S    
d n = 0 + ( e1 e6 )   + ( g1 g6 )   (2.51)
l d  d 
 6i   6j 

1
where: er = ( y2 − y3 ) t1r ( x1 , y1 ) + ( x3 − x2 ) t2 r ( x1 , y1 )  and
l

1 ( y3 − y1 ) t1r ( x2 , y2 ) + ( x1 − x3 ) t2 r ( x2 , y2 ) 
gr =  .
l  + ( y1 − y2 ) t1r ( x3 , y3 ) + ( x2 − x1 ) t2 r ( x3 , y3 ) 

The strain energy of the normal contact spring forces is:

2
kn 2 kn  6 6
S 
 nc = d n =   er d ri +  g r d rj + 0  (2.52)
2 2  r =1 r =1 l 

By taking the derivatives to minimize the strain energy Πnc, the submatrices of contacts
are calculated as:
 2 (  nc ) kn  2  6
2

=   →  Kii 
er d ri  ⎯⎯ (2.53a)
d ri d si 2 d ri d si  r =1 
 2 (  nc ) kn  2  6  6 
d ri d sj
=  
2 d ri d sj  r =1
er d ri   g r d rj  ⎯⎯
 r =1 
→  K ij  (2.53b)

 2 (  nc ) kn  2  6  6 
d rj d si
=  
2 d rj d si  r =1
er d ri   g r d rj  ⎯⎯
 r =1 
→  K ji  (2.53c)

 2 (  nc ) kn  2  6
2

d rj d sj
=  
2 d rj d sj  r =1 
→  K jj 
g r d rj  ⎯⎯ (2.53d)

  nc ( 0 )
2
kS   6 
− =− n 0   →  Fi 
er d ri  ⎯⎯ (2.53e)
d ri l d ri  r =1 

  nc ( 0 )
2
kS   6 
−
d rj
=− n 0  
l d rj  r =1 
→  Fj 
er d rj  ⎯⎯ (2.53f)

54
 Chapter 2 Literature review

(2) Shear contact

Assume P0 with coordinates (x0, y0) is the projection of point P1 on the inter-penetration
line P2P3, the shear displacement is ds is:

1  x +u − x −u 
ds = ( x1 + u1 − x0 − u0 y1 + v1 − y0 − v0 )  3 3 2 2  (2.54)
l  y3 + v3 − y2 − v2 

Ignoring the 2nd order infinitesimal small terms and introducing Eq. (2.19b), Eq. (2.54)
was rearranged as:

 d1i   d1 j 
S    
d s = 0 + ( e1 e6 )   + ( g1 g 6 )   (2.55)
l d  d 
 6i   6j 

x −x 
where, S0 = ( x3 − x2 y3 − y2 )  1 0  ,
 y1 − y0 
1
er = ( y3 − y2 ) t1r ( x1 , y1 ) + ( x3 − x2 ) t2 r ( x1 , y1 )  and
l

1 ( 2 x0 − x1 − x3 ) t1r ( x2 , y2 ) + ( 2 y0 − y1 − y3 ) t2 r ( x2 , y2 ) 
g r =  .
l  + ( x1 + x2 − 2 x0 ) t1r ( x3 , y3 ) + ( y1 + y2 − 2 y0 ) t2 r ( x3 , y3 ) 

The strain energy of the shear contact spring forces is:

S 
2
k k  6 6
 sc = s d s2 = s   er d ri +  g r d rj + 0  (2.56)
2 2  r =1 r =1 l 

By taking the derivatives to minimize the strain energy Πsc, the submatrices of contacts
are calculated as:
 2 (  sc ) ks  2  6
2

=   →  K ii 
er d ri  ⎯⎯ (2.57a)
d ri d si 2 d ri d si  r =1 
 2 (  sc ) ks  2  6  6 
d ri d sj
=  
2 d ri d sj  r =1
er ri   g r d rj  ⎯⎯
 d
 r =1


→  Kij  (2.57b)

 2 (  sc ) ks  2  6  6 
d rj d si
=  
2 d rj d si  r =1
er ri   g r d rj  ⎯⎯
 d
 r =1


→  K ji  (2.57c)

55
Chapter 2 Literature review

 2 (  sc ) ks  2  6
2

d rj d sj
=  
2 d rj d sj  r =1
g r d rj  ⎯⎯

→  K jj  (2.57d)

  nc ( 0 ) k S   6
2

− =− n 0   →  Fi 
er d ri  ⎯⎯ (2.57e)
d ri l d ri  r =1 

  sc ( 0 ) k S   6
2

−
d rj
=− n 0  
l d rj  r =1
er d rj  ⎯⎯

→  Fj  (2.57f)

(3) Criterion to determine the contact status

Two types of iterative solver, namely directly equation solver and the Successive Over-
Relaxation (SOR) solver are used, depending the block number in the model. After
solving the equilibrium equations, the block displacement variables, the artificial
interpenetration distance and the contact forces among the blocks could be solved. The
possible contact status is shown in Table 2-14. The simultaneous equations Eq. (2.30)
are changed in accordance with the spring selection. The procedure of solving equations
and selecting springs are repeated until no penetration and tension constraints are
satisfied, which is also called “open-close (O-C) iteration”. Hatzor et al. (2018) has
mensioned that the numerical damping is essential for DDA analysis as it allows the
oscillations caused by contact forces. The numerical damping is also proportional to the
time step size. The high stiffness evoked by the penalty parameter means that the time
step size must be reduced. An allowable soft contact spring is particularly advantageous
in dynamic solutions.

Table 2-14 Contact status in DDA

Contact Lock/Spring Illustration

Contact forces
states applied

Rn = −kn d n   t No springs and

Open
forces applied.

Rs  Rn tan  + c, Rn   t Lock with normal

Close
and shear springs.
Lock with normal
Sliding Rs  Rn tan  + c, Rn   t spring and a pair
of friction forces.

56
 Chapter 2 Literature review

2.4.6. DDA coding framework

DDA is a displacement method where the unknowns in the equilibrium equations are
displacements as shown in Eq. (2.30). Using an incremental solution procedure, the
block movements are solved at each time step. The framework of the DDA calculation
is illustrated in Figure 2-30.

Figure 2-30 Framework of the multi-time step calculation adopted in DDA

2.5 Summaries and conclusions

The rockbolt load transfer mechanism, the rockbolt design principles and the rockbolt
field application are reviewed in this chapter. Although significant progresses have been
made in analytical, experimental, and numerical studies, the rockbolt design in
underground excavations is still “in the state of art” (Hoek, 2007). The problem in
applying these results is that the properties of the rock mass, particularly a jointed rock
mass, are not fully understood which has been recognized and are common in rock
mechanics. Practically, numerical modelling of specific geometrical and rock properties
can gain insights into the reinforcement mechanism of rockbolt system. This is
especially important for rockbolt design in underground construction. The following
conclusions can be drawn.

(1) Varies types of rockbolts have been developed to restrain the rock movements
in tunnels. However, the reinforcement of different rockbolts are results from
their load transfer capacities to the rock mass which have not been fully studied.
Further researches are required to develop a more general bond-slip model to
present the load transfer mechanism of different rockbolt elements.

57
Chapter 2 Literature review

(2) One of the challenges in underground construction is the large displacement at

the excavation surface. Numerical models using FEM may not get the ground
reaction curve due to its convergence problems. DEM has advantages in
simulating the large displacement of rock system, however, the rockbolt
elements are not comprehensive enough. A more general rockbolt model should
be developed to present various kinds of rockbolt with consideration of their
featured reinforcing behaviours.
(3) The design of rockbolt system in underground construction is complex. The
analytical methods, empirical methods and numerical methods are working
together to propose a reasonable design. Among these methods, numerical
methods have advantages in modelling the ground reactions. More simulations
should be carried out in the future to support and analyze the empirical methods.
(4) The characteristic curves generated using Convergence-Confinement Method
(CCM) are useful expressions to justify the reinforcement going to be adopted
during underground constructions. However, the application is limited by the
complex rock conditions in site. The characteristic curves can be further studied
and the reinforcement mechanism of rockbolt system in underground excavation
will be examined based on the rock/rockbolt interactions.

58
 Chapter 3 Determination of the load transfer of a rockbolt element

DETERMINATION OF THE LOAD TRANSFER

OF A ROCKBOLT ELEMENT

The load transfer of a rockbolt element is studied using 2D-DDA. The load
transfer mechanism is represented by the bond behaviour of a rockbolt element under
shearing in constant normal loading conditions. The bond stress versus slip
displacement curves of different rockbolt elements are simulated and observed. The
effects of the rib profile configurations and the confining pressure are presented.

3.1 Introduction

As discussed in Chapter 2, the load transfer concept is fundamental in understanding

reinforcement mechanism of rockbolt. Many experimental works have been carried
out to study the load transfer in the fully grouted rebars. However, few researches have
discussed the progressively de-bonding and the relationship between the bond-slip
model of the rockbolt element and its profile configuration.

A typical rib profile configuration of a rock element is shown in Figure 3-1. The
profile can be defined by a section along rockbolt axis, i.e., rib height H, rib width w,
rib face angle α and rib spacing s. Due to the variations of the configuration of lab tests,
three potential failure modes in mortar have been detected, such as the parallel shear
slip, the dilational slip, and the wedge slip as shown in Figure 3-1 (Hyett et al., 1992;
Yeih et al., 1997; Ito et al., 2001; Cao et al., 2014).

Several analytical solutions have been presented to judge the different failure modes of
the typical rockbolt profiles (Cao e al., 2014; Ghadimi et al., 2014). These failure
modes are examined and further discussed by numerical analyses. The numerical

59
Chapter 3 Determination of the load transfer of a rockbolt element

models can consider multiple parameters or/and avoid variability caused by sampling
in physical tests.

Figure 3-1 Sketch of the rib profile of a CMC rockbolt element

In this chapter, the two-dimensional Discontinuous Deformation Analysis (2D-DDA)

method is used to investigate the pullout performance of the CMC rockbolt element.
The CMC rockbolt element is modeled as three components, i.e. the rock, the rockbolt
and the mortar. The crack propagations at the rockbolt elements with particular rib
profile configurations are simulated using the sub-block approach (Lin et al., 1996;
Jiao et al., 2012). The material properties of mortar blocks are calibrated against the
experiment results presented by Yokota et al. (2018). Parameter studies are carried out
to investigate the effects of the rib profile configurations and the confining pressures
on the bond stress versus slip displacement curves.

3.2 Trilinear bond-slip model

The sketch of a section of the CMC rockbolt element with length of Δl is shown in
Figure 3-2(a). The rockbolt has a diameter of Db and a radius of rb. The axisymmetric
section can be simplified as shown in Figure 3-2(b). The slip displacement of the
rockbolt element is denoted as Δδ. The bond stress at the bolt-rock interface is written
as τ. The unit shear force Fs with unit of force per meter acting on the bolt-rock
interface for an element can be calculated as,

Fs = 2 rb (3.1)

where: rb is the radius of rockbolt;

τ is the bond stress at the bolt-rock interface.

The axial stress ∆σb on the cross-section of the rockbolt element is given as,

60
 Chapter 3 Determination of the load transfer of a rockbolt element

Fs ( l −  )
 b = (3.2)
Ab
where: ∆l is the length of rockbolt element section;
Ab is the cross-sectional area and Ab=πrb2.

Substituting Eq.(3.1) to (3.2) gives,

2 ( l −  )
 b = (3.3)
rb

If ∆δ << ∆l, Eq.(3.3) could be simplified as,

2l
 b = (3.4)
rb

The bond stress versus slip displacement curve can be described using a trilinear bond-
slip model proposed by Benmokrane et al. (1995). As shown in Figure 3-2(c), a
trilinear bond-slip model consists of an ascending branch up to the peak stress at (τmax,
Δδ1), followed by a softening branch down to (τres, Δδ2), and a horizontal branch
representing the non-zero residual frictional strength (τres) after complete debonding.
The mathematical expression of the trilinear bond-slip model is expressed as,
 =k  + cb (3.5)
where: Δδ is the slip between the bolt and rock, k and cb are constants which can be
given as (Ma et al., 2016),

 max
when 0 ≤ Δδ < Δδ1 : k = k1 = ，cb = c1 = 0 , (3.5a)
1
 res −  max   −  
when Δδ1 ≤ Δδ < Δδ2 : k = k1 = ，cb = c2 = res 2 max 1 (3.5b)
 2 − 1  2 − 1

when Δδ ≥ Δδ2: k = k3 = 0, cb = c3 =  res (3.5c)

where, k1, k2 and k3 are the bond stiffness at stage 1, stage 2 and stage 3 respectively;
Δδ1 and Δδ2 are the slips corresponding to the maximum bond strength and the residual
bond strength, respectively.

To ensure the uniformly distribution of the bond stress at the interface, the section
length of rockbolt element ∆l should be less than 8rb (Blanco Martin et al., 2013).
Using Eqs. (3.4) and (3.5), the relationship between the axial stress ∆σb and slip

61
Chapter 3 Determination of the load transfer of a rockbolt element

displacement Δδ could be determined and can be further used to represent the bond-
slip model of a rockbolt element.

(a) (b)

(c)

Figure 3-2 Sketch of a CMC element for (a) one element, (b) axisymmetric view, and
(c) a trilinear bond-slip model to simulate the bond stress versus slip
displacement curve

3.3 Crack propagation in 2D-DDA

To simulate the crack propagation in rock-like materials, artificial joints have been
introduced to divide a continuous domain into small blocks in 2D-DDA (Lin et al.
1996; Ning et al., 2010; Jiao et al., 2012). In this study, the flat joint contact model
(Potyondy and Cundall, 2004; Zhang and Wong, 2014; Zhang and Zhang, 2017) is
introduced in the 2D-DDA code to present the force versus displacement behavior of
an artificial joint. Blocks A and B as shown in Figure 3-3(a) are co-edged by an

62
 Chapter 3 Determination of the load transfer of a rockbolt element

artificial joint with a length of L. The joint aperture in the plane, w, is assumed to
approach zero in the model, while the joint aperture in the out-of-plane direction is
assumed as t = 1 unit. There are forces and moments arising at the co-edge once a
relative deformation occurs between the two blocks. By using the normal spring and
shear spring to represent the contacts as shown in Figure 3-3(b), the force and moment
at the joint can be calculated as:

Rn = kn dn (3.6)

Rs = −k s d s (3.7)

kn L2
M s = −  s (3.8)
12
where ∆Rn and ∆Rs are increments of normal and shear forces at the artificial joint,
respectively; kn and ks are normal and shear stiffness of the contact spring, respectively;
∆dn and ∆ds are increments of normal and shear displacements, respectively; ∆θs is
increment of shear rotation angle.

B
B
A
A

(a) Sub-blocks (b) Contacts

Figure 3-3 Contact at an artificial joint in a 2D-DDA model

The tensile stress σn and shear stress τs acting at an artificial joint can be calculated as:

 L
 n =  −kn d n + kn s  L (3.9)
 2

 s = ks d s L (3.10)

The fracturing of an artificial joint is determined by Mohr-Coulomb failure criteria.

The artificial joint opens to being a real one if σn > σt or τs > τs,max where t is the
tensile strength and τs,max is the shear strength and could be calculated as τs,max = σn×

tanφ + c. The properties of an artificial joint are assigned initially with the friction

63
Chapter 3 Determination of the load transfer of a rockbolt element

angle , cohesion c and tensile strength t. Once the artificial joint opens, its
properties are resigned with the residual friction angle φ’, residual cohesion c’ and
residual tensile strength t’, and usually c’ = 0, t’ = 0.

3.4 Numerical Modelling

3.4.1. 2D-DDA models

The numerical analyses are carried out to simulate the bonding behavior of the
rockbolt element using a direct shear model. As shown in Figure 3-4, the rock, mortar
and bolt blocks with width of 80 mm are modeled from top to bottom with their
heights of 4.0 mm, 12.0 mm and 12.0 mm, respectively. The left and right boundaries
of the rock block are restrained by horizontal rollers. A normal stress p is applied on
the top surface of the rock block. The mortar block is divided into many sub-blocks
and connected using artificial joints. To simulate the geometry of the rib configuration
of the CMC element, the rockbolt block is designed with the rib height h = 2.0 mm, rib
face angle α = 30°and rib spacing s = 17.8 mm. The rockbolt is pulled horizontally at
the right end of the rockbolt block with a constant speed v = 0.01 mm/s and restrained
the vertical movement by a series of rollers. The horizontal movement of the bolt
block at its right end is recorded as the slip displacement. As the force components
along x-axis are in equilibrium, the bond stress could be calculated as:

rb
=  xx (3.11)
l

where rb is the height of rockbolt block and ∆l is the length of bond interface.

Figure 3-4 Numerical model of the CMC rockbolt element in 2D-DDA (unit: mm)

64
 Chapter 3 Determination of the load transfer of a rockbolt element

3.4.2. Determination of the properties of artificial joint in mortar

The crack propagations in the artificial joints are simulated using the flat joint contact
model. The mortar properties in the DDA model are determined by comparing with
the unconfined compression (UC) tests conducted by Yokota et al. (2018). In their
study, the early curing staged mortar (right after 24 hr) has an average unconfined
compression strength (UCS) of 10.1 MPa and an average Young’s modulus of 3.4 GPa.
The specimen failure involves slip and shear rupture.

The UC test of the above mortar sample is simulated using 2D-DDA with a blocky
model composed by 1815 sub-blocks as shown in Figure 3-5(a). One rigid block on
the top surface of the specimen is used to apply a constant downwards movement at a
speed of 0.1 mm/s to simulate the loading in UC test. The right and left sides of the
rigid block are restrained on the horizontal movement. Another block is attached at the
bottom of the specimen to simulate the rigid base of the UC test. The vertical
displacement of point M1 at the top rigid plate is recorded as the displacement, while
the vertical stress of point M2 at the bottom plate is used to represent the stress. During
the simulation, the Young’s modulus, the uniaxial compression strength and the final
shear rupture from the lab test were used to calibrate the mortar parameters. The final
stress-strain curve obtained from the numerical analysis is shown in Figure 3-5(c), as
well as the percentages of broken bonds during the test. It shows the crack is initiated
at the vertical stress about 60% of UCS, followed by a sharply increases when the
stress approach to UCS. As approximately 80% of the broken bonds are caused by
shear failure, it shows the mortar sample has relatively low shear resistant (Shang et al.,
2018). The stress contour of mortar specimen at failure is shown in Figure 3-5(b). The
parameters used in DDA calculation are listed in Table 3-1. The determined properties
of the mortar block are summarized in Table 3-2.

Table 3-1 Parameter setting of 2D-DDA calculation

Step time, g2 0.0005
Max. disp. ratio, g1 0.0002
Normal spring stiffness, kn 5×108 N/m
Successive over-relaxation (SOR) factor 1.4

65
Chapter 3 Determination of the load transfer of a rockbolt element

(a) (b)

10 100
Percentage of broken bonds, %

8 80
Vertical stress, MPa

6 Stress-strain 60
Total broken

4 40
Shear broken
2 E = 3.54 GPa 20

0 0
0 1 2 3 4
Strain, %

(c)

Figure 3-5 Simulation of the UC test of the motor specimen to determine the
properties of motor (a) numerical model (unit: mm), (b) vertical stress
contours of the specimen at failure, and (c) stress-strain curve and the
relative percentage of broken bonds

66
 Chapter 3 Determination of the load transfer of a rockbolt element

3.4.3. Determination of the frictional properties at the M-B interface

A series of direct shear tests were carried out by Yokota et al. (2018) to determine the
frictional properties at the M-B interface. The schematic diagram of the direct shear
test is shown in Figure 3-6(a). The mortar and steel cube with dimensions of 80 × 80

× 12 mm3 are placed in the upper and lower shear boxes, respectively. Both the
contact surfaces of mortar and steel cube are flat. A normal stress p = 2.0 MPa is
applied onto the upper loading plate. The shear box is pushed horizontally with a
constant speed of 0.01 mm/s. The bond stress is calculated as the axial force F divided
by the surface area of steel cube (80 × 80 mm). The horizontal movement of the shear
box is recorded to represent the slip displacement at the interface and to determine the
bond stress-slip displacement curve.

The direct shear test is simulated using 2D-DDA to obtain the joint parameters at the
M-B interface as shown in Figure 3-6(b). The mortar is divided into sub-blocks
assigned with the material parameters determined in Section 3.4.2 as shown in Table
3-2. A rigid plate is placed next to the mortar block to restrain its horizontal movement.
One more horizontally fixed rigid plate is placed on the surface of mortar block to
apply the normal stress p = 2.0 MPa. The bottom steel plate is pushed with a constant
speed of 0.01 mm/s to simulate the direct shear test procedure. The properties of the
friction parameters at the M-B interface are determined using trial-and-error method
by comparing the calculated bond stress versus slip displacement curve and that from
direct shear test. The final stress-strain curve obtained from the numerical analysis and
that from the direct shear test are shown in Figure 3-6(c). It shows a reasonable
agreement between the two results. The determined the friction parameters along the
M-B interface are listed in Table 3-2.

Table 3-2 Physical parameter setting of the rockbolt element model

Blocks E (GPa) Density (×103 kg/m3) Poisson ratio
Rock (R) 25.6 2.13 0.26
Mortar (M) 5.0 1.96 0.27
Rockbolt (B) 200.0 7.80 0.30
Friction angle (°) Cohesion (MPa) Tensile strength (MPa)
Joints After After After
Initial Initial Initial
fracturing fracturing fracturing
M-B 30 20.0 3.0 0 1.0 0
M-M 40 21.5 6.0 0 4.0 0

67
Chapter 3 Determination of the load transfer of a rockbolt element

(a)

(b)

2.5

2.0
Bond Stress, MPa

1.5

1.0

0.5 Bond stress (DDA)

Shear test (Yokota et al. , 2018)
0.0
0.0 1.0 2.0 3.0 4.0
Slip displacement, mm
(c)

Figure 3-6 Simulation of the direct shear test using 2D-DDA (a) schematic diagram of
the lab test (unit in: mm), (b) numerical model (unit in: mm) and (c)
comparisons of the bond stress versus slip displacement curves

68
 Chapter 3 Determination of the load transfer of a rockbolt element

3.5 Numerical Results

The bond stress versus slip displacement curve of the CMC rockbolt element with rib
face angle  = 45°and rib spacing s = 17.8 mm under a normal stress p = 2 MPa is
plotted in Figure 3-7. Based on the progressive fracture development and the
displacement accumulation, the curve can be classified into three stages. Stage 1 starts
from the initial loading to the time when the bond stress is less than the maximum
bond stress τmax which can be further classified into the initial elastic stage (stage 1-1)
and the crack initiation stage (stage 1-2). In the initial elastic stage, there are no broken
bonds observed. Once there are broken bonds, stage 1-2 starts. The broken bonds are
firstly observed at the rib surface, such as the fracture developed at the point A (Δδ =
0.06 mm) in the bond stress versus slip displacement curve as shown in Figure 3-8(a).
The bond stress continues increasing with the accumulated slip displacement. Near the
end of stage 1, the cracks have grown from the rib tips at an average angle of 60°with
respect to the moving direction, such as fracture developed at the point B (Δδ = 0.3
mm) in the bond stress versus slip displacement curve as shown in Figure 3-8(b).
Three local fractures generate from three rib tips towards the R-M surface, dividing
the mortar area into four parts. There is approximately 60% of the total cracks
occurring during stage 1. After the maximum bond stress τmax, the bond stress
decreases with the accumulated slip which is denoted as stage 2 in the bond stress
versus slip displacement curve as shown in Figure 3-7. More cracks generate and
connect among the three local fractures, further dividing the mortar into smaller parts
(see Figure 3-8(c)). The new generated rupture surface in mortar block is at an average
angle of 34°with respect to the moving direction of the bolt block. The number of new
cracks in this stage accounts for about 25% of the total cracks.

The bond stress becomes stable with the residual bond strength τres, and the following
stage is defined as stage 3 where very few new cracks are generated. The model fails
in wedge failure mode (see Figure 3-8). The bond stress versus slip displacement
curve could be fitted using the trilinear bond-slip model. As shown in Figure 3-7, the
obtained parameters are k1 = 9.0 MPa/mm, k2 = -2.5 MPa/mm, k3 = -0.2 MPa/mm, τmax
= 3.4 MPa and τres = 2.4 MPa.

69
Chapter 3 Determination of the load transfer of a rockbolt element

Stage 1 Stage 2 Stage 3

4.0 100
Stage 1-1 Stage 1-2 τmax = 3.4 MPa

Percentage of broken bonds, %

B
80
Bond stress, MPa

3.0
C τ = 2.43 MPa
res

60
D
2.0 k1-2 =4.1 MPa/mm
k2 = -2.5 MPa/mm
k1 = 9.0 MPa/mm 40
A k3= -0.2 MPa/mm
1.0 k1-1 = 35.6 MPa/mm — Total broken 20
- - Shear broken

0.0 0
0.0 0.2 0.4 0.6 0.8 1.0
Slip displacement, mm

Figure 3-7 Simulated bond stress-slip displacement curve of the CMC element with
rib face angle α = 45°and rib spacing s = 17.8 mm under p = 2 MPa

Tension
60°
Shear
A – Δδ = 0.06 mm B – Δδ = 0.3 mm

(a) (b)

34°

C – Δδ = 0.7 mm D - Δδ = 0.9 mm

(c) (d)

Figure 3-8 Crack propagations in mortar of the CMC rockbolt element with rib face
angle α = 45°and rib spacing s = 17.8 mm under p = 2 MPa at the slip
displacements of (a) Δδ = 0.06 mm, (b) Δδ = 0.3 mm, (c) Δδ = 0.7 mm and
(d) Δδ = 0.9 mm

70
 Chapter 3 Determination of the load transfer of a rockbolt element

3.6 Parametric studies

To identify the major factors that affect the bond strengths of the CMC element,
parametric studies are carried out. The height and width of the rock, mortar and bolt
blocks are fixed as shown in Figure 3-4. The studied variables are the rib face angles α,
the rib spacing s, and the normal stress p.

3.6.1. Effects of the rib face angle

The rockbolt blocks with rib spacing s = 17.8 mm and rib face angles α = 30°, 60°and
90°are analysed to investigate the effect of the rib face angle on the bond-slip models
under the normal stress p = 2 MPa. The bond stress versus slip displacement curves
for the CMC rockbolt elements with different rib face angles are plotted in Figure 3-9.
The accumulated percentages of the broken bonds during the slip progress are also
included. In the case of α = 30°, the bond stress versus slip displacement curve (see
Figure 3-9(a)) shows the bond stress increases linearly with respect to the slip
displacement at beginning. In addition, the curve has no softening branch. One
possible reason is that the moving of the rockbolt element is sliding along the M-B
interface. This can be verified using the crack propagation contours as shown in Figure
3-10(a) and (b). It shows the dilation happens along the M-B interface and very few
cracks generated in the mortar block. Thus, the bonding effects of the CMC element
with rib face angles of less than 30°are not effective under the normal stress of 2 MPa.

The bond stress versus slip displacement curves of the CMC element with rib face
angles of 60°and 90°are shown in Figure 3-9(b) and (c), respectively. The obtained
curves are similar to that of the rib face angles α = 45° (see Figure 3-7). The
inclination angles of the initial cracking in the mortar from rib surface of the CMC
elements with rib face angles α of 60°and 90°are slightly increased (see Figure 3-10(c)
and Figure 3-10(e)), but those generating in stages 2 are not much different (see Figure
3-10(d) and Figure 3-10(f)). The bond stress versus slip displacement curve could be
fitted using the trilinear bond-slip model. The obtained parameters are summarized in
Table 3-3. It can be seen that the CMC element with rib face angles α of 90°has a
slightly higher maximum bond stress. However, the other parameters are similar. Thus,
if change the rib face angle of a CMC element from 45°to 90°under normal stress
p = 2 MPa, the changes of the parameters of the bond-slip models are limited.

71
Chapter 3 Determination of the load transfer of a rockbolt element

4 100
Total broken

Percentage of broken
Bond stress-slip 80
3
Bond stress, MPa

Shear broken

bonds, %
60
2
40
1
20

0 0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Slip displacement, mm
(a)

4 100

Percentage of broken bonds,

Total broken
Bond stress-slip
80
3
Bond stress, MPa

Shear broken
60

%
2
40
k2 = -2.6 MPa/mm
1 k3= -0.3 MPa/mm 20
k1 = 7.4 MPa/mm
0 0
0 0.2 0.4 0.6 0.8 1 1.2
Slip displacement, mm
(b)

4 100
Bond stress-slip
Percentage of broken bonds, %

Total broken
80
Bond stress, MPa

3
Shear broken
60
2
40
1 k2 = -2.7 MPa/mm
k3= -0.27 MPa/mm 20
k1 = 10.8 MPa/mm
0 0
0 0.2 0.4 0.6 0.8 1
Slip displacement, mm
(c)

Figure 3-9 Effects of the rib face angles on the bond stress-slip displacement curves
when (a) α = 30°, (b) α = 60°and (c) α = 90°

72
 Chapter 3 Determination of the load transfer of a rockbolt element

50°

(a) (b)

64°

(c) (d)

77°

(e) (f)

Figure 3-10 Effects of the rib face angles on the crack propagations when (a) α = 30°,
Δδ = 0.3 mm, (b) α = 30°, Δδ = 0.8 mm, (c) α = 60°, Δδ = 0.3 mm, (d) α =
60°, Δδ = 0.8 mm, (e) α = 90°, Δδ = 0.3 mm and (f) α = 90°, Δδ = 0.8 mm

Table 3-3 Parameters defining the trilinear bond-slip models under p = 2 MPa

Rib
Rib Average
face k1-1, k1-2, k2, k3, τmax, τres,
spacing, k1,
angle, MPa/mm MPa/mm MPa/mm MPa/mm MPa MPa
s MPa/mm
α
30° 32.1 - - - - 2.36 1.81
45° 35.5 4.2 9 -2.6 -0.2 3.34 2.23
17.8mm
60° 34.6 3.5 7.4 -2.6 -0.3 3.25 2.10
90° 30.0 6.2 10.8 -2.7 -0.27 3.55 2.16
30° 32.1 - - - - 1.84 1.81
45° 32.0 2.80 5.4 -2.2 -0.27 3.34 2.28
35.6mm
60° 34.7 2.01 5.0 -1.9 -0.2 3.20 2.00
90° 29.0 2.75 5.3 -2.1 -0.3 3.60 2.26

73
Chapter 3 Determination of the load transfer of a rockbolt element

3.6.2. Effects of the rib spacing

The rockbolt blocks with rib spacing s = 35.6 mm and rib face angles α = 30°, 60°,
and 90°are analysed to investigate the effect of the rib spacing. The normal stress p is
2 MPa. The bond stress versus slip displacement curves of the CMC rockbolt element
with rib face angles α = 30°and rib spacings s = 17.8 and s =35.6 mm are plotted in
Figure 3-11(a). It can be seen that the rib spacing has no much influences on the bond
stress versus slip displacement curves at the initial elastic stage (stage 1-1 as defined in
Figure 3-7). With the accumulated slip displacement, the bond stress versus slip
displacement curves show their differences. The larger the rib spacing of a rockbolt
element, the lower the bond stress the element could sustain at the same slip
displacement. As shown in Figure 3-12(a), the crack propagations and the stress
contours at the slip displacements Δδ ≈ 1.0 mm show that the failure mode of the
element is the sliding along M-B interface and the cracks are only occurring near the
end rib. The bond stress versus slip displacement curves of the CMC rockbolt
elements with rib spacing s = 17.8 and 35.6 mm, and rib face angles α = 45°, 60°and
90°, are plotted in Figure 3-11(b) - (d), respectively. It can be seen that the simulated
bond stress-slip displacement curves of the rockbolt element are similar as that of the
CMC element with rib spacing s = 17.8 mm and rib face angle  = 45°as shown in
Figure 3-7. The rib spacing has less influence on the bond stress versus slip
displacement curves at the initial elastic stage. Compared with the bond stress versus
slip displacement curves of rockbolt elements with rib spacing s = 17.8 mm, the curves
of rockbolt element with rib spacing s = 35.6 mm delay the crack development. The
bond stress versus slip displacement curves could be fitted using the trilinear bond-slip
models (see Table 3-3). The bond stiffness at stage 1-2 for the rockbolt element with
rib spacing s = 35.8 mm are approximately 57% lower than that of rockbolt element
with rib spacing s = 17.6 mm which results in the average bond stiffness at stage 1 are
approximately 67% of the latter one. The bond stiffness at stage 2 for the rockbolt
element with rib spacing s = 35.8 mm is also lower than that of rockbolt element with
rib spacing s = 17.6 mm, while the maximum and the residential shear strengths are
almost the same.

