Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

International Journal of Rock Mechanics & Mining Sciences 126 (2020) 104179

Contents lists available at ScienceDirect

International Journal of Rock Mechanics and Mining Sciences


journal homepage: http://www.elsevier.com/locate/ijrmms

Estimation of REV for fractured rock masses based on Geological


Strength Index
Huajie Huang a, Jiayi Shen a, b, c, *, Qiushi Chen c, Murat Karakus d
a
Institute of Port, Coastal and Offshore Engineering, Zhejiang University, Hangzhou, 310015, China
b
State Key Laboratory for GeoMechanics and Deep Underground Engineering, China University of Mining & Technology, Xuzhou, 221116, China
c
Glenn Department of Civil Engineering, Clemson University, Clemson, SC, 29634, USA
d
School of Civil, Environmental and Mining Engineering, The University of Adelaide, Adelaide, SA, 5005, Australia

A R T I C L E I N F O A B S T R A C T

Keywords: The representative elementary volume (REV) of fractured rock masses is a significant index to investigate the
GSI rock mass behaviors in the continuum mechanics. In this research, a new indicator to estimate the REV size of
REV fractured rock masses based on the Geological Strength Index (GSI) is proposed. For this purpose, a new method
Synthetic rock mass
that combines the PFC-based synthetic rock mass (SRM) model with the Hoek-Brown (HB) failure criterion is
Hoek-Brown
Numerical simulation
proposed to investigate the strength and deformation properties of fractured rock masses under biaxial stress
conditions. Extensive numerical analyses are carried out to estimate variation of the uniaxial compression
strength (UCS), deformation modulus (E) and GSI of the Brunswick mine rock mass with increasing the size of the
SRM models up to a REV size. Results show that the GSI-based indicator gives relatively larger REV size
compared with the traditional UCS or deformation modulus (E) based indicators. Compared with the traditional
indicators, the proposed GSI-based indicator has merits of not only reflecting the geometrical characteristics of
rock structures but also containing both geometrical and mechanical properties of discontinuities.

1. Introduction corresponding results can be applied to represent the overall rock mass
properties.
The determination of rock mass mechanical properties is critical for The concept of REV was first proposed by Bear6 to study fluid flow in
the design of many engineering infrastructures, such as slopes, tunnels a porous medium. Over the last decades, various indicators have been
and foundations.1,2 However, it is not easy to estimate the strength and proposed to estimate the REV size by many researchers. Generally, these
deformation characteristics of fractured rock masses because of the ex­ indicators can be categorized into three groups, namely, geometrical
istence of various discontinuities, like joints, fractures, faults and indicators, mechanical indicators and hydraulic indicators, as shown in
bedding planes in the rock masses. These discontinuities are often Table 1. The geometrical and mechanical indicators can be used for
different in orientation, trace length, waviness and aperture, which various engineering applications, while the hydraulic indicators are
make the rock mass inhomogeneous and anisotropic.3,4 specifically used for projects that consider water flow conditions.
Alejano et al.5 have shown that the strength and deformation prop­ In order to obtain realistic mechanical properties of jointed rock
erties of rock masses are significantly influenced by the rock structure, masses, large volumes of rock mass configurations should be tested.
discontinuity properties and stress states. It is also known that these rock However, as Bieniawski24 noted, investigating the effect of sample size
mass properties gradually change with the increase of rock mass sizes. on the mechanical properties of a large volume jointed rock mass using
When the rock mass size is relatively small, a slight increase in sample the in-situ testing approach is not only difficult but also time-consuming
size may result in an obvious change in mechanical characteristics. and very expensive. Therefore, numerical simulations were widely used
However, rock mass properties no longer change with an increase of by many researchers7,11,15,16 to estimate the large scale rock mass
sample size when the sample size is larger than a critical value. Such properties using an upscaling procedure which involves estimating the
critical value is named as the representative elementary volume (REV) variation of the mechanical properties with increasing the size of the
size. When the sample is greater than or equal to the REV size, the examined rock volumes up to the REV size. After the determination of

* Corresponding author. Institute of Port, Coastal and Offshore Engineering, Zhejiang University, Hangzhou, 310015, China.
E-mail address: jiayi@zju.edu.cn (J. Shen).

https://doi.org/10.1016/j.ijrmms.2019.104179
Received 15 January 2019; Received in revised form 2 November 2019; Accepted 15 December 2019
Available online 27 December 2019
1365-1609/© 2019 Elsevier Ltd. All rights reserved.
H. Huang et al. International Journal of Rock Mechanics and Mining Sciences 126 (2020) 104179

