Journal of Rock Mechanics and Geotechnical Engineering 9 (2017) 267e280

Contents lists available at ScienceDirect

Journal of Rock Mechanics and

Geotechnical Engineering
journal homepage: www.rockgeotech.org

Full Length Article

Explicit reinforcement models for fully-grouted rebar rock bolts

Navid Bahrani*, John Hadjigeorgiou
Lassonde Institute of Mining, University of Toronto, Toronto, M5S 1A4, Canada

a r t i c l e i n f o a b s t r a c t

Article history: This paper investigates the explicit use of rock reinforcement in a discontinuous stress analysis model. A
Received 28 April 2016 series of numerical experiments was undertaken to evaluate the performance of local and global rein-
Received in revised form forcement models implemented in universal distinct element code (UDEC). This was made possible by
7 July 2016
calibrating the reinforcement models to the laboratory behavior of a fully-grouted rebar bolt tested
Accepted 24 July 2016
Available online 9 December 2016
under pure pull and pure shear loading conditions. The model calibration focuses on matching different
loading stages of the forceedisplacement curve including the initial elastic response, the hardening
behavior and the bolt rupture. The paper concludes with a discussion on the suitability of the different
Keywords:
Rock reinforcement
reinforcement models in UDEC including their advantages and limitations. Finally, it addresses the choice
Fully-grouted rebar of input parameters required for a realistic simulation of fully-grouted rebar bolts.
Distinct element method Ó 2017 Institute of Rock and Soil Mechanics, Chinese Academy of Sciences. Production and hosting by
Local reinforcement Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/
Global reinforcement licenses/by-nc-nd/4.0/).
Pull-out test
Shear test

1. Introduction excavation in a rock mass. The design of reinforcement using nu-

merical models can be either implicit or explicit. An implicit design
Reinforcement is a means of improving the overall properties of process has been outlined by Wiles et al. (2004). In this approach,
a rock mass by using stabilizing elements such as rock bolts, cable the results of a stress analysis can be used to qualify the ground
bolts and ground anchors. Rock reinforcement is often used as a response into “broken ground” and “cracked ground”. The “broken
primary support element, applied during or immediately after ground” is a ground that has undergone stress-driven failure and
excavation, to stabilize the ground and ensure the safe working represents the dead weight that has to be supported by the rein-
conditions during subsequent excavation. forcement. The “cracked ground” that is determined by a rock mass
The design of reinforcement for underground excavations in damage threshold criterion defines where the reinforcement
rock has not evolved considerably since the 1990s. Design is often anchoring begins.
based on empirical rules and rock mass classification schemes. This An explicit design process implies that a representative rein-
is somewhat surprising given the development and accessibility of forcement has been implemented in a stress analysis model and the
sophisticated stress analysis tools. A potential reason for this may results of the stress analysis process are accounting for the role and
be the inherent limitations of how reinforcement is represented in influence of reinforcement in the design. There are several chal-
stress analysis software packages. A further reason is related to the lenges that have to be overcome in the explicit representation of
difficulties associated with calibrating the numerical models to gain rock reinforcement in stress analysis models. The first part is the
confidence on the implemented reinforcement tools. choice of the type of model to be used that meets the objectives of
Numerical modeling is a valuable tool in the design of under- the simulation and the problem definition. The next step is to
ground excavations. Continuum, discontinuum, and hybrid con- specifically address how reinforcement is implemented in the
tinuumediscontinuum codes are used to determine the resulting stress analysis models. The final step, and the most important one,
stresses and displacements following the introduction of an is how one can attain a successful level of calibration of stress
analysis models that can be used with confidence for design
problems.
This paper focuses on the explicit representation of reinforce-
* Corresponding author. ment in a distinct element stress analysis model. The major
E-mail address: [email protected] (N. Bahrani).
objective is to critically and systematically evaluate two types of
Peer review under responsibility of Institute of Rock and Soil Mechanics,
Chinese Academy of Sciences.
rock reinforcement models, i.e. local and global reinforcements. The

http://dx.doi.org/10.1016/j.jrmge.2016.07.006
1674-7755 Ó 2017 Institute of Rock and Soil Mechanics, Chinese Academy of Sciences. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND
license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
268 N. Bahrani, J. Hadjigeorgiou / Journal of Rock Mechanics and Geotechnical Engineering 9 (2017) 267e280

theoretical basis for these reinforcement models is reviewed as

well as a description of the required input parameters. The inves-
tigation was based on laboratory experiments of cement-grouted
rebar bolts. The paper addresses calibration issues and in-
vestigates the behavior of fully-grouted rebar bolts under pure pull
and pure shear loading conditions. An assessment of the advan-
tages and limitations of the results obtained using these techniques
can provide the basis for the selection of appropriate reinforcement
models for a realistic simulation of rock reinforcement in jointed
rock masses.

