Engineering Fracture Mechanics xxx (2017) xxx–xxx

Contents lists available at ScienceDirect

Engineering Fracture Mechanics

journal homepage: www.elsevier.com/locate/engfracmech

Numerical simulation of rock failure under static and dynamic

loading by splitting test of circular ring
Xibing Li, Fan Feng, Diyuan Li ⇑
School of Resources and Safety Engineering, Central South University, Changsha, Hunan 410083, China

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

Article history: The fracture propagation and failure process in rock are subjects of great interest in rock
Received 29 May 2017 mechanics. In the present study, an approach combining the finite element method
Received in revised form 13 August 2017 (FEM) and the discrete element method (DEM), named ELFEN, is adopted to simulate
Accepted 13 August 2017
and investigate the failure process of a typical hard rock (Carrara marble) under static
Available online xxxx
and dynamic splitting ring tests. The FDEM software is firstly validated by simulating
the test of a Brazilian disk under static loading. Then, the failure of circular ring specimens
Keywords:
is numerically studied under static and dynamic loading, with the ratio of internal to exter-
Numerical simulation
Splitting ring test
nal diameter (represented by k) of the specimen groups varying from 0.1 to 0.6, with incre-
Failure mode ments of 0.1. Under static loading, with the increase in internal diameter, the failure mode
Indirect tensile strength is transformed from diametrical splitting to four-fan-shaped failure (when k
0.4). Under
FDEM dynamic loading, four-fan-shaped failure occurs from geometric axial symmetry to axial
asymmetry, which means that, with increasing internal diameter, the tensile cracks along
the horizontal diametrical direction gradually deviate toward the top loading platen. The
peak load of the circular ring has a descending trend with the increase in internal diameter
under both loading conditions. A rational load value (peak load or crack initiation load) is
suggested to calculate the tensile strength of circular rock ring specimens under the split-
ting ring test. Finally, the numerical results are compared with previous experimental
results. The good agreement between the two sets of results, in terms of the failure modes
and variation trend of the indirect tensile strength with increasing internal diameters, ver-
ify the accuracy and applicability of the FDEM approach.
Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction

The failure process of brittle and hard rock has always been of great interest in rock mechanics and rock engineering,
especially regarding the rock failure modes and strength criteria. Tensile and shear fractures are the two main failure modes
for hard rocks [1]. The Brazilian test, deemed as a simple indirect testing method, has been used extensively to obtain the
tensile strength and measure the toughness of brittle materials such as rocks, rock-like materials, and concrete [2–4]. How-
ever, the conventional Brazilian test has also its own disadvantages and shortcomings in terms of the test principle and the
reliability of results, particularly for soft rocks. The splitting ring test, a disk with a small central hole, was proposed by Hobbs
to determine the indirect tensile strength of a rock [5]. It can ensure that the crack initiation starts from the periphery of the
hole in the ring specimen and thus avoid stress concentration close to the loading platen. Over the past 20 years, extensive

⇑ Corresponding author.
E-mail address: [email protected] (D. Li).

http://dx.doi.org/10.1016/j.engfracmech.2017.08.022
0013-7944/Ó 2017 Elsevier Ltd. All rights reserved.

Please cite this article in press as: Li X et al. Numerical simulation of rock failure under static and dynamic loading by splitting test of cir-
cular ring. Engng Fract Mech (2017), http://dx.doi.org/10.1016/j.engfracmech.2017.08.022
2 X. Li et al. / Engineering Fracture Mechanics xxx (2017) xxx–xxx

Nomenclature

A cross-sectional area of test specimen

B area of the element
d internal diameter of ring specimen
D external diameter of ring specimen
E Young’s modulus
Ed elastic damaged secant modulus
Et tangential softening modulus
ft tensile strength of the material
Gf fracture energy
K Ic fracture toughness in mode i
lc local control length in finite element context
nn local coordinate system associated with the principal stresses
P peak load
Pn normal penalty in FDEM approach
Pt tangential penalty in FDEM approach
r internal radius of circular ring
R external radius of circular ring
t thickness of the test specimen in laboratory tests
ri principal stress invariants
rt tensile strength of Brazilian disk
rtr indirect tensile strength of circular ring
rEtrs static tensile strength of circular ring based on previous experimental results
rNtrs static tensile strength of circular ring based on FDEM approach
rEtrd dynamic tensile strength of circular ring based on previous experimental results
rNtrd dynamic tensile strength of circular ring based on FDEM approach
rh loading rate
x damage parameter
w(e) function of the total strain
k ratio of internal diameter to external diameter of the circular ring
l Poisson’s ratio
c friction of newly generated cracks in FDEM approach

experimental and numerical studies have been carried out with the splitting ring test [6–11]. The boundary element method
(BEM) approach was used by Chen and Hsu to model the ring test for determining the indirect tensile strength of anisotropic
rocks [6]. The results showed that the tensile strength of anisotropic rocks determined by the ring test was influenced by
their elastic properties, the angle between the planes of rock anisotropy and the loading direction, the diameter of the central
hole, and the contact condition of loading. Zhu et al. [7] adopted the rock-failure process analysis (RFPA) software to model
the ring specimen, and proposed a new test scheme in which the crack initiation load was used rather than the peak load to
determine the indirect tensile strength of a rock based on the Hobbs’ equation. You et al. [8] found that the peak load for the
rock rings decreased exponentially with the increase in internal diameter, and the Hobbs’ equation based on the elastic
mechanics has great difference compared with the experimental results. In the splitting ring tests, the critical factor to deter-
mine the tensile strength of a rock is the ratio of ring’s radii. Li et al. [10] experimentally studied the static and dynamic fail-
ure characteristics of the splitting test of circular rings. It was found that the tensile strength of rock ring specimens is
dependent on the ratio of ring’s radii.
The fracture process of brittle materials such as rock and glasses has been a critical issue in fracture mechanics for some
decades [12,13]. Many researchers employ experimental, analytical, and numerical approaches to study the failure processes
of rock and rock-like materials. Numerical approaches, such as the finite difference method (FDM), finite element method
(FEM), discrete element method (DEM), and discontinuous deformation analysis (DDA) are the main tools used for simulat-
ing the fracturing processes of a rock and rock mass. Wong and Wu [14] used the numerical manifold method to investigate
the progressive failure in rock slopes. They found neither the continuous nor the discontinuous approaches, but a hybrid
method that combines both continuous and discontinuous methodologies can provide a comprehensive analysis. Based
on the physical observations of fracturing processes in brittle rocks, Li et al. [15] proposed numerical modeling schemes that,
based on the finite difference method, considered the material heterogeneity and initial microflaws at the element scale.
Wen et al. [31] studied the influence of different height to diameter ratio on the acoustic emission characteristics of coal rock
damage evolution using a microparticle flow PFC2D software platform.

