Journal of Rock Mechanics and Geotechnical Engineering 14 (2022) 1761—1781

Contents lists available at ScienceDirect

Journal of Rock Mechanics and

Geotechnical Engineering
journal homepage: www.jrmge.cn

Full Length Article

Amplification effect of near-field ground motion around deep tunnels

based on finite fracturing seismic source model
Qiankuan Wang a, b, Shili Qiu a, c, *, Yao Cheng b, Shaojun Li a, c, Ping Li d, Yong Huang e,
Shirui Zhang a, c
a
State Key Laboratory of Geomechanics and Geotechnical Engineering, Institute of Rock and Soil Mechanics, Chinese Academy of Sciences, Wuhan, 430071, China
b
Faculty of Engineering, China University of Geosciences, Wuhan, 430074, China
c
University of Chinese Academy of Sciences, Beijing, 100049, China
d
South-Central University for Nationalities, Wuhan, 430074, China
e
China Railway First Survey and Design Institute Group Ltd., Xi’an, 710000, China

a R t i c L E i n F o a b s t R a c t

Article history: Dynamic failure of rock masses around deep tunnels, such as fault-slip rockburst and seismic-induced
Received 8 August 2021 collapse, can pose a significant threat to tunnel construction safety. One of the most significant factors
Received in revised form that control the accuracy of its risk assessment is the estimation of the ground motion around a tunnel
17 October 2021
caused by seismicity events. In general, the characteristic parameters of ground motion are estimated in
Accepted 2 December 2021
Available online 10 March 2022
terms of empirical scaling laws. However, these scaling laws make it difficult to accurately estimate the
near-field ground motion parameters because the roles of control factors, such as tunnel geometry,
damage zone distribution, and seismic source parameters, are not considered. For this, the finite frac-
Keywords:
Near-field ground motion
turing seismic source model (FFSSM) proposed in this study is used to simulate the near-field ground
Amplification effect motion characteristics around deep tunnels. Then, the amplification effects of ground motion caused by
Seismic waves the interaction between seismic waves and deep tunnels and corresponding control factors are studied.
Deep tunnel The control effects of four factors on the near-field ground motion amplification effect are analyzed,
Rockburst including the main seismic source wavelength, tunnel span, tunnel shape, and range of damage zones. An
empirical formula for the maximum amplification factor (am ) of the near-field ground motion around
deep tunnels is proposed, which consists of four control factors, i.e. the wavelength control factor (Fλ ),
tunnel span factor (FD ), tunnel shape factor (Fs ) and excavation damage factor (Fd ). This empirical for-
mula provides an easy approach for accurately estimating the ground motion parameters in seismicity-
prone regimes and the rock support design of deep tunnels under dynamic loads.
ⓒ 2022 Institute of Rock and Soil Mechanics, Chinese Academy of Sciences. Production and hosting by
Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/
licenses/by-nc-nd/4.0/).

1. Introduction development, and expansion in the surrounding rocks of the tunnel

will be induced, resulting in a decrease in rock strength, which can
With rapid developments of underground works, seismicity form an excavation damaged zone (EDZ) and excavation fracture
hazards induced by high in situ stresses have become a challenging zone (EFZ). This would provide favorable conditions for the
issue, which can pose a substantial threat to the safety of engi- occurrence of seismic hazards (Kaiser et al., 1996; Vazaios et al.,
neering construction. On one hand, excavation unloading can 2019; Sun et al., 2021). Under this condition, the overstressed
induce stress redistribution and associated areas characterized by rock masses around tunnels are prone to fracturing, resulting in
high in situ stresses. On the other hand, the crack initiation, seismic events. As seismic waves propagate to the boundary of the
tunnel, the ground motion increases due to site amplification ef-
fects, and the additional vibration energy can then be fed into rock
* Corresponding author. State Key Laboratory of Geomechanics and Geotechnical masses. This will directly shake loose rocks or evoke rock ejection in
Engineering, Institute of Rock and Soil Mechanics, Chinese Academy of Sciences, these high burst-prone areas, thereby triggering three types of
Wuhan, 430071, China. rockbursts (Wang and Cai, 2014, 2015). The three types of rock-
E-mail address: [email protected] (S. Qiu). bursts are bulking caused by surrounding rock fracturing, rockfall
Peer review under responsibility of Institute of Rock and Soil Mechanics, Chi-
induced by ground motion, and ejection caused by seismic energy
nese Academy of Sciences.

https://doi.org/10.1016/j.jrmge.2021.12.024
1674-7755 ⓒ 2022 Institute of Rock and Soil Mechanics, Chinese Academy of Sciences. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-
NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
1762 Q. Wang et al. / Journal of Rock Mechanics and Geotechnical Engineering 14 (2022) 1761—1781