The crack propagations and stress contours of rockbolt elements with rib spacing s =
35.6 mm pulled at Δδ ≈ 1.0 mm are shown in Figure 3-12. For the case of rib face

74
 Chapter 3 Determination of the load transfer of a rockbolt element

angles α = 45° as shown in Figure 3-12(b), the crack grows from the two rib tips
towards the R-M interface with an inclination about 40°, dividing the mortar into 3
parts. More cracks are generated in mortar near the M-B interface resulted in wedge
failure mode. The crack propagations at failure in the mortar block are similar for the
rockbolt element with varied rib face angles α = 60° and 90° as shown in Figure
3-12(c) and Figure 3-12(d), respectively.

4
Bond stress, MPa

1
— s = 17.8 mm
— s = 35.6 mm

0
0.0 0.2 0.4 0.6 0.8 1.0 1.2
Slip displacement, mm

(a)

k2 = -2.2 MPa/mm
3
Bond stress, MPa

2 k1 = 5.4 MPa/mm

1
— s = 17.8 mm
— s = 35.6 mm
0
0.0 0.2 0.4 0.6 0.8 1.0 1.2
Slip displacement, mm

(b)

75
Chapter 3 Determination of the load transfer of a rockbolt element

k2 = -1.7 MPa/mm
Bond stress, MPa

k1 = 5.5 MPa/mm
2

1 — sb = 17.8 mm
— sb = 35.6 mm

0
0.0 0.2 0.4 0.6 0.8 1.0 1.2
Slip displacement, mm

(c)

4
k2 = -2.8 MPa/mm
Bond stress, MPa

2 k1 = 5.3 MPa/mm

1 — sb = 17.8 mm
— sb = 35.6 mm

0
0.0 0.2 0.4 0.6 0.8 1.0 1.2
Slip displacement, mm
(d)

Figure 3-11 Effects of the rib spacings on the bond stress versus slip displacement
curves when (a)  = 30°, (b)  = 45°, (c)  = 60°and (d)  = 90°

76
 Chapter 3 Determination of the load transfer of a rockbolt element

(a) (b)

(c) (d)

Figure 3-12 Effects of the rib spacing s on the fracturing in mortar block at slip
displacement Δδ ≈ 1.0 mm when (a) α = 30°, (b) α = 45°, (c) α = 60°
and (d) α = 90°

3.6.3. Effects of the normal stress

The rockbolt element with rib spacing s = 17.8 mm and rib face angles α = 30°, 45°,
60°and 90°under different normal stress p = 1 MPa, 2 MPa and 4 MPa are simulated
to investigate the effects of the normal stress. The simulated bond stress versus slip
displacement curves are shown in Figure 3-13. As expected, both the bond stiffness
and the bond strength are increasing with the increase of normal stress. For the case of
rockbolt blocks with rib spacing s = 17.8 mm and rib face angles α = 30° under
normal stress p = 4 MPa as shown in Figure 3-13 (a), the element could be restrained,
but the differences between the maximum bond strength and residual bond strength are
little. The bond stresses versus slip curves of the rockbolt element with rib face angle
 = 45°, 60° and 90° are shown in Figure 3-13(b), (c) and (d), respectively, the
maximum bond strengths are still much higher than their residential bond stresses.

The crack propagation contours of rockbolt element with rib spacing s = 17.8 mm
pullout at Δδ ≈ 1.0 mm are shown in Table 3-4. It can be seen that the higher the
normal stress, the more cracks are propagated in the mortar. This explains why both
the bond stiffness and the bond strength are increasing with the increase of normal
stress. It also shows that the inclination of the macro-fracture from the rib tips is

77
Chapter 3 Determination of the load transfer of a rockbolt element

increasing with the increase of normal stress. The cracks at failure propagated in the
mortar block of rockbolt element with rib face angle of α = 60°and 90°are evenlt
parallel to the M-B interface when the normal stress is high (i.e., p = 2 and 4 MPa).

6.0

4.0
Bond stress, MPa

2.0
30-3ribs-p=4MPa
30-3ribs-p=2MPa
30-3ribs-p=1MPa
0.0
0.00 0.25 0.50 0.75 1.00 1.25 1.50
Slip displacement, mm
(a)

6.0

k1 = 19.1 MPa/mm

4.0
Bond stress, MPa

k1 = 9.0 MPa/mm 45-3ribs-p=4MPa

2.0 45-3ribs-p=2MPa
45-3ribs-p=1MPa
Bond-slip, p=4MPa
Bond-slip, p=2MPa
k1 = 4.5 MPa/mm Bond-slip, p=1MPa
0.0
0.00 0.25 0.50 0.75 1.00 1.25 1.50
Slip displacement, mm
(b)

78
 Chapter 3 Determination of the load transfer of a rockbolt element

6.0

k1 = 17.7 MPa/mm

4.0
Bond stress, MPa

k1 = 7.4 MPa/mm 60-3ribs-p=4MPa

2.0
60-3ribs-p=2MPa
60-3ribs-p=1MPa
Bond-slip, p=4MPa
k1 = 4.9 MPa/mm Bond-slip, p=2MPa
Bond-slip, p=1MPa
0.0
0.00 0.25 0.50 0.75 1.00 1.25 1.50
Slip displacement, mm
(c)

6.0

k1 = 12.0 MPa/mm
Bond stress, MPa

4.0
k1 = 10.8 MPa/mm

90-3ribs-p=4MPa
90-3ribs-p=2MPa
2.0
90-3ribs-p=1MPa
Bond-slip, p=4MPa
Bond-slip, p=2MPa
k1 = 7.0 MPa/mm
Bond-slip, p=1MPa
0.0
0.00 0.25 0.50 0.75 1.00 1.25 1.50
Slip displacement, mm

(d)

Figure 3-13 Effects of the normal stress p in different rib faces angles (a) α = 30°, (b)
α = 45°, (c) α = 60°and (d) α = 90°

79
 Table 3-4 Effects of the normal stress on crack propagations in the mortar of CMC rockbolt element with rib spacing s = 17.8 mm
and different rib face angles at the slip distances ∆δ = 1.0 mm

80
Chapter 3 Determination of the load transfer of a rockbolt element
 Chapter 3 Determination of the load transfer of a rockbolt element

The effects of the rib face angle and the rib spacing of the rockbolt profile
configuration on the pullout performance of the rockbolt are summarized in Table 3-5.
Under the lower normal stress, the effect of the CMC rockbolt element with rib face
angles of less than 30°is not effective. If the normal stress is high, the rib face angle
doesn’t have much influences on the pullout performance of the rockbolt. The rib
spacing is one of the key parameters controlling the bond stiffness. The confining
conditions are the most important parameters affecting the bond-slip behaviour of a
rockbolt element. The higher the normal stress, the higher the bond stiffness and the
bond strength of the rockbolt and more cracks are propagated in the mortar block.

Table 3-5 Results of case studies

Bolt profile
Configuration Normal
Rib face angle Rib spacing, from
stress
from 30°to 90° 17.8 to 35.6 mm
Load transfer increase
capacity parameters
Bond stiffness before
No much effects Increase
cracking, k1-1
Average bond stiffness in
Slightly increase Decrease Increase
stage 1, k1,ave
Max. bond strength, τmax No much effects No much effects Increase
Bond stiffness in stage 2,
No much effects Slightly decrease Decrease
k2
Res. bond strength, τres Slightly increase No much effects Increase
Inclination of
Crack propagation in mortar

the crack
Slightly increase Decrease Increase
growing from
rib tips
Inclination of
the rupture No much effects Decrease Decrease
surface
Wedge failure, but
Main failure tend to parallel shear Wedge failure and Parallel
modes slip if the normal stress dilational slip shear slip
increases
Note: The effects of the rib face angle will influenced by the normal stress scales.

81
Chapter 3 Determination of the load transfer of a rockbolt element

The fully grouted rebar could be considered as combinations of CMC rockbolt

elements. Once the bond-slip model of the rockbolt element is obtained, the
performance of rockbolt could be simulated by discrete rockbolt element. An
application example is shown in Section 4.4.2. It should be noted that the above bond-
slip models are drawn by assuming the rockbolt and mortar were well bonded, thus, no
crack propagations are observed at the rockbolt and mortar interface in the above
simulations. The bond-slip models should also be adopted for the rockbolt element
under similar conditions.

3.7 Conclusions

In this chapter, 2D-DDA is used to investigate the pullout performance of the CMC
rockbolt element. The CMC element is modeled as three components, i.e. the rock, the
rockbolt and the mortar. The material properties of mortar blocks are adopted from the
experimental results proposed by Yokota et al. (2018), and the frictional properties
along rockbolt – mortar interface are calibrated with their experimental results. The
flat joint contact model is introduced into the 2D-DDA code to simulate the force
versus displacement behaviour of an artificial joint. The results show that a bond stress
versus slip displacement curve of the CMC element is generally exhibiting three stages
which are distinguished by five key parameters, i.e., the bond stiffness in three stages
(k1, k2, and k3), the maximum bond strength τmax, and the residual bond strength τres.

Parametric studies have been carried out to investigate the effects of the normal stress
and the rib profiles on the bond stress versus slip displacement curves and the
fracturing modes in mortar. It is found that the normal stress plays an important role in
the bond-slip modeling. The bond stiffness k1, the maximum bond strength τmax and the
residual bond strength τres are increasing with the increase of normal stress. A larger
rib spacing will result in a lower bond stiffness when other parameters are kept the
same. Under a lower normal stress, the interlock of the CMC rockbolt element with rib
face angles of less than 30°is not effective because the rockbolt elements are pulled
out at the rib faces and less cracks are generated in the mortar blocks.

82
 Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

DEVELOPMENT OF A ROCKBOLT MODEL IN

2D-DDA AND ITS VERIFICATIONS

Based on the load transfer theory, a rockbolt model is developed in 2D-DDA.

The rockbolt model is characterized by a relatively large displacement at rock joints.
Generally, the rockbolt is represented by several rockbolt elements. For each element,
the axial behaviour, bond behaviour and shear behaviour are introduced. Based on
the load transfer at the bond of a rockbolt element, the reinforcement of rockbolts
could be simulated.

4.1 Introduction

As reviewed in Chapter 2, a reinforcement system can be considered as the

combination of four components in a conceptual view (see Figure 4-1): rock, internal
fitting (or bond), rockbolt and external fitting. If tension occurs in the rockbolt rod, it
will be transferred to the rock by bond. The bond is referring to the bond resistance
along contact surface defined by two springs. In this chapter, a new developed
rockbolt model based on this conceptual view is introduced. Some verification works
will also be carried out based on the experimental results.

Figure 4-1 Sketch of the components of the proposed rockbolt model (Nie et al.,
2014a)

83
Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

4.2 Development of the rockbolt model

4.2.1. Implementation of the rockbolt model in 2D-DDA

A rockbolt model is proposed as a bar element with its direction of extension from the
free end to the other end where a faceplate is usually attached. To be integrated with
the block system, a rockbolt model is divided into several segments according to the
block boundaries. Each segment contains several rockbolt elements, while each
element is containing a pair of bolt nodes (Nie et al., 2014a, b). To consist with the
bond-slip models deduced from Chapter 3, the length of the rockbolt element is set as
0.1 m (except those near joints whose lengths might be determined by the hinge
length). For example, a rockbolt is going through three rock blocks A, B and C as
shown in Figure 4-2. Thus, this rockbolt is divided into segments A, B and C
accordingly. Take Segment C as an example, it contains four rockbolt elements, i.e.
Elements 1 to 4. One rockbolt element, such as Element 1, is a bar element. It contains
two rockbolt nodes, such as bolt node 1 and bolt node 2. To record the relative
movement between rock and rockbolt, there is a rock node which is initially with same
position as the bolt nodes, such as the rock node i’ is initially with the same position as
the bolt node i. Once the block moves, the relative movements between two nodes are
recorded as slip displacement which cause the rockbolt extension later. Thus, the
tensile force/shear force will restrain the block movement. This simulates the load
transfer in a rock and rockbolt system.

Segement A Segment B Segment C

Element 13 ... Element 4 ... Element 1

block A block B block C 14 13 12 11(10) 9 8 7 6(5) 4 3 2 1 (bolt node)

(Fixed)

14' 13' 12' 11'(10') 9' 8' 7' 6'(5') 4' 3' 2' 1' (rock node)

block A block B block C block A block B block C

(a) Rockbolt element in rock block (b) Nodes in the rockbolt and the related rock nodes

Figure 4-2 Rock block and rockbolt model in 2D-DDA program (Nie et al, 2014a)

84
 Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

4.2.2. Analytical solutions of a rockbolt model

The mathematical algorithm for a rockbolt element has been introduced in Section
2.1.2. It is assumed that only the slip at the interface of rockbolt and rock is considered.
Thus, the slip displacement is determined as ΔUrx – ΔUbx. By assuming a linear
relationship between the bond stress and the slip displacement, the increase of bond
stress Δτi could be calculated as:
 i  k  U rx  Ubx  (4.1)

where: k –bond stiffness at the interface, unit in Pa/m; k = k1, before decoupling
happens, k = k2, once the accumulated shear stress at the anchorages is
beyond the maximum shear strength or the relative displacement is greater
than δ1 but less than δ2, and k = k3, for the low frictional slipping along
interface if relative displacement is greater than δ2,
ΔUrx – displacement of the rock;
ΔUbx – displacement of the rockbolt.

The axial stress in the rockbolt element i induced by bond slip is written as:
 i li
 i   Db (4.2)
Ab

where: Ab – cross-sectional area of the rockbolt,

Δσi – incremental axial stress, and
li – length of rockbolt element i.

Assuming the stress-strain relationship of the rockbolt is in linear elastic, thus,

dU bx
  i  Eb  i  Eb (4.3)
dx

The governing equation of a rockbolt element is:

d 2 Ubx  i
2
 (4.4)
dx Eb Ab

d 2  Ubx  k k
or 2
  Urx    Ubx  (4.4a)
dx AE AE

Eq.(4.4) could be rewritten as:

Ab Eb AE x x 
xi  xi 1
    
Ubxi  Ubxi 1  b b Ubxi  Ubxi 1  k  i 1 i 1  U rxi  Ubxi
xi1  xi  2 
 (4.5)

85
Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

If the rockbolt is fixed by faceplate, then,

Ab E b
Tn 
x n 1  x n
U bxn  U bxn 1  (4.5a)

If the rockbolt end is free, we have

Ab Eb  x x 
xn  xn1
Ubxn  Ubxn1   k  n n1  Urxn  Ubxn 
 2 
(4.5b)

The increment of the restrain force provided by rockbolt at the rockbolt node i (i =
1, …, n-1) ΔPi could be calculated as:

 Pi  kli U rxi  U bxi  (4.6)

By defining the angle between the rockbolt direction and x-axis as θ, the variations of
point loads to the rock block induced by the restrain of rockbolt at a time step could be
calculated as:

 Fx   cos 
 F  =Pi   (4.7)
 y -sin  

where Fx and Fy are the point loads at the horizontal and vertical direction,
respectively.

Figure 4-3 Transformation from 1D bar element to 2D space

86
 Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

The extension directions of rockbolt elements are assumed as constant, i.e., the same
as the ones under the initial loading. However, the directions may change at the joint
place, such as the extension direction at node A may follow θ1 or θ2. If UbA < UrA, that
is the bolt is under unloading condition, θ1 is chosen. On the other hand, if UbA > UrA,
that is the bolt is under extension, θ2 is chosen.

4.2.3. Axial behaviour

A rockbolt is usually made of steel bar. An elastic, linear strain-hardening steel model
is adopted to simulate the axial behaviour of the rockbolt model. As shown in Figure
4-4(a), the rockbolt deforms according to elastic modulus Eb if the stress is less than
the yield stress σY. Once the stress is beyond the yielding stress (point D), the rockbolt
is in linearly plastic range. At the stage of initial yield, if the loading continues to
increase, the rockbolt element deforms according to tangent modulus ET. The rockbolt
will be broken by tension if the accumulated plastic strain is beyond the predefined
extreme strain εext as shown as point M. If the element is unloaded (see Figure 4-4(b)),
the rockbolt will deform with the elastic modulus of Eb. These parameters are easily
derived from the tensile experiments of rockbolt.

(a) (b)

Figure 4-4 Stress-strain curves for the elastic, linear hardening rockbolt material
(modified after Owen and Hinton, 1980)

4.2.4. Bond behaviour

The bond between rock and rockbolt is represented by a normal spring kn and a shear
spring k, as shown in Figure 4-1. The bond slip model during relative displacement is

87
Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

following the trilinear model as shown in Figure 4-5. Calculation of the relative
displacement or slip is based on the analytical solution in Section 4.2.2. Under a small
slip distance, the bond stress is increasing with the slip distance in a ratio of k1. Once
the bond stress achieves the maximum bond stress, τmax, the bond stress will decrease
with the slope k2. Once the bond stress is reduced to the residual bonding stress, τr, the
bonding stress keeps nearly constant if the slip continues. The maximum bonding
strength (τmax) and residual bonding strength (τr), are functions of the confining
pressure on the rockbolt element with respect to the normal spring kn. In a DDA block
system, the confining pressure is calculated as a stress component of the rock block
with direction normal to the rockbolt extension direction. By adopting various bond
slip models, different types of rockbolt elements (i.e., CMC, CFC and DFMC) could
be simulated. Figure 4-6 shows four types of rockbolt which will be discussed in the
later sections.

Figure 4-5 Trilinear model used to present bonding behaviour

Figure 4-6 Numerical models for four types of rockbolts (Nie et al., 2014a)

88
 Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

4.2.5. Shear behaviour

When a fully grouted rockbolt in rock is subjected to shear loading, two plastic hinges
might form at both sides of the joint plane as shown in Figure 4-7. By assuming the
intersection between the rock joint and the rockbolt element at the middle of the
rockbolt elements, the length of the rockbolt element through rock joint is 2le where le
is hinge length of rockbolt. If there is no experimental data, the hinge length le is often
taken as 1.0 ~ 2.0 Db (Gerdeen et al. 1981; Grasselli 2005; Jalalifar and Aziz 2010).
Under the action of shear loading, the axial load N, shear load Fs and bending moment
M are generated in the fully grouted rockbolt.

Figure 4-7 Simplified rockbolt element subjected shear movement

An example of a rockbolt through a rock joint with direction angle of θ, and incline
angle between rock joint and rockbolt of β is presented in Figure 4-10. The analytical
solution to consider the effects of combined tension and shear loading onto the yield
performance of rockbolt is given by Li et al. (2015) and shown as:

 le   
 0 0   
 Eb Ab N   0  U 
 le3 l 2
le  o
   q0le 4 2 
 q0le   
b

 0  e   Fso       Us 
(4.8)
 3Eb Ib Gb Ab 2Eb Ib     360Eb Ib 12Gb Ab   
Mo   
 le2 le  q0le3   o
 0  
 2Eb Ib Eb Ib   60Eb Ib 

where: No – axial load,

Fso – shear load,
Mo – bending moment,
Gb – the shear modulus of the rockbolt,

89
Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

Ib – the inertia moment of the cross section of the rockbolt element,

κ – a concentration coefficient of the shear stress distribution at the cross
section which is equal to 4/3 of the solid cross-section,
Ub – axial extension,
Us – lateral deflection, and
αo – deflection angle.

For a symmetrical homogenous beam element with very small diameter, the bending
moment Mo could be neglected. The force equilibrium at the rockbolt gives axial load
No and shear load Fso at a point O as:
Eb Ab
No  Ub (4.9)
le

240 Gb Ab Eb2 I b2  40Gb2 Ab2 Eb I b le2

Fso  U (4.10)
 6 Eb Ib le  Gb Able3 13Gb Able3  30 Eb I b  s

The relationship of the rockbolt deformation along its tangential and axial directions is
written as (Li et al., 2015):

sin  sin  (4.11)

Us  Ub  Ub
cos    o  cos  o sin 

where: Us and Ub – the rockbolt deformations along its tangential and axial directions,
respectively.

The force equilibrium on point A as shown in Figure 4-10 gives axial load NA, shear
load FsA and bending moment MA at point A, as:

Eb Ab
NA  Ub
le (4.12)

FsA  K sU s
(4.13)
M A  FsAle
(4.14)

where: Ks is shear stiffness of the rockbolt element which could be calcualted by:

240 Gb Ab Eb2 I b2  40Gb2 Ab2 Eb I ble2 (4.15)

Ks =
 6 Eb Ib le  Gb Able3 13Gb Able3  30 Eb Ib 

90
 Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

The increase of axial load ∆NA, shear load ∆FsA and bending moment ∆MA at the
rockbolt element in a time-step could be given as:

Eb Ab
NA  U b (4.16)
le

FsA  K s U s (4.17)

M A  FsAle (4.18)

The increase of the normal stress induced by bending moment acting at point A of
rockbolt element ∆σM is:

M A Db (4.19)
 M 
2I b

Combining Eqs. (4.2) and (4.19) gives the normal stress increase of rockbolt acting by
combined tension and shear loading as:

NA M A Db (4.20)
 A  
Ab 2Ib

Figure 4-8 A free body diagram of a rockbolt installed through a joint

For short span beams carrying large concentrated load, the reduction from shear is
significant (SSJ 1996). The Von Mises yield criterion is usually used to describe the
short span beam and shown as (see Figure 4-9):
91
Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

2 2
  A Ab   FsA  (4.21)
      1
 N y   Fs,max 

where: Ny – yield axial load of the rockbolt element in pure tension, and Fs,max  N y 3.

It is assumed that the shear force and moment of rockbolt element do not increase with
respect to the increasing of shear displacement once the rockbolt element is yielded.
The normal stress versus axial strain is still linear with tangent modulus of ET. As
shown in Figure 4-9, the yield criterion is of the short span beam is,

2 2
  A Ab   Fs ,y  (4.22)
      1
 Nf   Fs,max 

where: Fs,y – the yielded shear load of a rockbolt element, and

Nf – ultimate axial load of rockbolt element in pure tension.

Figure 4-9 Loading state judgement using shear stress versus axial load curves

The variations of point loads to the rock block induced by the restrain of rockbolt
under the combined tension and shear loads at a time step could be calculated as:

 Fx   cos   sin   (4.23)

 F  =Pi   +FsA  
 y -sin   cos 

The point loads as shown in Eq. (4.23) will be added to the submatrix of point loading
as shown in Eq. (2.37) to simulate the reinforcement effect of rockbolt to the rock

92
 Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

blocks. An example of the rockbolt element geometry in 2D-DDA is shown in Figure

4-10.

Figure 4-10 An example of a rockbolt element in 2D-DDA

4.2.6. Programming

The development rockbolt model used in 2D-DDA requires several input parameters
as shown in Table 4-1. It includes four parts:
A. Rockbolt system installation parameters,
B. Axial behaviours,
C. Bond behaviour, and
D. Shearing behaviour.

The parameters are usually determined by best fitting experiments tests, such as pull-
out tests. The default values are also listed based on previous experimental tests (Rong
et al., 2004; Stillborg, 1986).

The newly developed rockbolt model is integrated with 2D-DDA program to apply the
restrains for rock block movements. By considering axial behaviour, bond behaviour
and shear behaviour, the restrains are calculated in each time step, as shown in Figure
4-11. Based on the calculations, the proposed model could present four major failure
modes of rockbolt: (1) decoupling along the interface, (2) tensile failure, (3) faceplate
failure and (4) shear failure. Once type (2) or/and (4) failure occurs, the force restrains
are not applied to the rock blocks any more.

93
Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

The changes made in flowchart of 2D-DDA codes are shown in Figure 4-12. The
rockbolt models in tunnels start to apply restrains to the rock blocks movement if the
pre-defined time or step is achieved (sub-function dccable). It is usually activated after
excavation or core replacement functions (sub-function df29). There could be several
rockbolt groups which are installed in different time/step. The restrains made by
rockbolts are designed as point loads applying on rock blocks (sub-function df15c).
After each round of open-close iteration, the slip distances are updated (sub-function
df25), followed by the calculation of the restrained loading (sub-function dfcable)
which is going to be applied in the next time step.

Table 4-1 Rockbolt model properties

Category Input parameters Unit Default

A. Rockbolt Installation time/step step or time -

system
installatio
Rockbolt type - Fully rockbolt rebar
n

Elastic modulus, Eb Pa 200e9

B. Axial Yield stress, σy Pa 382e6

behaviour Tangent modulus, ET Pa 4e9

Extreme strain, εext - 15%

1st bond stiffness k1 Pa/m 54e9

2nd bond stiffness k2 Pa/m 6e9

C. Bond 3rd bond stiffness k3 Pa/m 5e6

behaviour
Max. bonding strength (no
Pa 5.24e6
confining effect), τmax

Residual bonding strength, τr Pa 0.5e6

D. Shearing Shear stiffness, Ks N/m Eq. 4.15

behaviour Ultimate shear force, Fs,max N Eq. 4.21

94
 Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

Geometry
Installation Geometry parameters:
time/step Rockbolt model Incline angle (β),
achieved Hinge length (le),
Diameter (Db), etc.

Rock movements, Ur

Loads in rockbolt element

Nodal axial Shear stiffness, Ks
displacement, Ub

Nodal shear
Update block contacts and movement

displacement, Us

Bond stress, τ

Variation of Variation of
axial load, P shear load, Fs

No
Yield?
Elastic stage
Yes

No Failed? Yes

Apply force No force

restrains restrains

Figure 4-11 Flowchart of the calculation of rockbolt forces

95
Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

Preparation:  Time-step:
df01 - Input df03 – Current time/step;
block dccable – Rockbolt element initialization if installation
information; time/step is achieved;
df02 - Physical
information;  Contact detection
df04 – Contacts judgment by distances;
df05 – Contacts judgment by angles;
df06 – Contacts transfer
df07 – Contacts initialization
df08 (di08) – Positions of non-zero storage
df09 – Time interpolation
 Build contact matrix
df10 – Initiation of contact matrix
df11 – Submatrix of inertia
df12 – Submatrix of fixed points
df13 – Submatrix of block stiffness
df14 – Submatrix of block stress
df15 – Submatrix of point loads
df15c – Submatrix of rockbolt induced point loads
df16 – Submatrix of volume force

 Open-close iterations:
df18 – Add and/or remove contracts
 Solves global equilibrium equations:
df20 (di20) – Solve global equilibrium equations

df22– Contact re-judgment after iteration

df24 – Calculation of displacement ratio

 Block system update and output the results

df25 – Compute step displacement (update the
rockbolt coordinates)
dfcable – Calculation of rockbolt element induced
point loads
df28 – Save files for post-processing
df29 – Do excavation/ core replacement if
time/step is achieved


Post-process:

Figure 4-12 Revised flowchart of 2D-DDA codes by integrating with rockbolt models
96
 Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

4.3 Verification of the rockbolt model by experimental results

4.3.1. Verifying bond behaviour by pull out test

The pullout model tests conducted by Rong et al. (2004) are adopted to evaluate the
accuracy of the present fully grouted rockbolt model. The maximum applied pullout
load in the test was 300 kN, inducing only necking at the head of rockbolt. The
dimensions and boundaries of the model in 2D-DDA is shown in Figure 4-13(a). The
rockbolt is divided into 10 elements each of which is 0.1 m long. The diameter and
length of the rockbolt are 32 mm and 1 m, respectively. The parameters of the trilinear
bond-slip model presented by Ma et al. (2014) are used: τmax = 5.24 MPa, τres = 0.524
MPa, δ1 = 0.096 mm and δ2 = 0.816 mm as shown in Figure 4-13(b). The elastic
modulus of the steel rebar Eb = 210 GPa before yield and ET = 4.2 GPa after yield as
shown in Figure 4-13(c). More details of the parameters used in the 2D-DDA model
are summarized in Table 4-2.
1.20m

P
0.1 m

1.0 m

(a) (b) (c)

Figure 4-13 The parameters of rockbolt model used in 2D-DDA (a) Dimensions and
boundaries of the model, (b) Bond-slip model and (c) Axial model of the
rebar

The axial loads versus displacement curves from the 2D-DDA simulation and model
test results are shown in Figure 4-14. It can be seen that the simulation results match
reasonably well with the model test data and the analytical method proposed by Ma et
al., (2014). In the present simulation, the first three nodes of rockbolt near loading
position are deboned when the axial force is 300 kN, and the extreme strain is
achieved at an axial load of 340 kN when five nodes are decoupled. The axial force
and bond stress distributions along the rockbolt from present simulation and model test
results are compared and shown in Figure 4-15(a) and (b), respectively. The pullout

97
Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

loads used in the simulation are 50 kN, 100 kN, 200 kN and 300 kN, respectively. It is
found that good agreements between the numerical results and the proposed model has
reasonable accuracy for the simulation of fully grouted rockbolt problems.