Please cite this article in press as: Li X et al. Numerical simulation of rock failure under static and dynamic loading by splitting test of cir-
cular ring. Engng Fract Mech (2017), http://dx.doi.org/10.1016/j.engfracmech.2017.08.022
 X. Li et al. / Engineering Fracture Mechanics xxx (2017) xxx–xxx 3

During the fracturing processes, the rock system is transferred from a stable state (continuum system) to an unstable
domain (fractured discontinuum system) owing to the crack initiation, propagation, and coalescence. The fracturing pro-
cesses are therefore better captured by the use of a numerical tool that is able to reproduce this transition from a continuum
to a discontinuum state [16,17]. In addition, the FDEM method, which integrates the FEM and DEM into a singular tool, has
provided a medium for both continuous and discontinuous problems. For example, Mahabadi et al. [19] used an innovative
combined finite-discrete element method to simulate numerically the behavior of Brazilian disk specimens as observed in
the laboratory during dynamic, high-strain rate, and indirect tensile tests. The Brazilian disk with a plane interface test
was also investigated by using the finite-discrete element called ELFEN by Cai [17]. In his numerical tests, the Rankine rotat-
ing crack constitutive model was adopted to achieve tensile failure in hard rocks. Hamdi et al. [18] used a combined finite-
discrete element approach (ELFEN software) to simulate the complete 3D fracture process during the conventional labora-
tory testing, including the Brazilian splitting, uniaxial and biaxial compression tests. The numerical results showed good
agreement between the final fracture patterns and damage obtained during 2D and 3D simulations of the Brazilian and uni-
axial compression tests. The numerical FDEM simulations are in agreement with the input values, showing the suitability of
this approach for realistically modeling the brittle fracture during laboratory testing.
In the present study, a FDEM approach (ELFEN, Rockfield Software Ltd.) is adopted to simulate the fracture process in the
splitting test of circular ring, with particular attention being paid to the internal diameter of the circular ring and the loading.
A typical homogenous rock (Carrara marble) is selected for the splitting ring test. The purpose of this study is to investigate
the failure mode (including crack initiation, propagation, and coalescence) and strength characteristics of brittle hard rocks
under static and dynamic loading by the simulation scheme. A rational load value is suggested to calculate the tensile
strength of the circular rock ring specimens under a splitting ring test. In addition, the numerical results based on the FDEM
approach are compared with the previous available experimental results in order to verify their accuracy and applicability.

2. Fundamental principles of FDEM method

The combined finite/discrete element method is a numerical approach that combines continuum mechanical principles
with discontinuum algorithms to simulate multiple interacting deformable solids [20]. Cai [17] has emphasized that there is
a distinction between the FDEM and the FEM/DEM hybrid methods. No new fractures are generated either in the DEM or in
the FEM domains in the hybrid method. On the contrary, new fractures and contact boundaries can be generated in FDEM
method model.
In this study, the FDEM approach is used to simulate the fracture process and to investigate the failure characteristics of
brittle rocks, in comparison with laboratory testing. ELFEN, a software developed by Rockfield Software Ltd. [21], contains a
variety of constitutive models and is capable of both explicit and implicit analyses of 2D and 3D types. The unique feature of
ELFEN is that it seamlessly simulates the transition of a rock mass from a continuum to a discontinuous state [17]. An obvi-
ous distinction from other numerical codes is its ability to allow the new cracks to cut across the original mesh (intra-
element fracturing) rather than merely along or around the element boundary (inter-element fracturing), as shown in
Fig. 1. Whether the fractures split the elements is determined by the value of the ‘‘smallest element” set under the discrete
contact parameters. If this value is the same as or larger than the original mesh size, the fractures will propagate only along
the element boundaries, as the minimum size has already been reached. However, the fractures may split elements, if con-
ditions dictate, if the ‘‘smallest element” is smaller than the original mesh.
ELFEN has a number of material modeling options that may be combined to create material models for different classes of
problems. In the present study, the Carrara marble is governed by the Rankine rotating crack constitutive model, which is
suitable for modeling the tensile failure of brittle materials such as rock, ceramic, and glass [21]. The most important param-
eters in the Rankine rotating fracture model are fracture energy and tensile strength. The failure mechanism of brittle rocks is
dominated by the formation of tensile cracks (Mode I) when the tensile stress exceeds the tensile strength of the material
[21]. For Mode I dominated problems, the initial failure surface for both models is defined by a tension failure surface, as
follows (Fig. 2):

Failure direction Intra-element fracturing Inter-element fracturing

Fig. 1. ELFEN crack insertion procedure (modified after [17]).

Please cite this article in press as: Li X et al. Numerical simulation of rock failure under static and dynamic loading by splitting test of cir-
cular ring. Engng Fract Mech (2017), http://dx.doi.org/10.1016/j.engfracmech.2017.08.022
4 X. Li et al. / Engineering Fracture Mechanics xxx (2017) xxx–xxx

Initial Yield Surface 2

ft

a b
Fig. 2. Yield Surface and Softening Curve for Rankie rotating fracture model (modified after [21]).

g¼tþDt ri tþDt f t ¼ 0 ð1Þ

where ri are the principal stress invariants and ft is the tensile strength of the material. In post initial yield, the rotating crack
formulation represents the anisotropic damage evolution by degrading the elastic modulus in the direction of the major
principal stress invariant:

rnn ¼ Ed enn ¼ ð1  xÞE ð2Þ

where x is the damage parameter and nn is the local coordinate system associated with the principal stresses, E is the
Young’s modulus, and Ed is the elastic damaged secant modulus. The damage parameter for the linear strain softening asso-
ciated with micro-fracturing is given by:
wðeÞ  1
x¼ ð3Þ
wðeÞ
where w(e) is a function of the total strain defined by:
rt
For e6 wðeÞ ¼ 1x ¼ 0
E

rt rt rt E2 e
For <e6 þ wðeÞ ¼ 0<x<1 ð4Þ
E E Et Et rt þ Ert  Et Ee

rt rt
For e> þ wðeÞ ¼ 1 x ¼ 1
E Et
where Et is the tangential softening modulus. The damage parameter is dependent on the fracture energy Gf, which is defined
as:
Z Z
G¼ rdu ¼ reðsÞds ð5Þ

Actually, the specific fracture energy (usually referred to as simply ‘‘fracture energy”) Gf is defined as the amount of
energy needed to create a continuous crack on a unit area. The fracture energy, Gf A, can be obtained through a simple uni-
axial tensile test [16,22]. The total fracture energy is the area under the load displacement curve. Note that A is the cross-
sectional area of the test specimen. Integrating over a localization band width lc for a constant slope softening model gives
[21]:
2
f t lc
ET ¼  ð6Þ
2Gf
where in the finite element context, the local control length lc = f(Be), where Be is the area of the element. Note that the rela-
tionship between fracture toughness in mode I (K Ic ) and fracture energy can be defined as [21]:

