Tunnelling and Underground Space Technology 66 (2017) 19–33

Contents lists available at ScienceDirect

Tunnelling and Underground Space Technology

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

Analysis of damage mechanisms and optimization of cut blasting design MARK

under high in-situ stresses
⁎
L.X. Xiea,b,c, W.B. Lua,b, , Q.B. Zhangc, Q.H. Jianga,b, M. Chena, J. Zhaoc
a
State Key Laboratory of Water Resources and Hydropower Engineering Science, Wuhan University, Wuhan 430072, China
b
School of Civil Engineering, Wuhan University, Wuhan 430072, China
c
Department of Civil Engineering, Monash University, Clayton, VIC 3800, Australia

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

Keywords: During excavation using the cut blasting method in deep rock masses, there are difficulties resulting from the in-
Cut blasting situ stress influences. This study uses numerical simulation methods to assess the causes of the difficulties
RHT model encountered in cut blasting. In order to overcome this difficulty, the Riedel–Hiermaier–Thoma (RHT) model in
Parameter determination the LS-DYNA software was employed. In the simulation, the parameter determination for the RHT model was
Optimized cut blasting design
first carried out based on existing experimental data. Additionally, the existing blasting experiment was used to
High in-situ stress
verify the determined parameters of RHT model. Second, the RHT model was adopted to investigate the damage
mechanisms of cut blasting under different hydrostatic pressures and different lateral pressure coefficients. The
simulation results indicate that the main causes of the complications arising in deep rock mass excavation are
resistance to in-situ stresses and anisotropy in the damage propagation direction. Third, in order to overcome
such difficulties, a cut blasting design optimization was conducted for a 2525 m depth of rock mass. According to
the numerical simulation of this optimization, a modified cut blasting design method applicable to deep rock
mass was proposed. This study can provide solutions to the cut blasting difficulties that are encountered during
the excavation of deep rock masses.

1. Introduction has been done on such mechanisms for deep rock masses under the cut
blasting method (Bäckblom and Martin, 1999; Chen et al., 2007;
Currently, there is an increasing demand for mineral resources, Cunningham and Goetzsche, 1990; Ramulu et al., 2009; Bruland and
hydropower resource exploitation, and the development of science and Zare, 2006). Ma and An (2008) investigated the influence of free face
technology. As a result, the excavation of rock masses has gradually in-situ stress and pre-existing joint planes on damage to rock masses
extended to greater depths. At present, the deepest underground cavern under blasting. Wang et al. (2007) studied tension and compression
is the China Jinping Underground Laboratory, where the average depth damage distributions under different charge structures using the TCK
is generally greater than 2000 m, and the maximum depth is approxi- model. Using a modified principal stress failure criterion, Zhu et al.
mately 2525 m. According to the back-calculated in-situ stresses, the (2008) applied the AUTODYN software to investigate the effects of
major principal stress is approximately 70 MPa. The other two principal boundary conditions, coupling mediums, borehole diameters, decou-
stresses are approximately 30 MPa in magnitude (Gong et al., 2012). plings, and joints on dynamic rock fractures. Considering the influence
When the drill and blast (D & B) method is adopted in rock excavation of in-situ stress on rock mass damage mechanisms, Yilmaz and Unlu
at great depths, cut blasting becomes difficult because of high in-situ (2013) investigated the effects of high anisotropic in-situ stresses on
stress influences (Xie et al., 2016). Cut blasting is important in blasting performance and blast-induced damage zones. Yang et al.
developing a free surface for subsequent blasting, and influences the (2015) researched the scope of damage to surrounding rock under
overall blasting procedure (Zhao et al., 2011). Therefore, it is important different blasting sequences. They accomplished this by applying
to investigate the damage evolution mechanisms of cut blasting. equivalent loads to an excavation surface. Zhao et al. (2011) used
Many researchers have investigated the damage evolution mechan- discontinuous deformation analysis (DDA) software to investigate
isms of rock under blasting loads (Deng et al., 2014; Li et al., 2013, dynamic rock response and rock fragmentation processes; however,
2014; Mitelman and Elmo, 2016; Zhu et al., 2010). However, little work they did not consider in-situ stress. When a deep rock mass is excavated

⁎
Corresponding author at: State Key Laboratory of Water Resources and Hydropower Engineering Science, Wuhan University, Wuhan 430072, China.
E-mail address: [email protected] (W.B. Lu).

http://dx.doi.org/10.1016/j.tust.2017.03.009
Received 5 November 2016; Received in revised form 8 March 2017; Accepted 21 March 2017
0886-7798/ © 2017 Published by Elsevier Ltd.
L.X. Xie et al. Tunnelling and Underground Space Technology 66 (2017) 19–33

Nomenclature P pressure of detonation products

Qv0,Qv released energy of 1 kg LFB-dynamite and released energy
List of symbols of 1 kg LFB-dynamite explosives
SL slope of linear relationship
A, n failure surface parameters of RHT model SANFO explosive weight strength relative to ANFO
A1,A2 ,A3 EOS constants of rock-like materials S weight strength relative to LFB-dynamite
As side width of rectangular opening v0,v initial specific volume of detonation products and specific
AJWL ,BJWL,R1,R2 JWL EOS constants volume of detonation products
Bb,B1,B2 cut hole burden, cut hole burden in the first quadrangle, STP standard atmospheric pressure and temperature
and auxiliary blasthole burden in the second quadrangle V0,V gas volume released by LFB-dynamite and explosives at
B,Q0 lode angle dependence factors STP
BΓ0,BΓ Gruneisen parameter coefficients VoD detonation velocity
c intact rock constant
d charge diameter Greek symbols
e0,e,e H initial internal energy per unit mass, internal energy per
unit mass, and internal energy per unit mass under α0,α initial porosity and porosity of porous materials
Hugoniot conditions α1,α2 angular deviation and collar deviation
E0,E initial specific internal energy of detonation products and εṗ ,εṗ c,εṗ t strain rate, reference tensile strain rate, and compressive
specific internal energy of detonation products strain rate
fc compressive strength ϕ diameter of the central hole
F blasthole deviation ρ0 ,ρ,ρe initial density of rock, rock density, and explosive density
Frc (εṗ ),Frt (εṗ ) compressive and tensile strain rate strength factor σ1,σ3 maximum and minimum principal stresses
H blasthole advance σn,τ normal effective stresses and shear stress
l linear charge density Γ(v ) Gruneisen parameter
lc,lc correction factor of compressive and tensile strength σf ,σf−C,σf∗ stress, corrected stress, and normalized stress on failure
N porosity index surface
PR,μ pressure of EOS in RHT model and volumetric strain ω JWL EOS constant