2. Laboratory tests on fully-grouted rebar

Rock bolts are the primary means of rock reinforcement for

excavations in rock. Rock bolts reinforce the rock mass by one or
more of the following methods: beam building, suspension of weak
fractured ground to more competent layers, pressure arch, and
support of discrete blocks (Hadjigeorgiou and Charette, 2001). The
in situ behavior of rock bolts can best be captured by pull tests. This,
however, is influenced by a multitude of parameters. A better un-
derstanding of the behavior of specific parameters can be obtained
under controlled laboratory experiments. From a numerical
perspective, it is convenient to investigate the representation of
reinforcement models to well defined experimental data. This can
be a prelude to modeling the in situ behavior of reinforcement.
Stjern (1995) conducted a series of laboratory tests to investi-
gate the loadedisplacement behaviors of different types of rock
bolts subjected to tensile (pull) and shear loading. Li et al. (2014)
provided a comprehensive review of the performances of both
conventional and energy-absorbing rock bolts based on the results Fig. 1. The test rig for static pull, shear and combined pull-shear tests: (a) the front
view of the test rig (after Chen, 2014); (b) an oblique sketch of the test rig (after Stjern,
of laboratory experiments conducted by Stjern (1995). The test rig
1995).
used for this purpose consisted of two concrete blocks of a uniaxial
compressive strength (UCS) of 65 MPa, which could be moved both
laterally (for shear test) and normally (for pull test) to the joint. The
sides of the concrete blocks were 0.95 m, which made testing of a rupture occurs. Table 1 compares the results of tests conducted by
full-sized rock bolt with standard anchorage element and bearing Stjern (1995) and Chen (2014) in terms of loading stages described
plate possible (Stjern, 1995). The pull and shear tests were con- above, including initial stiffness, yield load, peak load, displacement
ducted on fully-grouted rebar bolts, frictional bolts, cable bolts, and at the peak load, and rupture displacement. It is understood from
mechanical bolts. The complete loadedisplacement characteristics this table and the loadedisplacement curves shown in Fig. 2 that
for the bolts were obtained during the tests using various in- the yield loads are about 86% and 47% of the peak loads under pull
struments. The grout for the tests carried out on fully-grouted bolts and shear loading conditions, respectively. In this paper, the results
had a water/cement ratio of 0.33. The bolts were installed according of laboratory tests on the fully-grouted rebar bolt reported by Stjern
to normal field installation practice for each specific bolt type. To (1995) (i.e. blue curves in Fig. 2) are used for the evaluation of
minimize the influence of joint shear resistance during shear tests, reinforcement models in universal distinct element code (UDEC).
a 1 mm thick teflon film was attached to each joint surface.
More recently, Chen (2014) and Chen and Li (2015a, b) reported 3. Numerical modeling of rock reinforcement
the results of similar laboratory tests and evaluated the anchorage
performance of the rebar bolt and the D-Bolt under combined pull- Rock reinforcements can be simulated using either material
shear loading condition. Chen (2014) investigated the influence of models or structural elements. Both approaches have been
displacing angle (angle between the pull and the shear displace- demonstrated to be able to represent rock reinforcement behavior
ments), joint gap and host rock strength on the loadedisplacement under different loading conditions. However, when simulating an
behavior of the rebar bolt and the D-Bolt. Fig. 1a shows the front underground excavation and support system, the approach based
view of the test rig used by Stjern (1995) and Chen (2014), and on material model is computationally intensive, as very fine mesh
Fig. 1b shows a sketch of the modified test rig (Chen, 2014). elements (or zones) are required to properly simulate rock rein-
Fig. 2a and b compares the results of pure pull and pure shear forcement. Therefore, numerical representation of rock reinforce-
tests on 18 mm diameter fully-grouted rebar bolt conducted by ment using a material model is usually limited to the simulation of
Stjern (1995) and 20 mm diameter fully-grouted rebar bolt con- laboratory tests. Structural elements, however, can be used for the
ducted by Chen (2014). The rebar bolt tested by Chen (2014) is simulation of rock reinforcement under both laboratory and field
relatively stronger than that tested by Stjern (1995) in both pull and conditions.
shear tests. This could be due to the differences in the diameters of Examples of numerical representation of rock reinforcement,
the rebar bolts used. However, their loadedisplacement curves are using material models, include those by Ferrero (1995), Grasselli
very similar in pull and shear loading conditions and both consist of (2005), Aziz and Jalalifar (2007), Chen and Li (2015c), and Tatone
three main loading stages. Under both pull and shear loading et al. (2015), who simulated laboratory tests on various types of
conditions, the bolts elongate elastically, then yield and harden rock bolts. Ferrero (1995) used a three-dimensional (3D) finite
until they reach the peak load. The bolts continue to elongate until element code to simulate shear test on a rock joint system
 N. Bahrani, J. Hadjigeorgiou / Journal of Rock Mechanics and Geotechnical Engineering 9 (2017) 267e280 269

material model to capture the strain-hardening of the steel. Chen

and Li (2015c) used different models for the rockegrout and
groutebolt interfaces in the D-Bolt and the rebar bolt to explicitly
simulate their different bonding mechanisms. In the simulation of
a rebar bolt, the bonding between the steel and the rock was
defined by its high shear strength at the boltegrout interface. In
the simulation of the D-Bolt, the bonding of the bolt was defined
by a shear strength that is the same as the grout at the two anchor
positions, and a zero shear strength on the bolt section between
the anchors. Using this modeling approach, Chen and Li (2015c)
were able to realistically simulate the loadedisplacement re-
sponses of the rebar and the D-Bolt under pure pull, pure shear
and combined pull-shear loading conditions. Tatone et al. (2015)
simulated pull tests on reinforcement represented using both
material model and structural element in the two-dimensional
(2D) Y-Geo code, which is based on the hybrid finite-discrete
element method. The results of both approaches were found to
be consistent in terms of forceedisplacement response and dam-
age propagation. They also demonstrated the effectiveness of the
rock reinforcement represented by the structural element in
reducing the amount of damage around underground openings.
Some of the most popular commercially available stress analysis
numerical codes, such as FLAC, FLAC3D, UDEC and 3DEC developed
by Itasca Consulting Group Inc., allow for the use of structural el-
ements to simulate different types of rock reinforcements. In effect,
two types of rock reinforcement models, i.e. the “Cable” and the
“Rockbolt” elements, can be used to account for the properties of
both bolt and grout. For practical purposes, the main difference
between these two approaches is that the “Cable” element does not
provide any resistance to bending, therefore it is best suited for
simulating cable bolts, whereas the “Rockbolt” element provides
resistance against bending, making it appropriate for simulating
other reinforcement elements such as fully-grouted rebar bolts. The
Fig. 2. Results of (a) pull test and (b) shear test on 18 mm diameter fully-grouted rebar practical implications of these choices in reinforcement elements
(Stjern, 1995) and 20 mm diameter fully-grouted rebar (Chen, 2014). are discussed in more detail in the next section.
A review of the technical literature suggests that “Cable” ele-
ments are more popular than the “Rockbolt” element, even for the
reinforced by steel dowels and back-analyzed the stressestrain simulation of rock reinforcements other than cable bolts. This is
behavior of the specimen to assess the states of stress evolution in interpreted to be due to the complexity of the “Rockbolt” element
the reinforcement. Both the rock and steel were assumed to compared to the “Cable” element, in terms of input parameters and
behave in an elastic-perfectly plastic manner. Grasselli (2005) used calibration process. For example, Vardakos et al. (2007), Malmgren
a 3D finite element code to simulate shear tests conducted on and Nordlund (2008), Jiang et al. (2009), Li et al. (2012), Gao et al.
grouted rebar and Swellex. He used an elastic material model to (2015), and Shreedharan and Kulatilake (2016) all used the “Ca-
simulate the bolts, and an elastic-perfectly plastic interface ble” element for the simulation of laboratory pull test or support of
element to simulate the joint separating the two blocks. The nu- underground openings, in cases where the actual reinforcement
merical simulation provided some insights into the failure mech- used was not cable bolt. Ruest and Martin (2002) also used the
anisms of the bolts under shear loading conditions. Aziz and “Cable” element in FLAC2D but for the simulation of laboratory pull
Jalalifar (2007) also used a 3D finite element code to simulate tests conducted on instrumented cable bolts grouted in steel pipe.
the laboratory experiments conducted on the bolted rock joints The loads calculated along the “Cable” element were compared to
subjected to shearing and investigated the shear stress, yield stress the measured loads for a range of model grout properties, and the
and the change in the bolt strain at the bolt-joint intersection. The results showed very good agreement with those of laboratory tests.
steel was represented using a bilinear hardening model. Chen and Only, Nemcik et al. (2014) and Ma et al. (2014) have reported the
Li (2015c) used the continuum code FLAC3D to simulate the labo- use of the “Rockbolt” element for the simulation of fully-grouted
ratory tests conducted on the rebar bolt and the D-Bolt under rock bolts subjected to tensile loading. Nemcik et al. (2014)
varying displacing angles to the anchorage. They used a tri-linear ignored the forces perpendicular to the “Rockbolt” element and

Table 1
Comparison between the results of pull and shear tests on fully-grouted rebar conducted by Stjern (1995) and Chen (2014).