K 2Ic
Gf ¼ ð7Þ
E

Please cite this article in press as: Li X et al. Numerical simulation of rock failure under static and dynamic loading by splitting test of cir-
cular ring. Engng Fract Mech (2017), http://dx.doi.org/10.1016/j.engfracmech.2017.08.022
 X. Li et al. / Engineering Fracture Mechanics xxx (2017) xxx–xxx 5

3. Numerical modeling of splitting ring test under static and dynamic loading

3.1. Establishment of numerical models

The basic geometry of the circular ring specimen and loading platens used in the numerical model are shown in Fig. 3a.
For 2D simulations, the plane strain analysis in ELFEN assumes a unit thickness. The circular ring specimens are prepared
with an external diameter (D) of 50 mm and an internal diameter (d) varying from 5 mm to 30 mm, with intervals of
5 mm. Here, for convenience, we use a parameter k to illustrate the ratio of internal to external diameter of the circular ring,
which means that k varies from 0.1 to 0.6 with intervals of 0.1. The Brazilian disk test is also considered in the simulation.
Carrara marble is selected as the testing material to validate the previous experimental work based on Li et al. [10]. The
material properties of the Carrara marble [22] and loading platen, and their discrete contact parameters, are listed in Table 1.
The total number of unstructured triangle elements of the established models range from 19,752 to 12,620 for different k, as
shown in Fig. 3b.
An important feature of ELFEN is that an explicit fracture is introduced when all the fracture energy is consumed. Models
with higher fracture energy require more deformation before visible cracks appear. The fracture energy values used to study
hard and brittle materials range from 0.01 to 0.3 N/mm [23]. Based on the previous evaluation and research in terms of frac-
ture energy, the value of 0.03 N/mm is appropriate for describing the hard and brittle properties of Carrara marble. Penalty
parameters are used to evaluate the normal and tangential contact forces. The value of the normal penalty is usually in the
range of 0.5E to 2.0E, where E is the Young’s Modulus, and the tangential penalty is about 0.1 of the normal penalty [21].
Two loading types are considered for the splitting ring test: static and dynamic loading conditions. For the static loading,
displacement control loading is imposed on the top loading platen with a constant velocity. A ramp load style is applied, with
a total displacement of 0.3 mm (corresponding to a total time of 0.5 s) at the top of the specimen when the load factor
reaches 1, as shown in Fig. 4a. The strain rate is about 0.012 s1, which can be regarded as a quasi-static loading condition.
For the rock dynamic test, the strain rate or the loading rate are the main loading parameters that define the scope of
dynamic loading conditions. The loading rate is defined as rh, obtained from the time vs. tensile stress [24]. In this study,
a specified loading path representing the incident compressive stress is imposed at the top loading platen. It takes 10 ls
to reach the amplitude of the incident compressive stress (20 MPa); this level is kept for the following 40 ls and finally
decreases to 0 MPa after 60 ls. The dynamic loading path is shown in Fig. 4b. It can be seen from Fig. 4b that the loading
rate can be calculated by the slope of the linear rising stage, with a value of 2000 GPa/s. Compared with that in other studies
[24–27], the specified loading path belongs to the scope of dynamic loading. It should be mentioned that the circular ring
under this dynamic loading conditions is different from that in the splitting Hopkinson pressure bar (SHPB) test. It is known
that the two end stresses in the SHPB test should be maintained in equilibrium owing to the overlap of incident stress and
reflect stress signals [25,27]. However, such a loading path may still be useful for the dynamic analysis of the circular ring in
the present study. The bottom loading platen is fixed in the horizontal and vertical direction for both static and dynamic
loading conditions. The upper platen is a non-fixed boundary with applied loading velocities.

3.2. Under static loading

A series of splitting ring tests with different internal diameters are modeled under the static loading condition. For com-
parison with the circular ring and also for validation of ELFEN, a Brazilian disk specimen is also simulated in the present
study. The load-displacement curve of the Brazilian disk test is shown in Fig. 5, with the evolution of crack initiation, prop-
agation, and coalescence at different stages. It can be seen from Fig. 5 that the peak load is about 555 N, with a maximum
vertical displacement of 0.068 mm in the 2D Brazilian disk test. The suggested formula for the tensile strength of the Brazil-
ian disk is:
2P P
rt ¼ ¼ 0:636 ð8Þ
pDt Dt
where P is the peak load (N), D is the diameter of the test specimen (mm), and t is the thickness of the test specimen, which is
deemed as 1 mm in the 2D simulation. Substituting the peak load into Eq. (8), the tensile strength of the Brazilian disk is
calculated to be 7.01 MPa, which is almost equal to the testing result of 6.9 MPa [22]. In addition, visible cracks in the disk
may be observed only in the yield stage. In other words, cracks cannot be observed in the compaction and elastic stages. As
the load increases, the macroscopic cracks propagate from the center of the disk to the upper and lower platen and coalesce
together when the applied load is about 510 N. After the peak load, there is a sudden reduction of the load with no obvious
post peak curve. The comparison of the simulation and experimental results indicates that the ELFEN software is able to
reflect numerically the fracture process and mechanical behavior of brittle rocks.
Fig. 6 presents the crack evolution processes of the splitting ring test with different internal diameters (ranging from 5 to
30 mm) under static loading. The points from a to d indicate different stages of the crack evolution processes, representing
the transition from intact to fractured rock. Corresponding to these points, the static load values and evolution stages are
listed in Table 2. It can be seen that point a shows no cracks in each group, which belongs to elastic stage, whereas point
b represents the crack initiation stage, which is either in the yield stage that is quite close to the peak load point, or in
the post-peak stage. The reason for this case is that macroscopic cracks in ELFEN can only be obtained when the whole load

Please cite this article in press as: Li X et al. Numerical simulation of rock failure under static and dynamic loading by splitting test of cir-
cular ring. Engng Fract Mech (2017), http://dx.doi.org/10.1016/j.engfracmech.2017.08.022
6 X. Li et al. / Engineering Fracture Mechanics xxx (2017) xxx–xxx

Load

Top loading platen

d
x

y
D=50mm

Circular ring

Bottom loading
b=5mm
platen

a=20mm
Fixed in x and y direction

(a)

d=0 d=5 d=10

d=15 d=20 d=25

d=30

(b)
Fig. 3. Model geometry and meshes of splitting test of circular rings with different internal diameters.

Please cite this article in press as: Li X et al. Numerical simulation of rock failure under static and dynamic loading by splitting test of cir-
cular ring. Engng Fract Mech (2017), http://dx.doi.org/10.1016/j.engfracmech.2017.08.022
 X. Li et al. / Engineering Fracture Mechanics xxx (2017) xxx–xxx 7

Table 1
Material properties adopted in splitting test of circular ring specimens.