using a cut blasting method, in-situ stress causes difficulties. under blast loading, the rock mechanical parameters are first deter-
To overcome these difficulties, blasting design must be optimized mined based on existing experimental results (Banadaki and Mohanty,
based on damage evolution mechanisms. At present, the successful 2012). The model parameters and model rationality are verified by
excavation of rock masses requires appropriate blasting design of the comparing the model results to existing blasting test and simulation
drilling pattern, quantity and type of explosives, and initiation sequence results (Banadaki and Mohanty, 2012). To overcome the difficulties
(Zhao et al., 2011). It is important to determine the burden of cut holes, encountered during cut blasting in deep rock masses, the determined
which is key in the overall blasting procedure. The basic principles RHT model parameters are applied to a simulation of the damage
behind calculating patterns and charges for a four-section cut (known distribution around a blasthole under hydrostatic pressure and various
as the Swedish method) were first developed by Langefors and lateral pressure coefficients. Through the simulation, the causes of
Kihlström (1978). The method was later updated by Holmberg and difficulties encountered during cut blasting are analyzed, and the cut
then simplified by Olofsson (1990). These modified methods are blasting design is optimized to overcome these difficulties. This study
applied to the calculation of cut blasting design for shallow rock provides both a method for determining RHT model parameters in LS-
masses. However, an optimized method has not been reported for DYNA and a theoretical basis and reference for addressing excavation
determining the cut hole burden in a deep rock mass. difficulties related to cut blasting in deep rock masses.
In LS-DYNA, three main damage models are used to simulate the
damage evolution of rock mass under blasting load: the
2. Verification of the RHT model
Holomquist–Johnson–Cook (HJC) model (Holomquist et al., 1993),
the JH series model (Johnson and Holmquist, 1992, 1994), and the
The RHT model is a tensile-compressive damage model proposed by
Riedel–Hiermaier–Thoma (RHT) model (Riedel et al., 1999). In contrast
Riedel et al. based on a modified HJC model. In contrast to existing
to the HJC and JH series models, the RHT model considers strength
damage models, it considers the effects of confining pressure, strain
characteristics in the three-dimension stress space, along with deforma-
rate, strain hardening, and damage softening on the failure strength of a
tion and failure under high confining pressure. It can better reflect rock
rock material under blasting and dynamic load. To describe pore
mechanical performance under different confining pressures and high
compaction hardening effects, the pressure is governed by the
strain rates. When the D & B method is applied, the excavated deep rock
Mie–Gruneisen equation of state (EOS) together with a p-α model
masses are under dynamic-static coupling loading. At such a moment,
(Borrvall and Riedel, 2011). As a complement to the model in LS-DYNA,
strain rate, confining pressure, strain hardening, and damage softening
the RHT model in AUTODYN is modified and embedded into LS-DYNA.
have a significant influence on the mechanical performance of the rock
Based on the above-mentioned analysis, the RHT model is suitable
masses. These factors are comprehensively considered in the RHT
for describing the dynamic mechanical response of deep rock mass to
model (Borrvall and Riedel, 2011). To investigate the optimized
blast loading. Although, this AUTODYN model has been widely applied
method and determine the burden of cut holes applicable to deep rock
to simulate damage evolution processes in concrete, it is seldom used in
masses, the RHT model in LS-DYNA is chosen. At present, the RHT
rock-like material modelling. Additionally, the RHT model has not been
model has been widely used to simulate the damage evolution of
widely applied in LS-DYNA. To apply the RHT model in LS-DYNA to
concrete. However, because of a lack of rock mechanical parameters,
simulate the rock damage evolution process, mechanical parameters
few investigations have used LS-DYNA to study the damage evolution of
must be determined based on existing rock mechanical tests. The model
rock.
parameters and model applicability are verified by comparison with the
To apply the RHT model to the evolution of rock damage evolution
blasting test results.

20
L.X. Xie et al. Tunnelling and Underground Space Technology 66 (2017) 19–33

2.1. The Riedel-Hiermaier-Thoma (RHT) model (Riedel et al., 1999) 2.3.1. Determination of strain rate
Many researchers have tested the tensile or compressive strength of
In the RHT model, pressure is expressed using the Mie–Greisen rock under different strain rates. Zhang and Zhao summarized these
form, with a polynomial Hugoniot curve and a p-α compaction investigations (Zhang and Zhao, 2014). In this study, we used their
relationship (Borrvall and Riedel, 2011). A schematic description of research findings to perform a curve regression. The relationship
the p-α EOS is shown in Fig. 1(a). In a compaction model, when the between the dynamic strain rate and tensile or compressive strength
pressure value is below the pore crush pressure, the model is elastic. is obtained as shown in Fig. 3. Eq. (1) shows the increase factor for the
Once the pressure exceeds the pore crush pressure, pore collapse strain rate of rock material under (a) compression and (b) tension.
reduces the volumetric stiffness of the material, resulting in the
⎧
⎪ 0.512εp
1/3
̇ εṗ > 30 s−1
reduction of the effective bulk modulus. The relationship between the Frc (εṗ ) = ⎨
⎪ c 0.026
pressure and volumetric strain is non-linear. When the pressure exceeds ⎩ (εṗ / ε0̇ ) εṗ ⩽ 30 s−1 (1a)
the pore crush pressure, unloading occurs along the current elastic
stiffness. This results in a permanent volumetric strain at zero pressure. where, Frc (εṗ )
is the strain rate strength factor under compression, ε̇p is
Subsequent reloading occurs along the unloading curve. To characterize the strain rate, and ε̇0c is reference strain rate under compression,
pore collapse behavior, an internal variable α is used to represent the ε̇0c = 3.0 × 10−5 s−1.
porosity of the material as the fraction between the matrix material and ⎧ 1/3
⎪ 2.4εp ̇ εṗ > 10−1 s−1
the porous material. The porosity of α decreases with increasing Frt (εṗ ) = ⎨
⎪ t 0.007
pressure, making the loading irreversible. When the pressure reaches ⎩ (εṗ / ε0̇ ) εṗ ⩽ 10−1 s−1 (1b)
the pore crush pressure and compaction pressure, α equals 0 and 1,
respectively. where, Frt (εṗ ) is the strain rate strength factor under tension, ε̇p is the
In the RHT model, three stress limit surfaces are used to account for strain rate, and ε̇0t is reference strain rate under tension,
ε̇0t = 3.0 × 10−6 s−1.
both reductions in strength along different meridians as well as strain
rate effects. These limit surfaces include the initial elastic yield surface,
and residual friction surface (Borrvall and Riedel, 2011). A typical 2.3.2. Determination of strength parameters
loading scenario is shown in Fig. 1(b)—the arrows indicate that the To better estimate rock strength under high in-situ stresses, the
model is elastic until the stress reaches the initial yield surface. Beyond empirical equation of Hoek and Brown is used to obtain the strength of
this surface, plastic strain begins to accumulate. The plastic strain and rock under various confining stresses (Hoek and Brown, 1980). Based
hardening properties of the material are applied to form an effective on the data presented in Banadaki and Mohanty (2012), the fitting
yield surface, which is created from an interpolation between the initial equations are as follows:
yield surface and failure surface. As the stress reaches the failure
⎛ σ ⎞1/2
surface, damage from plastic strain accumulates, which causes the σ1 = σ3 + 167.1 ⎜24 3 + 1⎟
⎝ 167.1 ⎠ (2a)
failure surface to soften. This results in the formation of the post-failure
stress limit surface, which is created by interpolating values between
the failure surface and residual friction surface. P
A detailed introduction of the RHT model can be seen in Borrvall
and Riedel (2011).