Table 1 � �a
Indicators for estimating the REV of rock masses. σ3
σ 1 ¼ σ 3 þ σci mb þs (1)
σci
Authors Indicators

Group 1 Esmaieli et al.7 Fracture intensity P32 where σ1 and σ3 are the major and minor principal stresses respectively.
Geometrical Chae and Seo8 Ratio of the micro-cracks mb, s, and a are HB input parameters, which depend on the fracturing
indicators Zhang et al.9 Fracture intensity P32
degree of the rock mass and can be estimated from the intact HB con­
Xia et al.10 Blockiness
Ni et al.11 Fracture intensity P32 stant mi, GSI and D, given by
Li et al.12 Geometrical connectivity
mb ¼ ​ mi eðGSI 100Þ=ð28 14DÞ
(2)
Group 2 Min and Jing13 Deformation modulus and
Mechanical Poisson’s ratio
indicators Pariseau et al.14 Deformation modulus and shear
s ¼ eðGSI 100Þ=ð9 3DÞ
(3)
modulus
Esmaieli et al.7 UCS and deformation modulus 1 eð GSI=15Þ
eð 20=3Þ

Khani et al.15 Deformation modulus and a¼ þ (4)


2 6
Poisson’s ratio
Yang et al.16 UCS and deformation modulus The details of calculation and selection of mi and D can be found in
Ni et al.11 Damage coefficient the papers by Hoek and Brown,27 Hoek et al.,26 Shen and Karakus,28
Farahmand et al.17 UCS and deformation modulus Wang and Shen29 and Hoek and Brown.4
Group 3 Kulatilake and Average block permeability The GSI is a rock mass classification system that has been developed
Hydraulic Panda18 in engineering rock mechanics to meet the need for estimating reliable
indicators Wang et al.19 Hydraulic conductivity
rock mass properties for designing and analyzing tunnels, slopes or
Min et al.20 Permeability tensor
Chen et al.21 Permeability tensor foundations. The value of GSI is estimated from visual examination of
Rong et al.22 Permeability tensor the rock mass exposed in outcrops, therefore, the heart of the GSI clas­
Wang et al.23 Permeability coefficient sification is a careful engineering geology description of the rock mass
which involves two factors, rock structure and discontinuity surface
condition.
the REV size, the overall strength and deformation characteristics of
However, due to the lack of measurable parameters and related in­
rock masses for a specific engineering project could be estimated and the
terval limits or ratings for describing the rock structure and the surface
rock mass with the size of REV could be regarded as isotropic materials.
conditions of the discontinuities, the use of the GSI chart involves some
Although there are various indicators available as listed in Table 1,
subjectivity. In order to facilitate the use of the classification system
one still must take into account the following limitations when using
especially by inexperienced engineers, many researchers30–32 proposed
them for the analysis of REV size for different rock projects: (1)
quantitative GSI charts by introducing measurable parameters. For
Geometrical indicators can only be used for initial estimation for the
example, Cai et al.31 made an attempt to provide a more quantitative
REV size in a time-efficient manner at an early stage of a project. As
approach to evaluate GSI. For this purpose, two terms, block volume
geometrical indicators can only reflect the rock mass structures, it
(Vb) and joint condition factor (Jc) were introduced into the existing GSI
cannot provide information of mechanical parameters of discontinuities,
classification scheme. In order to assign ratings to these terms, the use of
such as the mechanical effects of infilling of discontinuities, which
some easily measurable input parameters such as roughness, weath­
reduce its reliability. (2) The existing mechanical indicators for esti­
ering, infilling and volumetric joint count were suggested. According to
mating REV size are based on the uniaxial compression tests, which can
the selected rating intervals, the GSI chart has been modified, as shown
only represent the rock mass behavior under the unconfined stress state.
in Fig. 1. By adding measurable quantitative inputs for a quantitative
However, rock masses are generally in states of triaxial stresses and the
output, the system becomes less subjective while maintains its overall
properties will be significantly influenced by the triaxial stress states.
simplicity. Once a GSI value has been decided upon, it can be used in
With the aim of achieving a better understanding of the fundamental
conjunction with appropriate empirical equations to calculate the
rock mass mechanisms and improving the accuracy of the determination
strength and deformation of a rock mass.
of REV size, in this research we propose a new indicator to estimate the
However, it should be noted that the rock mechanical properties of
REV size of fractured rock masses based on the Geological Strength
fractured rock masses are generally anisotropic and stress path depen­
Index (GSI) classification which has merits of not only reflecting the
dent13、33. The HB failure criterion based on the GSI system is used for
geometrical characteristic of rock structures but also containing both
predicting isotropic rock mass strength which cannot reflect such
geometrical and mechanical properties of discontinuities.
anisotropic behavior and stress path dependent characteristics.
In order to establish such GSI-based REV indicator, we also propose a
As GSI not only describes the geometrical feature of rock masses but
new method that combines the PFC-based synthetic rock mass (SRM)
also relates the mechanical properties of the discontinuities, a GSI-based
model with the Hoek-Brown (HB) failure criterion to investigate the
indicator could be ideal to calculate the REV size for a rock mass,
strength and deformation of fractured rock masses under biaxial stress
overcoming limitations of the traditional indicators based on either
conditions.
geometrical or mechanical characteristics alone.
2. Hoek-Brown criterion based on Geological Strength Index
3. Synthetic rock mass for numerical analyses
In rock engineering, the Hoek-Brown (HB) criterion based on the
In this research, the synthetic rock mass (SRM) model of the Bruns­
Geological Strength Index (GSI) has been widely used for predicting
wick mine rocks was employed as an example to illustrate the method of
isotropic rock mass strength for many decades. The original HB crite­
estimating REV based on the GSI indicator using an upscaling process.
rion, which was proposed by Hoek and Brown25 for estimating the intact
The upscaling procedure involves estimating the variation of the me­
rock strength, requires two intact rock properties, namely, the uniaxial
chanical properties with increasing the size of the rock mass volumes up
compressive strength (UCS or σci) and a constant mi of the intact rock. It
to a REV in which stationary limits are reached for the properties under
was then extended to estimate the rock mass strength using the GSI and
study.
disturbance factor D.26 The latest version of the HB criterion presented
by Hoek and Brown4 is expressed in Eq. (1).