Name Carrara marble Loading platen

Mechanical parameters
Young’s modulus (E, GPa) 49 211
Possion’s ratio (l) 0.19 0.29
Shear modulus (G, GPa) 20.6 82
Density (q, N s2/mm4) 2.7  109 7.84  109
Tensile strength (rt, MPa) 6.9 —
Fracture energy (Gf, N/mm) 0.03 —
Normal penalty (Pn, N/mm2) 50,000 211,000
Tangential penalty (Pt, N/mm2) 25,000 100,000
Discrete contact parameters
Friction of newly generated cracks (c) 0.5 —
Mesh element size (mm) 0.5 0.5
Smallest element size (mm) 0.25 0.25
Contact type Node-Edge Node-Edge

Dynamic load functions

1 20

s Face loading (MPa)

Load factor

on
n cti σθ=2000GPa/s
d fu
oa
cl
ati
St

0 0.5 t (s) 0 10 50 60 t (μs)

a b

Fig. 4. Loading paths under static and dynamic conditions.

Yield stage
Load in Y direction (N)

Elastic stage

Post-peak

Compaction
stage

Vertical displacement (mm)

Fig. 5. Load-displacement curve of Brazilian disk test under static loading. Note that the unit of Direct Stress Y-Y in each legend is MPa.

capacity across the localized failure band decreases to zero, which would lag slightly. Similar phenomena can be also found
in Cai [17] and Hamdi et al. [17,28]. Points c and d represent the crack propagation and coalescence in each group, which
belong to post-peak stages. It can be seen from Fig. 6 that the failure modes of circular rings experience a typical transfor-

Please cite this article in press as: Li X et al. Numerical simulation of rock failure under static and dynamic loading by splitting test of cir-
cular ring. Engng Fract Mech (2017), http://dx.doi.org/10.1016/j.engfracmech.2017.08.022
8 X. Li et al. / Engineering Fracture Mechanics xxx (2017) xxx–xxx

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

λ=0.1

λ=0.2

λ=0.3

λ=0.4

λ=0.5

λ=0.6

Fig. 6. Crack evolution processes of circular rings with different internal diameters under static loading.

Please cite this article in press as: Li X et al. Numerical simulation of rock failure under static and dynamic loading by splitting test of cir-
cular ring. Engng Fract Mech (2017), http://dx.doi.org/10.1016/j.engfracmech.2017.08.022
 X. Li et al. / Engineering Fracture Mechanics xxx (2017) xxx–xxx 9

Table 2
Static load values and evolution stages corresponding to the points in Fig. 6.

Type Peak Load a b c d

(N)
Load Free of Load Crack Load Crack propagation or secondary crack Load Crack
(N) crack (N) initiation (N) initiation (N) coalescence
k = 0.1 351.1 114.5 Elastic 347.5 Yield 292.3 Post-peak 52.75 Post-peak
k = 0.2 212.2 99.4 Elastic 209.1 Yield 178.4 Yield 34.3 Post-peak
k = 0.3 133.3 66.2 Elastic 132.9 Yield 120.4 Yield 37.4 Post-peak
k = 0.4 106.4 60 Elastic 105.3 Yield 96.1 Yield 93.4 Post-peak
k = 0.5 60.5 35.4 Elastic 55.0 Yield 43.4 Post-peak 17.7 Post-peak
k = 0.6 32.4 20.1 Elastic 30.5 Yield 18.3 Post-peak 8.3 Post-peak

mation with the increase in internal diameter. When k  0.3, the final failure mode shows a diametrical splitting, in which
only a vertical diametrical crack initiates and propagates toward the upper and lower platen. When k = 0.4, the vertical dia-
metrical splitting crack is generated, and in addition, a horizontal diametrical crack occurs and propagates in one of the half
rings. When k > 0.4, the horizontal diametrical cracks occur in both half rings almost at the same time. This failure mode is
called four-fan-shaped failure in the present study. Note that the horizontal diametrical crack always lags behind the vertical
diametrical splitting crack.
The peak load of the circular ring shows some regularity with the change of internal diameter. It can be seen that the peak
load decreases exponentially with the internal diameter of the ring specimens. The fitting curve of peak load and k is shown
in Fig. 7 (including the Brazilian disk specimen). With the increase in k, the reduction in peak load becomes flatter, indicating
that the influence of the internal diameter to the peak load in a circular ring gradually decreases.
The load-displacement curves for circular rings with different internal diameters under static loading are shown in Fig. 8.
For circular ring specimens, the post-peak curves also show differences with the different internal diameters. When k  0.3
(including the Brazilian disk test), no residual load remains in the post-peak stage, where the peak load decreases abruptly.
When k
0.4, a certain residual load exists in the post-peak stage, and the load does not rapidly decrease to zero. The vari-
ation of post-peak curves corresponds to the transformation of the final failure modes of the ring specimens with different
internal diameters, which can be interpreted by the released kinetic energy of the circular rings. Fig. 9a shows the kinetic
energy variation trend against the loading time of ring specimens with different internal diameters. With the increase in
k, the kinetic energy will decrease dramatically, which means that it is easier to fracture the circular ring owing to the
increase in the internal diameter of the hole. The kinetic energy maintains a relatively low value during the pre-peak stage
but it surges when the peak load is reached. In addition, the evolution of kinetic energy when k  0.3 experiences only one
augment stage, indicating that nearly all the energy will be released once the peak load is achieved. Therefore, the peak load
will experience a sudden reduction, and no residual load is shown in the post-peak stage. For those circular rings where
k
0.4 (shown in Fig. 9b), not only the first augment of kinetic energy will appear in the formation of the vertical diametrical
splitting crack, but also the second augment of kinetic energy will occur with the initiation, propagation, and coalescence of
the horizontal diametrical crack. It reflects a certain bearing capacity (residual strength) of the circular ring to form the hor-
izontal diametrical crack in the post-peak stage. The evolution regularity of the kinetic energy might provide insights into the
post-peak behavior of rocks.

600

500
y = 554.36e-4.546x
400 R2 = 0.9914
Peak load y (N)

300

200

100

0
0 0.1 0.2 0.3 0.4 0.5 0.6

Fig. 7. Relationship between the peak load and k for static splitting test of circular rings.

Please cite this article in press as: Li X et al. Numerical simulation of rock failure under static and dynamic loading by splitting test of cir-
cular ring. Engng Fract Mech (2017), http://dx.doi.org/10.1016/j.engfracmech.2017.08.022
10 X. Li et al. / Engineering Fracture Mechanics xxx (2017) xxx–xxx

— ¬=0.1 — ¬=0.2
— ¬=0.3 — ¬=0.4
— ¬=0.5 — ¬=0.6

Load in Y direction (N)

Displacement (mm)

Fig. 8. Load-displacement curves of circular rings with different internal diameters under static loading.