Pcomp
2.2. Numerical model for physical tests

To verify the RHT model and apply it to blasting design optimiza-

tion, parameters must be determined from existing lab tests. A series of Pcrush
physical tests were conducted by Banadaki and Mohanty (2012) on
Barre granite. The specimen was cylindrical in shape, with a diameter
of 144 mm and height of 150 mm. A Φ 6.45 mm hole was created as a
blasthole. During the test, a copper tube was installed in the blasthole to vol

prevent detonation gas from entering any cracks. The outer diameter of (a)
the copper tube was 6.45 mm and the inner diameter was 5.15 mm. Air
was taken as the coupling medium for tests. The test model is shown in
Fig. 2.
A plane numerical model is adopted for comparison with the test
Failure surface
results. The model consists of five materials: granite, copper case, resin,
air, and PETN dynamite. The size of the numerical model is the same as
that of the physical model. In SOLID, 164 elements are adopted, and the D>0
total number of meshes is 460,000. The numerical model is shown in
Fig. 2.
p>0
Yield surface

2.3. Determination of RHT model parameters Residual surface

There are 38 total parameters in the RHT model. Among these

parameters, 18 are determined based on the experimental data
presented in Section 2.3. The minimum damaged residual strain (εpm) Pt Pu Pc P
is obtained from the inverse calibration of the blasting tests. Literature
values are used to determine the 19 remaining parameters, which are (b)
not sensitive to the simulation results (Borrvall and Riedel, 2011). The Fig. 1. RHT model (Borrvall and Riedel, 2011): (a) Schematic description of the p-α EOS;
determined parameters are summarized in Table 2. (b) stress limit surfaces and loading scenario.

21
L.X. Xie et al. Tunnelling and Underground Space Technology 66 (2017) 19–33

Fig. 2. Test setup and numerical model: (a) Test setup (Banadaki and Mohanty, 2012); (b) overall numerical model; (c) local magnified model; (d) material location.

⎛ σ −6.93 ⎞0.57 6.83 MPa. According to Eq. (3b), the pure shear strength is 35 MPa.
τ = 1.32 × 167.1 ⎜ n ⎟
⎝ 167.1 ⎠ (2b) Considering the effects of the strain rate, this value can be taken as
35.14 MPa.
where σ1 and σ3 are the maximum and minimum effective stresses at
failure, σn is the normal effective stress, and τ is the shear stress.
Rock strength is influenced by strain rate. Because of this, the strain 2.3.3. Determination of parameters for failure surface equation
rate used in the rock tests is different from the reference strain rate, and The failure surface parameters A and N must be determined such
the rock strength must be converted to a value that corresponds to the that the normalized pressure satisfies 3P∗ ≥ F. Both parameters are
reference strain rate. The strain rate in the compression tests is obtained as follows (Borrvall and Riedel, 2011):
controlled at 4.63 × 10−5 s−1 and the equivalent strain rate is σf∗ (P∗,Fr ) = A (P∗−Fr /3 + (A / Fr )−1/N )N 3P∗ ⩾ Fr (4)
2.59 × 10−5 s−1 (Banadaki and Mohanty, 2012). To obtain the
strength at the reference strain rate ε̇0c/t , the Hoek-Brown intact strength where σf∗ ∗
(P , Fr) is the normalized strength relative to the compressive
needs to be corrected for, which is done by multiplying the following strength, σf∗ = σf/ fc; P∗ is the normalized pressures; Fr is the dynamic
correction factor. strain rate increase factor; and A, N are the failure surface parameters,
P∗ = P/fc.
1
lc = . When the material is in a quasi-static state, ε̇p = 3.0 × 10−6 and
Frc (εṗ ) (3a) Fr = 1. It can be seen from Table 1 that the normalized pressures
1 corresponding to the confining stresses of 10 MPa and 100 MPa satisfy
lt = . Eq. (4). Substituting the normalized pressure P∗ and normalized
Frt (εṗ ) (3b)
equivalent stress σ∗f (corresponding to confining stresses of 10 MPa
where, lc and lt are the correction factors for compressive and tensile and 100 MPa) into Eq. (4), A = 2.44 and N = 0.76 can be achieved.
strength, respectively. In the RHT model, the ratio between the radii of the tensile and
The mechanical parameters of the rock under various confining compressive meridians is
pressures can be obtained from Eqs. (1a), (1b), (2a), (2b), (3a), and Q = Q (P∗) = Q0 + BP∗. (5)
(3b). These values are presented in Table 1.
As presented in Table 1, under the reference strain rate, the where Q0 and B are lode angle dependence factors.
compressive strength is 167.8 MPa and the tensile strength is The suggested values of Q0 and B for simulations of concrete are

22
L.X. Xie et al. Tunnelling and Underground Space Technology 66 (2017) 19–33

(a)

(b)

Fig. 3. Curve regression for strain rate increase factor: (a) Curve fitting for dynamic compressive strain rate (data from (Zhang and Zhao, 2014)); (b) curve regression for dynamic tensile
strain rate (data from (Zhang and Zhao, 2014)).