2
H. Huang et al. International Journal of Rock Mechanics and Mining Sciences 126 (2020) 104179

Fig. 1. Quantification of GSI chart (Cai et al.31).

3
H. Huang et al. International Journal of Rock Mechanics and Mining Sciences 126 (2020) 104179

3.1. PFC2D-based synthetic rock mass model Table 2


Fracture set field data for fracture system generation.
The SRM model, which was constructed in PFC2D software in this Fracture characteristic Set 1 Set 2 Set 3
research, is based on two well-developed numerical techniques, the
2D-dip (deg) 90 89 22
discrete fracture network (DFN) and the bonded particle model (BPM). Trace length (m) 1.40 1.44 1.16
A DFN is generally generated using the Monte-Carlo simulation. P21 (m 1) 0.63 1.80 1.95
Several statistical parameters, including the mean and standard devia­
tion of joint set dip angle and trace length, are required to control the
orientation, size and intensity of each joint set in the DFN. scale effect of intact rocks in PFC2D, initial micro defects need to be
In BPM, rigid particles are bonded together to form the basic ele­ added into intact rocks,35 which is not considered in this research.
ments of a hard rock. The bond represents the rock cohesive component Based on the geometrical parameters of fractures listed in Table 2, a
and breaks when the external stress exceeds the prescribed bond big DFN model with the size of 100 � 200 m was generated using the
strength. The particle movement is influenced by frictional resistance. Monte Carlo method. 12 small DFNs with the same sizes corresponding
As the BPM explicitly considers the particle-scale mechanical behavior, to the 12 intact rock samples (see Fig. 3) were extracted at the center of
it requires no continuum-scale constitutive model (such as the Mohr- the big DFN, as shown in Fig. 4. Moreover, in order to avoid the random
Coulomb and Hoek-Brown criteria) to control the macroscopic me­ effect in the DFN model, we generated 10 big DFN models and repeated
chanical rock behaviors. The micro-parameters in BPM can be identified the procedure of extracting small samples from the big DFN models,
following a calibration process to assure the similarity of the macro- resulting in a total of 120 DFN models for subsequent SRM model gen­
properties can be obtained. erations and numerical simulations.
The SRM is then constructed by inserting a DFN into a BPM assembly Finally, the intact rock samples (Fig. 3) were combined with DFN
and a “smooth joint” logic is developed to assert smooth particle sliding models (Fig. 4) to generate the SRM models. Fig. 5 shows a typical set of
along joints. The smooth joint model (SJM), which resolves the local 12 SRM models with various sizes. In total, 120 SRM samples were
force-displacement law onto the plane of the joint, could be used to generated in PFC2D.
simulate the behavior of such “smooth joint”. By this approach, bonded
behavior, Coulomb sliding and tensile opening can be realistically rep­ 4. Calibration of synthetic rock mass models
resented.34 Fig. 2 shows the main components of a synthetic rock mass.
The micro-parameters of intact rocks and fractures in the SRM
samples need to be identified following a calibration process to assure
3.2. SRM model of the brunswick mine rock mass the rock mass strength and deformation properties estimated from SRM
models are the same as that from laboratory and field data.7
Rock mass data was collected from Brunswick Mine,7 which is
located in northeastern New Brunswick, Canada. The deposit is hosted 4.1. Micro-parameters of the bonded particle model for intact rocks
mainly by metamorphosed volcaniclastic sediments and tuffs that
overlie felsic volcanic rocks and have a high specific gravity (4.3) and Uniaxial compression tests were carried out to calibrate the micro-
stiffness (UCS up to 205MPa). The elastic modulus of the intact massive parameters of the bonded particle model (BPM) for intact rocks to
sulphide is 104GPa and the Poisson’s ratio is 0.29. make sure the BPM model can provide the same mechanical properties
In this case, the mean values of joints dip angles in 2-D were calcu­ as from laboratory tests. In uniaxial compression tests, a symmetric