Test Reference Stiffness Yield load (kN) Peak load (kN) Load ratio Displacement at Rupture displacement (mm)
(kN/mm) (yield/peak) peak load (mm)

Pull Stjern (1995) 20 150 175 0.86 35 39

Chen (2014) 29 181 211 0.86 30 43
Shear Stjern (1995) 23 84 177 0.47 39 43
Chen (2014) 14 90 195 0.46 44 53
270 N. Bahrani, J. Hadjigeorgiou / Journal of Rock Mechanics and Geotechnical Engineering 9 (2017) 267e280

subsequently its resistance to bending since they were only dealing element can be used for simulating cable bolts in which the cable
with tensile loading condition. In their analysis, the “Rockbolt” can generate very little bending resistance under shear loading and
element was in fact behaving similar to the “Cable” element. therefore fails in tension, while the bonding agent may fail in shear
Similarly, Ma et al. (2014) used the “Rockbolt” element in FLAC2D to over some length of the reinforcement. The “Rockbolt” element,
simulate the laboratory pull-out test and to investigate the inter- however, can simulate rock bolts, in which the bolt can resist
action between rock bolt and rock mass in a roadway tunnel. against bending under shear loading.
It follows that there are both technical and practical reasons to It would appear that the “Cable” element has been used in
advance the simulation of rock reinforcement using a numerical various research projects as reviewed in the previous section, while
representation that is capable of resisting against bending (i.e. the “Rockbolt” element has not been widely acknowledged by re-
“Rockbolt” element). This paper is a contribution to the simulation searchers. This is partially due to the large number of input pa-
of laboratory tests on fully-grouted rebar bolt under both pull and rameters required in the “Rockbolt” element, which makes its
shear loading conditions using the “Rockbolt” element. calibration difficult and a time-consuming process. In this paper,
the local reinforcement and the shearing- and bending-resistant
4. Rock reinforcement models in distinct element models global reinforcements (“Rockbolt” element) are used to simulate
the behavior of the fully-grouted rebar bolt under pure pull and
The numerical simulations of fully-grouted rebar bolts were pure shear loading conditions conducted by Stjern (1995) (i.e.
conducted using the distinct element method. The method has forceedisplacement curves in Fig. 2). In the following, these two
inherent advantages as it permits block deformation and move- reinforcement models are reviewed.
ment of blocks relative to each other. The method has been suc-
cessfully employed to model complex behavior and mechanisms of 4.1. Local reinforcement
jointed rock masses (Karampinos et al., 2015). The explicit repre-
sentation of reinforcement and support is, however, much more As the local reinforcement only considers local effect of rein-
complex and has not received the same level of attention. This is forcement where it intersects the discontinuities, it is therefore
possibly associated with the challenges in implementing rock most applicable to cases where the deformation of individual rock
reinforcement in a stress analysis model and in demonstrating that blocks is small relative to the deformation of the reinforcing sys-
successful numerical calibration can provide significant confidence tem. Its formulation is based on simple forceedisplacement re-
in the results. lationships that describe both the shear and axial behaviors of
This paper addresses issues of implementation of reinforcement reinforcement across discontinuities.
in a 2D distinct element model (UDEC) and calibration based on The axial and shear forceedisplacement behavior of local rein-
documented laboratory tests. The explicit representation of rock forcement are illustrated in Fig. 3a. The axial forceedisplacement
reinforcement in UDEC is possible, using both local and global re- behavior is described by axial stiffness (Ka), ultimate axial capacity
inforcements (Itasca, 2014). The local reinforcement can be applied max Þ and axial failure strain, and the shear forceedisplacement
ðFa;b
to both rigid and deformable blocks, while the global reinforcement behavior is represented by shear stiffness (Ks) and ultimate shear
can only be applied to deformable blocks. The local reinforcement max Þ.
capacity ðFs;b
considers only the local effect of reinforcement where it intersects The forceedisplacement relations that describe the axial and
existing discontinuities. The global reinforcement considers not shear responses are given by the following equations:
only the local effect of reinforcement where it intersects the dis-
continuities, but also the restraint to rock block that may experi- DFa ¼ Ka jDUa jf ðFa Þ (1)
ence inelastic deformation surrounding an excavation.
The global shearing-resistant reinforcement (“Cable” element) DFs ¼ Ks jDUs jf ðFs Þ (2)
and the global shearing and bending-resistant reinforcement
(“Rockbolt” element) are the two types of global reinforcement where DFa and DFs are the incremental changes in axial and shear
implemented in UDEC. The “Cable” element is a one-dimensional forces, respectively; DUa and DUs are the incremental changes in
element with two degrees of freedom (two displacements), while axial and shear displacements, respectively; and f (Fa) and f (Fs) are
the “Rockbolt” element is a 2D element with three degrees of the functions that describe the path by which the axial (Fa) and
freedom (two displacements and one rotation). The “Cable” shear (Fs) forces approach the ultimate axial ðFa;b max Þ and shear

force force es or ea = 0

maximum pull/
rupture
shear force

Ka or Ks
es or ea > 0
1
displacement displacement

(a) ( b)

Fig. 3. (a) Forceedisplacement behavior of local reinforcement in pull and shear tests; (b) influence of stiffness exponent (axial or shear) on forceedisplacement response of local
reinforcement (after Itasca, 2014).
 N. Bahrani, J. Hadjigeorgiou / Journal of Rock Mechanics and Geotechnical Engineering 9 (2017) 267e280 271

max Þ forces, according to the values of axial (e ) and shear (e )

ðFs;b a s Reinforcing element
(steel)
stiffness exponents. The axial and shear stiffness exponents are
used to define the curvature of the forceedisplacement behavior,
by controlling the rate at which the bounding forces are reached Grout annulus
(Fig. 3b). By default, the values of stiffness exponents are zero,
which means that the forceedisplacement relations follow a con- Reinforcement
Excavation
stant (linear) stiffness until the ultimate capacity (axial or shear) is nodal point