3.3. Under dynamic loading

In order to simulate the mechanical behavior of a circular ring under dynamic loading, a simplified compression trape-
zoidal stress wave is applied at the top loading platen, as shown in Fig. 4b. The crack evolution and failure processes in
the splitting ring tests with different internal diameters (ranging from 0 to 30 mm) under dynamic loading are shown in
Fig. 10. Similarly, the points from a to e indicate the crack evolution processes under the dynamic loading condition, which
include the stages of free of crack, vertical diametrical splitting crack initiation, propagation, horizontal diametrical crack
initiation, and final coalescence. Note that no horizontal diametrical crack appears in the Brazilian disk test. It can be seen
that the initial crack is always located in the surface of the internal hole along the vertical loading diametrical direction, then
propagates and coalesces to the upper and lower platen. Following this, the process is accompanied by the initiation and
propagation of the horizontal diametrical cracks. The final failure modes of circular rings under dynamic loading are also
obviously influenced by the ratio of internal to external diameter. The numerical results show that even though the circular
rings are all eventually split into four parts, the horizontal diametrical crack tends to deviate toward the upper loading platen
when k
0.5. It shows an asymmetrical four-fan-shaped ring. In addition, the degree of deviation of the crack becomes
increasingly apparent with the increasing internal diameter. Another significant feature is that the final failure fractures
become simpler and less numerous when k
0.5 compared with other specimens when k < 0.5. The asymmetrical phe-
nomenon and simpler failure cracks for rings specimens with larger internal diameter ring might be the result of smaller
bearing capacity of the specimens. The impact load can lead to the entire failure, even without reaching a stress balance
at the two ends of the specimens. This seems to be a structural response of the circular ring instead of a material property
under impact load.
Fig. 11 shows the relationship between the peak load and k under dynamic loading for the splitting ring tests. The overall
variation trend is similar to that under the static splitting test, showing a regular descent of peak load with increasing k. The
ratio of peak load of the specimens between k = 0 and 0.6 in the static splitting test is about 10.8, while it is about 8.2 in
dynamic splitting ring test (seen in Table 3). In addition, it can be seen in Fig. 7 and Fig. 11 that the peak load under dynamic
loading is higher than that under static loading at the same corresponding k value, which is in good agreement with the
available results [10–11,25]. However, the ratio of dynamic peak load to static peak load ranging from k = 0 to 0.6 changes
from 7.23 to 9.52. The difference in the specific values suggests that the structural effect by the internal hole has a significant
influence on the strength of ring specimens under dynamic loading.
The load-displacement curves for the circular ring with different internal diameters under dynamic loading are shown in
Fig. 12. It is worth noting that some vibrations occur far before the peak point, and the vibrations should be caused by the
dispersion effect during the propagation of the stress wave, rather than by the rock failure (crack initiation and propagation).
The dispersion effect of the induced stress wave attenuation finally leads to a series of oscillation phenomena under dynamic
loading conditions, though the ‘‘contact damping” function is currently utilized. Actually, the high frequency oscillations, i.e.
‘‘noise,” which are unavoidably introduced with the high-speed interaction could be minimized by specifying the contact
damping acting as a modifier to the normal contact force, based on the relative velocity of the interacting bodies. In the pre-
sent study, the default algorithms, namely velocity/momentum, for contact damping are recommended for the general
mechanical interactions. Because the vibrations before the peak point may not affect the subsequent analysis regarding
the strength and failure modes of a rock ring sample, a general value of about 0.1 for geomechanical projects is typical
[21], and it is adopted in the present study. Different from that under static loading, the post-peak curves of circular rings
with different internal diameters under dynamic loading have all a certain value of residual load, which is accompanied by

Please cite this article in press as: Li X et al. Numerical simulation of rock failure under static and dynamic loading by splitting test of cir-
cular ring. Engng Fract Mech (2017), http://dx.doi.org/10.1016/j.engfracmech.2017.08.022
 X. Li et al. / Engineering Fracture Mechanics xxx (2017) xxx–xxx 11

4.0

3.6 — λ=0 — λ=0.1

— λ=0.2 — λ=0.3
3.2 — λ=0.4 — λ=0.5
— λ=0.6
2.8

Kinetic energy (J)

2.4

2.0

1.6

1.2

0.8

0.4

0.0
0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.20 0.22 0.24
t (s)
(a)
0.28

0.24 First augment Second augment — λ=0.4 — λ=0.5

— λ=0.6

0.20
Kinetic energy (J)

0.16

0.12

0.08

0.04

0.00
0.00 0.05 0.10 0.15 0.20 0.25
t (s)
(b)

Fig. 9. Kinetic energy-time curves of ring specimens with different internal diameters under static loading. Note that Fig. 9b is the partial enlarged drawing
of Fig. 9a.

consistent fluctuations, showing the persistent effect caused by the dynamic incident compressive stress waves. Fig. 13
shows the kinetic energy variation trend against the loading time of ring specimens with different internal diameters. No
obvious fluctuations and surge of kinetic energy values are observed, but instead, there is a consistent increase in the entire
failure process. Therefore, the post-peak behavior (especially the post-peak strength) cannot be explained by the kinetic
energy in the condition of dynamic loading, because the steady growth of kinetic energy is unable to illustrate the relation-
ship between horizontal crack propagation and post-peak behavior. Once an impact load has been applied to the circular
ring, it can produce relatively high kinetic energy, even before the primary cracks begin to initiate and propagate (including
the compaction and elastic stages). The increasing kinetic energy induced by impact load and crack propagation will be dif-
ficult to distinguish during the failure process of the circular ring specimens. On the contrary, the fluctuation and augment of
kinetic energy under static loading can be a good index to illustrate the variation trend of post-peak strength because it is
merely induced by crack initiation, propagation, and coalescence.

Please cite this article in press as: Li X et al. Numerical simulation of rock failure under static and dynamic loading by splitting test of cir-
cular ring. Engng Fract Mech (2017), http://dx.doi.org/10.1016/j.engfracmech.2017.08.022
12 X. Li et al. / Engineering Fracture Mechanics xxx (2017) xxx–xxx

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

λ=0

λ=0.1

λ=0.2

λ=0.3

λ=0.4

λ=0.5

λ=0.6

Fig. 10. Crack evolution processes of circular rings with different internal diameters under dynamic loading.

Please cite this article in press as: Li X et al. Numerical simulation of rock failure under static and dynamic loading by splitting test of cir-
cular ring. Engng Fract Mech (2017), http://dx.doi.org/10.1016/j.engfracmech.2017.08.022
 X. Li et al. / Engineering Fracture Mechanics xxx (2017) xxx–xxx 13

3500

3000
y = 3246.6e-3.601x
2500 R2 = 0.9594

Peak load y (N)

2000

1500

1000

500

0
0 0.1 0.2 0.3 0.4 0.5 0.6

Fig. 11. Relationship between k and peak load of dynamic splitting test of circular ring.