lated from the mean direction of fractures in 3-D onto the East-West vertical constant loading velocity of 0.02 m/s was applied on top and
plane. The trace length was assumed to follow a normal distribution bottom boundaries of rock models, as shown in Fig. 6.
with a specific fracture intensity P21 (total length of fracture traces per The details of calibration procedure can be found in the paper by
unit area). Table 2 shows the basic information of the identified three Zhou et al.36 Firstly, the Poisson’s ratio v was obtained by calibrating the
sets of fractures used for building the synthetic rock mass models. The normal-to-shear stiffness ratio kn/ks; then the effective modulus Ec was
shear strength parameters of all fractures are assumed to be cohesionless modified to acquire the target elastic modulus; the last stage was the
with cj ¼ 0 and angle of friction ϕj ¼ 30� .7 In addition, the infill of joints modification of the bond tensile strength t and bond cohesion cb to
was not considered in this case. obtain the desired value of UCS.
In order to build the SRM models, firstly, 12 intact rock samples with Fig. 6 shows a typical failure pattern and the associated stress and
different sizes were generated based on the BPM, as shown in Fig. 3. The strain responses of the 10 � 20 m intact rock specimen modeled by BPM
height-width ratio of samples was set to be 2. under the uniaxial compression test. The values of micro-parameters of
It should be noted that the scale effect of intact rocks was not this intact rock model with ID ¼ k are listed in Table 3. The values of
considered in this research, which means the UCS is assumed to be the mechanical properties (UCS, elastic modulus and Poisson’s ratio)
same for all intact rock samples. It is known that the strength of intact generated by the BPM using these micro-parameters are shown in
rocks is scale-dependent, which is caused by the existence of micro- Table 4, which are close to laboratory test results in Section 3.2. It
cracks and flaws in the rocks. Therefore, in order to investigate the should be noted that the value of UCS was obtained directly from the

Fig. 2. Synthetic rock mass components.

4
H. Huang et al. International Journal of Rock Mechanics and Mining Sciences 126 (2020) 104179

Fig. 3. Intact rock samples with different sizes.

between the fracture and the minor principal stress direction. Four rock
mass samples with θ ¼ 40� , 45� , 50� and 55� were used for biaxial tests.
In these samples, the intact rock was modeled with the BPM using the
micro-parameters listed in Table 3, and the strength and deformation
behaviors of fractures were modeled with the SJM.
During the biaxial test, the lateral stress was loaded to the samples
with a confining stress σ3 ¼ 1MPa in horizontal directions. A symmetric
vertical constant loading velocity of 0.02 m/s was applied on top and
bottom boundaries of rock mass models and the peak strength σ1 was
recorded when the sample fails, as shown in Fig. 7.
Then, the normal stress σn and shear stress τ acting on the fracture
plane can be calculated using the critical stress state (σ1, σ 3). Parametric
studies were carried out to get the optimal micro-parameters of the
fracture to ensure that the rock sample can generate the same response
Fig. 4. The big DFN model (right) and the corresponding 12 small DFN models as that from field data. Fig. 7 presents the final stress-strain curves of
(left) extracted and used for generating SRM samples. four rock samples after the final calibration. These optimal values of
input micro-parameters (knj ¼ 2500GPa/m, ksj ¼ 500GPa/m, μj ¼ 0.2)
maximum strength of the axial stress-strain curve of the uniaxial could give proper macro-properties of the fracture with cj ¼ 0 and ϕj ¼
compression tests. The elastic modulus was estimated using the tangent 29.78� , as shown in Fig. 8, which is identical to field data with cj ¼ 0 and
modulus at the 50% UCS. The Poisson’s ratio was estimated by taking ϕj ¼ 30� .
the slope of the radial strain-axial train curve between 20% and 30% of
maximum strength. 5. The proposed method for estimating REV based on GSI