transfer (Kaiser et al., 1996). The rockburst mechanisms induced by found that the field PPV is 4—10 times that calculated by the
the seismic event around the deep tunnel are significantly empirical scaling law. Ortlepp (1993) and Stacey and Rojas (2013)
complicated. Two factors are dominant accounting for the strong also found that the initial ejection velocity of rock blocks on tun-
ground motion parameters, e.g. the peak particle velocity (PPV) of nel boundary exceeds 10 m/s and is much higher than that esti-
seismic events and the amplification effect of the excavation mated by the empirical scaling law indicating the existence of
structure. amplification effect. For this, Kaiser et al. (1996) pointed out that
The parameter PPV is widely used in strong ground motion the increase of ejection velocity of rock blocks is derived from the
prediction, commonly as an indicator of rockburst induced by a release and transformation of stored strain energy around the
seismic event to identify the damaged zones of rock masses or rock tunnel and that the amplification factor (a, which refers to the PPV
supports and to delineate their corresponding damage influence ratio as between the excavation model and the unexcavated back-
radius (Feng et al., 2012; Morissette and Hadjigeorgiou, 2019). ground model at the same location) should be 1—4. However,
Generally, the PPV is estimated by simple empirical scaling laws Durrheim et al. (1996) attributed the amplification effect to reso-
proposed by McGarr et al. (1981), McGarr (1984), Spottiswoode nance, and Linkov (2000) believed that the amplification effect is
(1984), Hedley (1988, 1992), Kaiser et al. (1996), Durrheim et al. caused by energy release due to rock softening.
(2005) and Potvin and Wesseloo (2013), as listed in Table 1. Most For Long-Victor Mine, Mikula (2013) stated that the average
of the PPV empirical scaling laws (Nos. 1—5 in Table 1) are proposed amplification factor can be set to 3. Zhang et al. (2015) used a one-
using far-field ground motion monitoring data without considering dimensional (1D) model with a fracture zone to describe the frac-
the near-field saturation and amplification effect. As a result, they tured rock mass. The amplification factor in terms of PPV shows
may not be suitable for estimation of the near-field ground motion that the PPV of the model’s free end is 2—3.6 times that of the
effects around deep tunnels (Kaiser et al., 1996; Wang and Cai, 2014, model’s input end due to the magnification of the fracture zone.
2015). Through combining the far-field PPV scaling law with the Based on the rockburst cases in South African mines, Wesseloo
near-field saturation effect, Durrheim et al. (2005) and Potvin and (2018) proposed PPVR = aPPVB for assessing the real peak veloc-
Wesseloo (2013) proposed two empirical formulas applicable to ity (PPVR ) of the tunnel surface where PPVB is the theoretical value
both far- and near-field PPV calculations (e.g. Nos. 6—7 in Table 1). of the peak velocity without considering the interaction between
However, these two relations do not consider the dynamic frac- seismic waves and tunnels. In the absence of engineering param-
turing process of seismic events, uneven radiation of seismic waves, eters, considering the wave reflection on the tunnel surface, a is the
and dynamic effects between seismic waves and underground theoretical amplification factor usually with a value of 2.
excavation structures. There are significant differences between the actual and the
In other words, by using empirical scaling laws in Table 1, the above-mentioned amplification factors for practical cases. This is
estimated near-field PPVs and their distribution around deep tun- due to multiple factors, including the radiation patterns, complex
nels and caverns usually may not be precise. Based on the study of interactions of seismic waves with geological structures and exca-
dynamically triggered rockbursts induced by seismic events in vation structures, influence of surface waves, frequency of incident
deep tunnels, due to the existence of site effect, there are some seismic waves, and the shape and span of excavation structure
significant differences between the measured ground motions in (Kaiser et al., 1996; Tang and Xia, 2010; Diao et al., 2011; Chen et al.,
the field and the results calculated from the empirical scaling laws. 2013; Potvin and Wesseloo, 2013; Wang and Cai, 2015; Almadani
Milev et al. (1999) and Durrheim (2012) analyzed the rockburst et al., 2020; Wang et al., 2021). Based on the literature review of
damage on the tunnel surface in South African mines, and they the site effect of ground motion and the authors’ investigations, it

Table 1
A summary of the scaling laws to assess the PPV.

No. Formulation of PPV scaling laws Parameter Source Remark

1 log 10 (R PPV) = 0 57MN + 1 95 R is the hypocentral distance in m; PPV is the McGarr (1984) Based on 12 mining tremor events
peak particle velocity in mm/s; MN is the
magnitude of a seismic event with Nuttli
magnitude scale
2 log10 (R PPV) = 0 5M + 2 81 M is the magnitude of a seismic event with Spottiswoode (1984) Based on ground motion monitoring
Richter scale; other parameter symbols are the
( )-1 6 same as that of No. 1
R
3 PPV = 4000 The parameter symbols are the same as those of Hedley (1988, 1992) Based on monitoring data obtained from
10M/3
Nos.1 and 2 underground mines in Sudbury and Elliot
Lake, Ontario, Canada
CM0a
4 PPV = M0 is the seismic moment in GN m; a and C are Kaiser et al. (1996) To facilitate dynamic underground support
R
empirical constants that should be calibrated design
for each site, a = 0 5
CM aML
5 PPV = ML is the local magnitude of the seismic event; a McGarr (1984); McGarr (1984) proposed this scaling law,
R
and C are empirical constants that should be Kaiser et al. (1996) Kaiser et al. (1996) recommended that for
calibrated for each site pre-feasibility type studies that a = 0 5 and
C = 0.1 or 0.25 m2/s (which correspond to
50% or 90—95% confidences, respectively)
6 ■ Vs △σ Vs is shear wave velocity; △σ is static stress Durrheim et al. (2005) The original work of McGarr (1984) showed
|
■ G (R ≤ r0 )
drop; G is shear stiffness of rock mass; r0 is the a saturation of the ground motion in the
PPV =
■ Vs △σ R
| near-field radius near-field
(R＞r0 )
G r0
7 C100 5(ML +1 5) C = 0 2—0 3 is recommended for design Potvin and Based on the original far-field relationship
PPV = purposes; R0 is the source radius; a = Wesseloo (2013) presented by Kaiser et al. (1996) and the
R + R0
0 53—1 14 original work of Durrheim et al. (2005)
1
(ML + 1 5)
R0 = a103
 Q. Wang et al. / Journal of Rock Mechanics and Geotechnical Engineering 14 (2022) 1761—1781 1763