reached.
Fig. 4a shows that the maximum shear force changes for various Steel compressive/
tensile strength
orientations of reinforcement relative to the discontinuity. As
shown in this figure, the maximum shear force decreases from a Axial stiffness
maximum value at q0 ¼ 0 to 50% of the peak shear load at q0 ¼ 90 of steel Shear spring
cohesive strength
(Itasca, 2014). The local reinforcement was implemented by Itasca Normal spring
cohesive strength
(2014) in such a way to be consistent with the results by Azuar Shear spring
Normal spring stiffness
et al. (1979), who found that the maximum shear force was about stiffness
half the product of the uniaxial tensile strength of the reinforce-
ment and its cross-sectional area for reinforcement perpendicular
to the discontinuity. Fig. 5. Conceptual mechanical representation of the global reinforcement (“Rockbolt”
The rupture strain under shear loading cannot be directly element), which accounts for shear behavior of grout annulus and bending resistance
defined, however, it can be adjusted by defining the active length of of the reinforcement (after Itasca, 2014).
the reinforcement. Active length is the short length of reinforce-
ment, which spans the discontinuity and changes the orientation
element interacts with the UDEC model via shear and normal
during shear displacement (Fig. 4b). The active length can be
coupling springs, which are connectors that transfer forces and
measured in the laboratory from the reinforcement deformed or
motion between the “Rockbolt” element and the grid points asso-
failed after the shear test. In the case of lack of data, the value of
ciated with the block zone, in which the nodes are located (Fig. 5).
active length can be adjusted until the rupture shear displacement
The “Rockbolt” element segments are treated as a linearly elastic
on the shear forceedisplacement plot is matched with that from
material that may yield in the axial direction either in tension or
the shear test result. The latter will be used in this paper for the
compression. The tensile and compressive yield strengths are used
calibration of local reinforcement to the behavior of rebar bolt in
to define these strength limits. Inelastic bending is simulated in the
the shear test.
“Rockbolt” element by specifying a limiting plastic moment. This
The input parameters required for the simulation of a rock bolt
means that “Rockbolt” elements behave elastically until they reach
using the local reinforcement model can be obtained from labora-
the plastic moment. In addition, segments may break and separate
tory pull and shear tests. In the case of lack of experimental data,
at the nodes, based on a user-defined tensile failure strain limit. A
empirical relations can be used to estimate bolt axial and shear
strain measure, called the total plastic tensile strain, based on
stiffnesses and pull and shear capacities.
adding the axial and bending plastic strains, is evaluated at each
node. If this strain exceeds the tensile failure strain limit, the forces
4.2. Global reinforcement (“Rockbolt” element) and moment in this segment are set to zero and the “Rockbolt”
element is assumed to have failed. The shear and normal behaviors
Both “Rockbolt” and “Cable” elements allow for the simulation of the “Rockbolt” element/grid point interface are represented by
of a shearing resistance along their length, as provided by the shear linear springs as shown in Fig. 6.
resistance (bond) between the grout and either the cable/rock bolt The shear behavior of the interface during relative displacement
or the host rock. The advantage of the “Rockbolt” element over the between the element nodes and the grid points is described
“Cable” element in modeling a rebar bolt is that it provides resis- numerically by the coupling spring shear stiffness (CSsstiff in Fig. 6a)
tance against bending. Furthermore, the “Rockbolt” element can according to the following equation:
simulate the actual bolt breakage based upon a user-defined tensile
failure strain limit. Similar to the “Cable” element, the “Rockbolt”

Fs =L ¼ CSsstiff up  um (3)
element is divided into a number of segments of length L, with
nodal points located at each segment end (Fig. 5). The “Rockbolt”

ks Direction of shearing
1 θ0

θ0 = 90°
θ
θ0 = 135° Discontinuity

(a) (b) ∆us

Fig. 4. (a) Shear behavior of local reinforcement element for various orientations of reinforcement relative to the discontinuity; (b) assumed reinforcement geometry after shear
displacement (DUs) and illustration of active length (after Itasca, 2014).
272 N. Bahrani, J. Hadjigeorgiou / Journal of Rock Mechanics and Geotechnical Engineering 9 (2017) 267e280

(a) Shear force / length (b) Fsmax / L

Fsmax / L
CSsfric
CSss ff

relative shear
displacement CSscoh

Fsmax / L σ′c × perimeter

(c) Compressive force / length (d)

Fnmax / L
Fnmax / L CSsfric

CSns ff
CSncoh

relative normal CSncoh σ′c × perimeter

displacement

Fnmax / L

Fig. 6. Material behavior of shear and normal coupling springs for “Rockbolt” element (after Itasca, 2014): (a) shear force versus shear displacement; (b) shear strength criterion for
the shear coupling spring; (c) normal force versus normal displacement; and (d) normal strength criterion for the normal coupling spring.

where up and um are the axial displacements for the “Rockbolt” 5. Numerical simulation of pull and shear tests
element and the medium (soil or rock), respectively; and L is the
“Rockbolt” element segment length. The limiting shear force ðFsmax Þ 5.1. Model geometry and boundary conditions
that can be developed along the “Rockbolt” element/grid point
interface is a function of the cohesive strength of the interface Fig. 7 shows the UDEC model used to simulate the pull and shear
(CSscoh) and the stress-dependent frictional resistance (CSsfric) along tests using the local and global (“Rockbolt” element) re-
the interface (Fig. 6b) according to following equation: inforcements. The model consists of two elastic blocks with a
Young’s modulus of 24 GPa and a Poisson’s ratio of 0.25, separated

Fsmax L ¼ CSscoh þ s0c l tan CSsfric (4) by a joint with zero cohesion and friction angle. The left block was
chosen to be 0.95 m  0.95 m to be consistent with the block di-
where l is the exposed perimeter of the rock bolt (i.e. the length of mensions in the actual test reported by Stjern (1995). The right
the rock bolt surface that is in contact with the medium), and s0c is block in this figure, however, was made slightly larger so that its
the mean effective confining stress normal to the “Rockbolt” boundary conditions (rollers at the left corners of the top and
element. bottom boundaries) do not interfere with the left block.
Similarly, the normal behavior during the relative normal The boundary conditions of the UDEC models used for the
displacement between the “Rockbolt” element nodes and the grid simulations of pull and shear tests are illustrated in Fig. 8. In both
points (Fig. 6c and d) is described numerically according to the tests, rollers are applied to the boundaries, and pins are applied to
following equations: the right corners of the right blocks. The pull test is simulated by
applying a velocity boundary to the left side of the left block as
 
Fn =L ¼ CSnstiff unp  unm (5)


Fnmax L ¼ CSncoh þ s0c l tan CSnfric (6)

Elastic block Elastic block

where CSnstiff is the coupling spring normal stiffness; Fn is the
normal force that develops in the normal coupling spring; unp and
unm are the displacements of the “Rockbolt” element and medium 0.95 m

normal to the axial direction of the element, respectively; Fnmax is

joint
the limiting normal force; and CSncoh and CSnfric are the cohesive
strength and friction angle of the normal coupling spring,
respectively.
Other parameters required for the simulation of rock rein-
0.95 m
forcement using the “Rockbolt” element include the cross-sectional
area, second moment of area, density, and elastic modulus of the Fig. 7. Geometry of UDEC model used to simulate pull and shear tests on fully-grouted
rock bolt. rebar bolt.
 N. Bahrani, J. Hadjigeorgiou / Journal of Rock Mechanics and Geotechnical Engineering 9 (2017) 267e280 273

(a) Start (b) Start

Simulation of shear
Simulation of pull test
test

Calibration to pull test Calibration to shear

results test results

Calibrated Calibrated
parameters as input parameters as input
for shear test model for pull test model

Simulation of shear
Simulation of pull test
test

Comparison with Comparison with

laboratory shear test laboratory pull test
results results

Fig. 8. Boundary conditions of UDEC models used to simulate (a) pull and (b) shear
tests.
End End