Table 4-2 Parameters used in 2D-DDA to simulate the pullout test

Item Parameter Value Item Parameter Value

Dynamic factor 0 (static) Diameter, db (mm) 32
Step max. displacement Tangent modulus, 4.2
0.001
ratio, g2 ET (GPa) (≈Eb/50)
Upper limit of time Yield strength, σY
0.0005 360
interval, g1 (MPa)
DDA setting

Extreme strain, εext

SOR factor 1.4 Bolt 1.6
(%)
Normal spring stiffness, Maximum bond
3 × 109 5.24e6
kn (N/m) strength, Fs,max
Residual bond
kn/ ks 2.5 0.524e6
stress, Fs,res
Block unit weight, ×103 Sliding bond node,
26 1e5
kN/m3 k3 (Pa/m):

400
350
300
Axail force, kN

250 End of experiment

Displacement 0.7mm
200 Yield, not broken,
not apprearent decoupling
150
100 Present simulation
Model test (Rong et al, 2004)
50
Analytical study( Ma et al., 2010)
0
0.0 0.5 1.0 1.5 2.0
Displacement of rockbolt faceplate, mm

Figure 4-14 Comparisons of the axial load versus displacement curves from present
study and model test conducted by Rong et al.(2004)

98
 Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

350
300 Model test (Rong et al., 2004) P=300 kN
Present simulation
250
Axial force, kN
200 P=200 kN

150 P=100 kN
100 P=50 kN

50
0
0 200 400 600 800 1000
Distance to the rockbolt end, mm
(a)

6
Analytical soluation (Ma et al., 2010)
5
Present simulation
Bond stress, MPa

P=300 kN
4
P=200 kN
3
P=100 kN
2 P=50 kN

0
0 200 400 600 800 1000
Distance to the rockbolt end, mm
(b)

Figure 4-15 Comparisons of (a) the simulated axial force distribution with the results
of model test conducted by Rong et al. (2004) and (b) the simulated bond
stress distribution with the analytical solutions by Ma et al. (2010)

4.3.2. Verifying shear behaviour by shear test

The shear behaviour of fully grouted rebar has been tested in laboratory (Chen and Li,
2015). The corresponding 2D-DDA model are shown in Figure 4-16(a). The bottom
block is fixed, while the upper block can only move along the horizontal direction. An
rockbolt is installed to connect two rock blocks with angle β. A faceplate is attached at
the top end, while the other end at the bottom block is free. According to the block
boundary, the rockbolt is divided into two segments equally in length. In each segment,
the rockbolt model is divided into several elements each of which is 0.1 m except the

99
Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

elements next to the joint where the hinge length le is used to locate the rockbolt/rock
nodes. The bond-slip model is assumed as a trilinear model shown in Figure 4-16(b).
An elastic, linear strain-hardening steel model is adopted to simulate the rockbolt
material (see Figure 4-16(c)). The detailed parameter settings of rock, rockbolt and
DDA calculations are listed in Table 4-3. Three shear tests with shearing angle β = 90°,
120° and 140° are simulated. Another simulation of pull-out test is also conducted for
comparison purpose. As shown in Figure 4-17, the rockbolt fail at the elements
adjacent to joint in all four cases. The curves of the total loads versus displacement
from the 2D-DDA simulation results are shown in Figure 4-18. It should be noted that
the total force is defined as the squared sum of normal and shear forces, while the total
displacement is the horizontal movement of the upper block (Chen and Li, 2015).

Bond stress, MPa Axial stress, MPa

691 ----------------
5.24 --- 573 --
0.210 -----------

0.52 ----------------
0.024 ------

0.224 ------
Slip, mm Strain, %
0.816 ----

(a) (b) (c)

Figure 4-16 DDA model setting of the shearing tests (a) DDA model, (b) bond–slip
model and (c) material model

Table 4-3 Parameters setting in the simulation of shear tests

Item Parameter Values Item Parameter Values
Elastic modulus Er, GPa: 30 Diameter Db, mm 20
Rock

Poisson ratio, v 0.25 Elastic modulus Eb, GPa 210

Unit weight, ×103 kN/m3 26.0 Yield axial force, kN 180
Rockbolt

Step max. displacement

0.0004 Ultimate axial force, kN 217
ratio
calculation
DDA

Upper limit of time 30

0.0002 Hinge length le, mm
interval (~1.5Db)
SOR factor 1.4 Shear stiffness Ks, MN/m 66.3

100
 Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

Elastic

Yield

Broken

(a) (b) (c) (d)

Figure 4-17 Failures of the rockbolts in numerical simulations (a) Pull-out test and
shear tests at angles of (b) β = 90°, (c) β = 120° and (d) β = 140°

250

200
Total force, kN

150 Pullout (DDA)

Pull-out test (Chen and Li, 2015)
β=140° (DDA)
100 β=140° shear test (Chen and Li, 2015)
β=120° (DDA)
50 β=120° shear test (Chen and Li, 2015)
β=90° (DDA)
β=90° shear test (Chen and Li, 2015)
0
0 20 40 60 80 100
Total displacement, mm

Figure 4-18 Comparisons of load-displacement curves between experiments and

numerical simulations

Table 4-4 lists the comparison between the numerical and experiment results. In terms
of the yielding and failure forces, the simulation results of pullout test provide good
agreement with the experiment results as the errors are 1.0% and 0.7% respectively.
The total loads of the rockbolts either at yielding or at breaking moments of shear tests

101
Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

are lower than those in the pullout test. In the simulation of shear test with a shear
angle β = 90°, the rockbolt yields at the lowest force of 115 kN. The error between the
numerical result and the experiment result is about 2.7%. The rockbolt fails at the
lowest total load. The error is about 23.9% compared with the experiment result. Both
the yield and fail forces are increasing with the increase of shear angles. In cases of β
= 120° and β = 140°, the errors of resulted yielding forces in the simulations of shear
tests are 1.9% and 5.0%, respectively, while those of failure forces are 17.5% and
11.6%, respectively. The results show that the proposed rockbolt model could catch
the load variation during pulling and shearing. However, the simulations have
limitations that they do not well catch the slopes of the increasing force with respect to
the displacement in elastic. As listed in Table 4-4, the largest error occurs in the
simulation of shear test with a shear angle β = 90°. The error becomes less if the shear
angle increases. This is mainly because the displacement caused by yielding of grout is
not considered in the current model.

Table 4-4 Comparisons between numerical simulations and the experiment results
Items A. Pull-out B. β = 90° C. β = 120° D. β = 140°
Experiment 181.86 118.13 149.45 165.92
Yielding
axial force, DDA results 180.05 114.93 146.64 157.60
kN
Error, % -1.0% -2.7% -1.9% -5.0%
Experiment 215.38 196.70 205.49 209.34
Ultimate
axial force, DDA results 216.97 149.66 169.63 185.07
kN
Error, % 0.7% -23.9% -17.5% -11.6%
Rotation
DDA results 0.0 18.5 14.3 10.5
angle α, °

As shown in Figure 4-19(a), the shear force varies with the shear displacement in
different response to the shear angle. The shear force in the case of β = 90° is larger
than the other two cases, causing a larger rotation of rockbolt element at joint (see
Figure 4-19(b)). At the failure point, the rockbolt element rotated about 27° in the case
of β = 90° and about 13.9° in the case of β = 140°. Therefore, the deduction of the load
carrying compacity during shearing is mainly induced by the combination work done
by the axial load and shear load at the joint.

102
 Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

80 30
β=90° (DDA)
25
60 β=120° (DDA)

Rotation, degree
Shear force, kN
20
β=140° (DDA)
40 15
β=90° (DDA)
10
20 β=120° (DDA)
5
β=140° (DDA)
0 0
0 10 20 30 0 10 20 30
Total displacement, mm Total displacement, mm
(a) Shear force histories (b) Rotations

Figure 4-19 Simulated shear behaviour of rockbolt elements at joints

4.3.3. Verifying axial behaviour by pull out tests

As shown in Figure 4-20(a), the pullout test done by Chen and Li (2015) are adopted
to verify the axial behaviour of the developed rockbolt model. To simulate this kind of
rockbolt elements, different bond slip models are assigned to them. As shown in
Figure 4-20(b) and (c), the models of fully grouted rebar and D-bolt are considered
respectively. The 2m rockbolt is divided into two segments according to the joint
between two rock blocks. Each segment has 7 rockbolt elements (8 pairs of nodes)
each of which is 0.14 m long. There is one more rockbolt element connecting two
segments with length of 0.04 m (assuming the potential hinge length is 0.02 m).

For the fully grouted rebar, the nodes are all assigned by anchored nodes. However,
for the D-bolt, the nodes are assigned by sliding nodes as the smoothing parts except
the 2nd, 4th and 12th nodes from the rockbolt end which are assigned by special
anchored nodes with higher bonding stiffness and bond strength than the anchored
nodes. Therefore, there are four kinds of nodes in the model: (1) fix node where the
faceplate is attached, (2) sliding node as a rockbolt element featured with low bond
stiffness, (3) anchored node is a rockbolt element with trilinear bond behaviour as
shown in Figure 4-13(b), and (4) special anchored node is a rockbolt element with
strong bond behaviour. The detailed parameter settings are shown in Table 4-5.

103
Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

(a) Simplified pullout test model (after Li et al., 2010)

(b) Model of the fully grouted rebar

(c) Model of the D-bolt

Figure 4-20 Sketches of the pull-out test and the various rockbolt models

Table 4-5 Parameter settings in pullout models

Rockbolt parameters Bonding models
Diameter, db (mm) 20 Sliding nodes k1 = 1 MPa/mm
k1 = 54 MPa/mm
Young’s modulus, Eb (GPa) 210
Anchored nodes k2 = 6 MPa/mm
1 (Rong et al, 2004) τmax = 5.24 MPa
Tangent modulus, ET (GPa)
(≈Eb/200) τr = 0.524 MPa
Yield strength, σY (MPa) 520 Special anchored k1s = 80.0 MPa/mm
nodes (for D-bolt,
Extreme strain, εext (%) 16 assumed) τmax = 20 MPa

104
 Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

Figure 4-21 shows a comparison of the simulated load-deformation curves and the
experimental results from the pullout tests. It indicates a reasonable agreement of the
axial behaviours between the simulation and experiment results. The differences of the
initial slope of the curves may be due to the assumed linear stress-strain relationship
adopted in the numerical models.

250

200
Axisl laod, kN

150

100
Rebar pullout test (Chen and Li, 2015)
Simulation rebar pullout
50 D-bolt pullout test (Chen and Li, 2015)
Simulation D-bolt pullout
0
0 25 50 75 100 125 150
Displacement, mm

Figure 4-21 Comparisons of load-deformation curves between experiments and

numerical simulations of pullout tests

The comparison shown graphically in Figure 4-21 indicates the fully grouted rebar
model reaches to high load carrying capacity earlier than the D-bolt, but fails at a
much lower displacement. By examining the axial stresses and bond stresses along the
rockbolts (see Figure 4-22 (a)), the loads are concentrated in the central 4 elements
when a pullout load of 100 kN is applied. Before failure, the central 4 elements are
yielded (see Figure 4-22 (b)), and decoupling process has helped to defuse the load
laterally. This indicates the deformation occurs in the central 4 elements mainly. On
the other hand, the rockbolt elements between two special anchored nodes (8 elements)
are sustaining the pullout load in different loading scales (such as pullout load of 100
kN and 200 kN in Figure 4-22(c) and (d) separately). It indicates the D-bolt is equally
loaded in these elements, causing larger deformability. Thus, the deformability of the
D-bolt is relying much on the performance of the special anchored points.

105
Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

720 Bonding limits 6

Yield limits
480 4

Bond stress, MPa

Axial stress, MPa

240 2

0 0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2
-240 From bolt end to faceplate, m -2

-480 -4

Join
Bonding limits
-720 -6

(a) Fully grouted rebar at pullout load of 100 kN

720 6
Bonding limits
Yield limits
480 4

Bond stress, MPa

Axial stress, MPa

240 2

0 0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2
-240 -2
From bolt end to faceplate, m
-480 -4
Joint

Bonding limits
-720 -6

(b) Fully grouted rebar at pullout load of 200 kN

720.0 21.0
Yield limits
480.0 14.0
Axial stress, MPa

Bond stress, MPa

240.0 7.0

0.0 0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2
-240.0 -7.0
From bolt end to faceplate, m
-480.0 -14.0
Joint

-720.0 -21.0

(c) The D-bolt at pullout load of 100 kN

106
 Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

720.0 21.0
Yield limits
Axial stress, MPa 480.0 14.0

Bond stress, MPa

240.0 7.0

0.0 0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2
-240.0 From bolt end to faceplate, m -7.0

-480.0 -14.0

Joint
-720.0 -21.0

(d) The D-bolt at pullout load of 200 kN

Figure 4-22 Stress distributions along the length of rockbolt models

4.4 Parameters for rockbolt design

Table 2-4 has proposed the main parameters for a rockbolting design. In the following,
numerical simulations are conducted to study the influences of the design parameters
on the load-displacement of a single rockbolt under pulling/shear condition, such as,
the rockbolt type, the embedded length, the installation angles, the rock properties, etc.

4.4.1. 2D-DDA models

As shown in Figure 4-23, a rockbolt gets through two deformable rock blocks. Block
B is fixed at the left-side edge and can only deform horizontally by the constrains of
vertical rollers. Block A can move horizontally by applying horizontal movement at its
right-side edge, or have vertical displacement if there are any vertical movements
applied on its top-side edge. The co-edge between blocks A and B is set as a
frictionless joint. A fully grouted rebar is simulated using the developed rockbolt
model. A faceplate is attached as the right end, while its left end is free to move. The
nodes are numbered from the faceplate to the free end as 1 to n (n is varying in the
parameter studies according to the different embedded length adopted). The
performance of the rockbolt is tested by considering four different embedded lengths
conditions, namely, l0 = 0.5 m, 0.8 m, 1.0m, 1.2 m and 1.6 m. The rockbolt could be
pulled if Block A moves horizontally or be sheared at angle β = 90° if Block A moves

107
Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

vertically. The deformability of two rock blocks are also considered in two conditions,
such as Er = 30 GPa and Er =0.3GPa. The elastic modulus of the rockbolt material is
assumed as Eb = 210 GPa before yielding, ET = 1 GPa after yielding. The yield force
and the extreme plastic strain are 180 kN and 15%, respectively.

Figure 4-23 2D-DDA model of a rockbolt pullout/shear test

As shown in Figure 4-24, for a rockbolt model with an embedded length of 1.0 m in
Block B, the simulated yield pulling force (Fy) and ultimate tensile force (Fult) at the
joint are 180 kN and 217 kN, respectively. Block A has been moved 23mm towards
right when the rockbolt failed. In the following, the nodal force distributions along the
rockbolts are examined to understand the load-displacement curve of the simulated
tests.

108
 Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

250
l0 =1.0 m, pull-out l0 =0.5 m, pull-out
Er = 0.3GPa
200

150
Axial load, kN

l0 =1.0 m, pull-out, Ma's model

100 l0 =0.5 m, β=90◦shear l0 =1.0 m, β=90◦shear

50 l0 =1.0 m, pull-out, Model A

l0 =1.0 m, pull-out, Model B

0
0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0
Displacement, mm

250

l0 =0.5 m, pull-out
200
Axial load, kN

150 l0 =0.5 m, β=90◦shear

100
l0 =1.0 m, β=90◦shear

50 l0 =1.0 m, pull-out, Model A

l0 =1.0 m, pull-out, Model B

0
0.0 2.0 4.0 6.0 8.0
Displacement, mm

Figure 4-24 Comparisons of the simulated load-displacement curves

4.4.2. Effects of the types of rockbolt

In Chapter 3, the bond-slip model of the rockbolt elements have been studied. In fact,
different types of rockbolt element could be represented by using various bond-slip
models at the rockbolt and rock interface. Section 4.3.3 has showed an example of the
application of the rockbolt model to simulate the fully grouted rebar and the D-bolt.
Here, two kinds of fully grouted rebar are adopted to study the influence of the bond-
slip models on the load transfer mechanism of a rock/rockbolt system. The two bond-
slip models from previous section (see Fig. 3-11(b)) are adopted, including Model A

109
Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

as the simplified bond – slip model of a rockbolt element with rib angle α = 45°, rib
spacing sb = 17.8 mm and confining pressure p = 2 MPa, and Model B as the bond –
slip model of a rockbolt element with rib angle α = 45°, rib spacing sb = 35.8 mm and
confining pressure p = 2 MPa. The bond stiffness of Model B at stage 1 and stage 2 are
lower than those of Model A. Both models have the same maximum and residual bond
stresses. Another trilinear bond-slip model proposed and verified by Ma et al. (2014) is
also included in the simulation (short as Ma’s model) for comparison. The bond-slip
models are shown in Figure 4-25. For the rockbolt material, the elastic modulus is
assumed as Eb = 210 GPa before yielding, ET = 1 GPa after yielding. The yield force
and the extreme plastic strain are 113 kN (≈ 360 MPa) and 10%, respectively.

6
Reference Model by Ma et al. (2014)
5
Bond stress, MPa

Model B, α=45°, sb=35.6mm and p=2MPa

4 Model A, α=45°, sb=17.8mm and p=2MPa

3
2
1
0
0 0.2 0.4 0.6 Slip displacement,
0.8 1
mm
Figure 4-25 Simplified bond-slip models

The obtained axial load versus displacement curves of the pull-out tests are shown in
Figure 4-24. All three rockbolt models have similar load capacity in the elastic stage.
The rockbolt with Ma’s model reaches the yielding point at first, followed by the
rockbolt with bond-slip Model A and then that with Model B. Thus, the rockbolt with
Ma’s model is more rigid than the other two. Considering the deformability of the
rockbolt, the simulation result of the rockbolt using Ma’s model has the lowest
displacement when the rockbolt has broken due to tension. The displacements at the
tensile broken of the rockbolts using Model B and Model A are almost the same.

The distributions of the nodal bond stresses along the rockbolt at the pullout loading of
108 kN (0.6 × yielding force) are selected, as shown in Figure 4-26(a), to analyze the
load transfer between rockbolt and rock blocks. It can be seen that there are no

110
 Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

decoupling occurs along the rockbolts simulated using Model A and Model B. One
decoupled node is found at the rockbolt element near the joint when the rockbolt
element is simulated using Ma’s model. The nodal bond force distributions of the
bonded nodes along the rockbolt could be curve fitted using exponential model y = A
× exp(Bx) (Li and Stillborg, 1999). The nodal bond force is calculated as the product
of bond stress multiplied by the average length of rockbolt element nearby. It shows
the smaller the bond stiffness k1, the larger the values of the parameter A. The
exponential models of the nodal axial forces along the rockbolt are presented in Figure
4-26(b). It also shows the parameter A is increasing with decreasing of the bond
stiffness k1. The load transfer capacity along the rockbolt is influenced by the bond
stiffness k1 and the maximum bond strength τmax. The decoupling process will extend
inside of the rockbolt if the pull-out loads increases. Before the tension failure, the
distributions of the axial forces and the bond stresses along the rockbolt elements are
presented in Figure 4-27(a) and (b), respectively. It shows there are 2 decoupled nodes
when analyzed using the rockbolt with Ma’s model, and 4 decoupled notes when using
Model A, and 3 decoupled notes when using Model B. The curve fitting parameters of
the exponential models are summarized in Table 4-6, presenting the mechanical
coupling at the interface between rock and rockbolt for engineering practice using
similar rockbolt elements.

500

Decoupled
Nodel bond force, kN/m

400 node

Ma's model
300
y = 438.94e6.7737x
Model A R² = 0.9366
200 Model B y = 17918e2.3219x
y= 27861e1.6392x R² = 0.9768
R² = 0.9688
100

0
0.0 0.2 0.4 0.6 0.8 1.0
Distance from rockbolt end, m
(a)

111
Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

100

80 Ma's model
Nordal axial force, kN

y = 41.925e7.3655x
R² = 0.9947
60 Model A
y = 2329.4e3.5656x
Model B R² = 0.9778
40
y = 3694.4e3.1199x
R² = 0.964

20

0
0.0 0.2 0.4 0.6 0.8 1.0

Distance from rockbolt end, m

(b)

Figure 4-26 (a) Nodal bond force and (b) nodal axial force along the rockbolts at the
pull-out loading of 108 kN

500
Ma's model

y = 745.39e7.9449x
400
R² = 0.9935

Model B
Nodal bond force, kN/m

Decoupled
300 y = 66320e1.5308x nodes
Model A R² = 0.9786

y = 49136e2.1099x
200 R² = 0.9832

100
Decoupled
nodes

0
0.0 0.2 0.4 0.6 0.8 1.0
Distance from rockbolt end, m

(a)

112
 Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

200

Nodal axial force, kN

150 Model A
y = 6894.2e3.3314x
R² = 0.9599
Model B
100 y = 9241e3.0083x
R² = 0.9523

50
Ma's model
y = 119.63e7.4819x
R² = 0.9959
0
0.0 0.2 0.4 0.6 0.8 1.0
Distance from rockbolt end, m

(b)

Figure 4-27 (a) Nodal bond force and (b) nodal axial force along the rockbolts at the
extreme loading

4.4.3. Effects of the embedded length

The effects of the embedded lengths are examined using the pull-out simulations.
Figure 4-24 compares the load-displacement curves in cases of l0 = 1.0 m and 0.5m.
The displacement at tension fail in the case of l0 = 0.5 m is only 70% of that of l0 = 1.0
m indicating a reduction in the deformability of the rockbolt if a shorter embedded
length is adopted. This could be explained using the nodal axial loading distributions
along the rockbolt. Figure 4-28(a) presents the loading conditions when the pullout
force is 104 kN. As the head of rockbolt are fixed by a faceplate, there are no much
differences in the load distributions along the rockbolt elements in Block A. However,
for the rockbolt model with the embedded length l0 = 0.5 m, all the anchorage points
of the rockbolt in Block B are activated, transferring the axial loads through bonds to
the rock block B (see Figure 4-29a). When the rockbolt is close to failure, the
anchorage points in Block B are easily decoupled at the interface (see Figure 4-29(b)).
The bonding conditions of the rockbolts become strong in the case of l0 ≥ 1.2 m. Thus,
the reinforcement effort provided by a rockbolt in short embedded length (l0 < 1.2 m)
should be re-estimated.

113
Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

Table 4-6 Parameters of three bond-slip models and the fitted exponential models

Items Ma’s model Model A Model B

54 MPa/mm 9.0 5.4

Bond stiffness at stage 1, k1
Bond-slip model parameters

MPa/mm MPa/mm

Max. bond strength, τmax 5.24 MPa 3.34 MPa 3.4 MPa

6 MPa/mm 2.6MPa/mm 2.2

Bond stiffness at stage 2, k2
MPa/mm

Residual bond stiffness, τres 0.524 MPa 2.23 MPa 2.28 MPa

0.1 MPa/mm 0.2 0.2

Bond stiffness at stage 3, k2
MPa/mm MPa/mm

Numbers of the decoupled

1 0 0
nodes in Block B
at pull-out loading = 108 kN

A = 438.9 A = 17918 A = 27861

Exponential models:
y = A ×exp(Bx)

Bond forces (bonded

B = 6.774 B = 2.322 B = 1.639
rockbolt elements)
R2 = 0.937 R2 = 0.977 R2 = 0.969

A = 41.9 A =2329.4 A = 3694.4

Axial forces B = 7.365 B = 3.566 B = 3.120

R2 = 0.995 R2 = 0.978 R2 = 0.964

Numbers of the decoupled 2 4 3

nodes in Block B
at pull-out loading = 108 kN

A = 745.4 A = 49136 A = 66320

Exponential models:
y = A ×exp(Bx)

Bond forces (bonded

B = 7.945 B = 2.110 B = 1.531
rockbolt elements)
R2 = 0.994 R2 = 0.983 R2 = 0.979

A = 119.6 A = 6894.2 A = 9241

Axial forces B =7.482 B = 3.331 B = 3.008

R2 = 0.996 R2 = 0.960 R2 = 0.952

114
 Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

Block B Block A
120

l=1.6m
l0 = 1.6 m 100

Nodal axial load, kN

l=1.2m
l0 = 1.2 m
80
l=0.8m
l0 = 0.8 m

Joint
60
l=0.5m
l0 = 0.5 m
40

20

0
28 25 22 19 16 13 10 7 4 1

Node number

(a)

Block B Block A
250

l0l=1.6m

Nodal axial load, kN

= 1.6 m
200
l0l=1.2m
= 1.2 m
150
l0l=0.8m
= 0.8 m
Joint

l0l=0.5m
= 0.5 m 100

50

0
28 25 22 19 16 13 10 7 4 1

Node number

(b)

Figure 4-28 Simulated nodal axial loads along the rockbolt models when the tensile
loads at joint are (a) 60% of Fy and (b) Fult

115
Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

Block B Block A
400

Joint
l=1.6m
l0 = 1.6 m
300
l=1.2m
l0 = 1.2 m

Nodal bond force, kN/m

l0 = 0.8 m
l=0.8m 200
l0 = 0.5 m
l=0.5m
Node number 100

0
28 25 22 19 16 13 10 7 4 1
-100

-200

-300

-400

(a)

Block B Block A
400
l=1.6m
l0 = 1.6 m
Joint

300
l=1.2m
l0 = 1.2 m

Nodal bond force, kN/m

l0 = 0.8 m
l=0.8m 200
l0 = 0.5 m
l=0.5m
Node number 100

0
28 25 22 19 16 13 10 7 4 1
-100

-200

-300

-400
(b)

Figure 4-29 Simulated nodal bond forces along the rockbolt models when the tensile
loads at joint are (a) 60% of Fy and (b) Fult

116
 Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

4.4.4. Effects of the shear angle

The deduction of the load carrying compacity during shearing is mainly due to the
combination work done by the axial load and shear load at the joint, as discussed in
Section 4.3.2. By examining the nodal axial loads along the rockbolt, the load
distributions are almost the same as shown in Figure 4-28(a) and Figure 4-30(a) when
the rockbolts are behaving in elastic. However, the maximum axial load is observed at
an inner element from the joint in shear test when the rockbolts are yield. Compared
with Figure 4-28(b) and Figure 4-30(b), the axial load at the node at the right of the
joint is reduced significantly, and much less reinforcement could be provided by the
relative rockbolt element under shear conditions.

4.4.5. Effects of the rock deformability

The main reason why the deformability of a rockbolt reduces in soften rock mass
comes from the contribution of deformation in rock blocks. As shown in Figure 4-31,
at the initial loading, the deformation occurs in the movable rock block causing a
uniformly distributed axial stress along the rockbolt in Block A. At the same time, the
embedded end of the rockbolt is slightly loaded. Once the joint opens, the axial load of
the rockbolt at the joint starts to concentrate and activates the anchorages at the
embedded end. The failure occur at the joint, but compared with the axial load
distribution of rockbolt in rock mass with Er = 30 GPa, the deformability of rockbolt
reduces due to the initially deformation with the deformable rock blocks whose Er =
0.3 GPa (see Figure 4-24).

117
Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

100
l=1.6m
l0 = 1.6 m
80 l=1.0
l0 = 1.2 m
Nodal axial load, kN

l0 = 0.8 m
l=0.8m
60
l0 = 0.5 m
l=0.5m

40

Joint
20

0
-2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0
Along the bolt lenght, m

(a)

250
l=0.5m
l = 0.5 m
0
200
l=1.6m
Nodal axial load, kN

l0 = 1.6 m
l=1.2m
l0 = 1.2 m
150
l=0.8m
l0 = 0.8 m
100

50
Joint

0
-2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0
Along the bolt lenght, m

(b)

Figure 4-30 Simulated nodal axial loads in the shear tests of rockbolt with shear angle
β = 90° when the loads at joint are (a) 60% of Fy and (b) Fult

118
 Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

250

Joint
Axial load = 30 kN,
Er = 0.3 GPa
200
Axial load = 80 kN,
Er = 0.3 GPa

150 Axial load = 210 kN,

Er = 0.3 GPa

Axial load = 210 kN,

100 Er = 30 GPa

50

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
Rockbolt nodes

Figure 4-31 Axial loads variations in case of deformable rock blocks

4.5 Conclusions

In this chapter, a rockbolt model is developed in 2D-DDA based on the load transfer
theory. A rockbolt is divided into several segments according to rock block boundaries.
Each segment has a number of rockbolt elements. Three major behaviours of rockbolts,
including axial behaviour, bonding behaviour and shear behaviour, are considered in
this model. By varying these three major behaviours, different kinds of rockbolt could
be simulated.

To integrate the rockbolt model under 2D-DDA frame work, several sub-functions are
developed. The rockbolt applies their restrains as point loads in the system. There will
be limited or no restrains if the rockbolt is yielding or failed. In such a way, the
rockbolt could be installed in groups at different times/steps.

In summary, the proposed rockbolt models could present four major failure modes of
rockbolt, i.e., decoupling along the interface, tensile failure, faceplate failure and shear
failure. Verifications of the proposed rockbolt model have been carried out.

119
Chapter 4 Development of a rockbolt model in 2D-DDA and its verifications

(1) The bond behaviour is verified using pullout tests. The simulation results show
good agreements between the proposed rockbolt model and experiments results.
The rockbolt model can reveal the decoupling along the interface, tensile
failure of rockbolt elements, faceplate failure under pulling out load.
(2) By comparing with shear tests, the simulation results show reasonable
agreements in the reduction of reinforcement effects. The shear failure of the
rockbolt element at joint are presented.
(3) The axial behaviour is verified using pullout tests results. Two types of
rockbolts are simulated under pullout condition. The simulated load-
deformation curves show good agreements with the experimental ones. It also
confirms the larger deformability of the D-bolt compared with the fully grouted
rebar as a result of equally loaded rockbolt elements between special anchored
nodes.
(4) Based on the above verifications, it can be concluded that the rock-rockbolt
reaction could be simulated with the proposed rockbolt model. In the following
chapter, these verified rockbolt models are adopted to simulate rock-rockbolt
interaction in various rock conditions.
(5) The bond-slip model controls the load transfer between rock and rockbolt. By
fitting the nodal axial forces and nodal bond force distributions along the
rockbolt using the exponential model, the results show the parameters of the
exponential models are heavily influenced by the bond stiffness k1 and the
maximum bond strength τmax. The curve fitting parameters of the exponential
models can be used for engineering practice under similar conditions.
(6) Using the given bond-slip models in the previous studies, parameter studies
show that the anchorage the rockbolts become strong enough for the case of l0
≥ 1.2 m. On the other hand, the reinforcement effort provided by a rockbolt
with a short-embedded length (l0 < 1.2 m) should be re-estimated.
(7) The installation angle will influence the load-displacement of a rockbolt as the
combination work done by the axial load and shear load at the joint may reduce
the load carrying compacity of the rockbolt.