5.1. Quantify fracturing for SRM using equivalent GSI


4.2. Micro-parameters of the smooth joint model for fractures
Properties of the intact rock and discontinuities are necessary and
In this research, the smooth joint model (SJM) was used to simulate identical input parameters for both the HB criterion and the SRM
the behavior of a planar interface. The model has three input micro- models. Therefore, if the isotropic condition for the rock mass properties
parameters: normal stiffness knj, shear stiffness ksj and friction coeffi­ is satisfied, then the jointed SRM model could be treated as a rock mass
cient μj. The selection of knj, ksj and μj was based on the inverse-modeling satisfying the HB criterion.
calibration approach based on biaxial compression tests to ensure that In general, the GSI is used for quantifying the fracturing of large rock
the rock mass model can generate the same response as that from site masses. In order to quantify the degree of fracturing of these synthetic
data with cj ¼ 0 and ϕj ¼ 30� . rock mass samples with the size ranging from 0.05 m to 12 m (Fig. 5), a
Synthetic rock mass samples used for numerical simulation have a concept of the equivalent GSI proposed by Alejano et al.5 was used in
width of 50mm and a height of 100mm. θ is the inclination angle this research.

5
H. Huang et al. International Journal of Rock Mechanics and Mining Sciences 126 (2020) 104179

Fig. 5. Synthetic rock mass samples with different sizes.

The strategy to calculate the equivalent GSI for SRM models is to use
the HB failure criterion to fit biaxial data using a regression method. The
procedure to calculate the equivalent GSI is introduced as follows:

Table 4
Mechanical properties calculated from BPM models.
Sample ID Sample length (m) UCS (MPa) Ei (GPa) v ρ (g/cm3)
a 0.05 205 105 0.28 4.3
b 0.1 205 106 0.31 4.3
c 0.2 204 104 0.30 4.3
d 0.5 205 105 0.31 4.3
e 1.5 205 105 0.29 4.3
f 3.5 205 104 0.30 4.3
g 5 205 106 0.29 4.3
h 7 205 106 0.30 4.3
i 8 205 105 0.29 4.3
Fig. 6. The mechanical response of the intact rock under the uniaxial
j 9 205 105 0.30 4.3
compression test.
k 10 205 105 0.31 4.3
l 12 205 105 0.31 4.3

Table 3
Micro-parameters of the BPM models.
Sample ID Sample length (m) Rmin (m) Rmax/Rmin kn/ks Ec (GPa) t (MPa) ϕb (� ) cb (MPa) μ e

a 0.05 0.0005 1.66 2.3 73.0 116.4 20 116.4 0.4 0.16


b 0.1 0.0010 1.66 2.3 73.2 110.5 20 110.5 0.4 0.16
c 0.2 0.0015 1.66 2.3 73.0 123.2 20 123.2 0.4 0.16
d 0.5 0.003 1.66 2.3 72.8 126.2 20 126.2 0.4 0.16
e 1.5 0.008 1.66 2.3 72.9 118.2 20 118.2 0.4 0.16
f 3.5 0.020 1.66 2.3 72.6 122.8 20 122.8 0.4 0.16
g 5 0.026 1.66 2.3 73.1 118.9 20 118.9 0.4 0.16
h 7 0.035 1.66 2.3 73.0 114.8 20 114.8 0.4 0.16
i 8 0.040 1.66 2.3 73.2 126.2 20 126.2 0.4 0.16
j 9 0.045 1.66 2.3 73.0 114.0 20 114.0 0.4 0.16
k 10 0.050 1.66 2.3 73.0 120.2 20 120.2 0.4 0.16
l 12 0.055 1.66 2.3 73.1 123.6 20 123.6 0.4 0.16

6
H. Huang et al. International Journal of Rock Mechanics and Mining Sciences 126 (2020) 104179

model.