seems that the amplification effect is affected mainly by structural engineering field because associated theory and methods were
factors and seismic source factors. The influence of seismic sources insufficient.
is primarily manifested by the wavelength (St John and Zahrah, Given this, this study introduced a recently proposed finite
1987; Hashash et al., 2001; Wang and Cai, 2014, 2015), while fracturing seismic source model (FFSSM) to characterize the radi-
structural factors mainly include three factors: tunnel span, tunnel ation seismic wave field evoked by dynamic fracturing processes of
shape and distribution of damage zones (Kaiser et al., 1996; Milev finite dimension sources around deep tunnels. Based on the radi-
and Spottiswood, 2005; Wang and Cai, 2014, 2015; Zhang et al., ation seismic wave field, the amplification effect of seismic waves,
2015). For this, the study focuses on the control effect of four fac- caused by its interaction with the underground excavation struc-
tors (i.e. wavelength, tunnel span, tunnel shape, and range of ture, was simulated by the spectral element method (SEM) soft-
damage zones) on the amplification effect and distribution char- ware Specfem2D. The near tunnel’s strong ground motion
acteristics of ground motion. distribution law was also analyzed and revealed. The emphasis
The strong ground motion in rocks around deep tunnels occurs focused on the influence of four control factors on the amplification
usually near a fracture source, or even within the fracture source effect of near-field ground motion. Finally, an empirical formula
region. Hence, a near-field finite fracturing seismic source should was established to estimate the amplification factor of near-field
be introduced to analyze the strong ground motion characteristics ground motion around deep tunnels.
around the tunnel. However, it is often assumed in the empirical
scaling laws (see Table 1) that the seismic sources are far-field point 2. A brief introduction to FFSSM and Specfem2D
sources with spherical radiation patterns. Only the geometry and
material attenuation of seismic waves are analyzed (i.e. the PPV is a 2.1. The FFSSM and parameters calculation procedures
function of the seismic source distance). In other words, the site
effects such as the amplification and shielding of the excavation When analyzing the ground motion of far-field seismic events,
structure, which are created by reflection, refraction, and diffrac- the effect of source size on the seismic response between seismic
tion of seismic waves around the excavation structure, are ignored waves and tunnels is almost negligible, so a point source model can
(Milev et al., 2001; Durrheim, 2012). The assumptions are incorrect be used to characterize the seismic source. However, when simu-
in the near-field ground motion analysis. The distance between the lating wave propagation induced by near-field seismic events, the
concentrated seismic events and the tunnel structure is usually 2—3 seismic source must be expressed as a finite dimension seismic
times the seismic source size. The seismic source size of large- source. The size of the seismic source, dynamic fracturing pro-
magnitude seismic events is generally relatively large, even com- cesses, and wave interactions must be considered. For this, the
parable to the above-mentioned distance (Hudyma et al., 2003; FFSSM was developed, which references the point seismic source
Mutke, 2008; Zhang et al., 2012; Ma et al., 2015). It means that model and the finite fault model (Beresnev and Atkinson, 1997; Aki
when analyzing near-field ground motions induced by near-tunnel and Richards, 2002; Li et al., 2014). Considering the dynamic frac-
seismic events, the controlling role of source sizes should be turing process of the seismic source, this model can effectively
considered, rather than simply using a point source model to es- characterize the seismic wave field characteristics of the near-field
timate ground motion parameters. For this, the source size effect seismic sources.
should be considered and a finite-size fracture source should be According to seismic wave propagation, the observable seismic
used to characterize the focal mechanism. Furthermore, the dy- energy of the pressure-shear vertical wave system (P-SV system) is
namic fracturing process of a finite-size seismic source, the prop- much greater than that of the pressure-shear horizontal wave
agation law of seismic waves excited by a finite-size seismic source system (P-SH system) (Hashash et al., 2001). Thus, a two-
in the near-field area, and the amplification effect of ground motion dimensional (2D) plane approximation of the P-SV system can be
in the excavation structure area should be considered. However, in used to analyze the seismic wave propagation in deep tunnels.
previous studies, the finite fracturing seismic source was rarely Fig. 1a is the basic principle of a 2D FFSSM: the dynamic fracturing
used in near-field ground motion analysis in the underground process of the main seismic source is characterized by dividing a

Fig. 1. A conceptual diagram of the FFSSM: (a) A conceptual diagram for the FFSSM—2D with the characteristics of central epicenter excitation and bidirectional rupture propagation
in P—SV system; and (b) A conceptual diagram for the FFSSM—3D in the X—Y—Z coordinate system. 4s , gr and δd represents the strike, rake, and dip of the fault, respectively. R1 and
R2 are two receivers at a certain distance away from the FFSSM. Ux , Uy , Uz represent the pressure (P), shear horizontal (SH), and shear vertical (SV) component of the displacement at
the position of the receivers.
1764 Q. Wang et al. / Journal of Rock Mechanics and Geotechnical Engineering 14 (2022) 1761—1781