120
 Chapter 5 Simulation of the reinforced rock unit in jointed rock mass

SIMULATION OF THE REINFORCED ROCK

UNIT IN JOINTED ROCK MASS

The load carrying capacity of a rockbolt can be evaluated using reinforced rock
unit (RRU). Generally, the RRU is a model considering the rockbolt as a suspension
device transferring loads in rock strata. In jointed rock mass, the RRU has to be
considered with regards to the loading scale and the condition of rock mass
discontinuities. In this chapter, a parametric study has been carried out using the newly
developed rockbolt element in 2D-DDA. The results are used to the estimation of RRUs
in jointed rock mass and the corresponding loads in rockbolt element.

5.1 Introduction

The reinforced rock unit model is a modified rockbolt suspension theory by considering
the rockbolt as a suspension device transferring the weight of the weaker rock strata to
the stronger strata. As the rock at the end of the rockbolt may be fractured, the rockbolt
can only influence the rock mass located between the fracture plane and the head of
rockbolt (Li, 2006). This part of the rock is often called Reinforced Rock Unit (RRU)
(Lang and Bischoff, 1982; Choquet and Hadjigeorgiou, 1993). A typical example for a
rockbolt to supply local support in homogeneous rock mass is shown in Figure 5-1. The
boundary of the RRU of the rockbolt is usually assumed in a cone shape with the apex
angle of β, which is also denoted as the reinforcement angle (Hobst and Zajíc, 1983; Nie
et al., 2016). The maximum axial stress of the rockbolt is determined by the properties
of rock mass and the installation method as shown in Table 5-1. For the rockbolt
installed in a homogeneous rock mass in line grid with spacing of S, see Figure 5-1b,
the maximum axial stress of the rockbolts is also influenced by the interaction between
rockbolts as summarized in Table 5-1. Calculation methods for maximum axial stress

121
Chapter 5 Simulation of the reinforced rock unit in jointed rock mass

of the rockbolts installed in squared grid with spacing of S, see Figure 5-1c, are
summarized in Table 5-1. For the rockbolt installed in the irregular and dense irregular
jointed rock mass, the properties of joints also influence the maximum axial stress of
the rockbolt, see Table 5-1. However, to the best knowledge of the author, there are still
not enough studies to analyse the relationships between fully grouted rockbolt and
jointed rock mass.

(a) RRU of a rockbolt (b)RRU of a line of rockbolts (c) RRU of rockbolt grid

Figure 5-1 Sketches of RRUs of rockbolts in homogeneous rock mass (after Hobst and
Zajic, 1983)

Table 5-1 The maximum axial stress of the rockbolt installed in different rock mass
(modified after Hobst and Zajìc, 1983)
Type of Rock
Single rockbolt Line Grid Square Grid Notes
Mass
2 ult L2 2 2 ult SL 4 ult LS
Homogeneous = = = β = 90°
sf A sf A sf A

Irregular  L3 tan 2  S  L2 tan   LS 2

= = = β = 2φ
jointed 3s f A sf A sf A

Dense irregular  h L2 tan 2  S h L tan  4S h L tan 

= = = β = 2φ
jointed 3s f A sf A sf A

Note: τult − shear strength of rock mass;

φ − the friction angle of discontinuity;
σ − maximum axial stress of the rockbolt;
σh − lateral field stress;
L − the rockbolt length;
sf − the safety factor;
A − the area of rockbolt cross-section;
S − the spacing of the rockbolt.

122
 Chapter 5 Simulation of the reinforced rock unit in jointed rock mass

5.2 2D-DDA model

Numerical analyses have been conducted to analyse the performance of a single rockbolt
installed in a jointed rock mass. The numerical model used in 2D-DDA software is
shown in Figure 5-2. The rock mass is 2.5 m deep and 10 m wide with grid of rock
blocks size in 0.1 × 0.1 m2. The left, right and bottom boundaries of the rock mass are
fixed in a rigid box. Two joint sets are dominant in the model. The primary joint set is
persistent to simulate the continuous joints or rock laminations with dip angle of α, and
the second joint set is non-persistent and perpendicular to the primary one. The rock
block is simulated as a linear elastic material with density of 2600 kg/m 3, Young’s
modulus of 0.3 GPa and Poisson’s ratio of 0.25. All the joints follow the Mohr-Coulomb
law with friction angle of 20°and cohesion of zero.

The rockbolt is 1.0 m long and fully embedded into the rock mass. The rockbolt is
divided into several segments according to the block boundaries. A rigid block, as block
A shown in Figure 5-2, with a height of 0.1 m and a width of 0.2 m are attached on the
head of the rockbolts to simulate the faceplate and apply pullout loads. The details of
the parameters used in the 2D-DDA model are summarized in Table 5-2. The numerical
analyses are divided into three steps. The first step is to build the initial equilibrium
between the gravity and deformation of rock blocks. The second step is to install the
rockbolt into the rock mass in the designed positions. The third step is to apply pullout
loads onto the faceplate with a constant speed of 1 mm/s.

Table 5-2 Parameters used in 2D-DDA to simulate the pullout test

Item Parameter Value Item Parameter Value
Dynamic factor static Diameter, db (mm) 32
Step max. displacement 0.001 Tangent modulus, ET 4.2
ratio, g2 (GPa) (≈Eb/50)
Upper limit of time interval, 0.0005 Yield strength, σY (MPa)
360
DDA setting

g1
Bolt

SOR factor 1.4 Extreme strain, εext (%) 1.6

Normal contact spring Maximum shear strength
3×107 5.24e6
stiffness, kn (N/m) (Pa)
kn/ ks 2.5 Residual shear stress (Pa) 0.524e6
Block unit weight, ×103 26 Sliding bond node, k3
(Pa/m): 1e5
kN/m3

123
Chapter 5 Simulation of the reinforced rock unit in jointed rock mass

(a)

(b)

Figure 5-2 DDA models of pulling rockbolt

5.3 RRU of single rockbolt

5.3.1. Influence of the joint orientation

Five jointed rock mass models containing the primary discontinuous with dip angles α
= 0°, 30°, 45°, 60°and 90°are used in the simulation to investigate the influences of the
joint orientations to the rockbolt. The pullout resistances versus displacements of
rockbolt curves are plotted in Figure 5-3. For the rock model with primary joint dip
angle α = 0°, the pullout resistance increases substantially before pullout resistance of
21.5 kN with respect to the increasing of displacement. Here, the threshold is denoted
as the critical pullout resistance when rockbolt displacement reaches 1 mm as the
increase ratio of pullout load with displacement becomes smaller after that. With the
increase of dip angle from 0°to 45°, the critical pullout load increases from 21.5 to 41
kN. However, when dip angle increases to 90°, the critical pullout resistance decreases
to 5.8 kN. This implies that the pullout resistances are heavily affected by the
discontinuities of jointed rock mass.

124
 Chapter 5 Simulation of the reinforced rock unit in jointed rock mass

The simulated displacement contour maps of the jointed rock blocks are summarized in
Table 5-3. The RRU region was determined as the rock domain which were moved with
the rockbolt continuously. The displacements of rockbolt faceplate UA are 5, 10 and 20
mm, respectively. For rockbolt displacement UA = 5 mm, only surrounding rock blocks
are moved together with rockbolt. With the increase of the rockbolt movement, more
rock blocks are moved together with rockbolt, see UA = 10 mm. The RRUs are heavily
influenced by the joint orientations and featured with triangular cross sections. For the
rock model with primary joint dip angle α = 0°, the RRU has a reinforced angle β = 53°
and the RRU area s = 0.495 m2. With the increase of dip angle α, the area of RRU
increase to the peak values s = 1.009 m2, when α = 45°. For dip angle α increasing from
45 to 90°, the reinforcement angle β is decreased from 90°to 0°and the RRU area
changes from 1.009 to 0.100 m2.

Figure 5-3 Effects of the rock joint orientations on the pullout resistance of rockbolt
(for Er = 0.3 GPa, φ = 20°, L = 1.0 m and α = 0˚, 30˚, 45˚, 60˚ and 90˚,
respectively)

These observations can be explained by the load transfer process between the rockbolts
and the rock blocks. Figure 5-4 shows the contour of the major principal stresses of the
rock blocks when the faceplate was pullout 10 mm from the original place. To reduce
the influence from the initial principal stresses induced by gravity 0, the major principal
stresses σ1 are normalized as σ1/σ0 in the contour maps where a warmer colour represents

125
Chapter 5 Simulation of the reinforced rock unit in jointed rock mass

a larger increasing ratio and a cooler colour presents a larger decreasing ratio. The
contour of the major principal stress for the rock mass with the dip angle α = 0°is
symmetric along the rockbolt axis because the rock structures are symmetric. There are
increasing transverse stresses in the adjacent rock blocks resulting from the
compressions in RRU. For the rock mass with dip angles α = 30°, 45°, and 60°, the rock
blocks along the persistent joint or lamination direction (right sides of rockbolts in
Figure 5-4(b) – (d)) have slightly smaller changing ratios than those along the non-
persistent one (left sides of rockbolt in Figure 5-4(b) – (d)) because of the influence from
the asymmetry rock orientations. For dip angle α = 90°, the RRU is in a column and
disturbs little amount of rock blocks nearby as shown in Figure 5-4(e).

Table 5-3 Displacement contours in rock blocks during rockbolt pullout process

The upward deformation of ground surface during the pullout process of rockbolt is
shown in Figure 5-5 where x-coordinate is the distance from rockbolt and y-coordinate
is the upwards displacement of the rock blocks. Generally, the upward deformation of

126
 Chapter 5 Simulation of the reinforced rock unit in jointed rock mass

ground surface increases with respect to pullout distance UA. The rock block at the
surface of rock mass with x = 0 has the same magnitude as UA because it is grouted to
the rockbolt. For the case of α = 0°, the surface movements of the rock mass are
symmetrically distributed along the axis of the rockbolt as shown in Figure 5-5(a). The
ground surface has small deformation when |x| > 0.5 m. The influence range of the
rockbolt on the upward deformation of the ground surface is about 2.0 m. For the rock
mass with the dip angle α = 30°, 45°, and 60° as shown in Figure 5-5(b) – (d),
respectively, the deformations of ground surface are not symmetric but dependent on
the angle between the rockbolt and the persistent joint. For the rock mass with the dip
angle α = 90°as shown in Figure 5-5(e), the deformation of the ground surface is limited
to the rock blocks where the rockbolt was penetrated through.

(a) (b)

(c) (d)

σ1/σ0
0.1 1 10

(e) Contour bar

Figure 5-4 Contours of normalized major principal stresses σ1/σ0 in rock mass for Ua =
10 mm, Er = 0.3 GPa, φ = 20°, L = 1.0 m and (a) α = 0°, (b) α = 30°, (c) α =
45°, (d) α = 60°, and (e) α = 90°

127
Chapter 5 Simulation of the reinforced rock unit in jointed rock mass

Figure 5-5 Deformation at the rock surface during the pullout process of rockbolts for
Er = 0.3 GPa, φ = 20°, L = 1.0 m and (a) α = 0°, (b) α = 30°, (c) α = 45°, (d)
α = 60°, and (e) α = 90°

5.3.2. Influence of the friction angle and Young’s modulus of rock mass

In order to evaluate the influence of frictional angle on the area of RRU and the axial
load carrying capacity of the rockbolt, cases with the friction angle of 20°and 40°are
analyzed using 2D-DDA. The other basic parameters are the same as those discussed in
Section 5.3.1. Figure 5-6(a) shows the variation of the RRU areas with respect to the dip
angle of rock mass. The ratios of increase of the RRU areas are 14%, 23%, 6% and 9%
for the dip angles α = 0°, 30°, 45°and 60°, respectively. However, for the dip angle of
90°, the effects of the friction angle on the areas of RRU are limited. Similar
observations are also obtained when comparing the results of axial loads of rockbolts as

128
 Chapter 5 Simulation of the reinforced rock unit in jointed rock mass

shown in Figure 5-6(b). To evaluate the inflluence of elastic modulus of rock mass on
the area of RRU and the axial load of rockbolt, cases with the elastic modulus of rock
mass of 0.3 and 30 GPa and friction angle of 20°are analyzed using the numerical
method. The RRU areas and axial load carrying capacity of rockbolt versus dip angles
plots are shown in Figure 5-6(a) and (b), respectively. It can be seen that the ratios of
increase of the RRU areas are less than 10% for the dip angles ranging from 0°to 60°,
and zero for dip angle α = 90°. Similar observations are also obtained for the axial load
carrying capacity of rockbolts as shown in Figure 5-6(b). It can be concluded from this
parametric study that the increase of friction angle and elastic modulus of the rock mass
can increase the axial load and RRU area of rockbolt, especially for the low-angle
laminated rock blocks, such as the dip angle ranging from 0°to 45°.

3.0
L=1m, Er=0.3GPa, ø=20˚
L=1m, Er=0.3GPa, ø=40˚
Area of RRU, m2

2.0 L=1m, Er=30GPa, ø=20˚

L=2m, Er=0.3GPa, ø=20˚

1.0

0.0
0 30 60 90
Dip angle of primary joint (α in degree)
(a) Area of RRU

200
L=1m, Er=0.3GPa, ø=20˚
L=1m, Er=0.3GPa, ø=40˚
Axil load at the head of

150 L=1m, Er=30GPa, ø=20˚

L=2m, Er=0.3GPa, ø=20˚
rockbolt, kN

100

50

0
0 30 60 90
Dip angle of primary joint (α in degree)
(b) Axial load on the head of rockbolt
Figure 5-6 Parametric studies on the key influence parameters

129
Chapter 5 Simulation of the reinforced rock unit in jointed rock mass

5.3.3. Influence of the rockbolt length

The rockbolts with their lengths of 1 and 2 m installed in rock mass with the elastic
modulus of 0.3 and 30 GPa and friction angle of 20°are analyzed. Figure 5-6(a) shows
shows that the ratios of increase of the RRU areas are 130%, 202%, 266% and 84% for
the dip angles = 0°, 30°, 45°and 60°, respectively. However, for the dip angle of 90°,
the area of RRU only increases 20%. Mathematically, the RRU area should increase
300% for triangle shape and 100% for column shape when the length of rockbolt is
doubled. One of the possible reasons for the lower ratios is that the rock joints restrict
the load transfer from rockbolt to the surrounding rocks. Figure 5-6(b) shows the
variations of the axial loads of rockbolts with respect to the rockbolt lengths. The axial
loads increase 153%, 175%, 179% and 160% for the dip angles of 0°, 30°, 45°and 60°,
respectively, but 100% for the dip angle of 90°. It can be concluded from this parametric
study that increasing the length of the rockbolt is one effective method to increase the
load transfer capacity and RRU area, especially for the low-angle laminated rock blocks,
such as the dip angles ranging from 30°to 45°.

5.4 Grouped rockbolts

Parametric studies are also conducted to investigate the group effect of the rockbolts
installed in grid spacing. The numerical model is shown in Figure 5-2(b). Three
rockbolts named as A, B and C are installed in the rock mass and are fully penetrated the
rock mass. Three grid spacings S = 0.3L, 0.5L and 0.8L are used in the simulation where
L is the length of rockbolt. Five jointed rock mass models containing primary joints with
dip angles of α = 0°, 30°, 45°, 60°and 90°are used to investigate the influences of the
rock lamination orientations on the grouped rockbolts. The pullout speed during the
simulation is kept as 1 mm/s.

Table 5-4 summarizes the influence of the rock lamination orientations α and the grid
spacing of grouped rockbolts S on the displacement contour of rock blocks when the
rockbolts are pullout 10 mm from their original locations. The RRUs are also calculated
and shown in Table 5-4. It can be seen that the RRUs are heavily influenced by the rock
orientations and the grid spacing of the rockbolts. The smaller the spacing S, the larger
overlap between the RRUs is generated. The RRUs have less overlap when the rockbolt
installed in spacing of 0.8L in the rock mass with dip angle α = 0°and 60°but still have

130
 Chapter 5 Simulation of the reinforced rock unit in jointed rock mass

relative larger overlap in rock mass with dip angle α = 45°. This phenomenon can also
be observed from the uplift deformation of the rock surface as shown in Table 5-5.

Table 5-4 Displacement contour of rock blocks when UA = 10 mm

Figure 5-7(a) shows the axial loads versus displacement curves of the rockbolts installed
at spacing of 0.3L in rock mass with dip angle α = 0°. It shows that the axial loads of the
rockbolt B and C increase with respect to the displacement. However, the axil load in
rockbolt A increases with the displacement at the beginning of loading test, but decreases
slowly after a peak value. Similar observations are also obtained for the case of S = 0.5L
as shown in Figure 5-7(b). However, for the case of S = 0.8L, all the axial loads of the
three rockbolts increase with respect to their displacement except those of the middle
rockbolt are slightly smaller (see Figure 5-7(c)). One reason for this observation is the
group effects of rockbolts installed in a grid. In the grouped rockbolt, the reinforced rock
blocks nearby the middle rockbolt may also be reinforced by the lateral rockbolts which
are consistent with the observation in the overlap area of the RRUs as shown in Table
5-4. In such cases, the overlap of RRU may be presented as a function of grid spacing
of rockbolts (see Figure 5-7(d)). To investigate the interaction of RRUs as a function of
grid spacing, the axial load developments of the middle rockbolt in other laminated rock
models are presented in Figure 5-8.

131
Chapter 5 Simulation of the reinforced rock unit in jointed rock mass

Table 5-5 Deformation of rock surface

α (a) S = 0.3 L (b) S = 0.5 L (c) S = 0.8 L

0°

30°

45°

60°

90°

Figure 5-8 shows the axial loads versus displacement of the middle rockbolt in
laminated rock models with dip angles α = 0°. It can be seen that the peak values of the
axial load of the middle rockbolt are heavily influenced by the grid spacing, decreasing
in value as grid spacing increases. For the grid spacing of 0.8L, the axial load in middle
rockbolt versus displacement curve is similar to that of single rockbolt. This can be
explained as the group effect of the three rockbolts is ignorable. However, decreasing
the grid spacing to 0.5L and 0.3L, the axial loads in the middle rockbolt is significantly
smaller than that of single rockbolt. Similar observations were also obtained from the
axial loads versus displacement curves of the middle rockbolt installed in the jointed
rock models with dip angles α = 30°, 45°and 60°as shown in Figure 5-8(a) – (c),
respectively. However, the grid spacing has no obvious influence on the axial loads
versus displacement curves of the rockbolt installed in the laminated rock models with

132
 Chapter 5 Simulation of the reinforced rock unit in jointed rock mass

dip angles α = 90°, see Figure 5-8(d). This is because there are little interactions between
the rockbolts or among the induced RRUs.

40 40

Left Right
Axial loads, kN

30 30

Axial loads, kN
Right 20 Left
20
Middle Middle
10 10

0 0
0 2 4 6 8 10 12 0 2 4 6 8 10 12
Displacment, UA, mm Displacment, UA, mm

(a) S = 0.3L (b) S = 0.5L

40 50
Single rockbolt
Right S=0.3L
40 S=0.5L
30
Axial loads, kN

Axial loads, kN

S=0.8L
30
20 Left Middle
20
10
10
0 0
0 2 4 6 8 10 0 2 4 6 8 10
Displacment, UA, mm Displacement, UA, mm

(c) S = 0.8L (d) Comparisons of axial loads

Figure 5-7 Axial loads vs. displacement curves for the rockbolt in jointed rock mass
with dip angle of 0°

By selecting the axial loads in the middle rockbolt at the displacement UA = 10 mm from
Figure 5-8 and comparing their values with those of single rockbolt, the percentages of
the axial load reduction in the middle rockbolt can be derived and plotted in Figure 5-9.
For α = 0°, the axial load in the middle rockbolt is reduced by as higher as 24.65% and
63.29% for grid spacing S = 0.3L and 0.5L, respectively. However, for α = 45°, these
values are reduced to 10.61% and 30.1% for grid spacing of 0.3L and 0.5L, respectively.
It can also be seen that the axial load in the middle rockbolt installed with grid spacing
of 0.8L are very close to single rockbolt for α > 60°or α < 30°.

133
Chapter 5 Simulation of the reinforced rock unit in jointed rock mass

80
80
Single rockbolt Single rockbolt
S=0.3L S=0.3L
60 S=0.5L 60 S=0.5L
Axial load, kN

Axial load, kN
S=0.8L S=0.8L

40 40

20 20

0 0
0 2 4 6 8 10 0 2 4 6 8 10
Dispacement, UA, mm Displacement, UA,mm

(a) (b)

50 10
Single rockbolt Single rockbolt
S=0.3L S=0.3L
40 S=0.5L 8 S=0.5L
S=0.8L
Axial load, kN
Axial load, kN

S=0.8L
30 6

20 4

10 2

0 0
0 2 4 6 8 10 0 2 4 6 8 10
Displacement, UA, mm Displacement, UA, mm

(c) (d)

Figure 5-8 Axial loads of the middle rockbolt in cases of primary joints with dip angles
of (a) 30°, (b) 45°, (c) 60°and (d) 90°
94.54%

99.14%

100.00%
100.00%
100.00%

100%
70.34%

68.40%
63.29%

61.62%

80%
Ratio of axial load

36.59%
36.17%

60%
30.13%
24.65%

40%
12.91%

10.61%

20%

0%
α=0° α=30° α=45° α=60° α=90°
S = 0.3L S = 0.5L S = 0.8L

Figure 5-9 Percentages of the axial loads of middle rockbolts

134
 Chapter 5 Simulation of the reinforced rock unit in jointed rock mass

5.5 Discussion

The rockbolt model is implemented into the 2D-DDA code to analyse the single and
grouped rockbolts installed in jointed rock mass. The performances of the rockbolts are
evaluated by pulling them out from five jointed rock mass models with the primary
joints in dip angles of 0°, 30°, 45°, 60°and 90°. The RRU is used to define the influenced
rock region affected by the pulling of the rockbolt. Based on current study, the
traditional method to calculate the area of RRU by assuming it is cone-shaped with
isosceles right angle is not always accurate especially for the jointed rock mass
conditions. The cross-sections of the RRU are influenced by the loading scale and the
condition of rock mass discontinuities. Take models with reinforcement angle less than
90°for example, the RRU areas are much smaller than those estimated by the traditional
method which may overestimate the reinforced area of the rockbolts.

To evaluate the cross-section of RRUs of grouped rockbolts, an overlap ratio r1 is

defined as:


r1 = 1 −
( RRU g − ( n − 1)  RRU i )  100%
(5.1)
 RRU i 

where n is number of rockbolts in the rockbolt group (here n = 3), RRUg and RRUi is the
RRU area of the grouped and individual rockbolts, respectively.

Figure 5-10 shows the calculated overlap ratio versus dip angle curves for the rockbolts
installed in grid spacing S = 0.3L, 0.5L and 0.8L, respectively. It shows that the smaller
the grid spacing of the rockbolts, the larger the overlap ratio is generated, see S = 0.3L,
which indicates that the capacity of rockbolt has not been fully utilized. The overlap
ratio is also influenced by the rock properties. For the same grid spacing but dip angle
α from 0°to 45°, the overlap ratios increase slowly till their peak value. However, for
the dip angle α from 45°to 90°, the overlap ratio decreases very quickly till close to
zero, see α=90°, where there is no overlap among rockbolts for all spacings. It can be
concluded that the maximum spacing of the grouped rockbolt has to take into
consideration the rock joints orientations.

135
Chapter 5 Simulation of the reinforced rock unit in jointed rock mass

1.5

Overlap ratio, r, %
Cx4
S = 0.3L
Cx5
S = 0.5L
1.0
Cx6
S = 0.8L

0.5

0.0
01 302 3
45 4
60 5
90
Dip angle, α

Figure 5-10 Overlap ratio of the RRUs

Based on this study, the free body diagram of the RRU can be simplified as shown in
Figure 5-11 by considering the effect of the rock joints orientations. Equilibrium of the
force along the vertical direction yields the following equations:
P = kW
(5.2)
k = 1 + sin 1 cos 1 tan 1 + cos 2  2 tan  2

where W is the dead weight of RRU, α is the dip angle of RRU boundaries and φ is the
friction angle at the boundaries.

Figure 5-11 Illustration of the relationship between RRU and axial load

In the simulated rock model, the load transfer capacity of a rockbolt is depending on the
resistance of the rock mass and affected by the discontinuities in rock masses. As the
results of single rockbolt pullout shown in Figure 5-12(a), a larger RRU results in a
larger axial load. Considering the cases with same joint properties, linear regression is
carried out and show in Figure 5-12(b). The result shows that the axial load is about 1.4
times of the dead weight of RRU which may be caused by the friction. In field, the rate

136
 Chapter 5 Simulation of the reinforced rock unit in jointed rock mass

k may vary in different conditions, such as the complex ground, the displacement ratio,
etc.

(a) Comparisons of common cases

(b) Linear regression of the relationship between RRU area and axial load

Figure 5-12 Comparison of the RRU areas with the maximum axial loads of rockbolt
at UA = 10 mm

5.6 Pressure arch estimation using RRUs

5.6.1. 2D-DDA model

Numerical studies are conducted to investigate the formation of pressure arch by

considering different RRUs in jointed rock mass. The sketch of the models used in 2D-
DDA is shown in Figure 5-13. The laminated rock mass is considered with an inclination
angle of 0º (see Figure 5-13(a)) and 90º (see Figure 5-13(b)), respectively. The
numerical model is 11.75 m wide and 3.25 m high modelled by 280 blocks with size of

137
Chapter 5 Simulation of the reinforced rock unit in jointed rock mass

0.4 × 0.4 m2. Two perpendicular joint sets are setup in the model. Five 3-m-long
rockbolts, named as RB1 to RB5, are preinstalled into the rock mass. The lateral sides
of the model are restrained using two rigid blocks as shown in Figure 5-13(a). Linear
pressure q2 are applied on rigid blocks towards each other to simulate the in-situ
horizontal stress. The top blocks of the model are restrained by one row of rigid block
elements. The line pressure q1 is applied on the blocks downwards to simulate the in-
situ vertical stress. Three blocks are used to constrain the bottom movement. The two
blocks on the right and left ends are fixed during the simulation, while the middle long
block is used to maintain the in-situ stress during stress balance stage and removed at
the excavation stage. In the numerical modelling, the initial in-situ stress state is
achieved at 1.0 s after the initial loadings are applied. After another 1.0 s balance time,
the middle long block is removed. It should be noted that the shear resistance along
discontinuities is assumed to be purely frictional. The cohesion and tensile strength of
the discontinuities are neglected in the simulations.

(a)

(b)
Figure 5-13 DDA models for pressure arches in layered rock with inclined angles of
(a) α = 0ºand (b) α = 90º

138
 Chapter 5 Simulation of the reinforced rock unit in jointed rock mass

5.6.2. Influence of the horizontally persistent joints

(1) Natural pressure arch

The in-situ stress ratio k is defined as the ratio between horizontal and vertical pressure,
q2/q1. Figure 5-14(a) shows the simulation results of the vertical displacement of point
M1 near the middle of the excavation boundary (see Figure 5-13(a)) with respect to the
different in-situ stress ratios. It can be seen that the larger the k value, the smaller the
vertical displacement at point M1, and a more stable natural pressure arch might be
induced. For the cases of k = 2.0 and 3.0, point M1 is movable after excavation. But, M1
become stable after some adjustments for the cases of k = 3.5, k = 4, and k = 6. The
displacement at point M1 is lowest for the case of k = 6.0. Figure 5-14(b) shows the
simulated contour maps of the horizontal stress, σxx, with respect to different in-situ
stress ratios. Generally, the pink contours determine the range of the pressure arch. The
simulation shows the pressure arch is not observed graphically in the studied domain if
k = 3.0, resulted in the crush of the model. It means the displacement at point M1 could
not be controlled. However, the contour maps show that the pressure arches could be
observed for the cases of k = 4 and k = 6 and the height of the former one is higher than
that of the latter one. The results could illustrate the phenomena that the higher stress
ratio k, the displacements of the immediate roof are controllable as the heights of the
pressure arch are lower.

k=6, t=15 s
Vertical displacement, m

k=4, t=15 s

k=3, t= 30s

(a) (b)

Figure 5-14 Effect of the in-situ stress ratios on (a) the vertical displacement of M1, and
(b) the contour maps of the transverse stresses

139
Chapter 5 Simulation of the reinforced rock unit in jointed rock mass

(2) Artificial pressure arch

A so-called artificial pressure arch will be formed after rock excavation due to the
reinforcement provided by the rockbolts. The in-situ stress ratio k of 4.0 and 2.0 are used
in the simulation. Figure 5-15(a) shows the simulation results of the vertical
displacement of the same point M1 with respect to the different number of rockbolts and
in-situ stress ratios. It can be seen that the roof may have some displacement when k =
2.0 even 5 rockbolts are installed in the rock mass. For the case of in-situ stress ratio k
= 4.0, the displacement of M1 after installed 5 rockbolts are much more stable than no
rockbolts. The vertical displacement at point M1 is slightly smaller for the case of 1
rockbolt than that for the case of 2 rockbolts. This is because the rockbolt go through
the block in the former case where M1 located. Furthermore, it also indicates that the
location of rockbolt may also influences the formation of pressure arch.

5 rockbolts

1 rockbolt

Time-displacement curves 2 rockbolts

(a) (b)

Figure 5-15 Effect of the number of rockbolts on (a) the vertical displacement of M1
and (b) contour maps of the transverse stresses

Four vertical cross-sections, marked as A, B, C and D in Figure 5-13(a), from left

abutment to midspan are used to investigate the distributions of the stresses and
displacements of rock blocks in the pressure arch. The distributions of the x - (horizontal)
and y - (vertical) stresses and y - (vertical) displacements of the rock blocks in the
pressure arch are shown in Figure 5-16. It can be seen that the rock blocks are under
compression in x- direction (σxx < 0) but the values become more unevenly distributed
when approaching to midspan. Above the abutment (cross section A), the distribution

140
 Chapter 5 Simulation of the reinforced rock unit in jointed rock mass

of σxx is in rectangular with its pressure of 2 MPa that is the same as the applied pressure.
When approaching to the midspan (from cross-section B to D), σxx decreases at the roof
and increases with the depth far away from the roof resulting in trapezoidal distributions
with a higher maximum σxx near the top edge and a lower σxx near the bottom edge. The
σxx at the midspan (cross section D) is not linear but in parabolic shape, which means
the pressure arch is moving to the depth. Connecting the points with σxx = 2.0 MPa in
the four cross sections, it gives parabolic curves in convex shapes. The heights of the
intersections are also shown in Figure 5-16(a). It can be seen that the heights are
proportional to the displacements of M1. The heights for the case without rockbolt is
17.7% higher than the case of 5 rockbolts. The lower value of the heights means a more
reliable pressure arch will be generated.