Fig. 9. Calibration procedures for the local and global (“Rockbolt” element)
shown in Fig. 8a, and the shear test is simulated by applying a ve- reinforcements.
locity boundary to the lower side of the left block as shown in
Fig. 8b. In the shear test, rollers are also used in the left side of the
left block in order to avoid block rotation.
Two methods are used to monitor the pull and shear forces. In
the first method, the force (both pull and shear) is determined from
the sum of reaction forces that develop on the boundaries of the
right block. In the second method, the forces are obtained directly
from the reinforcement models and then compared with the forces
determined using the first method.
The calibration of UDEC model was conducted in two inde-
pendent stages as illustrated in the flowcharts in Fig. 9. First, the
model is calibrated to the forceedisplacement behavior of the rebar
bolt from the pull test (Fig. 9a). The shear test is then simulated
with the parameters obtained from the calibration of the model to
the pull test results. Next, the forceedisplacement behavior ob-
tained from the simulation of shear test is compared with that of
laboratory shear test.
In the second stage (Fig. 9b), the model is first calibrated to the
forceedisplacement behavior of the rebar bolt from the shear test.
Fig. 10. Illustration of calibration criteria on the forceedisplacement plot.
The parameters obtained from the calibration of the model to the
shear test results are then used to simulate the pull test and the
results in terms of forceedisplacement behavior are compared with the UDEC’s reinforcement models, which can capture the forcee
that of laboratory pull test. The criteria used for the calibration of displacement curves of fully-grouted rebar bolts under both pull
local and global (“Rockbolt” element) reinforcements to the results and shear loading conditions.
of pull and shear tests are different loading stages on the forcee
displacement plot, including the initial stiffness, yield load, peak
load, and rupture displacement, as illustrated in Fig. 10. 5.2. Calibration of local reinforcement
Rock reinforcement elements used to support underground/
surface excavations may experience pure pull load, pure shear load, The results of calibration of local reinforcement to the laboratory
or a combination of both, depending on the location and orienta- behavior of pull and shear tests on fully-grouted rebar bolt are
tion of reinforcements relative to discontinuities. The objective of presented in Fig. 11. During the calibration process, the best results
the two independent calibration stages illustrated in Fig. 9 is to were obtained using an active length of 40 mm (see Fig. 4b for the
investigate whether there exists a unique set of input parameters in illustration of active length). In the simulation of pull test, the axial
274 N. Bahrani, J. Hadjigeorgiou / Journal of Rock Mechanics and Geotechnical Engineering 9 (2017) 267e280

Fig. 11. Calibration of local reinforcement to the results of laboratory tests on fully-grouted rebar by Stjern (1995): (a) calibration to pull test result; (b) result of shear test on local
reinforcement with input parameters obtained from calibration to pull test result; (c) calibration to shear test result; and (d) result of pull test on local reinforcement with input
parameters obtained from calibration to shear test result.

Table 2
Local reinforcement input parameters obtained from calibration to pull and shear tests.

Calibration Axial stiffness Maximum axial Rupture Shear stiffness Maximum shear 1/2 active Axial stiffness Shear stiffness Bolt spacing (m)
(kN/mm) force (kN) strain (kN/mm) force (kN) length (mm) exponent exponent

Pull test result 20 180 1 23 160 20 0.5 0 1

Shear test result 20 180 0.6 23 160 20 0.8 0 1

stiffness, maximum pull load, rupture strain and axial stiffness under pull load condition can be successfully captured using the
exponent were adjusted. The values of local reinforcement shear local reinforcement. The shear test was then simulated using the
stiffness and maximum shear load were chosen so that the shear calibrated parameters. The results shown in Fig. 11b indicate an
stiffness and yield load in the shear forceedisplacement curve overestimation of the rupture strain by about 14 mm (an over-
match with those of fully-grouted rebar bolt. Fig. 11a demonstrates estimation by 33%) and an overestimation of the peak load by about
that the forceedisplacement curve of the fully-grouted rebar bolt 75 kN (an overestimation by 41%).
Next, the local reinforcement was calibrated to the shear test
results by adjusting the rupture strain and the shear stiffness
exponent, and keeping the rest of parameters constant. The pull
test was then simulated using the calibrated parameters. The re-
sults shown in Fig. 11c and d reveal that when the local reinforce-
ment is calibrated to the results of shear test, the rupture strain in
the pull test is underestimated by about 15 mm (an underestima-
tion by 38%). The input parameters obtained from the two cali-
bration stages are listed in Table 2.
As discussed in Section 3, the forceedisplacement behavior of
the fully-grouted rebar during the shear test consists of three main
stages: elastic behavior (i.e. increase in shear force with a slope
equal to the shear stiffness until the yield load), hardening behavior
(i.e. reduction in the slope of forceedisplacement curve until the
peak load), and bolt rupture. Interestingly, the shear forcee
displacement determined from the sum of reaction forces on the
right block (e.g. Fig. 11c) realistically captures these three stages,
although the shear forceedisplacement behavior implemented to
the local reinforcement model does not represent the hardening
Fig. 12. Shear and axial forces generated in the local reinforcement element during effect (see Fig. 3a). This was investigated by obtaining the shear and
shear test.
 N. Bahrani, J. Hadjigeorgiou / Journal of Rock Mechanics and Geotechnical Engineering 9 (2017) 267e280 275

axial forces directly from the local reinforcement element itself (i.e.
not from reaction forces generated on the block boundaries), and
the results are shown in Fig. 12. It can be seen from this figure that
during the shear test, the axial force is also generated in the local
reinforcement, but after about 5 mm of shear displacement when
shear force reaches its maximum value. From this point on, with the
increase in the shear displacement, the axial force gradually in-
creases until rupture (defined by the local reinforcement active
length) occurs. In fact, the combination of shear and axial forces
generated in the local reinforcement during the shear test results in
a shear forceedisplacement curve, which is similar to that of lab-
oratory shear tests conducted by Stjern (1995) and Chen (2014), as
shown in Fig. 2b.

5.3. Calibration of global reinforcement (“Rockbolt” element)

During the calibration process, it was realized that the number

of segments along the “Rockbolt” element (in fact the number of
segments near the joint) has a significant impact on the results,
especially in the shear test. The results presented in this section are
from a model in which the “Rockbolt” element consists of 80 seg-
ments. This was selected to ensure that at least one node falls inside
each triangular zone near the joint, as shown in Fig. 13.
The results of calibration of the “Rockbolt” element to the lab-
oratory pull and shear tests are presented in Fig. 14. The values for
the cross-sectional area, exposed perimeter, density, elastic
modulus, and tensile yield strength were obtained from Stjern
(1995) and the known steel properties. The forceedisplacement
behavior of the rebar in the pull test was matched by adjusting the
values of the shear coupling spring stiffness and cohesive strength
Fig. 13. “Rockbolt” element consisting of 80 segments. and the tensile failure strain limit of the “Rockbolt” element. It was
found that the stiffness and cohesive strength of the normal
coupling spring, plastic moment and tensile failure strain limit