5.2. Comprehensive results of equivalent GSI for SRM models

We carried out extensive numerical analyses to identify the equiva­


lent GSI values of 120 SRM samples and presented results in Fig. 10. The
black dots in Fig. 10 represent the mean value of GSI for a given size of
SRM models. It is shown that the values of GSI scatter at small sizes
while tend to converge with the increase of the sample size. This is
because when the sample size is smaller, there might be no fractures or
few fractures in the sample, resulting in the obvious differences of GSI
values. When the sample size is relatively large, the fracture patterns in
the samples tend to be similar. Thus, the value of GSI tends to converge
to a stable level.
In this research, the Coefficient of Variance (CV) was used to
Fig. 7. The stress-strain curves of rock mass samples under biaxial tests.
calculate the REV size of rock masses. CV could quantify the random
influence of a group of data by calculating the value of Standard Devi­
ation/Mean7、10、13、33. According to Esmaieli et al.7 and Mahnaz
et al.,33 when the value of CV was less than 20%, the sample size could
be regarded as REV. The results of REV in Table 6 show that the sample
(j) with the size of 7 � 14 m gives the REV size of the Brunswick mine
rock masses.

5.3. Comparison with the mechanical REV size

In order to test the reliability of the proposed GSI-based REV size, we


also calculated the REV size of this Brunswick mine rock mass using
traditional mechanical indicators, the UCS and deformation modulus E.
The calculation results are shown in Figs. 11 and 12. The results of the
CV are listed in Table 7. It is found that the UCS and E-based indicators
give relatively small REV sizes (5 � 10 m) compared with the GSI-based
indicator (7 � 14 m). This could be explained by the fact that the UCS
and E-based indicators consider only one type of mechanical informa­
tion, compared with the GSI-based indicator, which contains two types
Fig. 8. The σn-τ curve of the fracture under biaxial tests. of information: geometrical information of rock structure and mechan­
ical information of the discontinuity. As stated by Zhang et al.,9 if the
Firstly, substitute Eqs. (2) to (4) into Eq. (1), the HB failure criterion can REV calculation accounts for more information of fractures, such as
be expressed as σ1 ¼ f (GSI, mi, σ ci, D and σ3). The value of blasting density, length, dip direction and dip angle, the REV can have more
damage factor D was assumed to be zero in this research. Then, the obvious practical applications. That means, in general, more informa­
values of σci and mi were estimated from the numerical biaxial tests of tion included in the indicator for estimating REV values, the better rock
the intact rocks (Fig. 3) over a confining stress σ3 ranging from 0 to mass properties will be achieved. In this regard, the proposed GSI-based
0.5σci by using regression analyses suggested by Hoek and Brown27. indicator would be useful for practical engineering application.
The values of mi and σci_fitted of various scales of intact rocks (Fig. 3)
are listed in Table 5. 6. Conclusions
Finally, the equivalent GSI for SRM models can be calculated from
the HB failure criterion, which can be expressed as σ1 ¼ f (GSI, mi, In this research, we propose a new indicator to estimate the repre­
σ ci_fitted, D and σ3), to fit biaxial data using a regression method. Take the sentative elementary volume (REV) of jointed rock masses based on the
SRM model (ID ¼ k) as an example, the values of mi ¼ 3.28, σci_fitted ¼
211.56MPa and D ¼ 0. The values of biaxial data (σ3, σ 1) and the final
fitting result are shown in Fig. 9, which gives GSI ¼ 50 for this SRM

Table 5
mi and σci_fitted of the intact rock models.
Sample ID Sample size (m) mi σci_fitted (MPa)
a 0.05 2.84 211.75
b 0.1 2.95 212.85
c 0.2 3.06 217.51
d 0.5 3.37 218.02
e 1.5 2.83 214.04
f 3.5 3.22 212.30
g 5 3.56 203.72
h 7 2.79 214.03
i 8 3.36 210.44
j 9 3.14 209.10
k 10 3.28 211.56
l 12 3.03 217.32
Fig. 9. GSI-equivalent result for the SRM model (ID ¼ k).

7
H. Huang et al. International Journal of Rock Mechanics and Mining Sciences 126 (2020) 104179

Table 7
CV results for estimating REV size based on UCS and E.
Sample ID Sample size (m) CV of UCS CV of E

a 0.05 0.51 0.16


b 0.1 0.73 0.25
c 0.2 0.88 0.30
d 0.5 1.93 0.29
e 1.5 1.11 0.55
f 3.5 0.56 0.62
g 5 0.19 0.19
h 7 0.18 0.17