The y stress σyy distributions along the four cross-sections are shown in Figure 5-16(b).
The stress concentration at the cross-section B has the highest level σyy which is about
threefold applied loading of 0.5 MPa. When approaching to the midspan (from cross-
section B to D), σyy decreases along the whole cross section. At the cross-section D, the
distribution of σyy is not linear. For the cases of no rockbolt and 2 rockbolts are installed,
only the top end of the blocks are compressed, while the bottom rocks are in zero stress
(σyy = 0) or even in tension (σyy < 0). When the rock blocks are reinforced by 3 or 5
rockbolts, more rock blocks on the top are compressed.

A B C D A B C D D

0 bolt
2.103

1 bolt
(RB1) 2.025

2 bolts
(RB2+RB4) 2.074

3 bolts
(RB1+3+5) 1.822

5 bolts
(RB1+2+3+4+5) 1.787

2 MPa 0.5 MPa 5mm

x-stress y-stress y-disp
(a) (b) (c)

Figure 5-16 Distributions of stresses and displacements in rock blocks along vertical
cross sections (a) x-stress, (b) y-stress and (c) y-displacement

141
Chapter 5 Simulation of the reinforced rock unit in jointed rock mass

5.6.3. Influence of the vertically persistent joints

(1) Natural pressure arch

Four cases are designed to investigate the formation of natural pressure arches in cases
of the in-situ stress ratio k = 2.0, 3.0, 4.0 and 6.0. The vertical displacements of M2 are
measured to see the stability of roof, as shown in Figure 5-17(a). When k = 6.0, the
natural arch is well established and no rock blocks from roof falls. On the other hand,
even the vertical displacement of M2 indicates a stable roof at k = 4.0, several blocks
fall down. When k decreases, more blocks in the roof become moveable, such as the
whole model may collapse in case of k = 2.0. The x-stress contours (see Figure 5-17(b))
show that there is no pressure arch formed in case of k = 2. Compared with the
displacement in the horizontal jointed rock mass, the present models show the vertically
persistent joints are unfavorable for roof stability.

(a) (b)

Figure 5-17 Effect of the stress ratios on the vertical displacement of M2 in verticlly
jointed rock mass (a) Displacement histories of measuring point M2 and
(b) Contour maps of the stress ratio k = 6.0, 4.0 and 2.0

(2) Artificial pressure arch

Figure 5-17 has showed the loss of local stability in the middle roof in case of stress
ratio k = 4.0. The moveable roof blocks are not well stabilized even some rockbolts are
added in the same model as shown in Figure 5-18. The rockbolts only reinforce the rock
blocks which the rockbolts are getting through indicating very limited reinforced unit of

142
 Chapter 5 Simulation of the reinforced rock unit in jointed rock mass

single rockbolt. Taking the displacements at M2 as examples, the case with installation
of 5 rockbolts has the smallest displacement at M2, but the loose roof may expand to
lateral blocks showing unfavorable reinforce forces for artificial arch formation. In
contrast, the displacements are doubled when 2 rockbolts are installed compared with
the displacement of M2 in natural arch formation. But the whole roof is more stable as
less moveable rock blocks than that with 5 rockbolts. The distributions of the stresses
and displacements along four vertical cross sections, marked as A, B, C and D from left
abutment to midspan, are showing in Figure 5-19.

5 rockbolts

1 rockbolt

2 rockbolts

(a) (b)

Figure 5-18 Vertical displacement of M2 in cases of artificial roof arches formed in

vertically jointed rock mass (a) Displacement histories of measuring point
M2; (b) Contour maps of the transverse stresses in cases of 5 rockbolts, 1
rockbolt and 2 rockbolts are installed

Considering the variation of σxx along four vertical sections, they are generally varying
from rectangular distribution along section A to trapezoid distribution along section B,
followed by triangular distributions along C and D. That is to say, σxx is increasing when
x approaches to midspan at the top edge and is decreasing at the bottom edge. But the
increaments may vary in different cases. Connecting the points with σxx = 2.0 MPa in
four cross sections, the depths are reversely proportional to the displacements of M2.
The depth in the case of 2 rockbolts installed is the smallest value in all five cases. The
distributions of σyy along four cross sections from left abutment to midspan are varying
from trapezoid along A to triangular distributions along D (except for the 5 rockbolts

143
Chapter 5 Simulation of the reinforced rock unit in jointed rock mass

case). This is because σyy has concentration near the bottom edge at abutment, but the
σyy concentration is limited and not affecting section B and other sections.

A B C D A B C D D

0 bolt
1.552

1 bolt
(RB1) 1.552

2 bolts
(RB2+RB4) 1.478

3 bolts
(RB1+3+5) 1.489

5 bolts
(RB1+2+3+4+5) 1.559

2 MPa 0.5 MPa 5mm

x-stress y-stress y-disp
(a) (b) (c)

Figure 5-19 Changes in the stresses and vertical displacements in various rockbolts
reinforced conditions (a) x-stress, (b) y-stress and (c) y-displacement

5.6.4. Discussion and limitation

Figure 5-20 plots the axial loading along the rockbolt with indication of the influence
area of single rockbolt as same as the areas showing in pullout test (Table 5-3). It can
be seen that the supposed RRU of rockbolt in horizontally laminated rock mass is much
larger than that in vertically laminated rock mass under the same number of rockbolts
reinforcement condition. In horizontally laminated rock mass, two rockbolts are
vertically installed with a spacing of 4.0 m on the roof, both rockbolts have low axial
forces of 4.0 kN evenly distributed along the whole rockbolts (see Figure 5-20a). The
influence areas are not overlapped indicating only local reinforcement are achieved by
each rockbolt, such as the local stress variation. When 5 rockbolts are installed with a
spacing of 2.0 m (see Figure 5-20b), the axial forces are unevenly distributed along the
rockbolt RB3 and RB2. The rockbolt RB3, located near the left abutment, has a yield
axial force at the rockbolt elements embedded in the immediate roof and very low axial
force (near 0) at the rest parts of rockbolt elements embedded in the deep roof. RB2 has
similar axial loading distribution as RB3. The axial forces of RB1 located at the midspan
are in tension along the rockbolt. The maximum axial force along RB1 is only 15 kN,
located at a joint with a depth of 0.8 m from roof. By the reinforcement of 5 rockbolts,

144
 Chapter 5 Simulation of the reinforced rock unit in jointed rock mass

the influence areas of five rockbolts are overlapping to each other resulted in a stiff rock
roof arch. In vertically laminated rock mass, the rockbolts are under evenly tension
along the whole length with axial loading of 20.0 kN evenly distributed along the whole
rockbolts, when two rockbolts are installed with a large spacing of 4.0m (see Figure
5-21(a)). The columns of each rockbolt stabilize the rock blocks which the rockbolt get
through. In five rockbolts cases (see Figure 5-21(b)), even RB3 bears large tensile force
at the elements in roof, and it does not help to change the stresses of roof too much due
to its limited reinforcement area. As the numerical models in the examples were
designed under specified geology conditions, the comparisons between the RRUs and
the pressure arches were only for illustration purpose. In the future, more detailed work
should be carried out to explore the relationship between RRU and the geometry of
pressure arch, as well as the possible loading on the rockbolt.

(a) (b)

Figure 5-20 Axial load of the rockbolt and its sketched RRU in the horizontally
laminated rock blocks (a) two rockbolts and (b) five rockbolts

(a) (b)

Figure 5-21 Axial load of the rockbolt and its sketched RRU in the vertically laminated
rock blocks (a) two rockbolts and (b) five rockbolts

145
Chapter 5 Simulation of the reinforced rock unit in jointed rock mass

5.7 Conclusions

A fully grouted rockbolt model is implemented into 2D-DDA to analyse the

performance of the rockbolts installed in jointed rock mass. The reinforcement area of
the reinforcement rock units (RRU) is evaluated by pulling rockbolts out from five
different jointed rock models.

For a single rockbolt, the pullout resistances are heavily affected by the discontinuities
of jointed rock mass. With the increase of dip angle from 0 to 45°, the critical pullout
resistance increases from 21.5 to 41 kN. However, when dip angle increases to 90°, the
critical pullout resistance decreases to 5.8 kN. The RRUs are heavily influenced by the
discontinuity orientations with triangular cross sections. Mathematically, the RRU area
should increase 300% for triangle-shaped cross section and 100% for column-shaped
when the length of rockbolt is doubled. However, the ratios of increase of the RRU areas
are 130%, 202%, 266%, 84% and 20% for the dip angles of 0°, 30°, 45°, 60°and 90°,
respectively. This is because the rock joints restrict the load transfer from rockbolt to
the surrounding rocks. The changing of principal stress shows the elongation of the
rockbolt induces compression in RRU. For the special model with discontinuity dipping
at 90°, the RRU of a single rockbolt is limited to a column of rock blocks.

For grouped rockbolts, the traditional method to calculate the area of RRU by assuming
it is cone-shaped with isosceles right angle is not always accurate especially for the
complex joint conditions. The RRUs are heavily influenced by the rock orientations and
grid spacing of the rockbolts. The smaller the spacing S or the rock mass with dip angle
closer to 45°, the larger overlap between the RRUs is generated which indicates that the
capacity of rockbolt has not been fully utilised. For the same grid spacing but dip angle
α from 0°to 45°, the overlap ratios increase slowly till their peak value. However, for
dip angle α from 45°to 90°, the overlap ratios decrease very quickly till zero. For α =
90°, there is no overlap among RRUs for all spacings. The maximum grid spacing of
the grouped rockbolt must take into consideration the rock joints orientations.

146
 Chapter 6 Stabilization of roof wedge using rockbolts

STABILIZATION OF ROOF WEDGE USING

ROCKBOLTS

Unstable wedges might be exposed on the roof of the underground excavation. The
design parameters of the rockbolt are usually determined by the weight of the unstable
blocks. More advanced research should be carried out to investigate the rockbolt
designs if the deformations induced by the re-adjustment of the stress field in the rock
is comparable to those induced by the weight of loosen rock blocks. In this chapter, 2D-
DDA is used to evaluate the rockbolt design to stabilize roof wedges for underground
opening. To represent a deformable rock mass under stresses, the joint relaxation
method is adopted in the 2D-DDA models.

6.1 Introduction

For tunnels excavated in jointed rock mass, rock wedges formed by intersecting
structural discontinuities may fall from the roof or slide out of the sidewalls as shown
in Figure 6-1(a). As there might be no restraints from the boundary, the rock wedge in
the tunnel roof may fall as soon as its base is fully exposed by the excavation of opening
(Hoek et al., 1995). The falling of the roof wedge will reduce the restraints and interlocks
in the surrounding rock mass. Unless proper supports are used to reinforce the loose
wedges, the stability of the surrounding rock may deteriorate rapidly.

The spot rockbolt and sparsely spaced pattern rockbolt are recommended in the
empirical methods in jointed rock mass to stabilize the rock wedges. The rockbolts for
wedge stabilization are usually designed based on the deadweight of the rock wedge (Li,
2017; Low and Einstein, 2013). This is because the major driving force of the wedge
falling is its deadweight, especially if the tunnel is excavated in hard jointed rock mass
under the low field stress state. The design parameters of the rockbolt include the bolt
147
Chapter 6 Stabilization of roof wedge using rockbolts

strength, length, diameter, spacing, patterns, and so on. The recommendations and
suggestions for the design parameters are summarized in Table 2-4.

In practice, the relative locations of the rockbolt to the major joints are also important
for the effectiveness of the rockbolt supports. The embedded length (l0), which is
defined as the length of rockbolt in the stable rock behind the wedge, is one of critical
parameters. As shown in Figure 6-1(b), the embedded length changes with the relative
location of the rockbolt to the wedge boundaries. It should be sufficient long to ensure
the anchorages in the stable interior rock. Otherwise, the rockbolt would be pulled out
easily. For the fully grouted rockbolt, the required embedded length is usually about 1.2
m (Li et al., 2016). The angle between the rockbolt and the joint is another important
factor. As shown in Figure 6-1(c), the rockbolts may be subjected to tension and/or shear
loads depending on the angle of rockbolt installation. The shear components may reduce
the load capacity of the rockbolt (Chen and Li, 2015).
l01

l02

(a) (b) (c)

Figure 6-1 Examples of the definitions of rockbolting design parameters in wedge

problems (a) sketch of unstable wedges formed in an underground opening,
(b) changes of embedded lengths and (c) changes of response modes of
rockbolts

The stable state of rock wedge might change during the stress re-adjustment in the
surrounding rock mass during the excavation process (Dwivedi et al., 2014). The rock
wedge may be partly loaded by surface forces caused by the residual internal pressure
(Wittke, 2014). To estimate the magnitudes of the surface forces, the joint relaxation

148
 Chapter 6 Stabilization of roof wedge using rockbolts

method is suggested by introducing the joint stiffness and wedge displacement caused
by joint deformation (Sofianos, 1986; Nomikos et al., 2002; Nomikos et al., 2006).
Using the above method, the required support forces from rockbolts are calculated as
the difference between the deadweight of rock wedge and the vertical deriving force
(Brady and Brown, 2006). The calculation often involves sensitivity studies to find out
the influences of each significant parameter on the displacement after relaxation. It will
be more efficient if a numerical method could be used (Nomikos et al., 2006).

Verification studies on the stability of rock blocks using 2D-DDA have been conducted
by MacLaughlin and Doolin (2006) and Jiao et al. (2014). They have concluded that the
solutions of the 2D-DDA matched the limit equilibrium solution with high precision.
As the 2D-DDA method uses an incremental solution procedure, the balanced contact
forces could be presented in the simulation progress. The stress variations after
excavation could also be presented in the simulation of rock tunnelling (Tsesarsky and
Hatethodzor, 2006). By using 2D-DDA method, it is possible to assess the force-
displacement relationships of rock blocks in the simulations and compose ground
reaction diagram to predict the interaction between rock wedge and rockbolt (Windsor,
1997; Aydin et al., 2014 and Nie et al., 2018a).

In this chapter, the stability of the rockbolt reinforced rock wedges in tunnel roof
reinforced is analysed using 2D-DDA. The change of stress state in the surrounding rock
blocks after excavation is represented using the joint relaxation method. Parameter
studies are carried out to investigate the effects of the horizontal pressure, the spacing
between rockbolts and the installation angle on the reinforcement effort of the rockbolts
for the rock wedge stabilization. A case study is also conducted to illustrate the usage
of the proposed method.

6.2 2D rock wedges stability in tunnel

For 2D problems, the geometry of rock wedge is assumed not to be changed in the
direction perpendicular to the discontinuity (Wittke, 2014). It is also assumed that the
forces acting on the wedge allow the movement only parallel to the discontinuity. The
rock block is generally subjected to its own weight, surface forces, and some support
load. The water pressure is not considered in this study.

149
Chapter 6 Stabilization of roof wedge using rockbolts

6.2.1. 2D-DDA model

A symmetric wedge block is used to simulate the unstable rock wedge as shown in
Figure 6-2(a). It has a height of 2.0 m and a semi-apical angle of 30°. The surrounding
rock mass are simulated using two blocks each of which with a height of 6.0m and a
width of 5.0 m. The joints between the wedge and two surrounding rock blocks are
purely frictional. The engineering properties of these three blocks are the same, i.e.
density ρ = 2.6 ×103 kg/m3, elastic modulus E = 30 GPa and Poisson’s ratio v = 0.1.
Two rigid blocks are added at the two sides to provide the horizontal confining pressure.
Roller supports are applied at the bottom of the rock blocks to restrain their vertical
movements.

The numerical analysis is conducted in three steps. The first step is to apply lateral
confining pressure p progressively onto the two lateral rigid blocks to achieve the initial
equilibrium stress state. The second step is to remove the roller at the central bottom of
rock wedge to simulate the excavation process. During this process, the deformation of
the joint defining the rock wedge is taken account. As shown in Figure 6-2(b), the
contact forces at points A to D between the rock wedge and the surrounding rock blocks
are recorded. It should be noted that points A and B are used to separate the contacts of
the top vertex to the right and left joints, respectively. The third step is to add two
symmetrical rockbolts into the model right after the second step to support the rock
wedge. As shown in Figure 6-2(c), the length, embedded length, inclined angle and
spacing of rockbolt are denoted as Lb, l0, θ and sb, respectively. To provide enough
anchorage force, the embedded length of a rockbolt is designed as l0 = 1.2 m as
suggested by empirical method given by Li (2017a). The rockbolt model used in Section
4.3.2 is employed here. The length of the rockbolt element in the rock is set as 0.1 m,
while that near the rock joint is set as 2le. The parameter setting is listed in Table 6-1.

6.2.2. Model verification

To verify the accuracy of the wedge model without rockbolt, the simulation result is
compared with that from analytical solution given by Brandy and Brown (2006). Before
relaxation progress as shown in Figure 6-3(a), the force equilibrium in the vertical
direction gives the vertical deriving load P of the rock wedge as:

150
 Chapter 6 Stabilization of roof wedge using rockbolts

P = 2 N 0 sec  sin ( −  ) (6.1)

where: P – Vertical deriving load;
N0 – Surface forces normal to the joint, assume N0 = N0,1 = N0,2;
S0 – Surface forces tangent to the joint; and S0 = S0,1 = S0,2;
 – Wedge semi-apical angle of joint,  = 1 = 2, and
 – Friction angle of joint,  = 1 = 2.

Boundary blocks：

ρ = 2600 kg/m3

E = 30 GPa
H = 6.0 m

v = 0.1

p
Normal
Wedge block：
locking

Roller supports Roller removed at the 2nd stage

(a)

0.1 m
2.0 m

0.1 m

(b) (c)

Figure 6-2 2D-DDA model for a symmetric wedge with straight wedge free face (a)
geometry model, (b) contacts at the wedge block and (c) parameters of
rockbolts

151
Chapter 6 Stabilization of roof wedge using rockbolts

Table 6-1 Parameters setting in the numerical simulation

Item Parameter Values Item Parameter Values
Elastic modulus Er, GPa: 30 Diameter Db, mm 20
Rock

Poisson ratio, v 0.1 Elastic modulus Eb, GPa 210

Unit weight, ×103 kN/m3 26.0 Yield axial force, kN 180

Rockbolt
Step max. displacement
0.0004 Ultimate axial force, kN 217
calculation

ratio
DDA

Upper limit of time 30

0.0002 Hinge length le, mm
interval (~1.5Db)
SOR factor 1.4 Shear stiffness Ks, MN/m 66.3

By introducing the joint stiffness k n and k s as shown in Figure 6-3(b), the limiting
vertical load Pl is calculated as:

M
H sin (j - a )
(6.2)
Pl = 2
D 0
where: H0 – The horizontal force before joint relaxation, H = N 2 + S 2 .
0 0 0

D and M are constants which are calculated as:

D = k s cos  cos  + kn sin  sin  (6.2a)

M = (k s cos 2  + kn sin 2  ) (6.2b)

(a) (b)

Figure 6-3 Free-body diagrams of a 2D roof wedge in (a) rigid joint condition and (b)
relaxed joint condition

Comparing Pl with the wedge weight W, the stability of the roof wedge could be
estimated by assessing the support force R and potential displacement or displacement

152
 Chapter 6 Stabilization of roof wedge using rockbolts

uy caused by joint relaxation. Some results are summarized in Table 6-2. It shows the
rock wedge could be self-supported after relaxation if Pl ≥ W. Assuming H0 = pH, where
p is the horizontal pressure, and substituting it to Eq. (6.2) gives the critical horizontal
confining pressure pcr as:

WD
pcr 
2MH sin ( −  )
(6.3)

Assume the joint has an area of A, the reaction forces due to the joint deformation could
be calculated as

Rn = kn Ad n (6.4)

Rs = ks A d s (6.5)

where: dn and ds – normal and shear deformation of a joint, respectively;

Rn and Rs – normal and shear reaction forces, respectively, and
k n and k s –joint stiffness at the normal and shear directions, respectively.

Based on the analytical solution, the horizontal pressure, pcr, versus friction angle, φ, are
plotted in Figure 6-4. It shows for a certain rock wedge, the critical horizontal confining
pressure pcr should vary with the ratio of normal and shear joint stiffness. The ratio of
joint stiffness on the critical horizontal confining pressure could also be considered as:

kn Rn d n kn
=  (6.6)
ks Rs d s ks

Here, the shear contact stiffness ks is assumed as 0.4kn. During the simulation, the
boundary of joint will be pushed back as no-penetration and no-tension are allowed
between blocks in 2D-DDA. In the following, three normal contact spring stiffness kn
are adopted in the numerical simulations to analyse the critical horizonal confining
pressure after joint relaxation, i.e., kn = 3×109, 30 ×109 and 300×109 N/m, or 1.3 Er×L,

13 Er×L and 130 Er×L, where Er is the Young’s modulus of the rock block and L is the

length of the line across which the contact springs are attached. The dimension of the
joint out-of-plane is assumed as 1 unit. The results of numerical analysis agree
reasonable with those obtained by analytical solutions, especially for the simulation
using normal contact stiffness kn = 3×109 N/m. The mismatch between the two results

153
Chapter 6 Stabilization of roof wedge using rockbolts

is because the unsymmetrical contacts forces are generated during the simulation. This
could be further explained using the numerical models with friction angle φ = 40° and

three contacts.

Table 6-2 Roof wedge stability based on the analytical solutions

Semi- Internal Stability assessment

Case apical horizontal Displacement,
angle,  force, H0 Pl Support force, R
uy
Potentially
 close to Any decrease of  and
stable if
0 < Pl  decrease of H0 will result in
A  and   Low sufficiently
W the increase of required
 support forces.
support
provided.
No or less support Potentially
B  Low Pl > W
required (for safety). stable.
No support required if the
displacement is allowable; Potentially
stable after a
C  High Pl > W Require reinforcement to large amount
increase the joint stiffness if of joint
the joint displacement is relaxation.
large.
• A larger Pl indicates a more stable condition.
Note: • uy affected by H0 and the joint stiffness, and
• H0 creates confining on the rock wedge

1.5
α=30°,
Critial horizontal pressure, pcr, MPa

E = 30GPa/m and
kn/ks=2.5
1.0

DDA, kn=10E
DDA, kn=1.0E
0.5 DDA, kn=0.1E
Ana. pcr

0.0
25 30 35 40 45 50 55 60 65 70

Friction anlge, φ, ⁰

Figure 6-4 Comparisons between the simulated critical horizontal pressures and the
analytical solutions at different friction angles

154
 Chapter 6 Stabilization of roof wedge using rockbolts

As shown in Figure 6-5, the magnitudes of the displacements in the 2nd step decrease
with the increase of contact stiffness. The model with a normal contact stiffness as kn =
3×109 N/m shows progressive displacement before the new balance achieves. The
normal contact forces of the wedge to the boundary blocks in the two simulation steps
are also investigated. The normal contact force versus the time step curves at the contact
points A to D are plotted in Figure 6-6(a). Even the load is supposed to be applied
linearly with time step at the 1st step, the obtained curves are not purely linear. The
contact forces are symmetrically distributed. The differences between upper contact
points (A and B) and lower contact points (C and D) are very little. For the case of kn =
30×109 N/m (see Figure 6-6(b)), the normal contact forces increase linearly with the
increase of time step during loading time. In both stages, the normal contact forces are
under and symmetric solution. However, the contact forces of the upper contacts (A and
B) are less than those of the bottom contacts (C and D). In case of kn = 300×109 N/m
(see Figure 6-6(c)), the differences between the normal forces at upper contact points (A
and B) and those at bottom contact points (C and D) become larger. The forces at the
upper contact points are only 70% of those of the bottom contacts. It also shows that the
contact forces have oscillations in the simulations leading to unsymmetrical loadings at
two sides of rock wedge. This might be the reason that a block model with soft contacts
(i.e., kn = (3 ~ 30) ×109 N/m or (1.3 ~ 13) ×E×L ) are resulted in a close agreement with
the analytical solutions.

Time step, t
0 0.5 1 1.5 2 2.5 3
0.00
Vertical displacement, mm

-0.01
Stage 1 Stage 2

-0.02

------- kn = 3×109 N/m

-0.03 − − − kn = 30×109 N/m
─── kn = 300×109 N/m
-0.04

Figure 6-5 Comparisons of the vertical displacements versus time curves obtained from
numerical analysis with different contact stiffness

155
Chapter 6 Stabilization of roof wedge using rockbolts

(a) (b) (c)

Figure 6-6 Comparisons of the normal contact forces versus time step curves when
friction angle φ = 40°(a) the vertical displacements and the normal contact
forces in the cases of (b) kn = 3×109 N/m, (c) kn = 30×109 N/m and (d) kn =
300×109 N/m

6.2.3. Parameter studies

Parametric studies are carried out to investigate the key parameters that might influence
the reinforced effects of rockbolts when they are used to stabilize the rock wedge. The
2D-DDA model with the contact stiffness kn = 3×109 N/m and ks = 0.4kn is adopted to
assess the stability of rock wedge. The friction angle of the joints between the wedge
and two surrounding rock blocks are assumed as 40°. The studied variables are the initial
horizontal pressure (p), the spacing between two rockbolts (sb) and the incline angles of
the rockbolt with respect to the direction of rock falling (θ).

The effects of the horizontal pressure, p, are investigated by changing its magnitude and
keeping sb = 2.0 m and θ = 0°. As shown in Figure 6-7, the rockbolts are slightly loaded
due to the joint displacement if the horizontal pressure p > pcr, which means the wedge
could be self-supported after stress redistribution. However, for p ≤ pcr, the wedge is
unstable. The support force provided by each rockbolt is about half of the wedge weight
(30 kN). The horizontal stress in the wedge after stress re-balance is about 6.2 kPa.

Three spacing between the two rockbolts are adopted to analyse the effects of the
spacing, i.e., sb = 0.5, 1.0 and 2.0 m and kept p = 80 kPa (< pcr = 93 kPa) and θ = 0°.

156
 Chapter 6 Stabilization of roof wedge using rockbolts

The variations of rockbolt force in three cases are shown in Figure 6-8. In general, the
axial load at the rockbolt element is decreasing with the decrease of spacing (sb). The
support force of each rockbolt when sb = 0.5 m is about 77% of that when sb = 2.0 m.
The horizontal stress in the wedge after stress re-balance in the former case is about 27%
of that of the latter case. This could be explained as the rockbolts with narrow spacing
induce less deformation in the deformable wedge.

20 40

p = 20 kPa
W/2

Axial load of the bolt element

-10 30
σxx of rock wedge, kPa

p = 40 kPa
-40 20

at joint, kN
p = 60kPa
p = 80 kPa
p = 100 kPa
-70 10
Ana. pcr = 93 kPa
-100 0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
Time step, t

Figure 6-7 Effects of the horizontal pressure p on the reinforced efforts of rockbolts to
stabilize the 2D wedge

20 30
sb = 2.0 m and θ = 0°
Axial load of the bolt element

0
σxx of rock wedge, kPa

sb = 1.0 m and θ = 0° 20
at joint, kN

-20

-40 sb = 0.5 m and θ = 0°

10
-60

-80 0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
Time step, t

Figure 6-8 Effects of the bolt spacing sb on the reinforced efforts of rockbolts to stabilize
the 2D wedge

157
Chapter 6 Stabilization of roof wedge using rockbolts

Two cases are analysed to investigate the effects incline angles of the rockbolt with
respect to the direction of rock falling. If keep p = 80 kPa (< pcr = 93 kPa) and sb = 1.0
m, the required support force when θ = 45° will increases with a magnifying factor of

sec(θ) from that when θ = 0°, as shown in Figure 6-9.

30 40

sb = 1.0 m

Axial load of the bolt element

σxx of rock wedge, kPa

0 and θ = 45° 30

at joint, kN
-30 20
sb = 1.0 m
and θ = 0°
-60 10

-90 0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
Time step, t

Figure 6-9 Effects of the installation angle θ on the reinforced efforts of rockbolts to
stabilize the 2D wedge

6.3 Case study

The Jurong Rock Caverns (JRC) cavern was constructed beneath the seabed at the
Banyan Basin in Jurong Island, Singapore, for hydrocarbons storage and located
approximately 150m below ground in sediment rock. As shown in Figure 6-10, the
cross-section of the storage galleries is in a horse-shoe shape with its height of 27 m and
span of 20 m and excavated using the top heading and bench technique. The height of
the top heading is about 9.0 m. A site investigation indicated that the ratio of maximum
principal horizontal stresses to maximum principal vertical stress is approximately 2.0
(Zhou et al. 2017). The geological map showed there is one bedding plane and three
joint sets on the site as shown in Figure 6-10 and summarized in Table 6-3. The rock
mass was composed by dark grey mudstone interbedded with light grey sandstone. It
also recorded that the bedding planes were not significant and there were other joints
which were rough, dry water condition, fresh and strong. The rockbolt support method
is used to restrain the unstable wedge in roof as shown in Figure 6-10. The 5.0 m long

158
 Chapter 6 Stabilization of roof wedge using rockbolts

rockbolt is installed in rectangular pattern with in-line spacing of 2.2 m and out-line
spacing of 2.3 m. The shotcrete with thickness of approximately 80 mm is used to
stabilize the excavation surface. One metric point in storage gallery, is selected in this
study to illustrate methods of using proposed rockbolt model for rock wedge
stabilization in the roof of rock cavern.

(a)

(b)
Figure 6-10 Site information after top heading (a) geological and (b)support conditions

Table 6-3 Characteristic of the discontinuities according to the site geological

mapping after top heading
Bedding Joint 1 Joint 2 Joint 3
Dip, ° 20 ~ 25 70 ~ 90 70 ~ 90 80 ~ 90
Dip direction, ° 320 170 ~ 180 80 ~ 90 30 ~ 50
Friction angle (assumed), ° 30 30 30 30

159
Chapter 6 Stabilization of roof wedge using rockbolts

A spreadsheet is developed to estimate the removable key blocks in the cavern surface
using key block theory proposed by Goodman and Shi (1985). Moreover, the envelopes
of the possible removable pyramids at the roof and wall are projected on the cross
section of the rock caverns which is also presented in the worksheet as shown in Figure
6-11. By gathering the geometry data and extending the boundaries of the envelopes, an
approximated 2D blocky model could be constructed as shown in Figure 6-12(a). After
invoking the 2D-DDA code, a 2D plane strain model could be adopted to simulate the
wedge stability with or without rockbolts. The rock block has a Young’s modulus of 12
GPa and density of 2.6 ×103 kN/m3. The joints are assumed with same properties as
friction angle φ = 30°, cohesion c = 2MPa and tensile strength σt = 1MPa. According to
the bolting design, two kinds of bolt arrangements are considered. Block A with an area
of 10.7 m2 on the immediate roof is chosen as the study target. According to the relative
positions between rockbolts and block A, two reinforcement situations could be
identified. One case, defined as Case A, is the rockbolts at metric axis WP0-71-89.2 are
installed through block A as shown in Figure 6-12(b). The other case, defined as Case
B, is the rockbolts at metric axis WP0-71-92.0 are installed through block A as shown
in Figure 6-12(c).