fault or fracture (main seismic source, the length is L0 ) into a finite parameter research, setting different seismic cases to study the
number of seismic sub-sources (see SS #1—SS #5 in Fig. 1a) with the dynamic wave interaction between seismic waves and under-
same length (Le ), different dynamic corner frequencies (f0 ) and ground excavation structures (including the tunnel and the damage
time shifts (ti ). ti represents the moment of excitation of the ith zones formed during the excavation of deep tunnels). Thus, Path #1
seismic sub-source. The number of seismic sub-sources (n) should in Fig. 2 is followed.
be determined according to the size of the main seismic source.
The FFSSM has two special modes: bilateral fracturing and 2.2. SEM and Specfem2D
delayed fracturing modes. The bilateral fracturing mode means that
the fracture propagates from the initial fracturing seismic sub- Many numerical methods can be used to simulate the propa-
source (SS #1 in Fig. 1a) to the follow-up activated seismic sub- gation of seismic waves, including the SEM, finite element method
sources (SS #2—SS #5 in Fig. 1a) on both sides simultaneously. In (FEM), finite difference method (FDM), and pseudo-spectral
other words, the fracturing sequence is SS #1/SS #2—#3/SS #4— method (PSM). The SEM proposed by Patera (1984) was first used
SS #5. The red and orange arrows in Fig. 1a show the main direction in fluid mechanics calculations, and then gradually applied to the
of fracturing propagation of the seismic sub-source. The delayed numerical simulation of seismic wave propagation (Seriani et al.,
fracturing mode means that the seismic sub-sources fracture 1995; Faccioli et al., 1997; Komatitsch, 1997). The SEM combines
sequentially, with different time shifts according to the fracturing the ideas of the FEM and PSM and has the advantages of both
sequence. The initially activated sub-source (SS #1, with crack methods, that is, the flexibility of FEM and the accuracy of PSM.
initiation time t1 = 0 s) fractures firstly and starts to radiate seismic This paper uses an advanced numerical code, Specfem2D, to
waves. As the fracture spreads towards other seismic sub-sources, explore the dynamic wave interaction between seismic waves and
these follow-up seismic sub-sources are activated one after deep tunnels. The Specfem2D software package based on the SEM
another (For example, SS #2—#5 are excited at t2 —t5 ). After all the has two main features, i.e. the tensor-product Lagrange inter-
seismic sub-sources are activated, the dynamic fracturing process is polants and Gauss-Lobatto quadrature (Komatitsch and Tromp,
completed. In particular, it is noted that the FFSSM utilizes a 1999, 2003). Specfem2D is a powerful tool for simulating forward
moment tensor point source model to quantify the seismic source and adjoint seismic wave propagation in 2D acoustic, (an)elastic,
characteristics and seismic wave radiation patterns of a seismic poroelastic, or coupled acoustic-(an)elastic-poroelastic media, with
sub-source. convolution perfectly matched layer (PML) absorbing layers as well
Although only a 2D FFSSM (FFSSM-2D) is used in this paper, the as higher-order time schemes (Basabe and Sen, 2007). It has been
FFSSM can naturally be extended to a three-dimensional (3D) proven that Specfem2D has high accuracy and convergence prop-
FFSSM (FFSSM-3D), as shown in Fig. 1b. The fracture surface of the erties for simulating seismic wave propagation and can accurately
main source is divided into many rectangular source sub-fracture capture the ground motion around tunnels (Seriani and Oliveira,
surfaces of the same size. The red seismic sub-source fracture 2008; Lin et al., 2014; Wang and Cai, 2015; Pan et al., 2020;
surface, where the dark red hexagram is located, is the fracturing Rohnacher et al., 2021).
initiation site, and the fracturing start time is t1 . The orange and
yellow arrows indicate the fracturing direction of seismic source 3. Simulation schemes and numerical models
fracture; that is, the fracturing sequence of the rectangular seismic
sub-sources is from red rectangular sub-source (SS #1) to orange 3.1. Numerical model configuration and layout of receivers
rectangular sub-sources (SS #2—#9) and then to yellow rectangular
sub-sources (SS #10—SS #21), as shown in Fig. 1b. The activating The numerical model of the P—SV system constructed in this
and fracturing of these rectangular sub-sources are used to simu- paper is shown in Fig. 3a. The size of the computation domain is
late the dynamic fracturing process of the fracture surface of the 104 m × 104 m, the tunnel is located in the center of the model
main seismic source. Compared with the FFSSM-2D, the FFSSM-3D (X = 50 m, Z = -50 m). The four edges of 2 m thickness are set as
has many advantages: (1) The 3D model can study more complex the PML absorbing layers to eliminate the unreal seismic wave
geometric conditions of the seismic source, i.e. a spatial relation- reflection. This study uses the engineering background of the
ship between the seismic source and the tunnel; (2) More complex headrace tunnel #2 at Jinping II Hydropower Station. The rock mass
geological conditions such as faults, joints, soft and hard rock in the numerical model is homogeneous Jinping marble (Zhang
contact surfaces can be characterized so that the control factors of et al., 2012; Qiu et al., 2014). The Young’s modulus, density, and
site effect of ground motion can be analyzed; (3) The 3D model Poisson’s ratio of Jinping marble are 56 GPa, 2780 kg/m3 and 0.27,
research object is no longer just an excavation section, but a tunnel respectively.
or cavern group with a certain length. It is more convenient to Three sets of receivers are placed in the model to capture the
assess the ground motion risk along the tunnel. However, the PPV distribution around the tunnel, covering the entire simulation
FFSSM-3D also has its limitations. Most importantly, the increase in area around the tunnel (Fig. 3b). The Set #A receivers are arranged
the number of computational elements and the complexity of in a 25 m × 25 m square region centered on the tunnel. The interval
geological conditions and spatial relationships of the 3D model will between two adjacent receivers is 0.5 m; The Set #B receivers are
inevitably lead to a substantial increase in modeling period and arranged on the tunnel surface. Since the ground motion on the
calculation time. Moreover, for multi-factor study, the FFSSM-3D tunnel surface is relatively strong, the interval between two adja-
model modification also is time-consuming and laborious. This cent receivers in Set #B is 0.25 m. The Set #C receivers are arranged
paper adopts a FFSSM-2D model considering time cost and in a 100 m × 100 m square region surrounding the Set #A receivers,
research requirements. with an interval of 1 m. Since the simulated tunnels in different
The parameters to be determined in the FFSSM-2D include the schemes have varied shapes and sizes, the location and number of
seismic sub-source length (Le ), the number of seismic sub-sources the three-receiver sets in different schemes are slightly modified.
(n), the sub-seismic moment (Me ), the dynamic corner frequency
(f0 ), and the time shift (ti ). There are two different calculation 3.2. Simulation schemes
procedures for the parameters of the FFSSM that are selected ac-
cording to the different research contents and interpretation This paper focuses on the control effect of four factors, i.e. the
methods (Fig. 2). This paper focuses on the forward simulation and main seismic source wavelength (λs ), the tunnel span (D), the
 Q. Wang et al. / Journal of Rock Mechanics and Geotechnical Engineering 14 (2022) 1761—1781 1765

Fig. 2. Calculation flowchart of model parameters for the FFSSM. Ae represents the area of the fracture surface of the seismic sub-source (Ae = Le 2 in the FFSSM); MW and MW e

represent the moment magnitudes of the main seismic source and a seismic sub-source, respectively; Vf denotes the seismic source fracture velocity, which is usually taken as 0.6—
0.85 times of the shear wave velocity (Vf = 0 8Vs in the FFSSM); NR represents the number of seismic sub-sources that have ruptured; Di denotes the distance between the ith
seismic sub-source (SS #i) and the initially activated sub-source (SS #1), and Di = ni Le ; s(t) is the standard definition of the second derivative of Gaussian of Ricker wavelet;
m(X s , t) and mij (X s , t) represents the moment density tensor of the main seismic source and a seismic sub-source, respectively; M ij denotes moment tensor component of the
seismic sub-source; δ(X -X s ) denotes the Dirac Delta distribution function of the seismic sub-source; w, u, and σ represent an arbitrary test vector, the displacement vector the
stress tensor, respectively; S represents the fracturing surface of seismic source; n represents the unit outward normal to the boundary (Γ); Ω represents an elastic medium; ρ
represents the density. These formulas are derived from Berkhout (1987), Faccioli et al. (1997), Beresnev and Atkinson (1999), Somerville et al. (1999), Aki and Richards (2002),
Chapman (2004), Motazedian and Atkinson (2005), Carcione (2007), Chaljub et al. (2007), Potvin et al. (2010), and Fichtner (2011).