Table 1 Hugoniot curve and a p-α compaction relation (see Appendix A). The
Mechanical parameters of rock under various confining pressures. EOS is expressed as
σ2 = σ3/MPa σ1/MPa P/MPa σf/MPa σf-C/MPa P* σ*f (BΓ0 + BΓ μ) α0 ρ0 e + A1 μ + A2 μ2 + A3 μ3
PR =
α (6)
0 6.93 −2.31 6.93 6.83 −0.01 0.04
0 167.1 56 167.1 167.8 0.34 1.00
where BГ0 and BГ are material constants; α0 is the initial porosity of
5 224 78 219 219.9 0.47 1.31
10 271 97 261 262.0 0.58 1.56 porous materials; ρ0 is the initial density of porous materials; A1, A2,
50 528 209 478 480.0 1.25 2.86 and A3 are the polynomial coefficients; PR is the EOS pressure in the
100 735 318 655 657.6 1.90 3.92 RHT model; and μ is volumetric strain.
Compared with concrete, Barre granite has lower porosity
(Neithalath et al., 2006; Yavuz et al., 2010). Additionally, deep rock
0.68 and 0.01, respectively. In the study, Q0 = 0.68 and B = 0.05 are mass is subject to high in-situ stress, which makes the porosity much
obtained from the Yu curve regression results (Yu, 1998). The regres- lower. Therefore, in the calculation, assuming that α0 = 1 (based on Eq.
sion curve is shown in Fig. 4. (A17)), A1, A2, and A3 are 25.7, 37.84 and 21.29 MPa, respectively.
When the pressure exceeds the pore crush pressure, pore collapse
2.3.4. EOS parameters(Zhang et al., 2017) causes inelastic deformation in the rock material. According to Riedel
In the RHT model, the shear and pressure components are coupled. et al. (2009), the elastic limit pressure is taken as 2/3 of the uniaxial
The pressure is described by a Mie–Gruneisen form with a polynomial compressive strength, which is 125 MPa.

23
L.X. Xie et al. Tunnelling and Underground Space Technology 66 (2017) 19–33

Table 2
Input parameters for RHT model for rock.

Parameter Value Parameter Value

Mass density RO (kg/m3) 2660 Porosity exponent NP 3.0

Initial porosity ALPHA 0 Reference compressive strain-rate EOC 3.E−5
Crush pressure PEL (MPa) 125 Reference tensile strain rate EOT 3.E−6
Compaction pressure PCO (GPa) 6.0 Break compressive strain rate EC 3.E+25
Hugoniot polynomial coefficient A1 (GPa) 25.7 Break tensile strain rate ET 3.E+25
Hugoniot polynomial coefficient A2 (GPa) 37.84 Compressive strain rate dependence exponent BETAC 0.026
Hugoniot polynomial coefficient A3 (GPa) 21.29 Tensile strain rate dependence exponent BETAT 0.007
Parameter for polynomial EOS B0 1.22 Volumetric plastic strain fraction in tension PTF 0.001
Parameter for polynomial EOS B1 1.22 Compressive yield surface parameter GC* 0.53
Parameter for polynomial EOS T1 (GPa) 25.7 Tensile yield surface parameter GT* 0.7
Parameter for polynomial EOS T2 0.0 Erosion plastic strain EPSF 2.0
Elastic shear modulus SHEAR (GPa) 21.9 Shear modulus reduction factor XI 0.5
Compressive strength FC (MPa) 167.8 Damage parameter D1 0.04
Relative tensile strength FT* 0.04 Damage parameter D2 1.0
Relative shear strength FS* 0.21 Minimum damaged residual strain EPM 0.015
Failure surface Parameter A 2.44 Residual surface parameter AF 0.25
Failure surface Parameter N 0.76 Residual surface parameter AN 0.62
Lode angle dependence factor Q0 0.68 Gruneisen gamma GAMMA 0.0
Lode angle dependence factor B 0.05

0.80 attenuated to stress waves. However, because the rock has relatively
Experiment low tensile strength, the tensile component of the stress waves can
Fitting Q=0.68+0.05P* cause plastic tensile deformation (Hustrulid, 1999), which can exceed
0.75 the tensile failure criterion and lead to tensile damage. Fig. 5(b) shows
the tensile failure zone under stress waves.
A free surface encountered during the propagation of stress waves
0.70 will reflect radial compressional waves. Tensile waves are then
Q

generated, and propagate into the rock from the free surface. These
tensile waves are superimposed on the compressional waves. If the
0.65 superimposed stress waves are tensile, tensile plastic deformation
occurs. When the tensile plastic deformation exceeds the tensile failure
criterion, spalling occurs, as shown in Fig. 5(c). After spalling, the speed
0.60
of shear waves is lower than that of compressional waves. As a result,
0.0 0.2 0.4 0.6 0.8 1.0 1.2
tensile stress in the tangential direction leads to the further tensile
P*
plastic deformation of the rock. The deformation exceeds the tensile
Fig. 4. Relationship between the radius ratio of tensile and compressive meridians and failure criterion, which leads to the further extension of damage in the
the stress. radial direction. Finally, these damage cracks reach the free surface, as
shown in Fig. 5(d). As seen in the numerical simulation and test results
2.4. Jones–Wilkins–Lee (JWL) EOS shown in Fig. 5(d) and (e), the compressive, tensile, and spalling failure
zones are similar. The scale and number of main damage cracks are also
When explosives are detonated in a blasthole, an instant chemical the same. Therefore, the simulation results are in good agreement with
reaction occurs. The explosives are transformed from concentrated solid the test results, and the model can accurately simulate the damage and
materials to a high-temperature, high-pressure gas. The JWL EOS (Kury failure of the rock.
et al., 1965) is a high-energy combustion model with the capacity to
predict a large pressure range for an explosion, and is widely used in
numerical simulation. In this study, JWL EOS is chosen to describe the 3. Damage mechanisms under high in-situ stress
explosion, and is expressed as follows:
3.1. Cut hole blasting design and parameter determination
⎛ ω ⎞ −R1v ⎛ ω ⎞ −R2 v ωE
P = AJWL ⎜1− ⎟e + BJWL ⎜1− ⎟e +
⎝ R1 v ⎠ ⎝ R2 v ⎠ v (7) Previous studies have indicated that the magnitude of in-situ
where P is the pressure of the detonation products, v is the relative stresses and horizontal pressure coefficients affects the damage char-
specific volume of the detonation products, E is the specific internal acteristics of rock masses under blasting load (Xie et al., 2016; Yilmaz
energy of the detonation products, and AJWL, BJWL, R1, R2, and ω are and Unlu, 2013). To resolve the difficulties associated with cut blasting
constants. The JWL EOS parameters are selected from Banadaki (2010), under high in-situ stresses, damage mechanisms must be investigated.
as listed in Table 3. First, the maximum burden must be determined. In this study, the
formula proposed by Per-Anders Persson is adopted (Langefors and
Kihlström, 1978):
2.5. Comparison of experimental and numerical results
Table 3
As seen in Fig. 5(a), when blast waves propagate in rock, large Parameters for the explosive material and JWL EOS (from Banadaki, 2010).
plastic deformation occurs in the radial direction, due to the high
intensity of shock waves in rock. When the deformation exceeds the ρe (kg/ VoD PCJ (GPa) AJWL BJWL R1 R2 ω E0 (GPa)
compressive failure criterion, compressive damage occurs, and a m3) (m/s) (GPa) (GPa)

compressive crushed zone forms in the vicinity of the blasthole. Energy 1320 6690 16 586 21.6 5.81 1.77 0.282 7.38
is dissipated by the propagation of blast waves in the rock, which are