Table 3
Dynamic load values and evolution stages corresponding to the points presented in Fig. 11.

Type Peak a b c d e
Load
Load Free of Load Crack Load Crack propagation Load Horizontal Load Crack
(N)
(N) crack (N) initiation (N) (N) diametrical cracks (N) coalescence
initiation
k=0 2540 470.2 Elastic 2434.5 Yield 1760.4 Post-peak 1833.3 Post-peak (No 811.6 Post-peak
horizontal
diametrical crack)
k = 0.1 2335.2 488.5 Elastic 2280 Yield 1950.6 Post-peak 1605.1 Post-peak 807.7 Post-peak
k = 0.2 1936.6 450.1 Elastic 1844.8 Yield 450.9 Post-peak 607.5 Post-peak 537.4 Post-peak
k = 0.3 1242.3 484.1 Elastic 1234.1 Post- 1003.5 Post-peak 1217.5 Post-peak 881.4 Post-peak
peak
k = 0.4 849.4 209.9 Compaction 832.8 Yield 835.2 Post-peak 522.8 Post-peak 464.9 Post-peak
k = 0.5 514.8 120.4 Elastic 510.2 Post- 469.3 Post-peak 483.0 Post-peak 201.4 Post-peak
peak
k = 0.6 308.4 51.3 Compaction 305.1 Yield 294.5 Post-peak 130.8 Post-peak 84.8 Post-peak
(Horizontal
diametrical crack
initiation)

2750

2500 — =0 — =0.1
2250 — =0.2 — =0.3
— =0.4 — =0.5
2000 — =0.6
Load in Y direction (N)

1750

1500
1250

1000

750

500

250

0
0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10
Displacement (mm)

Fig. 12. Load-displacement curves of circular rings with different internal diameters under dynamic loading.

Please cite this article in press as: Li X et al. Numerical simulation of rock failure under static and dynamic loading by splitting test of cir-
cular ring. Engng Fract Mech (2017), http://dx.doi.org/10.1016/j.engfracmech.2017.08.022
14 X. Li et al. / Engineering Fracture Mechanics xxx (2017) xxx–xxx

70
— λ=0 — λ=0.1
60 — λ=0.2 — λ=0.3
— λ=0.4 — λ=0.5
— λ=0.6
50

Kinetic energy (J)

40

30

20

10

0
0 10 20 30 40 50 60
t (μs)

Fig. 13. Kinetic energy-time curves of ring specimens with different internal diameters under dynamic loading.

4. Discussion

4.1. Failure modes of circular ring specimens under static and dynamic loading

Based on the present study, the failure modes of circular ring specimens are influenced by the ratio of internal to external
diameters and the loading rate. Under static loading, the failure modes of the circular rings are transformed from diametrical
splitting style to four-fan shaped failure with the increase in internal diameter. Under dynamic loading, the failure modes
experience a shift from symmetrical four-fan-shaped failure to asymmetrical four-fan-shaped failure with the increase in
internal diameter.
Previous studies have shown that the failure modes are related to the internal diameter under static loading [7,10], and
the transition of failure modes of the circular ring specimens is similar to the present modeling results. Zhu et al. [7] used the
RFPA code to carry out numerical splitting tests of ring specimens. The results showed that when the r/R (ratio of internal
radius to external radius)
0.4, the circular specimen is finally split into four parts. Li et al. [10] experimentally studied the
tensile failure characteristics of rocks based on splitting ring tests under the condition of static and dynamic loading. In those
tests, marble was selected as a typical brittle rock to study the tensile strength and failure modes with the change in the ratio
k (r/R). The test results showed that under static loading the specimens split along the diametrical loading direction when
k  0.2, and formed a four-fan-shaped failure when k
0.3, as shown in Fig. 14. A slight difference is observed between these
two results. In the present study, horizontal diametrical cracks can be only observed when k
0.4. The difference might be
the result of a discrepancy in the static loading rate and material properties between these two tests. Nevertheless, the over-
all variation trends of the failure modes with the increase in internal diameter are extremely similar.
The dynamic characteristics of the Brazilian test have been studied by many researchers [25,26,29], but little attention
has been paid to the dynamic tensile failure characteristics of rocks based on the splitting ring test, especially by means
of numerical simulations. According to the experimental results by Cheng [30], two types of failure modes are observed
under impact loading by SHPB, including the symmetrical four-fan-shaped ring with small internal radius and the asymmet-
rical four-fan-shaped ring with large internal radius. The ring specimens usually break into four pieces. Fig. 15 shows the
comparisons of the final failure modes of circular ring specimens between the experimental results by Cheng [30] and
the present numerical results under dynamic loading. The symbol r in Fig. 15a indicates the ratio of internal diameter to
external diameter of ring specimen in laboratory testing, which has the same meaning with k in the current numerical sim-
ulation. Moreover, the symbols B3 to G2 in the left bottom corner are the specimen numbers, which mean r varies from 0.1 to
0.6 with an interval of 0.1, respectively. Since three specimens are tested in each group, the number B3, for instance, indi-
cates the third ring specimen when the ratio of internal to external diameter is 0.1. It can be seen that the failure modes of
specimens in the present numerical studies are in good agreement with the experimental results. The horizontal diametrical
cracks begin to deviate from the horizontal diametrical direction when k
0.5, which shows an asymmetrical four-fan-
shaped failure, and the degree of deviation gradually becomes more serious with the increase in internal diameter. Note that
the deviation of the horizontal diametrical cracks in previous research [30] also occurs when k
0.5, which is precisely in
good agreement with the present numerical results. Even though two different incident compressive stress waves are
imposed on the platen between the two tests, wherein the sine waves are performed in the SHPB test system and a simplified
trapezoidal wave is used in the present numerical study, the failure modes are almost the same. It indicates that the FDEM
method can reflect the dynamic characteristics and failure process of ring specimens under dynamic loading.

Please cite this article in press as: Li X et al. Numerical simulation of rock failure under static and dynamic loading by splitting test of cir-
cular ring. Engng Fract Mech (2017), http://dx.doi.org/10.1016/j.engfracmech.2017.08.022
 X. Li et al. / Engineering Fracture Mechanics xxx (2017) xxx–xxx 15

(a)

λ=0.1 λ=0.2 λ=0.3 λ=0.4 λ=0.5

(b)
Fig. 14. Comparisons of final failure modes of circular ring specimens between the results by (a) Li et al. [10] and (b) present numerical studies under static
loads.