Geological Strength Index (GSI). Compared with the existing two


geometrical and mechanical categories, the proposed GSI-based indi­
cator has merits of reflecting the geometrical characteristics of rock
structures, but also containing both geometrical and mechanical prop­
Fig. 10. Relationships between GSI and sample size of SRM models. erties of discontinuities.
For this purpose, the synthetic rock mass (SRM) models of the
Brunswick mine rocks, based on the discrete fracture network (DFN) and
Table 6
CV results for estimating GSI-based REV.
the bonded particle model (BPM) techniques, were constructed in
PFC2D to investigate the strength of fractured rock masses with
Sample ID Sample size (m) CV of GSI
increasing the size of the rock mass models up to a REV size. The micro-
a 0.05 0.20 parameters of the SRM model were calibrated using the published
b 0.1 0.33 experimental intact rock and field joint data to make sure the SRM
c 0.2 0.37
d 0.5 0.57
model is capable of capturing the mechanical behaviors of fractured
e 1.5 0.34 rock masses.
f 3.5 0.31 Biaxial tests on the 120 SRM models with different sizes were carried
g 5 0.22 out to get their failure envelopes, and then, the Hoek-Brown (HB) failure
h 7 0.13
criterion was used to fit the biaxial data to get the values of equivalent
GSI of the SRM models. Based on the evaluation of CV value of equiv­
alent GSI values, GSI-based REV size (7 � 14 m) is obtained, which is
relatively larger compared with the traditional UCS and E-based REVs
(5 � 10 m).

Declaration of competing interest

The authors declared that they have no conflicts of interest to this


work. We declare that we do not have any commercial or associative
interest that represents a conflict of interest in connection with the work
submitted.

Acknowledgment

This research has been funded by the National Natural Science


Foundation of China (No.51504218) and the State Key Laboratory for
GeoMechanics and Deep Underground Engineering, China University of
Fig. 11. Relationships between UCS and sample size of SRM models.
Mining and Technology (SKLGDUEK1808) and the Australian
Endeavour Research Fellowship.

References

1 Shen J, Priest SD, Karakus M. Determination of Mohr-Coulomb shear strength


parameters from generalized Hoek-Brown criterion for slope stability analysis. Rock
Mech Rock Eng. 2012;45:123–129.
2 Zheng H, Li T, Shen J, Xu C, Sun H, Lü Q. The effects of blast damage zone thickness
on rock slope stability. Eng Geol. 2018;246:19–27.
3 Zhang G, Karakus M, Tang H, Ge Y, Zhang L. A new method estimating the 2D joint
roughness coefficient for discontinuity surfaces in rock masses. Int J Rock Mech Min
Sci. 2014;72:191–198.
4 Hoek E, Brown ET. The Hoek-Brown failure criterion and GSI-2018 edition. J Rock
Mech.Geotech Eng. 2019;11(3):445–463.
5 Alejano LR, Arzúa J, Bozorgzadeh N, Harrison JP. Triaxial strength and deformability
of intact and increasingly jointed granite samples. Int J Rock Mech Min Sci. 2017;95:
87–103.
6 Bear J. Dynamics of Fluids in Porous Media. Amsterdam: Elsevier; 1972.
7 Esmaieli K, Hadjigeorgiou J, Grenon M. Estimating geometrical and mechanical REV
based on synthetic rock mass models at Brunswick Mine. Int J Rock Mech Min Sci.
2010;47(6):915–926.
Fig. 12. Relationships between deformation modulus and sample size of 8 Chae BG, Seo YS. Homogenization analysis for estimating the elastic modulus and
SRM models. representative elementary volume of Inada Granite in Japan. Geosci J. 2011;15(4):
387–394.