24
L.X. Xie et al. Tunnelling and Underground Space Technology 66 (2017) 19–33

˄a˅ ˄b˅ ˄c˅

˄d˅ ˄e˅(Banadaki 2010)

Fig. 5. Comparison of test and simulation results: (a) Compressive failure zone under blast waves; (b) tensile failure zone under stress waves; (c) spalling failure zone under stress waves;
(d) final failure pattern under blast waves and stress waves, obtained by numerical simulation; (e) final failure pattern under blast waves and stress waves, obtained by test.

Bb 1.5

l = 55d
( ) ϕ
ϕ c
(Bb− 2 )( 0.4 )
as 0.25 m. In the calculation, the rock constant c is equal to 0.4
(Langefors and Kihlström, 1978).
SANFO (8)

where l is the linear charge density, ϕ is the diameter of the central 3.2. Damage to rock mass from blast loading under different in-situ stress
hole, d is the diameter of the blasthole, c is a rock constant, Bb is the conditions
burden of cut hole, and SANFO is the explosive weight strength relative
to ANFO. Eq. (8) considers the effects on the burden of different Based on the blasting design, a numerical model is built, as shown in
lithologies, linear charge densities, blasthole diameters, and explosive Fig. 6. To investigate the effect of in-situ stress on the mechanisms of
strengths. Therefore, this equation can be used to determine the cut rock damage evolution during cut blasting, the hydrostatic stresses are
hole burden for various rock types. set as 0, 10, 20, and 30 MPa. In consideration of the exact in-situ stress
To calculate SANFO, the weight strength relative to a reference conditions, a horizontal stress of 30 MPa is used. The effects of the
explosive (LFB-dynamite) needs to be determined. In the paper, PETN is horizontal pressure coefficient on rock mass damage are investigated by
chosen as the explosive used for rock breaking. Its weight strength varying the magnitude of the vertical stress. The vertical in-situ stress is
relative to a reference explosive is obtained from the following equation varied as 10, 20, 40, and 70 MPa.
(Langefors and Kihlström, 1978): As indicated by the simulation results for cut blasting under an in-
5 Qv 1V situ stress of 0 MPa (see Fig. 7), shock waves create a compressive
S= + damage zone around the blasthole that is three times the size of that
6 Qv0 6 V0 (9)
blasthole. When the shock waves are attenuated to stress waves, the
where S is the weight strength relative to LFB-dynamite; Qv is the tensile stress component leads to tensile damage in the rock. As the
explosion energy of 1 kg of explosives; V is the volume of gas released at stress waves interact during superposition, the tensile stress component
standard atmospheric pressure and temperature (STP); Qv0 is the increases in the superposition zone. Tensile damage is produced and
released energy of the LFB-dynamite (5 MJ/kg); and V0 is the volume extends towards the boreholes. With the extension of tensile damage,
of gas released by LFB-dynamite at STP (850 L). the damage zones between the blastholes coalesce.
For ANFO, Qv = 3.91 MJ/kg and V = 970 L. Based on Eq. (9), the When stress waves propagate toward the free surface, tensile
weight strength of ANFO relative to LFB-dynamite (S) is 0.84 (Langefors damage cracks extend from the blastholes to the central hole. As the
and Kihlström, 1978). PETN is chosen as the explosive used to break the stress waves reach the free surface, wave reflection occurs and tensile
rock. For PETN, Qv = 6.38 MJ/kg and V = 717 L. Based on Eq. (9), waves are generated, which leads to spalling in the vicinity of the free
S = 1.2. Therefore, the weight strength relative to ANFO (SANFO) is surface. Then, the tensile damage cracks extend from the central hole to
1.43. the blastholes. When a free surface is present during stress wave
In the blasting design, a center hole is set with a diameter of propagation, energy shifts towards the free surface, leading to greater
100 mm. Four 42 mm diameter cut holes are arranged around the damage to the rock masses near the free surface. For instance, more
center hole. According to Eqs. (8) and (9), the burden can be calculated damage cracks appear in the rock masses between the free surface and

25
L.X. Xie et al. Tunnelling and Underground Space Technology 66 (2017) 19–33

stress acts on a rock mass, a stress field is generated that resists the
(a) stress waves induced by blasting, resulting in less damage to the rock
mass.
As shown in Fig. 9, when the horizontal stress is maintained at
30 MPa, the magnitude of vertical stress varies. When the vertical stress
is lower than 30 MPa, the horizontal damage cracks are longer than the
vertical damage cracks. When the vertical stress is higher than 30 MPa,
the vertical damage cracks are longer than the horizontal damage
cracks. In addition, because the difference in horizontal and vertical
stresses is great, the difference in the damage crack length for the two
directions is also larger. This shows that different lateral pressure
coefficients result in anisotropic rock damage. The larger the stress
difference, the more obvious the damage anisotropy of rock masses
becomes. When the horizontal stress is 30 MPa and the vertical stress is
20 MPa, fewer damage cracks coalesce between blastholes.