4.2. Tensile strength of circular ring specimens under static and dynamic loading

The splitting test of a circular ring, a disk with a small central hole, for determining the indirect tensile strength of rock
was first proposed by Hobbs [5]. The Hobbs equation for calculating the indirect tensile strength for a circular ring is

 
P r2
rtr ¼ 6 þ 38 2 ð9Þ
pRt R

where P is the peak load (N), r is the internal radius of the circular ring (mm), and R is the external radius of the circular ring
(mm). The above equation is appropriate to use for both static and dynamic loading.
The static tensile strengths of circular ring specimens with different internal diameters, in the present study and in the
experimental results based on Li et al. [10], are listed in Table 4. According to their results, the calculated tensile strengths of
the circular ring specimens are higher than the Brazilian tensile strength. The ratio of rEtrs =rt varies from 5.08 to 2.03 when k
increases from 0.1 to 0.5. Similarly, the ratio of rNtrs =rt using the FDEM method varies from 4.16 to 1.17 when k increases
from 0.1 to 0.6 (as seen in Fig. 16a). Zhu et al. [7] found that the tensile cracks would initiate and propagate immediately
after the maximum tensile strength surrounding the internal hole reached the local tensile strength of the rock. However,
the bearing capacity of the rock does not reach the peak strength. The applied load to the circular ring will continue to rise
with the propagation and coalescence of cracks. Therefore, the tensile strength calculated by Eq. (9) using the peak load may
lead to a higher value than that using the fracture initiation load.
Another notable phenomenon in the present numerical results is that the ratio of rNtrs =rt is equal to 1.69 and 1.16 when
k = 0.5 and 0.6, respectively. It indicates that the difference of static tensile strength between the Brazilian test and ring test
becomes smaller when k
0.5. Therefore, for those circular ring specimens whose internal diameters are small, especially
when k is smaller than 0.5, it is recommended to select a crack initiation load to calculate the indirect tensile strength of
rock under static loading. However, when the internal diameter is large, for example when k
0.5, the peak load may be
used to calculate the static tensile strength.
Similarly, the dynamic tensile strengths of circular ring specimens with different internal diameters using the FDEM
method and the experimental results based on Cheng [30] are listed in Table 5. It can be found that the variation trend of
the dynamic tensile strength is similar to that of the static tensile strength. The ratios of rEtrd =rt and rNtrd =rt vary from
4.31 to 1.35 and 5.87 to 2.39, respectively, when k increases from 0.1 to 0.6. The reason for higher dynamic tensile strength
values in the numerical simulations is that the peak loads are used for circular ring specimens in the same way as under
static loading. The dynamic tensile strength normalized by rt (tensile strength of the Brazilian disk) is shown in Fig. 16b
for illustration of the applicability of the P value (peak load or crack initiation load). It can be seen from Fig. 16b that for those
circular ring specimens with small internal diameters, especially when k  0.5, it is recommended to select the crack-

Please cite this article in press as: Li X et al. Numerical simulation of rock failure under static and dynamic loading by splitting test of cir-
cular ring. Engng Fract Mech (2017), http://dx.doi.org/10.1016/j.engfracmech.2017.08.022
16 X. Li et al. / Engineering Fracture Mechanics xxx (2017) xxx–xxx

(a)

λ=0.1 λ=0.2 λ=0.3

λ=0.4 λ=0.5 λ=0.6

(b)
Fig. 15. Comparisons of final failure modes of circular ring specimens between the results by (a) Cheng [31] and (b) present numerical studies under
dynamic loads.

Table 4
Tensile strengths of circular ring specimens in numerical simulation and experimental tests based on Li et al. [10] under static loading.

k 0 0.1 0.2 0.3 0.4 0.5 0.6

rEtrs 3.60 18.30 16.51 13.17 10.91 7.30 —
rNtrs 7.01 28.75 20.30 15.96 16.2 11.84 8.12

Note: rEtrs is the static tensile strength of circular ring based on Li et al. [10], and rN
trs is the static tensile strength of circular ring based on combined FDEM
approach.

Please cite this article in press as: Li X et al. Numerical simulation of rock failure under static and dynamic loading by splitting test of cir-
cular ring. Engng Fract Mech (2017), http://dx.doi.org/10.1016/j.engfracmech.2017.08.022
 X. Li et al. / Engineering Fracture Mechanics xxx (2017) xxx–xxx 17

6
Experimental results

t
Static tensile strength normalized by
5
Numerical results
4

0
0.1 0.2 0.3 0.4 0.5 0.6

(a)

6 Experimental results
t
Dynamic tensile strength normalized by

5 Numerical results

0
0.1 0.2 0.3 0.4 0.5 0.6

(b)

Fig. 16. Static and dynamic tensile strength of circular ring normalized by rt for the laboratory tests and combined FEM/DEM approach. Note that no result
is acquired in Li.et al. [10] when k is 0.6 under static loading.

Table 5
Tensile strengths of circular ring specimens in numerical simulation and experimental tests based on Cheng [30] under dynamic loading.

k 0 0.1 0.2 0.3 0.4 0.5 0.6

rEtrd 24.0 103.4 84.0 70.1 58.0 43.6 32.3
rNtrd 32.3 189.7 185.4 149.0 130.7 101.6 77.3

Note: r is the dynamic tensile strength of circular ring based on Li et al. [10], and r
E
trd
N
trd is the dynamic tensile strength of circular ring based on combined
FDEM approach.

initiation load to calculate the indirect tensile strength of rock under dynamic loading. Peak load should be adopted to cal-
culate the dynamic tensile strength when the internal diameter is large, especially when k > 0.5.

5. Conclusions

The FDEM approach (ELFEN) is applied to model the fracture process and failure characteristics of Carrara marble spec-
imens by the splitting ring test, including both static and dynamic loading conditions. The numerical results show that the
ratio of internal to external diameters (k) and loading conditions are the main factors that affect the final failure modes and
tensile strengths of the circular rock ring specimens. Under static loading, the final failure modes are transformed from dia-
metrical splitting to four-fan-shaped failure with the increase in internal diameter. The peak load shows an exponential
decrease with increasing k. The post-peak behavior experiences a shift from non-residual load to a certain degree of bearing
capacity, which is related to the kinetic energy in the entire process from continuum to discontinuum state. Under dynamic
loading, the final failure modes of a rock ring are transformed from a symmetrical to an asymmetrical four-fan-shaped failure
with the increase in internal diameters. Similarly, the dynamic peak load shows a decreasing trend with the increase in k.

Please cite this article in press as: Li X et al. Numerical simulation of rock failure under static and dynamic loading by splitting test of cir-
cular ring. Engng Fract Mech (2017), http://dx.doi.org/10.1016/j.engfracmech.2017.08.022
18 X. Li et al. / Engineering Fracture Mechanics xxx (2017) xxx–xxx

Different from those under static loading, the post-peak curves have all a certain degree of residual load, indicating the dif-
ference between static and dynamic loading conditions. The calculated static and dynamic tensile strengths of the circular
ring specimens are all greater than those of the Brazilian disk specimen. It is suggested to select a rational load value to cal-
culate the indirect tensile strength of a circular rock ring. The results of the FDEM numerical simulations, including both sta-
tic and dynamic loading conditions, are in good agreement with previous experimental results. The FDEM approach is an
effective and convenient way to study the failure processes of rocks or rock-like materials.