Figure 6-11 Method to estimate the removable key blocks in the cavern surface using
key block theory

160
 Chapter 6 Stabilization of roof wedge using rockbolts

The simulation is conducted in three steps. The 1st step is to obtain the initial stress state.
The 2nd step starts at time step t0 = 60 when the excavation core is removed, followed
by the 3rd step when the rockbolts are added in the model at time step tb.

(a)

(b) (c)

Figure 6-12 An approximated blocky model of (a) the cavern cross-section, (b) rockbolt
arrangement of Case A and (c) rockbolt arrangement of Case B

Numerical studies are carried out to identify the major factors that affect the stability of
block A for the case with no rockbolt reinforcement. Figure 6-13(a) plotted the vertical
displacement versus time curves and the stress versus vertical displacement curves of
block A for normal stiffness of the contact spring kn = 1.2×109, 6×109, 12 ×109, 120 ×

109 N/m, and ks = 0.4kn .It can be seen that the soften contact with kn = 1.2×109 and 6×
109 N/m will be resulted in a larger displacement after stress re-adjustment. As shown
in Figure 6-13(b), the vertical stress is quickly rebalanced for kn = 120 ×109 N/m, but

progressively balanced for kn < 12 ×109 N/m. To further explain the phenomena, Table

6-4 shows the horizontal stress contours σxx for kn = 6×109 and 120 ×109 N/m at the step

161
Chapter 6 Stabilization of roof wedge using rockbolts

time right after excavation, t = 60.05, and that at the step time after rebalanced, t = 100.
It can be seen that the stress re-balance at block A for kn = 120 ×109 N/m is much faster

than that for kn = 6 ×109 N/m. In the following, the simulation of kn = 6 ×109 N/m is
adopted to study the rockbolt performance during the stress redistribution and the wedge
progressively movement.

0.03
Vertical displacment uy, m

kn = 1.2×109N/m
0.02
kn = 6×109N/m

kn = 12×109N/m
0.01
kn =120×109N/m

0.00
60 80 100 120 140
Time step, t
(a)

1.5
kn=10E
kn =120×109 N/m
σyy of block A, MPa

1.0 kn =12×109 N/m

kn=1.0E
kn =6×109 N/m
kn=0.5E
n =1.2×10 N/m
kkn=0.1E 9

0.5

0.0
0 0.002 0.004 0.006 0.008 0.01
Vertical displacement uy ,m
(b)

Figure 6-13 Simulated reactions of (a) vertical displacement and (b) the variations of
stress versus displacement at block A under different joint relaxation
situations

162
 Chapter 6 Stabilization of roof wedge using rockbolts

Numerical studies are also carried out to study the influences of rockbolting pattern and
timing. For rockbolt installed in Case A and Case B as shown in Figure 6-12(b) and (c)
respectively, totally five rockbolts are installed through the block A in different position.
The total support forces of the five rockbolts acting on block A in the two cases are
shown in Figure 6-14. It can be found the total support forces for rockbolts installed in
Case A increase slowly with the time step than those for rockbolts installed in Case B
at time tb = 60. However, the ultimate support forces acting at the boundary of the block
A in both cases are matched. The total support forces for rockbolts installed in Case B
at step time tb=60.05 and 61.0 are much lower than those installed at step time tb=60. It
can be concluded that the rockbolts installed at tb=61.0 are too late.

Table 6-4 Contour maps of the horizontal stress σxx at different time step

Installation time step = 60.05 Installation time step =100

kn =120×109
N/m
kn =6×109
N/m

Note: Excavation occurs at step time 60.

The installation position of the rockbolt determines the maximum axial force along the
rockbolt. For rockbolt installed at step time tb=60 in Cases B and A as shown in Figure
6-15(a) and (b), respectively, the maximum axial force always occurs in the second
rockbolt from left to right. The difference of the maximum force in rockbolt installed in
Case A is 13% higher than those installed in Case B. The maximum axial force along
the rockbolt changes with respect to its installation time. For rockbolts installed at step
time tb=60, 60.05 and 61.0 in Case A as shown in Figure 6-14 (b), (c) and (d), the
maximum axial forces at the second rockbolt from left to right are 89 kN, 43 kN and 20
kN, respectively. This is to say the maximum axial force along the rockbolt installed at
step time tb=60 reduced 345% than that installed at step time tb=61.0.

163
Chapter 6 Stabilization of roof wedge using rockbolts

600
Sum of bolt forces acting at the kn=0.5Er tb=60
tb = 60, Case B type 2
boundary of block A, kN 500 kn=0.5Er tb=60
t = 60, Case A type 1
b
kn=0.5Er, tb=60.05, type1
400 tb = 60.05, Case A
kn=0.5Er, tb=61.0, type 1
tb = 61.0, Case A
300

200

100

0
60 62 64 66 68 70
Time step, t

Figure 6-14 The total support force acting on block A in Cases 5 to 9

The installed rockbolts restrain the movement of block A. Figure 6-16 shows the
numerical results of the stress versus vertical displacement curves of block A installed
in Case A at step time tb=60, 60.05 and 61.0. Results for the case of kn = 6×109 N/m

with no rockbolt reinforcement is also plotted in Figure 6-16 for reference purpose. It
can be seen that the earlier the bolting installed, the more restrains could be provided
and less movement of the block A. For rockbolt installed at step time tb=60, the
displacement of block A could reduce by 0.4 mm or 9%. For the rockbolts installed at
tb = 60.05, the displacement of block A could reduce by 0.2 mm or 4%. However, for
the rockbolts installed at tb = 61.0, the rockbolt is carrying few loads and did not reduce
the displacement of block A.

15 15
Max: 101.37 kN Max: 89.78 kN

12 12

9 9

6 6
-12 -7 -2 3 -12 -7 -2 3

(a) tb = 60, Case B (b) tb = 60, Case A

164
 Chapter 6 Stabilization of roof wedge using rockbolts

15 15
Max: 43.46 kN Max: 20.31 kN

12 12

9 9

6 6
-12 -7 -2 3 -12 -7 -2 3

(c) tb = 60.05, Case A (d) tb = 61.0, Case A

Figure 6-15 Simulated axial forces along the rockbolts through block A

1.2 120

Nodal axial load at segemnt 11 of bolt

kn = 0.5Er, without rockbolts
0.4 mm
1.0 tb = 60, with rockbolts 100

tb = 60.05, with rockbolts

0.8 80
σyy of block A, MPa

13, kN
0.6 0.2 mm 60

0.4 40

0.2 20

0.0 0
0.000 0.001 0.002 0.003 0.004 0.005
Vertical displacement of block A uy, m

Figure 6-16 Simulated reactions of block A with varies rockbolting time

It should be noted that the proposed 2D-DDA model is constructed according to the
projection of the 3D blocks on 2D tunnel cross-section based on Block Theory. The
spreadsheet could quickly find the possible falling wedges and the projections of
envelops. The average CPU time for the computation for the case study using 2D-DDA
is about 40 min/case (on a PC of Intel Core 2 Quad CPU @ 3.0 GHz and 4.0 GB of
RAM). Therefore, the proposed method provides a way to quickly access wedge
stability and axial forces along rockbolt.
165
Chapter 6 Stabilization of roof wedge using rockbolts

6.4 Conclusions

Unstable wedges might be exposed on the roof of the underground excavation. More
advanced research should be carried out to investigate the rockbolt designs if the
deformations induced by the readjustment of the stress field in the rock is comparable
to those induced by the weight of loosen rock blocks. In this chapter, 2D-DDA is used
to evaluate the rockbolt design to stabilize roof wedges in underground opening.

The joint relaxation method is adopted to represent a deformable rock mass under
stresses. The results are calibrated using the analytical solutions. Parametric studies are
carried out to investigate the key parameters that might influence the effects of rockbolts
when they are used to stabilize the rock wedge. The results show 2D-DDA could be
used to find the critical horizontal pressure to sustain the rock wedges using the contact
stiffness kn = (3 ~ 30) ×109 N/m or (1.3 ~ 13)×E×L and ks = 0.4kn where E is the Young’s

modulus of the rock block and L is the length of the line across which the contact springs
are attached. The wedge could be self-supported after stress redistribution if the
horizontal pressure p > pcr and   . In such cases, the rockbolts are slightly loaded
due to joint displacement. It also shows the bolt force slightly decreases when the
spacing between two rockbolts decreases. However, the bolt force increases if there is
shear components during the wedge deformation.

A case study to analyse the Jurong Rock Caverns (JRC), Singapore, is conducted to
show the application of the 2D-DDA models for rockbolt design to support roof wedges.
The block model is constructed based on the projection of the envelopes of possible
wedges on the cavern cross-section. A spreadsheet is developed to estimate the
removable key blocks in the cavern surface using key block theory. The reaction of the
wedge block at the roof is used to study the reinforcement efforts provided by rockbolts.
The result shows that the maximum axial force along the rockbolt is determined by the
installation pattern and time of rockbolts. The stress versus vertical displacement curves
showed clearly the effect of rockbolt on sustaining the movement of rock wedge. This
way to create characteristic diagram could be used by the site engineers to evaluate the
rockbolting design to sustain the possible roof wedges.

166
Chapter 7 Performance-based support design for horseshoe-shaped rock caverns
using 2D FEM

PERFORMANCE-BASED SUPPORT DESIGN

FOR HORSESHOE-SHAPED ROCK CAVERNS
USING 2D FEM

The previous work on rock reinforcement mechanism aims to understand the

rockbolt and rock mass interaction. It shows the rockbolt might be used as suspension
structure to connect the loosen layers to the stable rock mass. The pattern rockbolts
could be utilized to generate the artificial arch if the spacing is narrow enough. In this
chapter, a support design method was proposed based on the ground reactions after
excavation and support performances.

7.1 Introduction

An accurate knowledge of geological conditions is favourable for rock cavern

excavation design. In the design stage, there is uncertainty due to naturally variable
phenomena in time and lack of knowledge or understanding. The uncertainty also comes
from the excavation which may affect the stress field and engineering properties of the
surrounding rock mass. As shown in Figure 7-1, the influence zones due to excavation
in rock are normally classified as Excavation disturbed Zone (EdZ), Excavation
Damaged Zone (EDZ), and Highly Damaged Zone (HDZ) (Siren et al., 2015). The rock
mass in the EDZ and HDZ are often fully yielded with very limited self-support capacity.
The normal distance from the EDZ edge to the excavation surface is often defined as
damage depth (Dp) to characterise the damage of the rock mass. Another commonly
used parameter to represent the damage due to excavation is the convergence on the
boundary of the excavation surface. The two parameters are often used for economic

167
Chapter 7 Performance-based support design for horseshoe-shaped rock caverns
using 2D FEM

evaluations for the rock cavern performance and supports designs (Kwon et al., 2009;
Zhang and Goh, 2010; Hijazo et al., 2012; Siren et al., 2015; Feng et al., 2018).

Excavation disturbed Zone (EdZ)

• Reversible damage
• Minor property changes Normal pressure
distribution
Excavation Damaged Zone (EDZ)

• Significant property changes

• Fracturing development

Highly Damaged Zone (HDZ)

• Macro-scale fracturing or spalling

Dp – Damage depth.
• Significant property changes
• Wedge/Surface instability

Figure 7-1 Influence zones due to excavation in rock (modified after Siren et al., 2015)

To reduce the damage depth and improve the self-support capacity of the surrounding
rock mass, the sequential excavation method (SEM) is often used for excavation in rock
by deliberately controlling and adjusting the stress and deformation field. One of the
major steps in the SEM process is the selection of the sequential excavation parameters,
such as the subdivision of cavern cross-section, the round length (or maximum
unsupported excavation length), and the support installation locations (or support
installation time). The sequential excavation parameters and the geological conditions
determine the selection of the support systems which are usually classified using
empirical methods, i.e. the RMR support system (Bieniawski, 1989) and the Q-system
(NGI, 2015). The commonly used support system includes the rockbolt, steel set,
shotcrete, concrete lining or a combination of the above.

The support system design has to consider its installation time and their interaction with
the surrounding rock mass. The convergence confinement method (CCM) is a widely
used method to assess the stress relaxation in the exacavation surface, the pressure on
support and the progressive expansion of the EDZ at different excavation steps (Sinha,
1989; Carranza-Torres and Fairhurst, 2000; Alejano et al., 2010; Lüet al., 2011). The
CCM uses characteristic curves to define the support-ground interaction. The
characteristic curves include the Ground Reaction Curve (GRC), the Longitudinal

168
Chapter 7 Performance-based support design for horseshoe-shaped rock caverns
using 2D FEM

Displacement Profile (LDP) and the Support Characteristic Curve (SCC). As the stress
re-distributions in the rock mass are quite complex during the excavation process, the
characteristic curves in CCM are hard to be determined using analytical methods.

Numerical simulations for excavation in rock have been performed by many researchers
to determine the characteristic curves (Karakus, 2007; Pellet et al. 2009). As a full 3D
numerical analysis is usually too time-consuming for the preliminary support design,
the suitable 2D finite element (FE) plane strain models are often used to consider the
3D excavation effects (Karakus, 2007; Cai, 2008; Janin et al., 2015; Kitchah and
Benmebarek, 2016). The key difficulty in the application of the 2D model is how to
determine the correspondence between the simulation stage and distance from the
excavation face (Alejano et al., 2010). To solve this problem, an improved Longitudinal
Displacement Profile (LDP) has been proposed by Vlachopoulus and Diederichs (2009,
2014).

In this chapter, the excavation in rock was simulated using the 2D FE plane strain
models. Parametric studies were conducted to investigate the influence of SEM
parameters and the geological conditions. Using the numerical results, the artificial
neural network (ANN) models were built to identify the relationships among the
geological conditions, the sequential excavation parameters and the cavern
performances. An evaluation chart was proposed which could provide an evaluation of
a support design for the rock caverns. A case study was also conducted to illustrate the
process of using the proposed evaluation chart.

7.2 Theoretical background

7.2.1. Characteristic curves

The characteristic curves of CCM include GRC, LDP and SCC. The typical
characteristic curves for excavation in rock are shown in Figure 7-2. The GRC gives the
relationship between the pressure on ground (pi) and the displacement at excavation
surface towards the opening (displacement, u). The in-situ ground pressure before
excavation is denoted as p0. With the increase of displacement, the pressure on ground
decreases linearly in elastic ground or nonlinearly in plastic ground. The point D’ in
GRC is the threshold that the rock mass starts to loosen and thus the supports have to be

169
 Chapter 7 Performance-based support design for horseshoe-shaped rock caverns
using 2D FEM

installed before it. The LDP presents the relationship between the displacement (u) and
the distance from the excavation face (x) which can be used to establish the displacement
at excavation locations in the SEM process. The SCC is the plot of the pressure on
support (ps) versus the displacement (u) curve. The stiffness of the support determines
the slope of the SCC. The SCC of the rigid Support B has higher slope than that of the
flexible Support A.

Figure 7-2 Characteristic curves for excavation in rock (modified after Lüet al., 2012)

The intersection of the SCC and displacement axis (point N) could be used to determine
the installation position of the support using LDP (point N’). The intersection between
SCC and GRC means the support pressure and the ground stress are balanced which
could be used to determine the pressure on the supports. As shown in Figure 7-2,
supports A, B, C have to sustain ps,A, ps,B, ps,C, respectively. As the required pressure on
support ps,A is the smallest among the three and much less than the support capacity
ps,max, the flexible support (Support A) installed at point N gives the highest factor of
safety. The allowable displacement umax is another parameter to restrain the support
design which indicates that a rigid support (Support B) installed at point N’ is more
proper due to uB < umax < uA. In this way, the characteristic curves can be used to evaluate
the design of the support system.

170
Chapter 7 Performance-based support design for horseshoe-shaped rock caverns
using 2D FEM

7.2.2. Performance functions

To represent the performances, the support system has to satisfy the following three
design criteria, i.e. displacement criterion, support capacity criterion and rockbolt length
criterion. The performance functions (limit state functions) of the displacement criterion
g1(x), the support capacity criterion g2(x) and the rockbolt length criterion g3(x) are
shown as,

g1 ( x ) = umax − u (7.1)

g 2 ( x ) = ps ,max − p (7.2)

g3 ( x ) = Lb − ( D p + l0 ) (7.3)
where: x is the vector of random variables,
umax is the allowable displacement,
u is the displacement at the excavation surface to the opening,
ps,max is the bearing capacity of the support,
ps is the pressure on support,
Lb is the length of rockbolt,
Dp is the damage depth of EDZ from the excavation surface, and
l0 is the anchored length of rockbolt which is approximately 1.2 m.

The displacement coordinate of the intersection between SCC and GRC is the
displacement at the excavation surface to the opening that the support system is fully
functioned. However, the displacement at the installation position of support u’ is
different from that at the intersection of SCC and GRC. As shown in Figure 7-2, support
B will provide support ps,B at point B with its displacement of uB. The displacement at
installation point N is uN. The difference between uB and uN is the section in LDP.
Therefore, the relationship between the displacements at the installation point u’ and that
the support system is fully functioned is:
u’ = u - ∆ur (7.4)
and
ps
ur = (7.5)
Ks
where ps is the support stress and Ks is the support stiffness.

171
Chapter 7 Performance-based support design for horseshoe-shaped rock caverns
using 2D FEM

The load carrying capacity ps,max in Eq. (7.2) is not easy to be determined as the group
effects of the support system are hard to be defined. It is usually assumed ps,max as the
sum of the load carrying capacity of each individual support element (Hoek, 2007). The
support capacities of three commonly used support systems, i.e. the steel set, the
rockbolt and the shotcrete lining, and their stiffness on a circular tunnel with radius of
r0 have been proposed by Hoek (2007) and summarized in Table 7-1. The length of
rockbolt Lb in Eq. (7.3) is determined by cavern span B when the rockbolt is used to
suspend the failure zone to the natural arch. An empirical equation to estimate Lb (unit
in: m) in moderately jointed hard rock masses has been proposed by Li (2007) and
shown as:
Lb = 1.40 + 0.184B (7.6)

Table 7-1 Support capacities defined by Hoek (2007)

Support
Parameters Determined equations
elements

Max. support pressure, pmax,ss pmax, ss = As ys sl r0

Steel set
support Stiffness, Ks,ss K s , ss = Es As sl r02

End anchored Max. support pressure, pmax,b pmax,b = Tbf sb2

rockbolts in
pattern Stiffness, Ks,b K s ,b = Eb db2 4Lb sb2

 cc  ( r0 − tc ) 
2

Max. support pressure, pmax,sc pmax, sc = 1 − 2


2  r0 
Shotcrete
linings Ec  r02 − ( r0 − tc ) 
2

Stiffness, Ks,sc K s , sc =  
2 (1 −  ) ( r0 − tc ) r02
2

Note: σys is the yield strength of the steel, MPa; Es is the Young’s modulus of the
steel set, MPa; As is the cross-sectional area of the section, m2; sl is the
spacing of support along the tunnel axis; σcc is the UCS of the concrete or
shotcrete, MPa; Ec is the Young’s modulus of the concrete or shotcrete,
MPa; υc is the Poisson’s ratio of the concrete or shotcrete; tc is the
thickness of the shotcrete lining, m; db is the rockbolt diameter, m; sb is the
in-plane rockbolt spacing, m, here assumed sc = sb (out-of-plane spacing),
and Eb is Young’s modulus of the rockbolt.

172
Chapter 7 Performance-based support design for horseshoe-shaped rock caverns
using 2D FEM

7.3 Analysis of horseshoe cavern using numerical method

7.3.1. Numerical models

A rock cavern with a horseshoe-shaped cross section (short as ‘horseshoe cavern’) has
been successfully constructed in Singapore with span B of 20 m and height H of 27 m.
The 2D FE plane strain model is conducted based on the dimensions of this horseshoe
cavern using RS2 program (Rocscience Inc., 2010), as shown in Figure 7-3(a). The
height and the width of the horseshoe cavern are denoted as H and B, respectively. The
distance of the cavern wall to the model boundary is set to 4H. To study the effects of
the sequential excavation, the full-face (FF) excavations and the subdivisions are
considered in the numerical analysis.

(a) (b) (c)

Figure 7-3 Numerical models using RS2 program for (a) horseshoe cavern, (b) circular
tunnel, and (c) horseshoe cavern under subdivision

The cavern is assumed under an isotropic stress state of 10 MPa (i.e., σh = σv = 10 MPa).
Five ground classes are assumed based on Q-value, i.e., very good (Q = 40), good (Q =
10), fair (Q = 4), poor (Q = 1.0) and very poor (Q = 0.1). The elastic moduli of the rock
mass before peak are calculated as (Bieniawski, 1984),
Em = 2RMR – 100 (RMR > 50) (7.7)
Em = 10^((RMR-10)/40) (RMR  50) (7.8)
where RMR is the Rock Mass Rating (RMR) and calculated as,
RMR = 9.0lnQ + 44 (7.9)

The UCS of the rock mass is (Serafim and Pereira, 1983; Palmstrom, 2000),

173
Chapter 7 Performance-based support design for horseshoe-shaped rock caverns
using 2D FEM

UCS = RMR (7.10)

The Geological Strength Index (GSI) of the rock mass is (Marinos et al., 2005),
GSI = RMR - 5 (7.11)

The Hoek-Brown criterion (Hoek, 2007) is used to predict the yield of rock masses
which can be written as follows,

(7.12)

where ’1 and ’3 are the maximum and minimum effective principal stresses at failure,
respectively, mb is the value of the Hoek-Brown constant m for the rock mass, s and a
are constants which depend upon the rock mass characteristics, and ci is the UCS of the
intact rock samples.

For the residual properties of rock mass, such as Em, UCS, GSI, mi, mb, s and a, the linear
interpolation method is used to calculate their values based on the intervals of GSI
proposed by Alejano et al. (2010). All the properties of the rock mass used for the
numerical analysis are summarized in Table 7-2.

Table 7-2 Geotechnical parameters for numerical models

Q 0.4 1.0 4 10 40
RMR 36 44 56 65 77
UCS (MPa) 36 44 56 65 77
GSI 31 39 51 60 72
Before Peak

Em (GPa) 4 7 13 30 50
mi 12 15 15 20 20
mb 0.985 1.698 2.515 4.793 7.358
s 0.0004 0.0011 0.0039 0.0117 0.0445
a 0.522 0.512 0.506 0.503 0.501
GSI 2.3 3.2 3.5 6.5 10
Em (GPa) 20 27 30 35 40
Residual

mi 10 10 10 10 10
mb 0.574 0.737 0.821 0.981 1.173
s 0.0001 0.0003 0.0004 0.0007 0.0013
a 0.544 0.527 0.522 0.516 0.511

174
Chapter 7 Performance-based support design for horseshoe-shaped rock caverns
using 2D FEM

The excavation effects in rock are simulated using the progressive core replacement
method or stiffness reduction method proposed by Swoboda (1979). The excavation
core and surrounding rock mass are initially assigned with same mechanical properties
and calculated to get their initial balanced stress state. The properties of the excavation
core are then replaced and assigned by softened unstressed elastic material. The
surrounding rock mass will converge, while the stress filed will rebalanced. The process
of replacement is repeated till the excavation core is totally removed from the model.
Totally ten stages are used to simulate the excavation process. Point M1 right above the
middle of the roof surface (out of excavation core) is used to record the roof
displacement towards opening (u). Point M2 right below the middle of the roof surface
(in the excavation core) is set to record the normal stress on the ground (pi). The normal
stress pi versus roof displacement u is the GRC as shown in Figure 7-4.

Figure 7-4 Illustration of the progressive core replacement method (span = 20 m under
fair rock condition)

Another circular tunnel model with diameter D = H as shown in Figure 7-3(b) is used to
determine the LDP curve for horseshoe cavern by assuming it is the same as that of the
circular tunnel (Vlachopoulos and Diederichs, 2014). The damage depth (Dp) of each
simulation stage is recorded. The simulation stages are associated with the locations

175
Chapter 7 Performance-based support design for horseshoe-shaped rock caverns
using 2D FEM

from the excavation face along the tunnel axis based on the empirical equation
(Vlachopoulos and Diederichs, 2009) as:
u
for x < 0, in the rock mass: u* = = u0*  e x (7.13a)
umax
u0 1 −0.15 R*
for x = 0, at face: u0* = = e (7.13b)
umax 3
3x
−
for x > 0, in the cavern: u* = 1 − (1 − u0* )  e 2 R* (7.13c)

where R* is the normalized plastic radius, R* = RP RT , RT is the tunnel radius, RP is the

radius of plastic zone, umax denotes the maximum radial displacement, u0 is the radial
displacement at the face location.

The SEM parameters and the geological conditions are essential to determine the
support system in the preliminary design stage. The numerical model to analyse the
effects of the subdivision is shown in Figure 7-3(c). The dimensions of the horseshoe
shaped cross-section used for parametric study are scaled with the ratio B/H of 0.74
based on the rock cavern model as shown in Figure 7-3(a). The cavern cross-section is
evenly subdivided into multiple sections based on the cavern height by assuming the
height of top-heading equals to that of benching. It is assumed that the sequenced models
for FF excavation are also valid to simulate the 3D subdivision excavation
(Vlachopoulos and Diederichs, 2014).

7.3.2. Parametric study and characteristic curve

（1）Effects of the Q-value

The effects of Q-value on the EDZ development are investigated using the numerical
analysis of the horseshoe carven with B = 20 m and H = 27 m. The roof displacements
u versus the distance from the excavation face x curves (LDP) are plotted in Figure
7-5(a). It can be seen that the roof displacements increase nonlinearly with respect to
the distances from the excavation face when Q < 4. For Q > 10, the roof displacement
is not changing significantly along the excavation axis which indicates the ground is
strong enough for self-support of the horseshoe carven. If an allowable roof
displacement is set as umax = 20 mm, the unsupported distance for good to very good

176
Chapter 7 Performance-based support design for horseshoe-shaped rock caverns
using 2D FEM

ground (Q > 10) could be more than 10.0 m. For a fair ground condition (4 < Q < 10),
the unsupported distance is less than 6.0 m. For poor or even worse ground conditions
(Q < 4), the roof displacement at the excavation face cannot satisfy the allowable value.

The damage depths versus the distances from the excavation face curves are shown in
Figure 7-5(b). As expected, the smaller the Q-value, the deeper the damage depth is
generated in the surrounding rock mass. The damage depth increases nonlinearly with
respect to the distances from the excavation face. For the cavern span of 20 m, the
rockbolt length is recommended of 5.0 m according to the support categories of the Q-
system (NGI, 2015). According to Eq. (7.3), the depth of EDZ from the excavation
surface is calculated as 3.8 m which can be used to separate the purposes of the rockbolt
as suspension and arching. For Dp > 3.8 m, the rockbolt might not connect to the
undamaged rock mass which indicates the spacing of the rockbolt has to be reduced to
assist the formation of artificial arch in the surround rock mass.

The normal stress versus roof displacement curves (GRC) for the horseshoe cavern
excavated in five ground classes are plotted in Figure 7-5(c). It shows the normal stresses
reduce nonlinearly with respect to the roof displacements. The smaller the Q-value the
larger normal stresses are generated in the surrounding rock mass and thus larger support
pressures are required. Eq. (7.2) could be used to assess the support safety once the
maximum support capacity of the support is known.

0.10
Q=0.4
Roof displacement (u), m

Q=1
0.08
Q=4
Q=10
0.06
Q=40

0.04

umax = 20 mm
0.02

0.00
0.0 2.0 4.0 6.0 8.0 10.0
Distance from the excavation face (x), m

(a)

177
Chapter 7 Performance-based support design for horseshoe-shaped rock caverns
using 2D FEM

25 Q=0.4
Q=1
Damage depth (Dp ), m

20 Q=4
Q=10
15 Q=40

10

5
Dp = 3.8 m
0
0.0 2.0 4.0 6.0 8.0 10.0
Distance from the excavation face (x), m

(b)

10
Normal stress (pi), MPa

Q=0.4

8 Q=1

Q=4
6 Q=10

Q=40
4

0
0.00 0.02 0.04 0.06 0.08 0.10
Roof displacment (u), m

(c)

Figure 7-5 Effects of Q-value on the cavern performance (a) LDP, (b) damage depths
at the roof from the excavation face, and (c) GRC

(2) Effects of the SEM parameters

178
Chapter 7 Performance-based support design for horseshoe-shaped rock caverns
using 2D FEM

The effects of the subdivision of cavern cross-section and the round length or the
maximum unsupported excavation length are investigated. The horseshoe caverns with
span B of 5 m, 8 m, 10 m, 15 m, 20 m, 25 m and 30 m excavated in five different ground
conditions represented by Q-value of 40, 10, 4, 1 and 0.4 are considered. The ratios of
the damage depth to span (Dp/B) versus Q-values are plotted in Figure 7-6(a) which
shows nonlinear relationships. The larger the round length the larger the ratio Dp/B. The
roof displacement versus normal stress curves for the subdivisions using 2-sections, 3-
sections, 4-sections and full face (FF) are shown in Figure 7-6(b). It can be seen that the
normal stress reduces nonlinearly with respect to the roof displacement.

However, there is no obvious relationship for the four subdivision processes. The normal
stress versus u/B plots for certain subdivision excavation and Q-value could indicate the
differences of the subdivision method, see FF and 3 subdivided sections excavated in
rock mass with Q = 1.0 in Figure 7-6(c). More advanced function is still required to
present the relationship between the SEM parameters and the cavern performance.

7.4 Prediction of cavern performance using ANN

The ANN models are built to identify the relationships among the geological condition
parameters, the excavation design parameters and the cavern performances obtained
from the numerical analyses. The ANN is the multi-layer feed forward back-propagation
network which has been widely used in rock engineering for data analysis to find their
complex relationships (Zhao and Ren, 2002; Zhao et al, 2008; Tiryaki, 2008). In this
study, a 4-n-1 structure is used to map the relationships between a set of SEM design
parameters Pi (i.e., ground class P1, width of top heading P2, height of top heading P3,
round length P4) and support performance Oj (i.e., normal stress O1, damage depth O2
or roof displacement O3) as shown in Figure 7-7. The n varies in the three models to
predict the support performance Oj based on the training and testing errors.

179
Chapter 7 Performance-based support design for horseshoe-shaped rock caverns
using 2D FEM

Figure 7-6 Effects of the SEM parameters on cavern performance (a) the round length,
(b) the cavern size, and (c) the cavern size for 3-section excavation in case
of Q = 1.0

180
Chapter 7 Performance-based support design for horseshoe-shaped rock caverns
using 2D FEM

Figure 7-7 Architecture of the ANN models

The transfer function is tangent sigmoid transfer function, denoted as ‘tansig’ in Figure
7-7, for both the hidden layer and the output layer, which can be expressed as,
1
f ( Pi ) = (7.14)
arctan( Pi ) + 1
where f(Pi) is tangent sigmoid transfer function, Pi is the inputs, i = 1 for ground class,
i = 2 for width of top heading, i = 3 for height of top heading, and i = 4 for round length.