4. Optimization of blasting design

It can be seen from the simulation results for various hydrostatic

(b) pressure and non-hydrostatic pressure conditions that there are two
main causes of difficulties when excavating deep rock masses under
high in-situ stresses with the cut blasting method: 1) the resistance
effect of in-situ stresses and 2) damage anisotropy due to stress
differences in the horizontal and vertical directions. The higher the
in-situ stress, the smaller the damage range of rock masses for a given
blasting load. The greater the difference between the horizontal and
vertical stresses, the more obvious the damage anisotropy becomes.
Consequently, the length of damage cracks decreases, and the propaga-
tion path of cracks around the blasthole deviates, leading to excavation
difficulties. To solve this problem, adjustments are made to 1) the
distance between cut holes and 2) the layout of blastholes, with the aim
of achieving more efficient cut blasting excavation.
High in-situ stresses cause resistance to damage between cut holes.
As a result, the damage crack propagation length is reduced. This
problem can be mitigated by gradually decreasing the spacing between
cut holes. The spacing between blastholes was varied as 0.5, 0.4, and
0.3 m. Additionally, three different cut hole layouts are shown in
Fig. 10(a).
(c) As seen from the simulation results, varying blasthole spacing can
overcome excavation problems caused by the in-situ stress resistance.
When the horizontal stress is 30 MPa and the vertical stress is 70 MPa, a
blasthole spacing of 0.4 m is most appropriate. However, the blasthole
spacing must be corrected, because the blasthole location may deviate
from the desired location during drilling. The blasthole correction
formula is as follows (Langefors and Kihlström, 1978):

B1 = Bb−(α1 + Hα2 ) (10)

where B1 is the cut hole burden in the first quadrangle; α1 is the angular
deviation (0.02); α2 is the collar deviation (0.01); and H is the advance
length (3 m).
The simulation results in Fig. 11 indicate that for a square-shaped
blasthole layout, a spacing of 0.3 m is the most reliable. For a diamond-
shaped layout, it is the most reliable to use a spacing of 0.4 m in the
Fig. 6. Numerical model: (a) numerical model size; (b) numerical model under hydro- direction perpendicular to major principal stress, and 0.3 m in the
static pressure conditions; (c) numerical model under non-hydrostatic pressure condi-
direction of minor principal stress. The results for the square-shaped
tions.
layout are in agreement with those in the literature, given B1 = 1.5ϕ.
Therefore, the blasthole spacing recommended by the formula
the center lines of blastholes. Some damage cracks are caused by
B1 = 1.5ϕ is more suitable for cut blasting under high in-situ stresses.
reflected tensile waves, whereas others are formed by the energy shift
As seen from the simulation results, because of high in-situ stresses,
toward the free surface.
the free surface created by the first square-shaped cut holes is too small
As seen from the simulation results shown in Fig. 8, when in-situ
for the excavation of deep rock masses. To create sufficient free surfaces
stresses have hydrostatic pressure conditions of 10, 20, and 30 MPa, the
for subsequent blasting, auxiliary blastholes are usually designed. The
length of damage cracks gradually decreases with increasing hydro-
two different types of layouts for auxiliary blastholes are shown in
static pressure. When in-situ stress is higher than 30 MPa, the coales-
Fig. 10(b). The formula for auxiliary blastholes is as follows (Langefors
cence of damage cracks becomes more difficult. Thus, when in-situ
and Kihlström, 1978):

26
L.X. Xie et al. Tunnelling and Underground Space Technology 66 (2017) 19–33

Fig. 7. Damage evolution process under 0 MPa pressure conditions: (a) compressive damage induced by shock waves; (b) tensile damage induced by stress waves; (c) tensile damage
induced by stress wave superposition; (d) tensile damage induced by reflected stress waves.

⎡ (B −F ) lSANFO ⎤1/2 efficient excavation. A value of 0.20 m is more suitable. This is close to
Bb = 10.5 × 10−2 ⎢ 1 ⎥⎦
⎣ dc (11) the burden value suggested by Gustafsson for the design of auxiliary
blastholes (Gustafsson suggested B2 = 0.7A) (Langefors and Kihlström,
where B1 is the burden for the first square-shaped cut holes (0.15 m); F 1978). In other words, the burden size recommended by Gustafsson can
is the blasthole deviation caused by angular and collar deviation be used to achieve larger free surfaces under high in-situ stresses.
(0.05 m); l is the linear charge density (1.8 kg/m); SANFO is the
explosive strength relative to ANFO (1.03); d is the charge diameter
5. Conclusions
(0.042 m); and c is a rock constant (0.4). The calculated value of Bb is
0.32 m.
As the D & B method is applied to rock excavations at great depths,
Considering deviation during drilling, the burden of auxiliary
difficulties due to the influence of high in-situ stresses are encountered
blastholes in the second quadrangle is expressed as B2 = Bb - F
in cut blasting. To overcome these difficulties, the RHT model is used to
(Langefors and Kihlström, 1978). In the calculation of auxiliary blast-
simulate damage evolution mechanisms and optimize cut blasting
holes, these values should satisfy 0.5 As < B2 ≤ As (As is the side
design. Through the simulation, the causes of difficulties in the
width of the rectangular opening, which is equal to 21/2 B1).
excavation of deep rock masses are analyzed, and a method for
Based on the above analysis, two types of auxiliary blasthole
optimizing cut blasting design under high in-situ stresses is proposed.
burdens are considered, specifically, 0.25 m and 0.20 m, respectively.
This method provides a theoretical basis and reference for cut blasting
As shown in Fig. 12, because of high in-situ stress resistance and
rock excavation at great depths. Based on the simulation results, the
damage anisotropy, a burden of 0.25 m is not sufficient to achieve
following conclusions can be drawn:

27
L.X. Xie et al. Tunnelling and Underground Space Technology 66 (2017) 19–33

Fig. 8. Extent of damage under hydrostatic pressure conditions: (a) extent of damage for a hydrostatic pressure of 10 MPa; (b) extent of damage for a hydrostatic pressure of 20 MPa; (c)
extent of damage for a hydrostatic pressure of 30 MPa.

Parameters in the RHT model are determined based on experimental is higher than 30 MPa, and the coalescence of damage cracks becomes
data. Their rationality is verified by a comparison between the existing difficult. In-situ stress resistance and damage anisotropy due to stress
experimental and simulation results. The simulation results are in good differences in the horizontal and vertical directions are the main causes
agreement with experimental results. The damage processes for the of difficulties in the cut blasting excavation of deep rock masses.
rock in the blasting experiments were analyzed by simulation, the To overcome these difficulties, a cut blasting design is optimized.
results of which indicated that compressive damage zones, tensile The simulation results indicate that when the vertical and horizontal in-
damage zones, and spalling damage zones were induced by the shock situ stresses are 70 and 30 MPa, respectively, the burdens for the cut
waves, stress waves, and reflected tensile waves, respectively. holes and auxiliary blastholes are recommended by the formulas
During cut blasting in a deep rock mass, rock mass failure is caused B1 = 1.5ϕ and B2 = 0.7As, respectively. These optimized values over-
by a combination of shock waves, stress waves, stress wave super- come depth-related excavation difficulties. The proposed method is
position, and reflected stress waves. The length of damage cracks simple and can be easily applied to the cut blasting excavation of deep
gradually decreases with increasing hydrostatic pressure. The differ- rock mass.
ence between the lengths of horizontal and vertical damage cracks
increases with increasing stress differences in the horizontal and Acknowledgements
vertical direction. When the hydrostatic pressure is higher than
30 MPa, or the horizontal stress is maintained at 30 MPa, vertical stress The authors would like to acknowledge financial support from the