8
H. Huang et al. International Journal of Rock Mechanics and Mining Sciences 126 (2020) 104179

9 Zhang W, Chen J, Liu C, Huang R, Li M, Zhang Y. Determination of geometrical and 23 Wang Z, Li W, Bi L, Qiao L, Liu R, Liu J. Estimation of the REV size and equivalent
structural representative volume elements at the Baihetan dam site. Rock Mech Rock permeability coefficient of fractured rock masses with an emphasis on comparing the
Eng. 2012;45(3):409–419. radial and unidirectional flow configurations. Rock Mech Rock Eng. 2018;51(5):
10 Xia L, Zheng Y, Yu Q. Estimation of the REV size for blockiness of fractured rock 1457–1471.
masses. Comput Geotech. 2016;76:83–92. 24 Bieniawski Z. Engineering classification of jointed rock masses. Civ Eng South Afr.
11 Ni P, Wang S, Wang C, Zhang S. Estimation of REV size for fractured rock mass based 1973;15(12).
on damage coefficient. Rock Mech Rock Eng. 2017;50(3):555–570. 25 Hoek E, Brown ET. Empirical strength criterion for rock masses. J Geotech Geoenviron
12 Li Y, Chen J, Shang Y. Determination of the geometrical REV based on fracture Eng. 1980;106:1013–1035.
connectivity: a case study of an underground excavation at the Songta dam site, 26 Hoek E, Carranza-Torres C, Corkum B. Hoek-Brown failure criterion-2002 Edition.
China. Bull Eng Geol Environ. 2018;77(4):1599–1606. In: Proceedings of the Fifth North American Rock Mechanics Symposium. vol. 1. 2002:
13 Min KB, Jing L. Numerical determination of the equivalent elastic compliance tensor 18–22.
for fractured rock masses using the distinct element method. Int J Rock Mech Min Sci. 27 Hoek E, Brown ET. Practical estimates of rock mass strength. Int J Rock Mech Min Sci.
2003;40(6):795–816. 1997;34(8):1165–1186.
14 Pariseau WG, Puri S, Schmelter SC. A new model for effects of impersistent joint sets 28 Shen J, Karakus M. Simplified method for estimating the Hoek-Brown constant for
on rock slope stability. Int J Rock Mech Min Sci. 2008;45(2):122–131. intact rocks. J Geotech Geoenviron Eng. 2014;140(6), 04014025.
15 Khani A, Baghbanan A, Norouzi S, Hashemolhosseini H. Effects of fracture geometry 29 Wang W, Shen J. Comparison of existing methods and a new tensile strength based
and stress on the strength of a fractured rock mass. Int J Rock Mech Min Sci. 2013;60 model in estimating the Hoek-Brown constant mi for intact rocks. Eng Geol. 2017;224:
(8):345–352. 87–96.
16 Yang J, Chen W, Yang D, Yuan J. Numerical determination of strength and 30 Sonmez H, Ulusay R. Modifications to the geological strength index (GSI) and their
deformability of fractured rock mass by FEM modeling. Comput Geotech. 2015;64: applicability to stability of slopes. Int J Rock Mech Min Sci. 1999;36(6):743–760.
20–31. 31 Cai M, Kaiser PK, Uno H, Tasakab Y, Minami M. Estimation of rock mass deformation
17 Farahmand K, Vazaios I, Diederichs MS, Vlachopoulos N. Investigating the scale- modulus and strength of jointed hard rock masses using the GSI system. Int J Rock
dependency of the geometrical and mechanical properties of a moderately jointed Mech Min Sci. 2004;41(1):3–19.
rock using a synthetic rock mass (SRM) approach. Comput Geotech. 2018;95: 32 Hoek E, Carter TG, Diederichs MS. In: Quantification of the Geological Strength Index
162–179. Chart. 47th US Rock Mechanics/geomechanics Symposium. American Rock Mechanics
18 Kulatilake PHSW, Panda BB. Effect of block size and joint geometry on jointed rock Association; 2013.
hydraulics and REV. J Eng Mech. 2000;126(8):850–858. 33 Laghaei M, Baghbanan A, Hashemolhosseini H, Dehghanipoodeh M. Numerical
19 Wang M, Kulatilake P, Um J, Narvaiz J. Estimation of REV size and three- determination of deformability and strength of 3D fractured rock mass. Int J Rock
dimensional hydraulic conductivity tensor for a fractured rock mass through a single Mech Min Sci. 2018;110:246–256.
well packer test and discrete fracture fluid flow modeling. Int J Rock Mech Min Sci. 34 Poulsen B, Adhikary D, Elmouttie M, Wilkins A. Convergence of synthetic rock mass
2002;39(7):887–904. modelling and the Hoek-Brown strength criterion. Int J Rock Mech Min Sci. 2015;80:
20 Min KB, Jing L, Stephansson O. Determining the equivalent permeability tensor for 171–180.
fractured rock masses using a stochastic REV approach: method and application to 35 Zhang Q, Zhu H, Zhang L, Ding X. Study of scale effect on intact rock strength using
the field data from Sellafield, UK. Hydrogeol J. 2004;12(5):497–510. particle flow modeling. Int J Rock Mech Min Sci. 2011;48(8):1320–1328.
21 Chen SH, Feng XM, Isam S. Numerical estimation of REV and permeability tensor for 36 Zhou C, Xu C, Karakus M, Shen J. A systematic approach to the calibration of micro-
fractured rock masses by composite element method. Int J Numer Anal Methods parameters for the flat-jointed bonded particle model. GeoMech Eng. 2018;16(5):
Geomech. 2008;32:1459–1477. 471–482.
22 Rong G, Peng J, Wang X, Liu G, Hou D. Permeability tensor and representative
elementary volume of fractured rock masses. Hydrogeol J. 2013;21:1655–1671.

You might also like