Fig. 14. Calibration of global reinforcement (“Rockbolt” element) to the results of laboratory tests on fully-grouted rebar by Stjern (1995): (a) calibration to pull test result; (b) result
of shear test on “Rockbolt” element with input parameters obtained from calibration to pull test result; (c) calibration to shear test result; and (d) result of pull test on “Rockbolt”
element with input parameters obtained from calibration to shear test result.
276 N. Bahrani, J. Hadjigeorgiou / Journal of Rock Mechanics and Geotechnical Engineering 9 (2017) 267e280

influence the shear forceedisplacement behavior of the “Rockbolt”

coupling spacing (m)

element.
Fig. 14a and b shows that when the “Rockbolt” element is cali-

Normal Bolt
brated to the forceedisplacement behavior of the pull test, the

1
1
rupture displacement in the shear test is overestimated by about

angle ( )
20 mm (an overestimation by 50%). This is an improvement

friction
spring
compared to the results of simulation using the local reinforce-

0
0
ment, as the peak load was also overestimated by about 75 kN

Shear coupling
spring cohesion coupling spring friction
by the local reinforcement (compare Figs. 14b and 11b). On the
other hand, when the “Rockbolt” element is calibrated to the

angle ( )
forceedisplacement behavior in the shear test (Fig. 14c), the
rupture displacement in the pull test is underestimated by about

0
0
20 mm (an underestimation by 50%) (Fig. 14d). The values of input

cohesion
Shear coupling Normal coupling Shear coupling Normal
parameters obtained from the two calibration attempts are listed in

(kN/m)
spring

8000
8000
Table 3.
The failure modes of the “Rockbolt” element in the pull and
shear test simulations are shown in Fig. 15. As can be seen from this
figure, the failure mode of the “Rockbolt” element in the pull test is
bolt (i.e. steel) failure near the joint. In the shear test, failure in-

(kN/m)

1200
1200
volves both the “Rockbolt” element (steel) and the interface (grout)
near the joint. The failure modes obtained from numerical simu-
lations are consistent with the observations made by Stjern (1995)

moment spring stiffness spring stiffness

and Chen (2014) from their laboratory tests.

(GN/m/m)
It should be noted that the rock blocks in the numerical simu-
lations were considered to be an elastic medium. Consequently, the

10
10
yielding and plastic deformations that have been observed in the
shear test at the bolt-discontinuity contact were not captured.
Whether rock block yielding near the bolt-discontinuity contact

(GN/m/m)
has a significant impact on the shear loadedisplacement curve
needs further investigation. However, this was beyond the scope of

0.05
0.05
this paper.

strain limit (kN m)

Bolt tensile Plastic
5.4. Sensitivity analysis

1.9
2
The calibration of the “Rockbolt” element to the forcee
displacement behavior of the rebar bolt from the pull test is
failure

straightforward as the shear coupling spring stiffness, “Rockbolt”

0.35
0.75
yield strength, and tensile failure strain limit are directly related to
the forceedisplacement stiffness, peak strength and rupture
strength
tensile

displacement, respectively. The calibration of the “Rockbolt”

yield

(kN)
Bolt
“Rockbolt” element input parameters obtained from calibration to pull and shear tests.

180
180

element to the forceedisplacement behavior of the rebar bolt from

the shear test was, however, found to be complex. Therefore, a
(kg/m3) modulus
moment of perimeter (m) density elastic

series of sensitivity analyses was conducted on the “Rockbolt”

(GPa)
Bolt

8  104 200
8  104 200

element input parameters relevant to shear forceedisplacement

response, including the normal coupling spring stiffness and
cohesion and the plastic moment. The results of sensitivity analysis
Bolt

are presented in Fig. 16. The model calibrated to the behavior of the
rebar bolt under shear loading (Fig. 14c) was used as the base case
and its forceedisplacement response is shown in Fig. 16 with the
Bolt cross- Bolt second Exposed

dashed black curve.

0.062
0.062

Fig. 16a shows the influence of normal coupling spring stiffness

on the shear forceedisplacement behavior of the “Rockbolt”
7.9  109
7.9  109

element. In general, the increase in the value of normal coupling

area (m2) area (m4)

spring stiffness from 0.01 GN/m/m to 10 GN/m/m results in the

increase in the normal force, calculated from the sum of reaction
forces from the right block. In the case of normal coupling spring
stiffness of 0.01 GN/m/m and 0.1 GN/m/m, the generated normal
sectional

0.0003
Shear test result 0.0003

force is low, the forceedisplacement behavior is relatively linear,

and rupture does not occur in the “Rockbolt” element even after
80 mm of shear displacement. Fig. 16a demonstrates how the
Pull test result

generated normal force increases with increasing value of normal

Calibration

coupling spring stiffness. For the normal coupling spring stiffness

values of 1 GN/m/m and 10 GN/m/m, the shape of shear forcee
Table 3

displacement curve becomes similar to that from the laboratory

shear test. However, only with the normal coupling spring stiffness
 N. Bahrani, J. Hadjigeorgiou / Journal of Rock Mechanics and Geotechnical Engineering 9 (2017) 267e280 277

value of 10 GN/m/m, do the stiffness, yield load, peak load and the spring stiffness is similar to the “Cable” element under shear
rupture displacement become comparable to those of laboratory loading. It is therefore concluded that when the “Cable” element, or
shear tests. a “Rockbolt” element with a low normal coupling spring stiffness, is
The results of sensitivity analysis of normal coupling spring used for the design of an excavation to be supported with fully-
cohesion are presented in Fig. 16b. This figure shows that the grouted rebar bolts, the shear capacity of the rebar bolt is likely
stiffness, yield load and peak load decrease and the rupture underestimated. The consequence of these modeling decisions is a
displacement increases with decreasing value of normal coupling conservative design of support. This is a result of underestimating
spring cohesion from 8 MN/m to 2 MN/m. Fig. 16c indicates how the the shear capacity of the rebar bolts and incorrect interpretation of
decrease in the plastic moment (Pmom) from 2 kN m to 0.5 kN m their behavior under shear loading.
shifts the forceedisplacement curve downward by lowering the
yield and peak loads, and keeping the stiffness and the rupture
displacement constant.

6. Discussion

It is recognized that the explicit implementation of reinforce-

ment in numerical analyses program presents a considerable
challenge. Due to its relative simplicity, the so-called “Cable”
element is used to simulate rock reinforcement in both continuous
and discontinuous numerical codes, even if the type of reinforce-
ment used in the field or laboratory is not cable bolts. The conse-
quence of using the “Cable” element instead of a bending-resistant
reinforcement element such as the “Rockbolt” element has signif-
icant implications as demonstrated by the results of the present
investigation.
The results of pull and shear test simulations using the “Cable”
element, with the input parameters listed in Table 4, are illustrated
in Fig. 17. It can be seen from this figure that the “Cable” element
underestimates the shear capacity of the fully-grouted rebar bolt,
when it is calibrated to the pull test results. Note that it is not
possible to increase the shear capacity of the “Cable” element, as
this type of reinforcement element interacts with the UDEC model
through its shear coupling spring (i.e. “Cable” element does not
have normal coupling spring), and therefore it provides very little
resistance against bending. This small bending resistance arises
from the axial force generated during the shear test and the pull
capacity of the “Cable” element.
By comparing the results of sensitivity analysis of the normal
coupling spring stiffness on the “Rockbolt” element shear forcee
displacement response (Fig. 16a) and the shear forceedisplacement
response of the “Cable” element (Fig. 17b), it is understood that the
behavior of the “Rockbolt” element with low normal coupling

Pull-out test:

Shear test: Grout failure

p

Bolt failure Fig. 16. Influences of (a) normal coupling spring stiffness, (b) normal coupling spring
cohesion, and (c) plastic moment, on the forceedisplacement behavior of the “Rock-
Fig. 15. Failure modes of the “Rockbolt” element in the pull and shear tests. bolt” element in the shear test.
278 N. Bahrani, J. Hadjigeorgiou / Journal of Rock Mechanics and Geotechnical Engineering 9 (2017) 267e280

Table 4
“Cable” element input parameters obtained from calibration to pull test.