28
L.X. Xie et al. Tunnelling and Underground Space Technology 66 (2017) 19–33

Fig. 9. Damage extent in the different in situ stresses: (a) extent of damage for a horizontal stress of 30 MPa and vertical stress of 10 MPa; (b) extent of damage for a horizontal stress of
30 MPa and vertical stress of 20 MPa; (c) extent of damage for a horizontal stress of 30 MPa and vertical stress of 40 MPa; (d) extent of damage for a horizontal stress of 30 MPa and
vertical stress of 70 MPa.

Chinese National Programs for Fundamental Research and

Development (973 Program) (2011CB013501), Chinese National National Natural Science Foundation (51279135), and China
Science Fund for Distinguished Young Scholars (51125037), Chinese Scholarship Council (to the first author, No. 201506270066).

Appendix A. Appendix

A general equation of state (EOS) for a solid medium can be expressed by the following (An, 2010; Zukas, 1990):
Γ(v )
P = ρm γ (v ) e = e,
v (A1)
where Γ(v ) is the Gruneisen parameter, e is the internal energy per unit mass, ρ is the density of the solid medium, and v is the specific volume.
Assuming that the state of the materials subjected to shock pressure falls on the Hugoniot curve, and that the shock pressure and internal energy are
related to a specific volume on the Hugoniot curve, then the relationship between shock pressure and internal energy can be given by
Γ(v )
PH = ρm γ (v ) e H = e H,
v (A2)
where PH is the shock pressure and e H is the internal energy per unit mass. If the Hugoniot curve is considered to be the reference state, the EOS can
be written as

29
L.X. Xie et al. Tunnelling and Underground Space Technology 66 (2017) 19–33

(a)

(b)

Fig. 10. Optimization of cut blasting design for a horizontal stress of 30 MPa and vertical
stress of 70 MPa: (a) reducing the spacing of cut holes; (b) reducing the spacing of
auxiliary holes.

Γ(v )
P−PH = = (e−e H ).
v (A3)
According to the energy conservation equation, the following formula can be obtained:
PH
e H−e0 = (v0−v H ),
2 (A4)
where e0 is the initial internal energy per unit mass and v0 is the initial specific volume. Substituting Eq. (A4) into Eq. (A3) yields
Γ(v ) P ⎛ Γ ⎞ Γ(v )
P = PH + (e−e0 )− H (v0−v ) = PH ⎜1− μ⎟ + (e−e0 ),
v 2 ⎝ 2 ⎠ v (A5)
where u is the volumetric strain.
According to the momentum and mass conservation equations and the linear relationship between shock wave velocity and particle velocity, the
following formulae are obtained:
PH = ρm0 US uP (A6)

ρm0 US = ρm (US−uP ) (A7)

US = c0 + SuP, (A8)
where US is the shock velocity, uP is the particle velocity, c0 is the sound speed at ambient pressure and temperature, and S is the slope of the linear
relationship.
ρ
Let u = ρ m −1. Substituting Eqs. (A7) and (A8) into Eq. (A6) yields
m0

ρ c 2 μ (μ + 1)
PH = m0 0 .
[1−(S−1) μ]2 (A9)
B Γ0 + B Γ μ
Assuming Γ(v ) = 1+μ
(BΓ0,BΓ are material constants), based on the relationship between the internal energy per unit initial volume and

30
L.X. Xie et al. Tunnelling and Underground Space Technology 66 (2017) 19–33

Fig. 11. Optimizing cut blasting design by reducing cut hole spacing: (a) extent of damage for a spacing of 0.5 m; (b) extent of damage for a spacing of 0.4 m; (c) extent of damage for a
spacing of 0.3 m.

e
internal energy per unit mass, e v = v0
= ρm0 e , substituting Eq. (A9) into the Eq. (A5) yields
B B
ρm0 c02 μ (1 + (1− 2Γ0 ) μ− 2Γ μ2 )
P= 2
+ (B0 + B1 μ) ρm0 e.
[1−(S−1) μ] μ (A10)
In the RHT model, the shear and pressure components are coupled, and the pressure is described by a Mie-Gruneisen form with a polynomial
Hugoniot curve. According to the Mie-Gruneisen form in Eq. (A10), by expanding Eq. (A9) as a Taylor series and neglecting the third- or higher-order
terms, the EOS in the RHT model is expressed as follows:
P = (BΓ0 + BΓ μ) ρm0 e + A1 μ + A2 μ2 + A3 μ3, (A11)
Together with
A1 = ρm0 c02 (A12)

31
L.X. Xie et al. Tunnelling and Underground Space Technology 66 (2017) 19–33

Fig. 12. Optimizing blast design by reducing auxiliary hole spacing: (a) extent of damage
for a burden of 0.25 m; (b) extent of damage for a burden of 0.2 m.

A2 = ρm0 c02 (2S−1) (A13)

A3 = ρm0 c02 [(3S−1)(S−1)]. (A14)

To consider the impact of porosity on shock pressure, Carroll & Holt proposed the following modified equation:
1 1
P= f (ρm ,e) = f (αρ,e).
α α (A15)
ρ
where α = ρm ,α is porosity of the porous material, and ρ is the density of the porous material.
Therefore, Eq. (A11) can be rewritten as
(BΓ0 + BΓ μ) α0 ρ0 e + A1 μ + A2 μ2 + A3 μ3
P= ,
α (A16)
Together with
αρ
μ (αρ) = −1
α0 ρ0

A1 = α0 ρ0 c02

A2 = α0 ρ0 c02 (2S−1)

32
L.X. Xie et al. Tunnelling and Underground Space Technology 66 (2017) 19–33

A3 = α0 ρ0 c02 [(3S−1)(S−1)], (A17)

References induced by blasting. Tunnelling Underground Space Technol. 51, 354–361.