tunnel shape, and the damage zone (including the EDZ and EFZ) subsequent Schemes #2—#5. Second, the excavation models are
range on the amplification effect of seismic waves around deep simulated for seismic wave propagation, as shown in Schemes #2—
tunnels. A total of five simulation schemes were designed in the #5 in Table 2. Finally, a comparative study of the seismic wave field
study, as shown in Table 2. First, we performed the simulations of characteristics of the excavation model with its corresponding
seismic wave propagation excited by the near-field finite seismic background model is carried out to explore the amplification effect
source. A model without tunnel excavation (the background model) of seismic wave field around deep tunnels in consideration of the
is adopted. Its simulation results are considered background data amplification factor (a).
for the subsequent ground motion parameter analysis, as shown in The seismic source is located in the surrounding rocks at the
Scheme #1 in Table 2. Except for non-excavation scheme, each excavation’s lower-left corner and is simulated with the FFSSM. The
background model’s parameter settings are consistent with the seismic sub-sources are set to 3, marked as 1, 2, and 3, respectively.
1766 Q. Wang et al. / Journal of Rock Mechanics and Geotechnical Engineering 14 (2022) 1761—1781

Fig. 3. Numerical model configuration and layout of receivers: (a) Numerical model configuration with spectral element meshes and PML absorbing layers; and (b) Layout of three
sets of receivers in simulation region or around the tunnel.

Seismic sub-source 1 is located between 2 and 3, as shown in Fig. 4. Scheme #4: The seismic source parameters and the tunnel span
The fracturing direction of the seismic sub-source is specified as a remain unchanged, and only the tunnel shape is changed. Scheme
bidirectional fracturing mode, i.e. the fracture propagates from #4 includes two sets of cases (Set 1 and 2), and each set includes
seismic sub-source 1 to 2 and 3 simultaneously. The seismic sub- four cases (see Fig. 4c, Table 2). The cases in the same set have the
source 1 is activated firstly, and then the follow-up seismic sub- same seismic source parameters (4 m and 6 m for Le of Set 1 and Set
sources 2 and 3 are activated with a delay time of ti . The simu- 2, respectively; see Table 3 for cases Nos. 6 and 7) and tunnel span
lated main seismic source is a normal fault. The strike 4s , rake gr (D = 12 m), but have different tunnel shapes (circular, gate arch-
and dip δd of the seismic sub-sources are -90° , -90° , and 45° , shaped, rectangular, and arched tunnels for cases Nos. 1—4,
respectively. The Ricker time wavelet is used as the time function of respectively). Different tunnel shapes lead to different directions of
seismic sub-sources. The stress drop of the seismic sub-source is set reflection and diffraction of seismic waves, changing the charac-
to 5 MPa. Table 2 lists the study content and model configurations teristics of the seismic wave field and thus resulting in variations in
for each numerical scheme and the specific schemes are described the magnification and distribution of the amplification factor (a).
below. Scheme #5: The seismic source and tunnel parameters are fixed
Scheme #2: The tunnel is fixed as a gate arch-shaped tunnel in this scheme, and only the range of damage zones is changed. In
with a tunnel span D = 12 m. The main seismic source wavelength the simulation, the surrounding rocks are subdivided into the EFZ,
is changed by controlling the length of the seismic sub-source EDZ, excavation influence zone (EIZ), and original rock zone (ORZ).
(Fig. 4a). The seismic source is located in the rock masses below Numerous studies show that the excavation damage in over-
the lower-left corner of the tunnel. Seismic sub-source 1 is located stressed rock masses decreases sequentially from EFZ to ORZ until it
at the point of X = 30 m, Z = -70 m and 28.28 m from the center of disappears (Siren et al., 2015; Bao et al., 2020). The corresponding
the tunnel. rock wave velocity gradually increases to the original rock wave
The coordinates of seismic sub-sources 2 and 3 can be deter- velocity. The elastic wave velocities are assigned in each zone to
mined by the seismic sub-source length (Le ). Scheme #2 includes 11 simplify the simulation. In the EFZ and EDZ, the elastic wave ve-
cases with different Le values (Le = 0.5 m, 0.75 m, 1 m, 1.5 m, 2 m, locities of rock masses are taken as 0.4 and 0.7 times of the elastic
4 m, 6 m, 9 m, 12 m, 15 m, and 25 m). Table 3 lists the source pa- wave velocity of original rock masses in the ORZ, respectively. In
rameters relating to each case. Each parameter is determined ac- particular, the elastic wave velocity of rock masses in the EIZ
cording to Path #1 (Fig. 2). matches with the ORZ. According to the acoustic test results con-
Scheme #3: Keeping the source parameters constant ducted in headrace tunnel #2 at Jinping II Hydropower Station, the
(Le = 1.5 m, same as Case 4 in Scheme #2 in Table 3). The wave- rock masses in the ORZ, EIZ, EDZ, and EFZ are given different elastic
length of the S-wave is therefore fixed at 3 m. The cross-sectional wave velocities listed in Table 4.
shape of the tunnel remains circular, and its spans in 7 cases in Moreover, as the degree of damage increases, the range of the
Scheme #3 are set at 1.5 m, 3 m, 6 m, 10 m, 12 m, 16 m, and 20 m, damage zones increases, as shown in Fig. 4d. Scheme #5 includes
respectively (see Fig. 4b, Table 2). The change in tunnel span (D), two sets (Set 1 and Set 2), each including four cases (as shown in
implies a change in the interaction range between seismic waves Fig. 4d and listed in Table 2). The cases in the same set have the
and the tunnel section, leading to changes in reflection, diffraction, same seismic source parameters (4 m and 6 m for Le of Set 1 and Set
and refraction of seismic waves, ultimately affecting the site effect 2, respectively), tunnel span of D = 11.67 m and different EDZ and
of ground motion. EFZ ranges. The ratio parameter k is used to control the regions of
 Q. Wang et al. / Journal of Rock Mechanics and Geotechnical Engineering 14 (2022) 1761—1781 1767

Fig. 4. Layouts of seismic sub-sources in the proposed FFSSM for different parametric analysis schemes: (a) Scheme #2 (Cases 5—9); (b) Scheme #3 (Cases 2—6); (c) Scheme #4; and
(d) Scheme #5. Not all cases are drawn in the figure due to the limited size. However, only some typical cases in Table 2 are selected to illustrate the changing law of the parameters
of different schemes.