Cable cross-sectional Cable density Cable Young’s Cable tensile yield Cable extensional Grout stiffness Grout cohesive Cable spacing (m)
area (m2) (kg/m3) modulus (GPa) strength (kN) failure strain (GN/m/m) capacity (kN/m)

0.0003 8000 200 180 0.26 0.035 1200 1

This work has demonstrated that both the local and the global (2) The local reinforcement, when calibrated to the results of
(i.e. “Rockbolt” element) reinforcements are capable of capturing pull tests on the fully-grouted rebar bolt (i.e. forcee
the bending resistance of the fully-grouted rebar bolt under shear displacement response), overestimates the rebar shear ca-
loading. The following provides a comparison between these two pacity and rupture displacement. The “Rockbolt” element,
reinforcement models: however, only overestimates the shear displacement of the
rebar bolt, when it is calibrated to the rebar forcee
(1) The local reinforcement considers the local effect of rein- displacement response under tensile loading condition.
forcement where it intersects the discontinuities, and (3) The local reinforcement does not simulate the grout and its
therefore it should be used in situations where the block interaction with the bolt or ground. The grout and its me-
strength is high and discontinuity failure dominates the rock chanical properties are considered in the “Rockbolt” element.
mass behavior. However, the global reinforcement (“Rock- This allows for simulating a fully-grouted rebar with various
bolt” element) can be used for either strong or weak rock grout strengths and stiffness properties (i.e. water/cement
blocks as it simulates the boltegrout or grouterock behavior ratios).
as well. (4) The advantage of the “Rockbolt” element over the local
reinforcement is that it captures failure modes of the fully-
grouted rebar bolt (i.e. steel and grout failure) under both
pull and shear loading conditions observed in the laboratory
tests.

It is therefore concluded from the comparison of local and

global (“Rockbolt” element) reinforcement models that the
“Rockbolt” element provides a more realistic representation of
the behavior of fully-grouted rebar bolts under pull and shear
loading conditions. It should be noted that such a realistic rep-
resentation is obtained by assigning enough number of segments
along the “Rockbolt” element. This, however, may significantly
increase the computation time if large models (e.g. models of
excavations in a jointed rock mass) are used. On the other hand,
fewer input parameters are required in the local reinforcement
model compared with the “Rockbolt” element and they can be
directly obtained from laboratory pull and shear tests. This makes
the calibration of the local reinforcement computationally effi-
cient. It was demonstrated that the local reinforcement provides
less accurate response of the fully-grouted rebar bolt. Therefore,
the use of local reinforcement is suggested to be limited to the
initial stability analysis of underground excavations and support
design.
In this paper, the reinforcement models in UDEC were evaluated
under pure pull and shear loading conditions. An attempt was
made to obtain a unique set of input parameters for the rein-
forcement models that could capture the forceedisplacement
curves of the fully grouted rebar bolts under both pure pull and
pure shear loading conditions. It was concluded from the results
presented in Figs. 11 and 14 that there does not exist such a unique
set of values. However, for practical purposes, it is possible to find
the one that leads to reasonable loadedisplacement curves under
such a loading condition.

7. Conclusions

In order to explicitly represent the reinforcement in stress

analysis numerical models to be used as an engineering design tool,
it is necessary to demonstrate that it successfully reproduces the
Fig. 17. Forceedisplacement response of the “Cable” element under (a) pull and (b)
results of the reinforcement action in a controlled environment.
shear loading conditions and its comparison with laboratory test results by Stjern Only then can this work be extrapolated with confidence to in situ
(1995). conditions whereby more extraneous factors and variations come
 N. Bahrani, J. Hadjigeorgiou / Journal of Rock Mechanics and Geotechnical Engineering 9 (2017) 267e280 279