There are 700 data generated from the numerical analysis. Approximately 70% of them
are randomly chosen as training data while the rest of them are used for testing. The
input Pi are normalized using the minimum and the maximum magnitudes of the
parameter Pi based on the numerical results as summarized in Table 7-3. The equation
to normalize parameter Pi is shown as,

2 ( Pi − Pi ,min )
Pi ' = −1 (7.15)
Pi ,max − Pi ,min
where Pi,min and Pi, max are the minimum and the maximum magnitudes of the parameter
Pi, respectively.

The network is essentially trained using optimization methods by adapting the weights
and biases of neurons and minimizing the mean square error between the predicted and
the target values. The MATLAB code is attached in Appendix C. The performance of
the ANN models for training the numerical results to predict the damage depth and the

181
Chapter 7 Performance-based support design for horseshoe-shaped rock caverns
using 2D FEM

normal stress are shown in Figure 7-8(a) and (b), respectively. The comparisons between
target and predicted values of the damage depth and the normal stress are shown in
Figure 7-9(a) and (b), respectively. The results show very good agreements between the
predictions and the testing data.

(a) (b)

Figure 7-8 Performances of the ANN models to train the numerical results to predict (a)
the damage depth and (b) the normal stress

(a) (b)

Figure 7-9 Comparisons between the targets and the predicted values of (a) the damage
depth and (b) the normal stress

182
Chapter 7 Performance-based support design for horseshoe-shaped rock caverns
using 2D FEM

Table 7-3 Magnitudes of Pi and Oj for normalization based on numerical results

P1 P2 P3 P4 O1 O2 O3

Height of Roof Inner

Q Span, Round Damage
top heading, displacement, pressure,
value m length, m depth, m
m m MPa

Max. 40 40 54 50 0.3 10 50

Min. 0.1 3 5 0 0 0 0

7.5 Development of an evaluation chart based on ANN models

To directly use the ANN models for preliminary support design, an evaluation chart is
proposed and coded using Microsoft Excel as shown in Figure 7-10. The evaluation
chart contains the following five parts, the inputs of cavern geometry and ground class
(A. Inputs), the prediction of cavern performance using ANN models (B. Prediction),
the SEM design parameters (C. SEM Design), the support design parameters (D.
Support Design), and the support performance based on the predicted results (E.
Performance Functions).

The cavern geometry and ground class in part A have the maximum and minimum
boundaries as summarized in Table 7-3. For part B, the proposed three ANN models are
adopted to generate the response surfaces and predict the damage depth and the stress
on support. Based on the above ANN models, the support performance Oj (i.e., normal
stress O1, damage depth O2 or roof displacement O3) are calculated as,
 n 
O j = f   w ji Pi + b j  (7.16)
 i =1 
where: wji is the weights of the neuron j, and bj is the bias of the neuron j.

183
Chapter 7 Performance-based support design for horseshoe-shaped rock caverns
using 2D FEM

Figure 7-10 Evaluation chart for the preliminary support design

184
Chapter 7 Performance-based support design for horseshoe-shaped rock caverns
using 2D FEM

Part C lists the SEM design parameters Pi (i.e., Q-value P1, width of heading P2, height
of heading P3 and round length P4). The width of heading P2 and height of heading P3
are calculated based on the divisions of span and height respectively. If no subdivision
is considered, P2 is the cavern span, and P3 is the cavern height. The round length P4 is
set according to the NHI (2009). These parameters are the inputs for the ANN models
to predict the normal stress and roof displacement to plot the GRC. In Part D, the
estimations of support capacity provided by the end-anchored rockbolt, the shotcrete
linings and/or the steel set are given according to Hoek (2007). Note that all three
support elements are assumed to act independently, and the bearing capacity and the
stiffness of the compound support system are the accumulation of their bearing capacity
and the stiffness respectively (Özsan and Başarır, 2003). The SCCs of different support
types are plotted with GRC to present the ground-support interaction in the diagram.
The intersection between GRC and SCC indicates the displacement and the support
pressure when the equilibrium is achieved. Part E gives three performance functions as
shown in Eqs. (7.1) to (7.3) to indicate the support performances.

The evaluation chart could be used to estimate the support safety. A shotcrete lining
with thickness tc = 0.1 m is selected as an example. The cavern is full-face excavated in
rock with UCS and Ec of 35 MPa and 35 GPa, respectively. The calculated support
capacity criterion g2(x) > 0 means the support is suitable for the ground conditions and
cavern size. As shown in Table 7-4, the shotcrete lining support with tc = 0.1 m could
only be effective in good rock conditions with Q ≥ 10 or narrow span caverns with B <
10 m. The data used to generate Table 4 are also shown in Figure 7-10 where the GRC
and SCC curves are also included to illustrate the calculation process.

The evaluation chart could be used to assess the functions of the patterned rockbolt to
support rock cavern. The damage depth can be predicted using the ANN model.
Example of the calculated rockbolt length criterion function g3(x), see Eq. (7.3), is
shown in Table 7-5. The functions of the patterned rockbolts for rock cavern support
could be generally separated as suspension element when g3(x) > 0 and arching element
when g3(x) ≤ 0 (Li, 2017). Table 7-5 shows that the patterned rockbolts are suspension
elements for 4 < Q < 10 and 20 < B < 30 m. The pattern rockbolt are arching elements
for the rest cases. The rockbolt lengths are much less than the sum of damage depth and
anchored length, g3(x) < - 2 m, for Q = 0.4 and B > 8.0 m.

185
Chapter 7 Performance-based support design for horseshoe-shaped rock caverns
using 2D FEM

Table 7-4 Predictions of the support safety function g2(x) for shotcrete linings with
thickness tc = 0.1 m
Q
0.4 1 4 10
Span, m tc = 0.1 m
30 -1.15 -1.07 -0.14 -0.28 UCS = 35 MPa
25 -1.12 -0.90 -0.10 -0.09 Ec = 35 GPa
20 -0.74 -0.42 -0.09 0.07 Legend
15 -0.63 -0.38 0.03 0.25 g2(x)≥1
10 -0.23 0.04 0.45 0.60 0≤g2(x)<1
8 0.30 0.32 0.69 0.80 -0.5≤g2(x)<0
5 0.67 0.97 1.27 1.34 g2(x)<-0.5

Table 7-5 Predictions of the rockbolt length criterion function g3(x) for rockbolt support
Q
0.4 1 4 10
Span, m
30 -7.0 -0.3 0.8 1.3
25 -5.8 -0.6 0.4 0.5 Legend
20 -5.1 -1.3 -0.7 0.0 g3(x)>0
15 -3.7 -1.2 -0.8 -0.8 -1≤g3(x)<0
10 -2.6 -1.3 -1.2 -0.9 -2≤g3(x)<-1
8 -2.1 -1.0 -1.0 -0.8 -3≤g3(x)<-2
5 -1.2 -0.6 -0.4 -0.5 g3(x)<-3

The evaluation chart could be used to optimize the subdivisions of the excavation cross
section. The division of span and height would be optimized using displacement
criterion g1(x). This could be achieved using “Solver” function by minimizing cell D22,
and changing cell E11 and cell E12 with their magnitudes of integers and less than 4
(assumed the division of the segment is less than 4 parts). The width and height of
heading are obtained by dividing the span based on the optimized results. As shown in
Figure 7-10, cells E11 and E12 are the optimization results for the FF excavation. The
width and height of heading are then obtained as shown in cells C11 and C12,
respectively, in Figure 7-10. The displacement after balance is u = 0.018 m which is
close to the allowable displacement umax = 0.02 m. It should be noted that the evaluation
chart provides an estimation of the ground-support interaction for preliminary design as
the ANN models are build based on the 2D numerical modelling which might not be

186
Chapter 7 Performance-based support design for horseshoe-shaped rock caverns
using 2D FEM

exactly representing the excavation process and time effect in real 3 dimensional
conditions.

7.6 A case study

The evaluation chart is applied to evaluate a real design case of Kaletepe tunnel
presented by Sari and Pasamehmetoglu (2004). The highway tunnel with a wall height
of 9.6 m, a width of 12.7 m and a length of 2.5 km was excavated under a hill with the
maximum overburden of approximately 300 m. The test results of core specimen taken
from six borehole sections and site investigations were used as the input parameters to
assess the geological conditions. The support recommendation and excavation guide
preliminary support design proposed by Sari and Pasamehmetoglu (2004) are shown in
Table 7-6. In previous studies, the tunnel length was divided into seven sections (section
1 to 7), along its axis for the preliminary support design. Only sections 2, 5 and 7 are
selected in this study to illustrate the applications of the proposed evaluation chart. An
allowable displacement is assumed as umax = 0.02 m. The support capacities are based
on those listed in Table 7-1.

For section 2, the length and the spacing of the rockbolt are 3.9 m and 0.8 m, respectively,
which are the average data given in Table 7-6. The sequential excavation method is top
heading and benching. The thickness of the shotcrete lining is 0.25 m. The calculated
GRC and SCC are shown in Figure 7-11(a). The intersection between SCC and GRC
determines the displacement and the support pressure which are 0.024 m and 0.207 MPa,
respectively. However, the displacement of 0.024 m is larger than umax = 0.02 m
indicating the support design should be optimized. Moreover, the SCC curve of
shotcrete shows the pressure on support is much less than its bearing capacity which
means the support is not utilized wisely. Figure 7-12(a) gives the calculated GRC and
SCC curves of the shotcrete lining and rockbolt support with optimized design
parameter, such as round length of 5 m and thickness of shotcrete lining of 0.15 m. The
intersection between SCC and GRC curves shows the support system could restrain the
displacement close to 0.02 m within its bearing capacity.

187
Chapter 7 Performance-based support design for horseshoe-shaped rock caverns
using 2D FEM

(a) (b) (c)

Figure 7-11 Calculated GRCs and SCCs using evaluation chart to assess the support
designs proposed by Sari & Pasamehmetoglu (2004) for (a) Section 2, (b)
Section 7, and (c) Section 5.

The GRCs and SCCs are calculated using the evaluation chart to assess the support
design proposed by Sari & Pasamehmetoglu (2004) for Section 5 and Section 7. The
support parameters are also given in Table 7-6. The calculated GRCs and SCCs of the
support designs for Section 7 and Section 5 are shown in Figure 7-11(a) and (b),
respectively. The intersections between SCC and GRC show the displacements are
smaller than the allowable displacement umax = 0.02 m which means the support system
could satisfy the requirement. For Section 7, Figure 7-11(b) shows the shotcrete linings
and the rockbolt can provide sufficient support pressure to restrain the displacement.
However, the predicted support pressure is 0.027 MPa which is only 10% the bearing
capacity of the shotcrete lining. Less support might be also effective, such as set the
spacing of rockbolt to 2.0 m. As shown in Figure 7-12(b), the SCCs and GRCs show the
support can restrain the displacement within 0.02 m. For Section 5, the intersection
between SCC and GRC in Figure 7-11(c) shows the rockbolt has no effective support
which means the shotcrete is strong enough to provide the support pressure. The spot
rockbolt could be used to provide the support pressure as shown in Figure 7-12(c).

Table 7-6 Support design parameters (B – bolting, SL – shotcrete lining)

188
Chapter 7 Performance-based support design for horseshoe-shaped rock caverns
using 2D FEM

Current research according to

Proposed by Sari and Pasamehmetoglu (2004)
evaluation chart
Section 2 Section 7 Section 5 Section 2 Section 7 Section 5
Ground condition
Poor Fair Good Poor Fair Good
Round
10 20 20 5 20 20
parameter

length, m
SEM

Top Top
Sub-
heading & FF FF heading & FF FF
division
benching benching
Length, m 3.9 3.9 3.9 3.9 3.9 3.9
B

Spot
Spacing, m 0.5 - 1.0 1 - 1.5 2.5 - 3 0.8 2
bolt
SL

Thickness, m 0.2 - 0.3 0.05 0.03 0.15 0.05 0.03

(a) (b) (c)

Figure 7-12 GRCs and SCCs to evaluate the revised support designs for (a) Section 2,
(b) Section 7, and (c) Section 5.

To further evaluate the performance of the support design parameters proposed by

current study and Sari & Pasamehmetoglu (2004), series of numerical simulations are
carried out. The displacement contours around the excavation face without support, with
support system in current study and with support system by Sari & Pasamehmetoglu
(2004) are shown in Table 7-7. The maximum displacement and the damage depth of
EDZ from numerical analysis and evaluation chart are also listed in the table for
comparison purpose. It can be seen that the support designs proposed by current study
and Sari & Pasamehmetoglu (2004) can restrain the roof displacement within umax =

189
Chapter 7 Performance-based support design for horseshoe-shaped rock caverns
using 2D FEM

0.02 m. The revised support design using evaluation chart proposed in this study adopts
less supports. The roof displacement and the damage depth of EDZ resulted in numerical
analysis and evaluation chart are further plotted in Figure 7-13(a) and (b), respectively.
The average difference between the numerical analysis and evaluation chart to calculate
the maximum displacement and the damage depth of EDZ are 12% and 1.2%,
respectively.

0.04
u from proposed eveluation

Without spport

0.03 Support by current study

chart, m

Support by Sari and y = 0.8786x

0.02 Pasamehmetoglu (2004) R²= 0.9776

0.01

0
0 0.01 0.02 0.03 0.04
u from numerical method, m
(a)
6
Without spport

Support by current study

Dp from proposed eveluation

5
Support by Sari and
Pasamehmetoglu (2004) y = 0.989x
4
R²= 0.8345
chart, m

2
2 3 4 5 6
Dp from numerical method, m
(b)

Figure 7-13 Comparisons between the results from numerical analysis and evaluation
chart in terms of (a) displacement and (b) damage depth of EDZ from the
excavation surface

Table 7-7 Displacement contours around the tunnel with/without supports

190
Chapter 7 Performance-based support design for horseshoe-shaped rock caverns
using 2D FEM

Section 2 Section 7 Section 5

Without support

ui Dpi ui Dpi ui Dpi

Num. 0.033 5.259 Num. 0.018 4.432 Num. 0.0084 4.123
Chart 0.028 4.849 Chart 0.016 4.823 Chart 0.0082 3.968
Error 15.15% 7.80% Error 11.11% 8.82% Error 2.38% 3.76%
With support as reference

ui Dpi ui Dpi ui Dpi

Num. 0.025 4.321 Num. 0.016 3.934 Num. 0.0074 3.438
Chart 0.022 4.558 Chart 0.0136 3.967 Chart 0.0082 3.217
Error 12.00% 5.48% Error 15.00% 0.84% Error 10.81% 6.43%
Revised support design

ui Dpi ui Dpi ui Dpi

Num. 0.023 4.171 Num. 0.015 4.031 Num. 0.0072 3.427
Chart 0.020 4.155 Chart 0.0137 3.967 Chart 0.0082 3.217
Error 13.04% 0.38% Error 8.67% 1.59% Error 13.89% 6.13%

7.7 Conclusions

A support design method for horseshoe-shaped rock caverns is proposed in this chapter
with considerations of the progressive damage of the rock mass using the 2D numerical
modelling and the artificial neural network (ANN). The excavation effects of the rock

191
Chapter 7 Performance-based support design for horseshoe-shaped rock caverns
using 2D FEM

cavern are simulated using the 2D finite element plane strain models. The performances
of the rock cavern during excavation are investigated based on the convergence-
confinement method (CCM).

Parametric studies are conducted to analyse the effects of the Q-values on the EDZ
development of rock cavern with a span of 20 m and a height of 27 m under isotropic
stress of 10 MPa. It is found that the roof displacements versus the distances from the
excavation face curves increase nonlinearly when Q < 4 and change insignificantly for
Q > 10. Furthermore, the smaller the Q-value the deeper the damage depth is generated
in the surrounding rock mass. The spacing of the rockbolt has to be reduced for poor
ground classes (Q ≤ 1) to assist the formation of the artificial arch in the surround rock
mass. The effects of sequential excavation parameters are also investigated by changing
the size of rock cavern and the subdivision of cross-sections. It is found that the
subdivision can reduce the range of the EDZ but there are no obvious relationships for
different subdivision methods. More advanced function is still required to present the
relationship between the SEM parameters and the cavern performance.

The ANN models are built using the numerical results to find the complex relationships
among the rock mass condition, the sequential excavation parameters and the cavern
performances. Totally 700 data are generated from the numerical analysis.
Approximately 70% of the results are randomly chosen as training data and the rest of
them are used for testing. Good agreements between the predictions and the targets are
obtained. An evaluation chart is proposed by integrating the ANN models into the
EXCEL software. The proposed evaluation chart provides an effective method to
evaluate the support safety, the functions of the patterned rockbolt and the optimization
of subdivisions of the excavation cross section. The evaluation chart is applied to
evaluate a real case design of Kaletepe tunnel. Comparing to the preliminary design
presented by Sari and Pasamehmetoglu (2004), the support designs proposed by the
current study can restrain the roof displacement. It should be noted that the evaluation
chart is feasible to estimate the support for the preliminary design as the ANN models
are built based on the 2D numerical results which might not be exactly representing the
complex excavation process and time effect.

192
 Chapter 8 Conclusions and future work

CONCLUSIONS AND FUTURE WORK

8.1 Conclusions

Underground space has been increasingly exploited in recent years. The potential for
instability in the rock mass surrounding the underground excavation is a threat to the
lives and safety of construction workers. The rockbolt has been the primary
reinforcement structure in underground tunnelling and mining. To mobilize the
reinforcement efforts and improve the design of the rockbolt, the reinforcement
mechanism of a rockbolt system in underground excavations is studied using 2D-DDA.
The following conclusions have been made through the studies:

(1) Pullout performance of the CMC rockbolt element

The 2D-DDA method is used to investigate the pullout performance of the CMC
rockbolt element which is modeled as three components, i.e. the rock, the rockbolt and
the mortar. The material properties of mortar blocks are adopted from the experiment
result proposed from Yokota et al. (2018). The frictional properties along rockbolt –
mortar interface are calibrated with their experimental results. The flat joint contact
model is introduced into the 2D-DDA code to simulate the force versus displacement
behavior of an artificial joint. The results show that a bond stress versus slip
displacement curve of the CMC element is generally exhibiting three stages and
representing by five key parameters, i.e., the bond stiffness in three stages (k1, k2, and
k3), the maximum bond strength τmax, and the residual bond strength τres. Parametric
studies have been carried out to investigate the effects of the normal stress and those of
the rib profiles on the bond stress versus slip displacement curves and fracturing modes
in mortar. It is found that the normal stress plays an important role in the bond-slip
modeling. The bond stiffness k1, the maximum bond strength τmax and the residual bond
strength τres are increasing with the increase of normal stress. The rib spacing may
193
Chapter 8 Conclusions and future work

influence the bond stiffness of a bond-slip model in the way that a larger rib spacing will
result in lower bond stiffness when the other parameters are the same. Under a lower
normal stress, the interlock of the CMC rockbolt element with rib face angles of less
than 30°is not effective because the rockbolt elements are pulled out along the rib faces
and less cracks are generated in the mortar blocks.

(2) Developed rockbolt model in 2D-DDA and verified the model

The developed rockbolt element is capable to simulate different types of rockbolt under
various load conditions with consideration of three behaviours: the axial behaviour, the
shearing behaviour and the bond behaviour. By adopting different models, the four
major failure modes of rockbolt could be simulated: (1) decoupling along the interface,
(2) tensile failure, (3) faceplate failure and (4) shear failure at joint. Verifications of the
rockbolt element are also carried out. The bond behaviour is verified using pullout test
results. The simulation results show good agreements between the proposed model and
the experiments results. By comparing with shear tests, the numerical results show good
agreements in the reduction of reinforcement effects. The shear failure of the rockbolt
elements at joint is also presented. The axial behaviour is verified using pullout test
results. Two types of rockbolts, i.e. the fully grouted rockbolt and the D-bolt, are
simulated under pullout condition. The simulated load-deformation curves show good
agreements with the experimental ones. It also confirms the larger deformability of the
D-bolt compared with the fully grouted rebar as a result of equally loaded rockbolt
elements between special anchored nodes. By verification of the proposed model, it
indicates that the proposed rockbolt model can simulate the rock reinforcement using
rockbolts. Especially, it can simulate several types of rockbolt restrains of jointed rock
blocks which are under large deformation conditions.

(3) Reinforced rock unit in jointed rock mass

The investigation of reinforced rock unit (RRU) of rockbolts in jointed rock mass is
carried out using the newly developed rockbolt element in 2D-DDA. In jointed rock
mass, the RRU should be considered with regards to the displacement and the conditions
of rock mass discontinuities. In general, the RRU is cone-shaped, but its boundary may
be limited by the existing joint sets, resulting in a non-triangular cross section in some

194
 Chapter 8 Conclusions and future work

cases. A parametric study has been carried out to show that the friction of joints and the
stiffness of rock blocks will affect the RRU area.

(4) Stability analysis of rockbolt reinforced rock wedge

The stability of the rockbolt reinforced roof wedges is investigated using 2D-DDA. The
change of stress state in the surrounding rock blocks after excavation is reparented using
the joint relaxation method. The movement of rock wedge induced by its dead weight
and the re-adjusted stress in field are presented. The influences of the horizontal pressure
on the wedge stability are investigated. It shows the roof wedge has potential to fall if
the semi-apical angle is less than the friction angle along the wedge boundary. The
critical horizontal pressure could be calculated using 2D-DDA that the wedge could be
self-supported after stress redistribution if the horizontal pressure is larger than it. The
results agree reasonably with the analytical solutions. The effects of the rockbolt spacing
and installation angle are also presented. The results show the axial forces of rockbolt
may slightly decrease when narrowing the rockbolt spacing but increase if the
installation angle increase.

A case study is presented to show the application of the 2D-DDA models for the
assessment of the rockbolting design to support roof wedges. Using the rock/rockbolt
interaction diagrams, the results could be used by the site engineers to evaluate the
rockbolting design to sustain the possible roof wedges.

(5) Performance-based support design for rock caverns

A support design method for horseshoe-shaped rock caverns is proposed with

considerations of the progressive damage of the rock mass using the RS2 and the
artificial neural network (ANN). The excavation effects of the rock cavern are simulated
using the 2D finite element plane strain models. The performances of the rock cavern
during excavation are investigated based on the convergence-confinement method
(CCM). Parametric studies are conducted to analyse the effects of the Q-values on the
EDZ development of rock cavern. The effects of sequential excavation parameters are
also investigated by changing the size of rock cavern and the subdivision of cross-
sections. The ANN models are built using the numerical results to find the relationships
among the rock mass condition, the sequential excavation parameters and the cavern

195
Chapter 8 Conclusions and future work

performances. An evaluation chart is proposed by integrating the ANN models into the
EXCEL software. The proposed evaluation chart provides an effective method to
evaluate the support safety, the functions of the patterned rockbolt and the optimization
of subdivisions of the excavation cross section. The evaluation chart is applied to
evaluate a real case design of Kaletepe tunnel. Comparing to the preliminary design
presented by Sari and Pasamehmetoglu (2004), the support designs proposed by the
current study can restrain the roof displacement. It should be noted that the evaluation
chart is feasible to estimate of the support for the preliminary design as the ANN models
are built based on the 2D numerical results which might not be exactly representing the
complex excavation process and time effect.

8.2 Future work

The present work attempted to improve the rockbolt design in complex rock conditions
and to have a good understanding of the reinforcement mechanism. Some
recommendations and suggestions for further studies are proposed as follows.

(1) As the expansion of the underground construction, many new types of rockbolts
have been developed to satisfy the support requirements. The developed rockbolt
model could be used to present types of rockbolts if the bond-slip models are
available for the rockbolt elements. The proposed rockbolt model can be used to
simulate the loading behavior of new proposed rockbolt element, such as CFC
element and DMFC element. As the current model is based on 2D numerical model
which could not consider the twist failure of the rockbolt, 3D numerical model with
spatial rockbolt arrangement should be developed to better understand their
reinforcement effects.

(2) The current studies focus on the reinforcement effects of rockbolts. However, other
structures are also used to support the underground opening, i.e. shotcrete lining,
steel set, etc. The interactions mechanisms between these support structures and
surrounding rock mass have not been fully understood yet. Future studies on the
working principles of the combined support system should be carried out. The
combined support stiffness and support capacity have to be used for support
designing.

196
 Chapter 8 Conclusions and future work

(3) One of the design principles of NATM is to take advantages of the inherent
geological strength available in the surrounding rock mass to stabilize the tunnel in
terms of economic consideration. The CCM method was use for preliminary
designing based only limited data. The program can be improved when the working
principles of the combined support system are better understood. The ground
reaction curves in complex ground conditions should be studied to increase the
reliability of the ANN models. Through building a more comprehensive database,
the rock/support diagram could be estimated during the construction progress and
help the site engineers to assess and mitigate the possible risks.

197
Chapter 8 Conclusions and future work

198
 References

REFERENCES

Alejano, L.R., Alonso, E., Rodrí

guez-Dono, A., Fernández-Manín, G. (2010).
“Application of the convergence-confinement method to tunnels in rock masses
exhibiting Hoek–Brown strain-softening behaviour”. International Journal of Rock
Mechanics and Mining Sciences 47, 150-160.
Alejano, L.R., Rodriguez-Dono, A., Alonso, E., Fdez.-Manín, G. (2009) “Ground
reaction curves for tunnels excavated in different quality rock masses showing
several types of post-failure behaviour”. Tunnelling and Underground Space
Technology 24, 689-705.
Aydin, A., Ozbek, A.,Acar, A. (2014) “Geomechanical characterization, 3-D optical
monitoring and numerical modeling in Kirkgecit-1 tunnel, Turkey”. Engineering
Geology 181, 38-47.
Aziz, N., Hossein, J., Hadi, M.S.N. (2005) “The Influence of Resin Thickness on Bolt
Bending”. The 19th International Mining Congress and Fair of Turkey,
IMCET2005, İzmir, Turkey, June 09-12, 65-71.
Aziz, N., Jalalifar, H., Remennikov, A., Sinclair, S., Green, A. (2008) “Optimisation of
the Bolt Profle Confguration for Load Transfer Enhancement”. In: AZIZ, N. (ed.)
Coal Operators' Conference. University of Wollongong & the Australasian Institute
of Mining and Metallurgy, pp. 125-131.
Bao, H. (2010) “Nodal-based discontinuous deformation analysis”, Ph.D. Thesis,
Nanyang Technological University, Singapore.
Barton, N., Lien, R., Lunde, J. (1974) “Engineering classification of rock masses for the
design of tunnel support”. Rock mechanics 6, 189-236.
Barton N. and Choubey V. (1978) Recent experiences with the Q-system of tunnel
support design: Oslo : Norwegian Geotechnical Institute.
Bawden, W.F. (2011) “Ground control using cable and rock bolting”, In: Darling, P.
(Ed.), SME Mining Engineering Handbook (3rd Edition). Society for Mining,
Metallurgy, and Exploration (SME).
199
References

Benmokrane, B., Chennouf, A., Mitri, H.S. (1995) “Laboratory evaluation of cement-
based grouts and grouted rock anchors”. International Journal of Rock Mechanics
and Mining Sciences & Geomechanics Abstracts 32, 633-642.
Bertuzzi, R., Douglas, K., Mostyn, G. (2016) “Comparison of quantified and chart GSI
for four rock masses”. Engineering Geology 202, 24-35.
Bieniawski, Z. T. (1978) “Determining rock mass deformability: experience from case
histories”. International Journal of Rock Mechanics and Mining Sciences &
Geomechanics Abstracts 15, 237-247.
Bieniawski, Z. T. (1984) Rock Mechanics Design in Mining and Caverning, A. A.
Balkema, Rotterdam, 97-133.
Bieniawski, Z. T. (1989) Engineering Rock Mass Classifications, John Wiley and Sons,
New York
Blanco Martín, L., Tijani, M., Hadj-Hassen, F. (2011) “A new analytical solution to the
mechanical behaviour of fully grouted rockbolts subjected to pull-out tests”.
Construction and Building Materials 25, 749-755.
Blanco Martín, L., Tijani, M., Hadj-Hassen, F., Noiret, A. (2013) “Assessment of the
bolt-grout interface behaviour of fully grouted rockbolts from laboratory
experiments under axial loads”. International Journal of Rock Mechanics and
Mining Sciences 63, 50-61.
Bobet, A. and Einstein, H.H., 2011. “Tunnel reinforcement with rockbolts”. Tunnelling
and Underground Space Technology 26, 100-123.
Brady, B.H.G. and Brown, E.T. (2006) “Rock support and reinforcement”. Rock
Mechanics for underground mining (3rd edition). Springer Netherlands, Dordrecht,
pp. 312-346.
Brown, E.T., ASCE, M., Bray, J.W., Branko Ladanyi, ASCE, F., Hoek, E. (1983)
“Ground Response Curves for Rock Tunnels”. Journal of Geotechnical Engineering
109, 15-39.
Cai, M. (2008) “Influence of stress path on tunnel excavation response – Numerical tool
selection and modeling strategy”. Tunnelling and Underground Space Technology
23, 618-628.
Cao, C., Nemcik, J., Aziz, N., Ren, T. (2013) “Analytical study of steel bolt profile and
its influence on bolt load transfer”. International Journal of Rock Mechanics and
Mining Sciences 60, 188-195.

200
 References

Cao, C., Ren, T., Cook, C., Cao, Y. (2014) “Analytical approach in optimising selection
of rebar bolts in preventing rock bolting failure”. International Journal of Rock
Mechanics and Mining Sciences 72, 16-25.
Carranza-Torres, C. and Fairhurst, C. (2000) “Application of the Convergence-
Confinement method of tunnel design to rock masses that satisfy the Hoek-Brown
failure criterion”. Tunnelling and Underground Space Technology 15, 187-213.
Chen, Y.and Li, C.C. (2015) “Performance of fully encapsulated rebar bolts and D-Bolts
under combined pull-and-shear loading”. Tunnelling and Underground Space
Technology 45, 99-106.
Choquet, P. and Hadjigeorgiou, J. (1993) The design of support for underground
excavations. Oxford; New York : Pergamon Press, 1993.
Crawford, A.M. and Bray, J.W. (1983) “Influence of the in-situ stress field and joint
stiffness on rock wedge stability in underground openings”. Canadian Geotechnical
Journal 20, 276-287.
Deb, D. and Das, K.C. (2011) “Modelling of fully grouted rock bolt based on enriched
finite element method”. International Journal of Rock Mechanics and Mining
Sciences 48, 283-293.
Deb, D. and Das, K.C. (2014) “A new doubly enriched finite element for modelling
grouted bolt crossed by rock joint”. International Journal of Rock Mechanics and
Mining Sciences 70, 47-58.
Deb, D., Gujjala, Y.K., 2018. “Extended finite element procedures for analysis of bolt
crossing multiple intersecting rock joints”. International Journal of Rock Mechanics
and Mining Sciences 107, 249-260.
Dias, D. (2011) “Convergence-confinement approach for designing tunnel face
reinforcement by horizontal bolting”. Tunnelling and Underground Space
Technology 26, 517-523.
Dwivedi, R.D., Singh, M., Viladkar, M.N., Goel, R.K. (2014) “Estimation of support
pressure during tunnelling through squeezing grounds”. Engineering Geology, 168,
9-22.
Farmer, I.W. (1975) “Stress distribution along a resin grouted rock anchor”.
International Journal of Rock Mechanics and Mining Sciences & Geomechanics
Abstracts 12, 347-351.