EDZ and EFZ. In this scheme, k is set to 0, 0.5, 0.8, and 1, respectively, wave induced ground motion is much smaller (Hashash et al., 2001;
as shown in Fig. 4d. Aki and Richards, 2002; Wang and Cai, 2015).
Fig. 5 shows snapshots of the vertical velocity components in
4. Model results the seismic wave field for seismic sub-source lengths of 0.5 m,
1.5 m, and 4 m at three-time points. The moment tA represents the
4.1. Influence of the main seismic source wavelength time when the leading edge of the S-wave first reaches the left foot
of the tunnel. tB is the moment when the S-wave fully interacts
Schemes #1 and #2 in Table 2 were implemented to explore the with the tunnel. tC is the time when the S-wave passes through the
effect of the main seismic source wavelength on the characteristics right arch of the tunnel. It should be noted that in Fig. 5, the specific
of the seismic wave field and the distribution of strong ground values of the above three moments are different, which is caused by
motions, which can be formed by the dynamic interaction of the propagation change of the induced seismic waves due to the
seismic waves with tunnels. This paper focuses only on the inter- difference in the wavelength of the seismic sub-source (Le ) in the
action between S-waves with tunnels, mainly because S-waves FFSSM. It can be seen from Fig. 5 that (1) As the λs in the FFSSM
transmit more energy and can cause a more significant ground increases (Table 3), the interaction times (tC - tA ) between S-waves
motion response in the surrounding rocks of tunnels, while the P- and tunnel increase accordingly. When Le is 0.5 m, 1.5 m, and 4 m,
1768 Q. Wang et al. / Journal of Rock Mechanics and Geotechnical Engineering 14 (2022) 1761—1781

Table 2
Simulation schemes for exploring influencing factors of the near-field ground motion amplification effect around deep tunnels based on the FFSSM.

Scheme No. Is the excavation Number of cases Seismic source parameters Figure of
carried out? numerical
Le (m) D (m) Tunnel shape k
model

#1 No Consistent with Consistent with Consistent with Consistent with Scheme #2—5 0 Fig. 4
Scheme #2—#5 Scheme #2—#5 Scheme #2—#5
#2 Yes 11 0.5, 0.75, 1, 1.5, 2, 4, 6, 9, 12 Gate arch-shaped tunnel 0 Fig. 4a
12, 15, 25
#3 Yes 7 1.5 1.5, 3, 6, 10, 12, 16, 20 Circular tunnel 0 Fig. 4b
#4 Yes 8 4 (Set 1), 6 (Set 2) 12 Circular, gate arch-shaped, 0 Fig. 4c
rectangular, arched
#5 Yes 8 4 (Set 1), 6 (Set 2) 11.67 arched 0, 0.5, 0.8, 1 Fig. 4d

Notes: k refers to the ratio of the modeled EDZ and EFZ of the tunnel to the actual EDZ and EFZ (Case 4 in Fig. 4d, taken from the Jinping II Hydropower Station, based on the
actual range of the EDZ and EFZ at 11 + 205 in the diversion tunnel #2, where a very strong rockburst event occurred).