into play. This investigation focused on the use of the distinct Acknowledgements
element method as it best captures the jointed nature of a rock
mass and the reinforcement mechanics of rock bolting. This project was supported by the Natural Science and Engi-
Two types of reinforcement models implemented in UDEC, neering Council of Canada. The authors would like to acknowledge
including local reinforcement and global shearing and bending- Jim Hazzard and Efstratios Karampinos for the discussion and
resistant reinforcement (called “Rockbolt” element), were used technical advice on modeling rock reinforcement in UDEC.
to capture the forceedisplacement response of a fully-grouted
rebar bolt under laboratory pull and shear loading conditions.
The results of laboratory tests conducted by Stjern (1995) were References
used for this purpose. The calibration of numerical models was
conducted in two stages. First, the model was calibrated to the Aziz N, Jalalifar H. Experimental and numerical study of double shearing of bolt
under confinement. In: Peng SS, Mark C, Finfinger G, Tadolini S, Khair AW,
pull test results. The parameters obtained from this calibration
Heasley K, Luo Y, editors. Proceedings of the 26th International Conference on
process were then applied to the shear test model. Next, the Ground Control in Mining, Morgantown, WV, USA; 2007. p. 242e9.
model was calibrated to the shear test results, and the parameters Azuar MM, Reuille D, Net PA, Embier S. Rock mass reinforcement by passive rebars.
obtained from this calibration process were then used in the In: Proceedings of the 4th ISRM Congress, vol. 1. Rotterdam: A.A. Balkema; 1979.
p. 23e30.
simulation of pull test. The calibration criteria included the initial Chen Y, Li CC. Performance of fully encapsulated rebar bolts and D-Bolts under
stiffness, yield load, peak load, and rupture displacement on the combined pull-and-shear loading. Tunnelling and Underground Space Tech-
forceedisplacement plot. Moreover, to be consistent with labo- nology 2015a;45:99e106.
Chen Y, Li CC. Influence of loading condition and rock strength to the performance
ratory test results, the parameters were chosen in a way that the of rock bolts. Geotechnical Testing Journal 2015b;38(2):208e18.
failure mode in the model with the “Rockbolt” element repre- Chen Y, Li CC. Experimental and three-dimensional numerical studies of the
sents failure in the bolt shank near the joint in both pull and anchorage performance of rock bolts. In: Proceedings of the 13th International
Congress of Rock Mechanics, Montreal, Canada; 2015.
shear tests. Chen Y. Experimental study and stress analysis of rock bolt anchorage performance.
Both local and global reinforcement models were demonstrated Journal of Rock Mechanics and Geotechnical Engineering 2014;6(5):428e37.
to be able to capture the overall forceedisplacement response of Ferrero AM. The shear strength of reinforced rock joints. International Journal of
Rock Mechanics and Mining Sciences & Geomechanics Abstracts 1995;32(6):
the fully-grouted rebar bolt under pull and shear loading condi- 595e605.
tions. However, when the local reinforcement is calibrated to the Gao F, Stead D, Kang H. Numerical simulation of squeezing failure in a coal mine
pull test results, the rupture displacement and the peak load in the roadway due to mining-induced stresses. Rock Mechanics and Rock Engineer-
ing 2015;48:1635e45.
shear test are overestimated. On the other hand, when it is cali-
Grasselli G. 3D behaviour of bolted rock joints: experimental and numerical study.
brated to the shear test results, the rupture displacement in the pull International Journal of Rock Mechanics and Mining Sciences 2005;42(1):13e
test is underestimated. The results of simulations using the 24.
“Rockbolt” element revealed an improvement in the calibration Hadjigeorgiou J, Charette F. Rock bolting for underground excavations. Chapter 63.
In: Hustrulid WA, Bullock RL, editors. Underground mining methods. Society of
results. When the “Rockbolt” element is calibrated to the pull test Mining Engineers; 2001. p. 547e54.
results, only the rupture displacement in the shear test is over- Itasca. UDEC (universal distinct element code), Version 6.0. Minneapolis: Itasca
estimated. When the “Rockbolt” element is calibrated to the shear Consulting Group Inc.; 2014.
Jiang Y, Li B, Yamashita Y. Simulation of cracking near a large underground cavern in
test results, its rupture displacement in the pull test is still a discontinuous rock mass using the expanded distinct element method. In-
underestimated. ternational Journal of Rock Mechanics and Mining Sciences 2009;46:97e106.
The advantage of the local reinforcement over the “Rockbolt” Karampinos E, Hadjigeorgiou J, Hazzard J. Discrete element modeling of the bulking
phenomenon in deep hard rock mines. International Journal of Rock Mechanics
element is the simplicity in defining its input parameters, which are and Mining Sciences 2015;80:346e56.
directly correlated to and can be obtained from laboratory pull and Li B, Qi T, Wang ZZ, Yang L. Back analysis of grouted rock bolt pullout strength
shear tests. The main advantage of the “Rockbolt” element over the parameters from field tests. Tunnelling and Underground Space Technology
2012;28:345e9.
local reinforcement is its explicit representation of the bolt and the
Li CC, Stjern G, Myrvang A. A review of the performances of conventional and
grout, which makes a realistic simulation of their failure mode (e.g. energy-absorbing rockbolts. Journal of Rock Mechanics and Geotechnical En-
bolt rupture or grout failure) under different loading conditions gineering 2014;6(4):315e27.
Ma S, Nemcik J, Aziz N. Simulation of fully grouted rockbolts in underground
possible.
roadways using FLAC2D. Canadian Geotechnical Journal 2014;51(8):911e20.
It was demonstrated that when the “Cable” element is used to Malmgren L, Nordlund E. Interaction of shotcrete with rock and rock bolts: a nu-
simulate a rebar bolt, the shear capacity of the rebar is under- merical study. International Journal of Rock Mechanics and Mining Sciences
estimated. This has implications for the support design and stability 2008;45(4):538e53.
Nemcik J, Ma S, Aziz N, Ren T, Geng X. Numerical modeling of failure propagation in
analysis of underground excavations in jointed rock masses; the use fully grouted rock bolts subjected to tensile load. International Journal of Rock
of the “Cable” element instead of the “Rockbolt” element may result Mechanics and Mining Sciences 2014;71:293e300.
in an underestimation of the rebar bolt shear capacity, which may Ruest M, Martin L. FLAC simulation of split-pipe tests on an instrumented cable bolt.
In: Proceedings of the 104th Annual General Meeting of the Canadian Institute
lead to a conservative design of the support system. of Mining, Metallurgy and Petroleum. Montreal: Canadian Institute of Mining,
The presented work demonstrated the use of an explicit repre- Metallurgy and Petroleum; 2002.
sentation of reinforcement in a stress analysis software. This Shreedharan S, Kulatilake PHSW. Discontinuum-equivalent continuum analysis of
the stability of tunnels in a deep coal mine using the distinct element method.
investigation has shown the implications of using different ap- Rock Mechanics and Rock Engineering 2016;49(5):1903e22.
proaches for simulation of fully-grouted rebar bolt. Both ap- Stjern G. Practical performance of rock bolts. PhD Thesis. Trondheim, Norway:
proaches were calibrated based on the results of a controlled University of Trondheim; 1995.
Tatone BSA, Lisjak A, Mahabadi OK, Vlachopoulos N. Incorporating rock reinforce-
experimental program. Further implication of the choice of ment elements into numerical analysis based on the hybrid finite-discrete
approach for in situ conditions has been recognized. element method (DFEM). In: Proceedings of ISRM Congress, Montreal, Can-
ada; 2015.
Vardakos SS, Gutierrez MS, Barton NR. Back-analysis of Shimizu Tunnel No. 3 by
Conflict of interest
distinct element modeling. Tunnelling and Underground Space Technology
2007;22(4):401e13.
We wish to confirm that there are no known conflicts of interest Wiles T, Villaescusa E, Windsor CR. Rock reinforcement design for overstressed rock
associated with this publication and there has been no significant using three dimensional numerical modeling. In: Villaescusa E, Potvin Y, edi-
tors. Ground support in mining and underground construction: Proceedings of
financial support for this work that could have influenced its the 5th International Symposium on Ground Support, Perth, Australia; 2004.
outcome. p. 483e9.
280 N. Bahrani, J. Hadjigeorgiou / Journal of Rock Mechanics and Geotechnical Engineering 9 (2017) 267e280

Navid Bahrani obtained his Bachelor’s degree from Azad Professor John Hadjigeorgiou holds the Pierre Lassonde
University (South Tehran Campus) in Iran in 2006, his Chair in Mining Engineering at the University of Toronto.
M.Sc. from the University of Alberta in 2009, and his Ph.D. John previously served as Head of the Department of
from Laurentian University in 2015. He worked as a Mining, Metallurgical and Materials Engineering at Uni-
Research Engineer for MIRARCO’s Geomechanics Research versité Laval in Quebec City. Dr. Hadjigeorgiou is a P.Eng.,
Centre (GRC) in Sudbury from 2008 to 2014, during which with over 25 years of international consulting and
he was involved in various rock mechanics projects research experience in mining engineering. John is a past
including in situ stress measurement, rock mass defor- recipient of the John Franklin Award from the Canadian
mation monitoring, laboratory testing, and numerical Geotechnical Society and the Rock Mechanics Award from
modeling. He is currently a Post-Doctoral Fellow at the the Canadian Institute of Mining. John is a Fellow of the
University of Toronto’s Lassonde Institute of Mining, where Canadian Institute of Mining and Metallurgy and holds the
he conducts research in the areas of ground support and ICD.D designation from the Institute of Corporate
numerical modeling. Directors.