201
References

Feng, X., Zhang, N., Yang, S., He, F. (2018) “Mechanical response of fully bonded bolts
under cyclic load”. International Journal of Rock Mechanics and Mining Sciences
109, 138-154.
Feng, X.T., Guo, H.S., Yang, C.X., Li, S.J. (2018) “In situ observation and evaluation
of zonal disintegration affected by existing fractures in deep hard rock tunnelling”.
Engineering Geology 242, 1-11.
Forbes, B., Vlachopoulos, N., Hyett, A.J., Diederichs, M.S. (2017) “A new optical
sensing technique for monitoring shear of rock bolts”. Tunnelling and Underground
Space Technology 66, 34-46.
Gerdeen, J.C., Snyder, V.W., Viegelahn, G.L.U.O. (1977) “Design criteria for roof
bolting plans using fully resin-grouted nontensioned bolts to reinforce bedded mine
roof”. Synthesis and design criteria 46(5), 129
Ghadimi, M., Shahriar, K., Jalalifar, H. (2014) “Analysis profile of the fully grouted
rock bolt in jointed rock using analytical and numerical methods”. International
Journal of Mining Science and Technology 24, 609-615.
Goodman, R. E. and Shi, G.H. (1985) Block Theory and Its Application to Rock
Engineering. Prentice-Hall, INC.
Goodman, R.E. (1995) “Block theory and its application”. Géotechnique 45, 383-423.
Grasselli, G. (2005) “3D Behaviour of bolted rock joints: experimental and numerical
study”. International Journal of Rock Mechanics and Mining Sciences 42, 13-24.
Gurung, N. (2001) “1-D analytical solution for extensible and inextensible soil/rock
reinforcement in pull-out tests”. Geotextiles and Geomembranes 19, 195-212.
Hatzor, H. Yossef, Ma, G., Shi, G.H. (2018) “Discontinuous deformation analysis in
rock mechanics practice”. ISRM Book Series, CRC Press, London, UK.
He, L., An, X.M., Zhao, Z.Y. (2015) “Fully grouted rock bolts: an analytical
investigation”. Rock Mechanics and Rock Engineering 48, 1181-1196.
He, L., An, X.M., Zhao, X.B., Zhao, Z.Y., Zhao, J. (2018) “Development of a unified
rock bolt model in discontinuous deformation analysis”. Rock Mechanics and Rock
Engineering 51, 827-847.
He, M., Gong, W., Wang, J., Peng, Q., Tao, Z., Du, S., Peng, Y. (2014) “Development
of a novel energy-absorbing bolt with extraordinarily large elongation and constant
resistance”. International Journal of Rock Mechanics and Mining Sciences 67, 29-
42.

202
 References

Hijazo, T. and González de Vallejo, L.I. (2012) “In-situ stress amplification due to
geological factors in tunnels: The case of Pajares Tunnels, Spain. Engineering
Geology, 137-138:13-20
Hobst, L. and Zajíc, J., (1983) "Anchoring in rock and soil", in: LeoŠ, H., Josef, Z.
(Eds.), Developments in Geotechnical Engineering. Elsevier, pp. 60-88.
Hoek, E., (2007) Practical rock engineering. E-book, https://www.rocscience.com.
Hoek, E., Kaiser P.K., Bawden, W.F. (2000) “Rockbolts and dowels”, Support of
Underground Excavations in Hard Rock. CRC Press, pp. 152-164.
Hyett, A. J., Bawden, W. F., Reichert, R. D. (1992) “The effect of rock mass
confinement on the bond strength of fully grouted cable bolts”. International Journal
of Rock Mechanics and Mining Sciences & Geomechanics Abstracts, 29, 503-524.
Hyett, A.J., Bawden, W.F., Macsporran, G.R., Moosavi, M. (1995) "A constitutive law
for bond failure of fully-grouted cable bolts using a modified hoek cell".
International Journal of Rock Mechanics and Mining Sciences & Geomechanics
Abstracts 32, 11-36.
Itasca Consulting Group, Inc., (2011) UDEC Ver3.0, Special features. Mineapolis:
Itasca Consulting Group, Inc.
Itasca Consulting Group, Inc., (2012) FLAC Ver5.0, Structural Elements. Mineapolis:
Itasca Consulting Group, Inc.
Ito, F., Nakahara, F., Kawano, R., Kang, S.-S., Obara, Y. (2001) “Visualization of
failure in a pull-out test of cable bolts using X-ray CT”. Construction and Building
Materials 15, 263-270.
Jalalifar, H., Aziz, N., Hadi, M., (2006) "The effect of surface profile, rock strength and
pretension load on bending behaviour of fully grouted bolts". Geotechnical &
Geological Engineering 24, 1203-1227.
Janin, J.P., Dias, D., Emeriault, F., Kastner, R., Le Bissonnais, H., Guilloux, A. (2015)
“Numerical back-analysis of the southern Toulon tunnel measurements: A
comparison of 3D and 2D approaches”. Engineering Geology 195, 42-52.
Jiao, Y.Y., Zhang, H.Q., Tang, H.M., Zhang, X.L., Adoko, A.C., Tian, H.N. (2014)
“Simulating the process of reservoir-impoundment-induced landslide using the
extended DDA method”. Engineering Geology, 182, 37-48.
Jiao, Y.Y., Zhang, X.L., Zhao, J. (2012) “Two-Dimensional DDA Contact Constitutive
Model for Simulating Rock Fragmentation”. Journal of Engineering Mechanics
138(2):199-209.
203
References

Jing, L. (2003) "A review of techniques, advances and outstanding issues in numerical
modelling for rock mechanics and rock engineering". International Journal of Rock
Mechanics and Mining Sciences 40, 283-353.
Kaiser, P.K., Yazici, S., Nosé, J. (1992) "Effect of stress change on the bond strength of
fully grouted cables". International Journal of Rock Mechanics and Mining
Sciences & Geomechanics Abstracts 29, 293-306.
Karakus, M., (2007) "Appraising the methods accounting for 3D tunnelling effects in
2D plane strain FE analysis". Tunnelling and Underground Space Technology 22,
47-56.
Kim, Y.L., Amadei, B., Pan, E. (1999) “Modeling the effect of water, excavation
sequence and rock reinforcement with discontinuous deformation analysis”.
International Journal of Rock Mechanics and Mining Sciences 36, 949-970
Kitchah, F. and Benmebarek, S. (2016) “Finite difference analysis of an advance core
pre-reinforcement system for Toulon's south tube”. Journal of Rock Mechanics and
Geotechnical Engineering 8, 703-713.
Kılıc, A., Yasar, E., Atis, C.D., (2003) "Effect of bar shape on the pull-out capacity of
fully-grouted rockbolts". Tunnelling and Underground Space Technology 18, 1-6.
Kılıc, A., Yasar, E., Celik, A.G., (2002) "Effect of grout properties on the pull-out load
capacity of fully grouted rock bolt". Tunnelling and Underground Space
Technology 17, 355-362.
Ko, H., Matthys, S., Palmieri, A., Sato, Y., (2014) "Development of a simplified bond
stress–slip model for bonded FRP–concrete interfaces". Construction and Building
Materials 68, 142-157.
Kwon, S., Lee, C.S., Cho, S.J., Jeon, S.W., Cho, W.J. (2009) “An investigation of the
excavation damaged zone at the KAERI underground research tunnel”. Tunnelling
and Underground Space Technology 24, 1-13.
Lang, T.A. and Bischoff, J.A. (1982) "Stabilization Of Rock Excavations Using Rock
Reinforcement". in: Proceedings of the Symposium on Rock Materials, American
Rock Mechanics Association, pp. 935–944.
Lang, T.A., Bischoff, J.A., (1983) "Research study of coal mine rock reinforcement":
US Bureau of Mines report OFR 72-82, Jan 1981, 227P. International Journal of
Rock Mechanics and Mining Sciences & Geomechanics Abstracts 20, A27-A28.

204
 References

Lee, S. W., Kang, S.-B., Tan, K. H., Yang, E.-H. (2016) “Experimental and analytical
investigation on bond-slip behaviour of deformed bars embedded in engineered
cementitious composites”. Construction and Building Materials 127, 494-503.
Li, C. and Stillborg, B. (1999) "Analytical models for rock bolts". International Journal
of Rock Mechanics and Mining Sciences 36, 1013-1029.
Li, C.C. (2006) "Rock support design based on the concept of pressure arch".
International Journal of Rock Mechanics and Mining Sciences 43, 1083-1090.
Li, C.C. (2010) “A new energy-absorbing bolt for rock support in high stress rock
masses”. International Journal of Rock Mechanics and Mining Sciences. 47(3),
396-404.
Li CC, Stjern G, Myrvang A. (2014) “A review on the performance of conventional and
energy-absorbing rockbolts”. Journal of Rock Mechanics and Geotechnical
Engineering 6(4), 315-327.
Li, C.C. (2017a) “Chapter Five - Rockbolting Design”, Rockbolting. Butterworth-
Heinemann, pp. 125-176.
Li, C.C. (2017b) “Principles of rockbolting design”. Journal of Rock Mechanics and
Geotechnical Engineering 9, 396-414.
Li, C.C., Kristjansson, G., Høien, A.H. (2016) “Critical embedment length and bond
strength of fully encapsulated rebar rockbolts”. Tunnelling and Underground Space
Technology 59, 16-23.
Li, C.C., Stjern, G., Myrvang, A., (2014) "A review on the performance of conventional
and energy-absorbing rockbolts". Journal of Rock Mechanics and Geotechnical
Engineering 6, 315-327.
Li, C.C, (2010) “A new energy-absorbing bolt for rock support in high stress rock
masses”. International Journal of Rock Mechanics and Mining Sciences 47, 396-
404.
Li, L., Hagan, P.C., Saydam, S., Hebblewhite, B., (2016) "Shear resistance contribution
of support systems in double shear test". Tunnelling and Underground Space
Technology 56, 168-175.
Li, X., Nemcik, J., Mirzaghorbanali, A., Aziz, N., Rasekh, H., (2015) "Analytical model
of shear behaviour of a fully grouted cable bolt subjected to shearing". International
Journal of Rock Mechanics and Mining Sciences 80, 31-39.

205
References

Lin, CT, Amadei, B, Jung, J, Dwyer, J. (1996) “Extension of discontinuous deformation analysis
for jointed rock masses”. International Journal of Rock Mechanics and Mining Sciences
and Geomechanical Abstract 33, 671–694.
Liu, J., Yang, H., Wen, H., Zhou, X. (2017) “Analytical model for the load transmission
law of rock bolt subjected to open and sliding joint displacements”. International
Journal of Rock Mechanics and Mining Sciences 100, 1-9.
Low, B.K. and Einstein, H.H. (2013) “Reliability analysis of roof wedges and rockbolt
forces in tunnels”. Tunnelling and Underground Space Technology 38, 1-10.
Lü, Q., Chan, C.L., Low, B.K., (2012) "Probabilistic evaluation of ground-support
interaction for deep rock excavation using artificial neural network and uniform
design". Tunnelling and Underground Space Technology 32, 1-18.
Lü, Q., Sun, H.-Y., Low, B.K. (2011) “Reliability analysis of ground–support
interaction in circular tunnels using the response surface method”. International
Journal of Rock Mechanics and Mining Sciences 48, 1329-1343.
Ma, S., Nemcik, J., Aziz, N., (2013) "An analytical model of fully grouted rock bolts
subjected to tensile load". Construction and Building Materials 49, 519-526.
Ma, S., Nemcik, J., Aziz, N., Zhang, Z., (2014) "Analytical model for rock bolts
reaching free end slip". Construction and Building Materials 57, 30-37.
Ma, S., Zhao, Z., Nie, W., Gui, Y. (2016) “A numerical model of fully grouted bolts
considering the tri-linear shear bond–slip model”. Tunnelling and Underground
Space Technology 54, 73-80.
Ma, S., Zhao, Z., Nie, W., Zhu, X. (2017) “An analytical model for fully grouted
rockbolts with consideration of the pre- and post-yielding behaviour”. Rock
Mechanics and Rock Engineering 50: 3019-3028.
MacLaughlin, M.M. and Doolin, D.M. (2006) “Review of validation of the
discontinuous deformation analysis (DDA) method”. International Journal for
Numerical and Analytical Methods in Geomechanics 30, 271-305.
McHugh, E., Signer, S., (1999) "Roof Bolt Response to Shear Stress: Laboratory
Analysis", in: S. Peng and C. Mark (Ed.), 18th International Conference on Ground
Control in Mining, Morgantown,WV, USA, pp. 232-238.
Mertoğlu, Ç., Anil, Ö., Durucan, C., (2016) "Bond slip behavior of anchored CFRP
strips on concrete surfaces". Construction and Building Materials 123, 553-564.

206
 References

Moosavi, M., Grayeli, R. (2006) “A model for cable bolt-rock mass interaction:
integration with discontinuous deformation analysis (DDA) algorithm”.
International Journal of Rock Mechanics and Mining Sciences 43(4), 661–670
Moosavi, M., Jafari, A., Khosravi, A., (2005) "Bond of cement grouted reinforcing bars
under constant radial pressure". Cement and Concrete Composites 27, 103-109.
Mousavi, S.S., Dehestani, M., Mousavi, K.K., (2017). "Bond strength and development
length of steel bar in unconfined self-consolidating concrete". Engineering
Structures 131, 587-598.
Napa-García, G.F., Santos, R.A., Beck, A.T., Celestino, T.B. (2018) “Improvement of
analytical factor of safety estimation of falling failure mode in roof wedge stability”.
International Journal of Rock Mechanics and Mining Sciences 103, 116-122.
NGI (2015) Using the Q-system: Rock mass classification and support design. Postboks,
Norway.
NHI (2009). Technical manual for design and construction of road tunnels – civil
elements. U.S. Department of Transportation, Federal Highway Administration.
Nie, W., Zhao, Z.Y., Ma, S.Q., Guo W. (2018) “Effects of joints on the reinforced rock
units of fully-grouted rockbolts”. Tunnelling and Underground Space Technology
71, 15-26.
Nie, W., Zhao, Z.Y., Ning, Y.J., Guo, W. (2014a) “Numerical studies on rockbolts
mechanism using 2D discontinuous deformation analysis”. Tunnelling and
Underground Space Technology 41, 223-233.
Nie, W., Zhao, Z.Y., Ning, Y.J. and SUN, J.P. (2014b) “Development of Rock Bolt
Elements in Two-Dimensional Discontinuous Deformation Analysis”. Rock
Mechanics and Rock Engineering 47, 2157-2170.
Nie, W., Zhao, Z.Y., Ma S.Q., (2016) "Numerical evaluation of rockbolt reinforcement
unit in joiuted rock mass by DDA method". In Reşat Ulusay, Ömer Aydan,Hasan
Gerçek, Ali Mehmet Hindistan and Ergün Tuncay(Eds), Rock Mechanics and Rock
Enineering: From the past to the future, Taylor & Francis Group, pp. 487-492.
Nie, W., Zhao, Z.Y., Ma, S.Q., Guo, W. (2018a) “Effects of joints on the reinforced
rock units of fully-grouted rockbolts”. Tunnelling and Underground Space
Technology 71, 15-26.
Nie, W., Zhao, Z.Y., Goh, A.T.C., Song, M.K., Guo, W., Zhu, X. (2018b). “Performance
based support design for horseshoe-shaped rock caverns using 2D numerical
analysis”. Engineering Geology 245, 266-279.
207
References

Ning, Y., Yang, J., An, X., Ma, G. (2011) "Modelling rock fracturing and blast-induced
rock mass failure via advanced discretisation within the discontinuous deformation
analysis framework". Computers and Geotechnics 38, 40-49.
Nomikos, P.P., Sofianos, A.I., Tsoutrelis, C.E. (2002) “Structural response of vertically
multi-jointed roof rock beams”. International Journal of Rock Mechanics and
Mining Sciences 39, 79-94.
Nomikos, P.P., Yiouta-Mitra, P.V., Sofianos, A.I. (2006) “Stability of Asymmetric Roof
Wedge Under Non-Symmetric Loading”. Rock Mechanics and Rock Engineering
39, 121-129.
Owen, D.R.J. and Hinton, E., (1980) Finite Elements in Plasticity: Theory and Practice.
Pineridge Press, Swansea, UK
Özsan, A. and Başarır, H. (2003). “Support capacity estimation of a diversion tunnel in
weak rock”. Engineering Geology 3, 319-331.
Palmstrom, A. (2000) “On classification systems”. Proceedings GeoEng2000,
Melbourne, Australia.
Pellet, F., Egger, P., (1996) "Analytical model for the mechanical behaviour of bolted
rock joints subjected to shearing". Rock Mechanics and Rock Engineering 29, 73-
97.
Pellet, F., Roosefid, M., Deleruyelle, F. (2009) “On the 3D numerical modelling of the
time-dependent development of the damage zone around underground galleries
during and after excavation”. Tunnelling and Underground Space Technology 24,
665-674.
Potyondy DO, Cundall PA. (2004) “A bonded-particle model for rock”. International
Journal of Rock Mechanics and Mining Sciences 41(8), 1329-64.
Ren, F.F., Yang, Z.J., Chen, J.F., Chen, W.W., (2010) "An analytical analysis of the
full-range behaviour of grouted rockbolts based on a tri-linear bond-slip model".
Construction and Building Materials 24, 361-370.
Rocscience Inc. (2011) Phase2 Version 8.0 - Finite Element Analysis for Excavations
and Slopes. www.rocscience.com, Toronto, Ontario, Canada.
Rong, G., Zhu, H.C., Zhou, C.B., (2004) "Testing study on working mechanism of fully
grouted bolts of thread steel and smooth steel". Chinese Journal of Rock Mechanics
and Engineering 23, 469-475. (in Chinese)

208
 References

Salcher, M. and Bertuzzi, R. (2018) “Results of pull tests of rock bolts and cable bolts
in Sydney sandstone and shale”. Tunnelling and Underground Space Technology
74, 60-70.
Sari, D. and Pasamehmetoglu, A.G. (2004) “Proposed support design, Kaletepe tunnel,
Turkey”. Engineering Geology 72, 201-216.
Serafim, J. L. and Pereira, J. P. (1983) “Considerations of the geomechanics
classification of Bieniawski”. Proc. of the International Symposium on Engineering
Geology and Underground Construction, Lisbon, A.A. Balkema, Rotterdam, the
Netherlands. Vol. 1, 1133-1142.
Shi, G.H. (1988) “Discontinuous deformation analysis: A new numerical model for the
statics and dynamics of block systems”. Ph.D. thesis, University of California,
Berkeley.
Singer, S.P., (1990) “Field verification of load transfer mechanics of fully grouted roof
bolts”. US Bureau of Mines, pp. 9301.
Sinha, R.S. (1989) Underground Structures: Design and Instrumentation. Elsevier,
U.S.A.
Siren, T., Kantia and P., Rinne, M. (2015) “Considerations and observations of stress-
induced and construction-induced excavation damage zone in crystalline rock”.
International Journal of Rock Mechanics and Mining Sciences 73, 165-174.
Stillborg, B., (1986). Professional users handbook for rock bolting. Clausthal-Zellerfeld,
W. Germany : Trans Tech Publications.
Sun, J., Ning, Y., Zhao, Z. (2011) “Comparative study of Sarma's method and the
discontinuous deformation analysis for rock slope stability analysis”.
Geomechanics and Geoengineering 6(4), 293-302.
Swoboda, G. (1979) “Finite element analysis of the New Austrian Cavernling Method
(NATM)”. In: Proceedings of the 3rd International Conference on Numerical
Methods in Geomechanics, Vol. 2, Aachen, 581–586.
Tan, C.H. (2016) “Passive bolts reinforcement around a circular opening in strain-
softening elastoplastic rock mass”. International Journal of Rock Mechanics and
Mining Sciences 88, 221-234.
Tiryaki, B. (2008) “Predicting intact rock strength for mechanical excavation using
multivariate statistics, artificial neural networks, and regression trees”. Engineering
Geology 99, 51-60.

209
References

Tsesarsky, M. and Hatzor, Y. H., 2006. “Tunnel roof deflection as a function of joint
spacing and friction in blocky rock masses – a parametric study using
Discontinuous Deformation Analysis (DDA)”. Tunnelling and Underground Space
Technology 21, 29-45.
Vlachopoulos, N., Diederichs, M.S. (2009) "Improved Longitudinal Displacement
Profiles for Convergence Confinement Analysis of Deep Tunnels". Rock
Mechanics and Rock Engineering 42, 131-146.
Vlachopoulos, N. and Diederichs, M.S. (2014) "Appropriate Uses and Practical
Limitations of 2D Numerical Analysis of Tunnels and Tunnel Support Response".
Geotechnical and Geological Engineering 32, 469-488.
Windsor CR, Thompson AG. “Rock Reinforcement - Technology, Testing, Design and
Evaluation”. New York : Pergamon Press, 1993.
Windsor, C.R. (1997) “Rock reinforcement systems”. International Journal of Rock
Mechanics and Mining Sciences 34, 919-951.
Wittke, W. (2014) “Stability of Rock Wedges and Excavation Surfaces”, in: Wittke, W.
(Ed.), Rock Mechanics Based on an Anisotropic Jointed Rock Model (AJRM).
Wu, J.H and Chen, C.H. (2011) “Application of DDA to simulate characteristics of the
Tsaoling landslide”. Computers and Geotechnics 38, 741-750.
Wu, X., Jiang, Y., Gong, B., Deng, T., Guan, Z. (2018). “Behaviour of rock joint
reinforced by energy-absorbing rock bolt under cyclic shear loading condition”.
International Journal of Rock Mechanics and Mining Sciences 110, 88-96.
Yazici, S. and Kaiser, P.K. (1992) "Bond strength of grouted cable bolts". International
Journal of Rock Mechanics and Mining Sciences & Geomechanics Abstracts 29,
279-292.
Yeih, W., Huang, R., Chang, J.J., Yang, C.C. (1997) "A pullout test for determining
interface properties between rebar and concrete". Advanced Cement Based
Materials 5, 57-65.
Yokota, Y., Zhao Z., Nie, W., Date K., Iwano K., Okada Y. (2018) “Interface behaviour
along the boundary between the rock bolt and bond material”, in: Eurock 2018 –
The ISRM European Rock Mechanics Symposium, pp. 607–615.
Zhandarov, S. and Mäder, E. (2004). “Determining of interfacial parameters in fiber-
polymer systems from pull-out test data using a bilinear bond law”. Composite
Interfaces 11, 361-391.

210
 References

Zhandarov, S., Mäder, E., (2016). "Determining the interfacial toughness from force–
displacement curves in the pull-out and microbond tests using the alternative
method". International Journal of Adhesion and Adhesives 65, 11-18.
Zhang, X.P. and Wong, L.N.Y. (2014) “Displacement field analysis for cracking
processes in bonded-particle model”. Bulletin of Engineering Geology and the
Environment 73, 13-21.
Zhang, X.P. and Zhang, Q. (2017). “Distinction of Crack Nature in Brittle Rock-Like
Materials: A Numerical Study Based on Moment Tensors”. Rock Mechanics and
Rock Engineering 50(10), 2837-2845.
Zhang, W.G. and Goh, A.T.C. (2015) “Regression models for estimating ultimate and
serviceability limit states of underground rock caverns”. Engineering Geology 188,
68-76.
Zhao, Z., Gu, J., Bao, H. (2007) “Understanding fracture patterns of rock mass due to
blast load – a DDA approach. Ju, Y., Fang, X. and Bian H. (Eds.) Proceedings of
the Eighth International Conference on the Analysis of Discontinuous Deformation.,
Beijing, China, 147-150.
Zhao, Z. Y., Zhang, Y., Liao H. J. (2008) “Design of ensemble neural network using
Akaike information criterion”. Engineering Application of Artificial Intelligence
21(8), 1182-1188.
Zhou, Y., Teo, T.Y., Cai, J.G. (2017) “Rock engineering practice for development of
underground caverns in Singapore”. GeoSS 10th Anniversary Conference,
Sinagpore.
Zhu, C., Chang, X., Men, Y., Luo, X. (2015) “Modeling of grout crack of rockbolt
grouted system”. International Journal of Mining Science and Technology 25, 73-
77.

211
References

212
 Appendix A

APPENDIX A
MATLAB PROGRAM FOR MAPPING THE
RELATIONSHIP BETWEEN THE SEM DESIGNS
AND THE SUPPORT PERFORMANCES

%%TITLE: ANN model for mapping the SEM parameters and performance
%Inputs: P1-ground class, P2-width of heading, P3-height of heading, P4-round length
%Target: O1-roof displacement (or O2-normal stress, O3-damage depth)
%Structure: 4-n-1, n is a variable in the program
%Treat the inputs and output with pre-set max. and min. values
% P1 P2 P3 P4 O1 O2 O3
% Max: 40 40 54 50 || 0.3 10 50
% Min: 0.1 3 5 0 || 0 0 0
% Outcomes: (a) the structure of the ANN model will be determined based on the R2
% (b) the weights and bias will be used in the further adoption of the model
%Note: The normalizations are preprocessed when generate data sets.
% Thus, this code is not including normalization

%%
clear all;
clc;
%%
X1=load('1.txt');
%---------------------------------- Examples ----------------------------------
% -1.00000 -0.85294 -0.99804 -0.99285 -0.91973
%-1.00000 -0.85294 -0.99804 -0.96978 -0.87113
%-1.00000 -0.85294 -0.99804 -0.91578 -0.82187

213
Appendix A

%-1.00000 -0.85294 -0.99804 -0.83644 -0.78487

%-1.00000 -0.85294 -0.99804 -0.71450 -0.68913
%-1.00000 -0.85294 -0.99804 -0.26610 -0.62327
%-1.00000 -0.98529 -0.99804 -0.99285 -0.94667
%…
%%
[M,N]=size(X1);
Y1=X1(:,1:4)'; %inputs
O1=X1(:,5)'; %target

[fid,message]=fopen('weight.txt','w');
if fid==-1
disp(message);
end
%%
trainRatio=0.7;
testRatio=0.3;
hiddenminno1=6;
hiddenminno2=6;
%%
maxepoch=500;
fb=10; %
if(trainRatio+testRatio>1.0)
disp('total ratio>1');
end

% hiddenminno (3 ~ 8) is a variable to determine the numbers of hidden nodes

% in each case, fb = 10 times of training and testing are carried out
for hiddenminno=3:1:8
for run=1:1:fb
[trainInd,valInd,testInd] = dividerand(M,trainRatio,valRatio,testRatio);
traininput1=Y1(:,trainInd);
traintarget1=O1(:,trainInd);
traininput=traininput1;
214
 Appendix A

traintarget=traintarget1;
% testing data
testinput1=Y1(:,testInd);
testtarget1=O1(:,testInd);
%assign same normalization as training
testinput = testinput1;
testtarget = testtarget1;
%%
% 'attrno' is the number of attributes,
%'trainexpno' is the number of training examples
[attrno,trainexpno] = size(traininput);
%%
% train the neural networks
net = newff(minmax(traininput),[hiddenminno 1],{'tansig' 'tansig'});
net.trainParam.epochs = maxepoch;
net.trainParam.goal = 0.0;
net.trainParam.max_fail=100;
net.trainParam.min_grad=1e-15;
net.trainParam.mu_max=1e20;
net.trainParam.mu_dec=0.7;
net.trainParam.mu_inc=1.03;
net.trainParam.lr=0.01;
net.trainParam.show=10;
[net,tr] = train(net,traininput,traintarget);
%%
fprintf(fid,'run = %d \n',run);
fprintf(fid,'span\tbenching-h\tQ-value\tdistance\tbias1\troof-disp\tbias2\n');
%The weights to layer i from input j are net.IW{i,j}.
w1=net.iw{1,1}
b1=net.b{1}
w2=net.lw{2,1}
b2=net.b{2}
%%
for i=1:1:hiddenminno
215
Appendix A

for j=1:1:4
fprintf(fid,'%f\t',w1(i,j));
end
fprintf(fid,'%f\n',b1(i));
end
for i=1:1:hiddenminno
fprintf(fid,'%f\t',w2(i));
end
fprintf(fid,'%f\n',b2);
%%
% test the model
[m,trexpno] = size(traininput);
output2 = zeros(trexpno);
output2 = sim(net,traininput);
[n,testexpno] = size(testinput); % 'testexpno' is the number of test examples
output1 = zeros(testexpno);
output1 = sim(net,testinput);
[l,allexpno]=size(Y1);
output = zeros(allexpno);
output = sim(net,Y1);
%%
% training data showing
[m1,b1,r1]=postreg(output_tr(1,:),traintarget1(1,:));
mse_te1 = mse(output_tr(1,:) - traintarget1(1,:)); % the mean squared error of
the ensemble
fprintf(fid,'Training single hidden=%g ',hiddenminno);
fprintf(fid,'r1=%-12.5g mse_te_O=%-12.5g ', r1, mse_te1);
fprintf(fid,'\n');%%
saveas(gcf,num2str(run*3-1),'jpg');
%testing data showing
[m2,b2,r2]=postreg(output_te(1,:),testtarget1(1,:));
mse_tr2 = mse(output_te(1,:) - testtarget1(1,:));
fprintf(fid,'Testing single hidden=%g ',hiddenminno);
fprintf(fid,'r2=%-12.5g mse_te_O=%-12.5g ', r2, mse_tr2);
216
 Appendix A

fprintf(fid,'\n');%%
saveas(gcf,num2str(run*3-2),'jpg');
%all data showing
figure;
[m3,b3,r3]=postreg(output,O1);
mse_t3 = mse(output-O1);
fprintf(fid,'All double hidden=%g',hiddenminno);
fprintf(fid,'r3=%-12.5g mse_t_O=%-12.5g ', r3, mse_t3);
fprintf(fid,'\n\n');
saveas(gcf,num2str(run*3),'jpg');
end
end
fclose(fid);
%%
%-------------Examples of the ‘weight.txt’ file--------

%....

217
Appendix A

218