Table 3 distribution of strong ground motions. In the background model,

List of seismic sub-source parameters of different seismic sub-source lengths used in the maximum seismic wave response region is mainly along the
Scheme #2.
two directions shown by the white dashed lines in Fig. 5, i.e. the
Case Le (m) Me Dynamic Dynamic Time shift Seismic λs (m) two directions with counterclockwise intersection angles of 45°
No. (MN m) corner corner of seismic wave and 135° with the X-axis. The propagation direction of the FFSSM-
frequency frequency sub- dominant
induced seismic waves changed due to the obstruction from the
of seismic of seismic sub- sources 2 frequency
sub- sources 2 and and 3, t02,3 of the main tunnel, which results in the fact that the area of the maximum wave
source 3, f02,3 (Hz) (10-4 s) seismic front in the background model is weakened in the excavation
1, f01 (Hz) source, f0m model to form a wave weakened area (green dashed oval in Fig. 5).
(Hz)
An enhanced area is formed on either side of the weakened wave
1 0.5 0.625 3,18 2205 2.24 2659 1.05 area.
2 0.75 2.11 2,12 1470 3.36 1790 1.56 Fig. 6 shows the PPV contours and amplification factor distri-
3 1 5 1,59 1102 4.48 1383 2.02
butions for Le = 0.5 m, 1.5 m, 4 m, 6 m, and 9 m. Fig. 6a and b shows
4 1.5 16.9 1,06 735 6.72 889 3.04
5 2 40 795 551 8.96 681 4.1 that PPVs increase with increasing Le for both the background
6 4 320 397 276 17.91 333 8.38 model and the excavation model. The zones with high-PPV values
7 6 1,08 265 184 26.87 222 12.57 in the background model are mainly distributed along the line
8 9 3,65 177 122 40.31 136 20.52
connecting the center of the tunnel and the center of the FFSSM
9 12 8,64 132 92 53.74 116 24.06
10 15 16,900 106 73 67.18 93 30.01 seismic source. The PPVs in the background model also follow the
11 25 78,100 64 44 111.97 62 45.02 pattern of decay with the epicenter distance (i.e. a distance to the
Notes: (1) f0m is a seismic wave dominant frequency of a main seismic source con-
center of the FFSSM seismic source), which show the maximum at
sisting of three seismic sub-sources. The seismic wave dominant frequency of a the FFSSM seismic source and decrease gradually with the increase
main seismic source is calculated by fast Fourier transformation and spectrum of the epicenter distance. In the excavation model, high-PPV zones
analysis of the seismic wave time-domain curve recorded by the receiver deployed (Zone-A and Zone-B) are formed due to the generation and prop-
at the bottom left corner of the tunnel (X = 45 m, Z = -55 m), where is the earliest
agation of reflected S-waves and diffracted S-waves. Zone-A is
exposure to seismic waves. (2) The main seismic source wavelength, λs , is obtained
from f0m with the relation of λs = Vs /f0m , where Vs is the S-wave velocity of rock located within a certain depth range of the surrounding rock on the
masses. tunnel’s left sidewall and left floor. It is formed by the superposition
of the reflected S-waves from the initial fracturing seismic sub-
source 1 and the incident S-waves of the follow-up activated
Table 4 seismic sub-sources 2 and 3. As shown in Fig. 6b, the extent of
Parameters of different surrounding rock zones for Scheme #5. Zone-A increases with the increase of Le and gradually expands
from the left sidewall and floor area to the left arch shoulder and
Rock mass Density of rock mass, Pressure wave Shear wave velocity,
zone ρ (kg/m3) velocity, Vp (m/s) Vs (m/s) right floor area. Zone-B appears far away from the left wall and floor
of the tunnel, which is formed by the joint superposition of re-
ORZ and EIZ 2780 4972 2791
EDZ 2780 3480 1954
flected S-wave, incident S-wave and diffracted S-wave from the
EFZ 2780 1989 1116 FFSSM seismic sub-sources. Additionally, the shielding effect of the
tunnel and diffraction loss of seismic waves lead to the formation of
a low-PPV zone, Zone-C, which locates on the right arch shoulder
the interaction times are 5.2 ms, 8.8 ms, and 14.4 ms, respectively. side of the tunnel, and a striped high-PPV zone, Zone-D, which is
(2) The increase in wavelength increases the delay time of seismic formed on both edges of Zone-C due to diffraction waves. As the
waves passing through the tunnel. However, more significantly, the FFSSM seismic sub-source length (Le ) increases, the extent of Zone-
seismic wave field formed by the interaction of seismic waves with C increases gradually, and Zone-B intermingles gradually with
the tunnel is more complex, as shown in Fig. 5c. The range of their Zone-D. Moreover, Fig. 6c and d shows the effect of the main
interaction is large. (3) When the maximum wave front of seismic seismic source wavelength on the distribution of PPV amplification
S-wave propagates to the tunnel, the reflected wave of S-wave is factor. Two types of PPV characteristic zones are formed: the PPV
generated on the left arch shoulder side and the right bottom amplified zone (a > 1, as shown in Fig. 6c and d for A1, A2 and A3,
corner side. It acts jointly with the original incident S-wave and its where A1, A2, and A3 are PPV amplified areas), and the PPV
diffracted wave to form a seismic wave superimposed area (white weakened zone (a < 1), which reflects the blocking and shielding
dashed ovals in Fig. 5), where the ground motion is superimposed effect of the tunnel on ground motion. These two PPV characteristic
and enhanced. This is quite different from the background model’s zones correspond well to the high- and low-PPV zones in Fig. 6b. As
 Q. Wang et al. / Journal of Rock Mechanics and Geotechnical Engineering 14 (2022) 1761—1781 1769

Fig. 5. Vertical velocity fields for FFSSM with different seismic sub-source lengths recorded at three-time points: (a) Le = 0.5 m; (b) Le = 1.5 m; and (c) Le = 4 m. The colors of the
seismic waves represent the positive or negative of the vertical velocity components, red represents positive while blue represents negative; the white dashed ovals and green
dashed ovals represent wave superimposed areas and wave weakened areas, respectively.

the seismic sub-source length (Le ) increases, zone A1 shifts grad- changes the excitation time of the FFSSM seismic sub-source. It
ually from the left sidewall and left floor corner areas of the tunnel causes the phase difference between the seismic sub-source waves,
to the left arch shoulder and right floor corner areas. Meanwhile, thus increasing the interaction time between the seismic waves
zones A2 and A3 also approach gradually each other and finally and the tunnel (Table 3), and changing the multi-wave super-
merge into a PPV amplification zone. This is caused by the change of position effect.
the maximum wavefront coverage area of seismic waves and their The maximum value of PPV, PPVm , is one of the most important
interaction region with the tunnel, accompanied with the change of parameters. For this, the relationship between the PPVm , the
the interaction of multiple waves (i.e. incident S-wave, reflected S- maximum amplification factor (am ) and the main seismic source
wave, and diffracted S-wave of the FFSSM seismic sub-sources). The wavelength (λs ) in the near-tunnel area are given in Fig. 7a and b. It
former is due to the increase in the seismic sub-source length, is worth noting that the main focus here is on the ground motion
which increases the length of the main seismic source. The latter is characteristics of the rock masses in a certain depth range around
caused by the increase in the wavelength of the main seismic the tunnels. Firstly, most seismic hazards often occur within the
source. The increase in the main seismic source wavelength shallow rock masses near the tunnel boundary (Feng et al., 2012;
1770 Q. Wang et al. / Journal of Rock Mechanics and Geotechnical Engineering 14 (2022) 1761—1781

Fig. 6. PPV contours and amplification factor distributions for Le = 0.5 m, 1.5 m, 4 m, 6 m and 9 m: (a) PPV contours in the background model; (b) PPV contours in the excavation
model; (c) Amplification factor distributions around the tunnel (25 m × 25 m); and (d) Amplification factor distributions for the entire computation domain (100 m × 100 m). Zone-
A, Zone-B, and Zone-D are the high-value zones of PPV. Zone-C is the low-value zone of PPV. A1, A2, and A3 are PPV amplified zones.