Acknowledgments

The authors would like to acknowledge the financial supports from the State Key Research Development Program of
China (No. 2016YFC0600706), the National Natural Science Foundation of China (No. 51474250) and the National Key Basic
Research Program of China (No. 2015CB060200).

References

[1] Ramsey JM, Chester FM. Hybrid fracture and the transition from extension fracture to shear fracture. Nature 2004;428(6978):63–6.
[2] Ayatollahi MR, Aliha MRM. Cracked Brazilian disc specimen subjected to mode II deformation. Eng Fract Mech 2005;72(4):493–503.
[3] Ayatollahi MR, Aliha MRM. On the use of Brazilian disc specimen for calculating mixed mode I-II fracture toughness of rock materials. Eng Fract Mech
2008;75(16):4631–41.
[4] Li DY, Wong LNY. The Brazilian Disc test for rock mechanics applications: review and new insights. Rock Mech Rock Eng 2013;46(2):269–87.
[5] Hobbs D. An assessment of a technique for determining the tensile strength of rock. Brit J Appl Phys 1965;16(2):259.
[6] Chen CS, Hsu S. Measurement of indirect tensile strength of anisotropic rocks by the ring test. Rock Mech Rock Eng 2001;34(4):293–321.
[7] Zhu WC, Feng D, Chau KT, Tang CA. Numerical test using ring specimen as medium to determine indirect tensile strength of rock. Dongbei Daxue
Xuebao/J Northeastern Univ 2004;25(9):899–902.
[8] You MQ, Chen XL, Su CD. Brazilian splitting strengths of discs and rings of rock in dry and saturated conditions. Chin J Rock Mech Eng 2011;30
(3):464–72.
[9] Kourkoulis S, Markides CF. Stresses and displacements in a circular ring under parabolic diametral compression. Int J Rock Mech Min Sci
2014;71:272–92.
[10] Li DY, Wang T, Cheng TJ, Sun XL. Static and dynamic tensile failure characteristics of rock based on splitting test of circular ring. Trans Nonferrous
Metals Soc China 2016;26(7):1912–8.
[11] Li XB, Wu QH, Tao M, Weng L, Dong LJ, Zou Y. Dynamic Brazilian splitting test of ring-shaped specimens with different hole diameters. Rock Mech Rock
Eng 2016;49(10):4143–51.
[12] Mehranpour MH, Kulatilake PHSW. Comparison of six major intact rock failure criteria using a particle flow approach under true-triaxial stress
condition. Geomech Geophys Geo-Energy Geo-Resour 2016;2(4):203–29.
[13] Lee B, Rathnaweera TD. Stress threshold identification of progressive fracturing in Bukit Timah granite under uniaxial and triaxial stress conditions.
Geomech Geophys Geo-Energy Geo-Resour 2016;2(4):301–30.
[14] Wong LNY, Wu ZJ. Application of the numerical manifold method to model progressive failure in rock slopes. Eng Fract Mech 2014;119:1–20.
[15] Li X, Konietzky H, Li XB. Numerical study on time dependent and time independent fracturing processes for brittle rocks. Eng Fract Mech
2016;163:89–107.
[16] Rougier E, Knight EE, Sussman AJ, et al. The combined finite-discrete element method applied to the study of rock fracturing behavior in 3D[C]//45th
US Rock Mechanics/Geomechanics Symposium. American Rock Mechanics Association; 2011.
[17] Cai M. Fracture initiation and propagation in a Brazilian disc with a plane interface: a numerical study. Rock Mech Rock Eng 2013;46(2):289–302.
[18] Hamdi P, Stead D, Elmo D. Damage characterization during laboratory strength testing: a 3D-finite-discrete element approach. Comput Geotech
2014;60:33–46.
[19] Mahabadi OK, Cottrell BE, Grasselli G. An example of realistic modelling of rock dynamics problems: FEM/DEM simulation of dynamic Brazilian test on
Barre granite. Rock Mech Rock Eng 2010;43(6):707–16.
[20] Munjiza A. The combined finite-discrete element method. John Wiley & Sons; 2004.
[21] Rockfield. ELFEN Explicit/Implicit Manual, Version 4.7.1. Rockfield Software Limited, West Glamorgan: UK; 2013.
[22] Wong LNY, Einstein HH. Crack coalescence in molded gypsum and Carrara marble: Part 1. Macroscopic observations and interpretation. Rock Mech
Rock Eng 2009;42(3):475–511.
[23] Klerck PA. The finite element modelling of discrete fracture in quasi-brittle materials. University of Wales Swansea, Wales; 2000.
[24] Huang S, Xia K, Yan F, Feng X. An experimental study of the rate dependence of tensile strength softening of Longyou sandstone. Rock Mech Rock Eng
2010;43(6):677–83.
[25] Zhu WC, Tang CA. Numerical simulation of Brazilian disk rock failure under static and dynamic loading. Int J Rock Mech Min Sci 2006;43(2):236–52.
[26] Zhang QB, Zhao J. Determination of mechanical properties and full-field strain measurements of rock material under dynamic loads. Int J Rock Mech
Min Sci 2013;60:423–39.
[27] Li XB, Zhou T, Li DY. Dynamic strength and fracturing behavior of single-flawed prismatic marble specimens under impact loading with a split-
Hopkinson pressure bar. Rock Mech Rock Eng 2017;50(1):29–44. doi: http://dx.doi.org/10.1007/s00603-016-1093-y.
[28] Hamdi P, Stead D, Elmo D. Characterizing the influence of stress-induced microcracks on the laboratory strength and fracture development in brittle
rocks using a finite-discrete element method-micro discrete fracture network FDEM-lDFN approach. J Rock Mech Geotech Eng 2015;7(6):609–25.
[29] Mahabadi O, Grasselli G, Munjiza A. Y-GUI: a graphical user interface and pre-processor for the combined finite-discrete element code, Y2D,
incorporating material heterogeneity. Comput Geosci 2010;36(2):241–52.
[30] Cheng TJ. Experimental study on dynamic strength characteristics and crack propagation behavior of rock specimens with a single hole under impact
loading. Central South University, Changsha; 2015.
[31] Wen ZJ, Wang X, Chen LJ, Lin G, Zhang HL. Size effect on acoustic emission characteristics of coal-rock damage evolution. Adv Mater Sci Eng
2017;2017:1–8.

Please cite this article in press as: Li X et al. Numerical simulation of rock failure under static and dynamic loading by splitting test of cir-
cular ring. Engng Fract Mech (2017), http://dx.doi.org/10.1016/j.engfracmech.2017.08.022