Neithalath, N., Weiss, J., Olek, J., 2006. Characterizing enhanced porosity concrete using
electrical impedance to predict acoustic and hydraulic performance. Cem. Concr. Res.
An, J.C., 2010. Soil Behavior Under Blasting Loading. The University of Nebraska-Lincoln. 36, 2074–2085.
Bäckblom, G., Martin, C.D., 1999. Recent experiments in hard rocks to study the Olofsson, S.O., 1990. Applied Explosives Technology for Construction and Mining.
excavation response: implications for the performance of a nuclear waste geological Applex.
repository. Tunnelling Underground Space Technol. 14, 377–394. Ramulu, M., Chakraborty, A., Sitharam, T., 2009. Damage assessment of basaltic rock
Banadaki, M.D., Mohanty, B., 2012. Numerical simulation of stress wave induced mass due to repeated blasting in a railway tunnelling project–a case study. Tunnelling
fractures in rock. Int. J. Impact Eng. 40, 16–25. Underground Space Technol. 24, 208–221.
Banadaki, M.M.D., 2010. Stress-Wave Induced Fracture in Rock Due to Explosive Action. Riedel, W., Kawai, N., Kondo, K.i., 2009. Numerical assessment for impact strength
University of Toronto. measurements in concrete materials. Int. J. Impact Eng. 36, 283–293.
Borrvall, T., Riedel, W., 2011. The RHT concrete model in LS-DYNA. In: Proceedings of Riedel, W., Thoma, K., Hiermaier, S., Schmolinske, E., 1999. Penetration of reinforced
the 8th European LS-DYNA Users Conference, Strasbourg, . concrete by BETA-B-500 numerical analysis using a new macroscopic concrete model
Chen, M., Lu, W.B., Yi, C.P., 2007. Blasting vibration criterion for a rock-anchored beam for hydrocodes. In: Proceedings of the 9th International Symposium on the Effects of
in an underground powerhouse. Tunnelling Underground Space Technol. 22, 69–79. Munitions with Structures, .
Cunningham, C., Goetzsche, A., 1990. The specification of blast damage limitations in Wang, Z.L., Li, Y.C., Shen, R., 2007. Numerical simulation of tensile damage and blast
tunnelling contracts. Tunnelling Underground Space Technol. 5, 193–198. crater in brittle rock due to underground explosion. Int. J. Rock Mech. Mining Sci. 44,
Deng, X.F., Zhu, J.B., Chen, S.G., Zhao, Z.Y., Zhou, Y.X., Zhao, J., 2014. Numerical study 730–738.
on tunnel damage subject to blast-induced shock wave in jointed rock masses. Xie, L.X., Lu, W.B., Zhang, Q.B., Jiang, Q.H., Wang, G.H., Zhao, J., 2016. Damage
Tunnelling Underground Space Technol. 43, 88–100. evolution mechanisms of rock in deep tunnels induced by cut blasting. Tunnelling
Gong, Q.M., Yin, L.J., Wu, S.Y., Zhao, J., Ting, Y., 2012. Rock burst and slabbing failure Underground Space Technol. 58, 257–270.
and its influence on TBM excavation at headrace tunnels in Jinping II hydropower Yang, J.H., Lu, W.B., Hu, Y.G., Chen, M., Yan, P., 2015. Numerical simulation of rock
station. Eng. Geol. 124, 98–108. mass damage evolution during deep-buried tunnel excavation by drill and blast. Rock
Hoek, E., Brown, E.T., 1980. Underground excavations in rock. Mech. Rock Eng. 48, 2045–2059.
Hustrulid, W., 1999. Blasting principles for open-pit mining. Vol. 2 Theoretical Yavuz, H., Demirdag, S., Caran, S., 2010. Thermal effect on the physical properties of
Foundations Rotterdam, Balkema. carbonate rocks. Int. J. Rock Mech. Mining Sci. 47, 94–103.
Johnson, G., Holmquist, T., 1992. A computational constitutive model for brittle Yilmaz, O., Unlu, T., 2013. Three dimensional numerical rock damage analysis under
materials subjected to large strains, high strain rates and high pressures. Shock Wave blasting load. Tunnelling Underground Space Technol. 38, 266–278.
and High-Strain-Rate Phenomena in Materials 1075–1081. Yu, M.H., 1998. Twin Shear Theory and its Application. Science Press, Beijing.
Johnson, G.R., Holmquist, T.J., 1994. An Improved Computational Constitutive Model for Bruland, A., Zare, S., 2006. Comparison of tunnel blast design models. Tunnelling
Brittle Materials, High-Pressure Science and Technology—1993. AIP Publishingpp. Underground Space Technol. 21, 533–541.
981–984. Zhang, Q.B., Braithwaite, C.H., Zhao, J., 2017. Hugoniot equation of state of rock
Kury, J., Hornig, H., Lee, E., McDonnel, J., Ornellas, D., Finger, M., Strange, F., Wilkins, materials under shock compression. Phil. Trans. R. Soc. A 375, 20160169.
M., 1965. Metal acceleration by chemical explosives. In: Fourth (International) Zhang, Q.B., Zhao, J., 2014. A review of dynamic experimental techniques and
Symposium on Detonation, ACR-126, . mechanical behaviour of rock materials. Rock Mech. Rock Eng. 47, 1411–1478.
Langefors, U., Kihlström, B., 1978. The Modern Technique of Rock Blasting. John Zhao, Z.Y., Zhang, Y., Bao, H.R., 2011. Tunnel blasting simulations by the discontinuous
Wiley & Sons. deformation analysis. Int. J. Comput. Methods 8, 277–292.
Li, J.C., Li, H.B., Ma, G.W., Zhou, Y.X., 2013. Assessment of underground tunnel stability Zhu, W.C., Li, Z., Zhu, L., Tang, C.A., 2010. Numerical simulation on rockburst of
to adjacent tunnel explosion. Tunnelling Underground Space Technol. 35, 227–234. underground opening triggered by dynamic disturbance. Tunnelling Underground
Li, X.B., Cao, W.Z., Zhou, Z.L., Zou, Y., 2014. Influence of stress path on excavation Space Technol. 25, 587–599.
unloading response. Tunnelling Underground Space Technol. 42, 237–246. Zhu, Z.M., Xie, H.P., Mohanty, B., 2008. Numerical investigation of blasting-induced
Ma, G.W., An, X.M., 2008. Numerical simulation of blasting-induced rock fractures. Int. J. damage in cylindrical rocks. Int. J. Rock Mech. Mining Sci. 45, 111–121.
Rock Mech. Mining Sci. 45, 966–975. Zukas, J.A., 1990. High Velocity Impact Dynamics. Wiley-Interscience.
Mitelman, A., Elmo, D., 2016. Analysis of tunnel support design to withstand spalling

33