He et al., 2018). Secondly, as revealed from Fig. 6, the PPV and with zones A2 and A3, the ground motion in zone A1 is much
amplification factor of the near-tunnel zone A1 are high. Compared stronger. As shown in Fig. 7a, as λs increases from 1.05 m to 45.12 m,
 Q. Wang et al. / Journal of Rock Mechanics and Geotechnical Engineering 14 (2022) 1761—1781 1771

despite the constant seismic source parameters. As D increases, the

time for the seismic wave to reach the lower-left boundary of the
tunnel is earlier, and the time to leave the upper-right boundary of
the tunnel is delayed, which means that the time difference (tC -
tA ) for the S-waves to interact with the tunnel structure increases.
The larger D, the shorter the time for the seismic S-wave to reach
the lower-left boundary of the tunnel, and the later it leaves the
upper-right boundary of the tunnel. It means that the time differ-
ence of S-wave interacts with the tunnel (denoted by tC - tA in
Fig. 8) increases with the increase of tunnel span. From Fig. 8, when
D is 6 m, 10 m, and 16 m, tC - tA are 4.8 ms, 6.4 ms, and 8.4 ms,
respectively. Moreover, as D increases, the seismic wave field
around the tunnel also has become more complex, such as the
appearance of stronger scattered and diffracted waves.
For D = 1.5 m, 3 m, 6 m, 10 m, and 16 m, the distributions of PPV
and amplification factor of ground motion around the tunnel are
illustrated in Fig. 9. The corresponding PPV in the background
model can be found in Fig. 6a (Le = 1.5 m) because the FFSSM source
parameters used in Figs. 6a and 9 are the same. In Fig. 9a, three
high-PPV zones, shown as Zones-A, B, and D, and one low-PPV
zone, Zone-C, could be found. Moreover, the PPV amplification
zones are marked as A1, A2, and A3, as demonstrated in Fig. 9b and
c. Similarly, a PPV weakened zone can be found in the tunnel upper-
right area. For the case of Le = 1.5 shown in Fig. 9a, analysis of the
PPV values for different tunnel spans shows that as D increases, the
PPV values also increase accordingly. For the PPVm illustrated in
Fig. 10a, D increases from 6 m to 16 m, and the corresponding PPVm
increases from 0.65 m/s to 0.95 m/s. The PPVs in the excavation
model increase significantly relative to those in the background
model of Fig. 6a, as indicated in Fig. 10a, and the site effect of the
tunnel is the major reason for the seismic wave amplification effect.
In Fig. 10a, when D increases from 2 m to 20 m, the PPVm in the
background model increases by approximately 20%, while the PPVm
in the excavation model increases by approximately 50%. The
amplification factor in Fig. 10b gives a more visual indication of the
amplitude, positively correlated with the increase of D. The am
increases from 1.61 to 1.97 when D increases from 2 m to 16 m.
However, as D increases, the am shows a trend of first increasing
rapidly and then leveling off, as shown in Fig. 10b. Additionally, for
the PPV weakened zone shown in Fig. 9b and c, it can be found that
Fig. 7. The relationship between the near-field ground motion in the near-tunnel area the weakened zone increases significantly with increasing tunnel
and the main seismic source wavelength: (a) The variation of the maximum PPV with
span because a larger tunnel span can lead to a longer tunnel
the main seismic source wavelength; and (b) The variation of the maximum amplifi-
cation factors with the main seismic source wavelength.
boundary to impede seismic wave propagation.

4.3. Influence of tunnel shape

PPVm increases from 0.16 cm/s to 30.72 cm/s in the excavation
The circular tunnel and the seismic source parameters used in
model and from 0.14 cm/s to 21.15 cm/s in the corresponding
Section 4.2 result in symmetry of the PPV distribution field. The
background model. In Fig. 7b, it can be seen that am can reach
symmetry axis of PPV distribution is the white dashed line shown
around 2.13. Regularly, as λs increases, the a shows a trend of first
in Fig. 8, which is a special scenario. However, by comparing Figs. 6
increasing and then decreasing. Especially, the PPV amplification
and 9, it can be observed that the symmetry of the PPV distribution
effect is the strongest at a specific S-wave wavelength, such as λs =
disappears for the gate arch-shaped tunnel. Thus, the tunnel shape
20 m in Fig. 7b.
controls the seismic wave field and the near-field ground motion
amplification effect. To explore the influence of tunnel shape on the
4.2. Influence of the tunnel span near-field ground motion amplification effect, Scheme #1 and
Scheme #4 in Table 2 are implemented.
This section focuses on the effect of the tunnel span (D) on the Fig. 11 shows snapshots of the vertical velocity component at
ground motion at a fixed wavelength (Le = 1.5 m, λs = 3 m). Seven 15.2 ms of four cases with different tunnel shapes in Set 1 (Le = 4 m)
tunnel spans (D = 1.5 m, 3 m, 6 m, 10 m, 12 m, 16 m, and 20 m) are of Scheme #4. Under the conditions of the same FFSSM seismic
chosen to perform the simulations. Details of the scheme are dis- source location, the variability of the four tunnel shapes slightly
cussed in Scheme #3 in Section 3.2. changes the distribution pattern of the seismic wave field. This is
For tunnel spans of 6 m, 10 m, and 16 m, Fig. 8 illustrates the mainly reflected in the slight difference in the interaction processes
distribution of the vertical velocity components in the seismic wave between the seismic waves and the tunnel boundary, making the
field at four—time points. It can be seen that the change in the composite seismic wave field among the incident wave, diffracted
tunnel span can induce significantly different seismic wave fields wave, and reflected wave of the seismic wave different. However,
1772 Q. Wang et al. / Journal of Rock Mechanics and Geotechnical Engineering 14 (2022) 1761—1781

Fig. 8. Vertical velocity fields for the circular tunnel with different tunnel spans recorded at three-time points: (a) D = 6 m; (b) D = 10 m; and (c) D = 16 m. The colors of the seismic
waves represent the positive or negative of the vertical velocity components, red represents positive while blue represents negative; the white dashed ovals, and green dashed ovals
represent wave superimposed areas and wave weakened areas, respectively.

Fig. 11 shows that the four tunnel shapes have a limited degree of the gate arch-shaped tunnels are also approximately symmetric.
alteration to the seismic wave field. Therefore, we further analyze Conversely, arched tunnels have some variability in zones A1 and
the PPV and its amplification factor distribution under four tunnel A2 + A3. The am of zone A1 on the left floor of the arched tunnel is
shapes, as shown in Fig. 12. Different tunnel shapes possess roughly 1.7, while that of the left wall is 1.6. However, in high-PPV regions,
the same PPV range (0.6—3.4 cm/s) and similar distribution pat- am is reached to 4.2 for the right-lower region and 2.4 for the left-
terns (Fig. 12a). Under the conditions of the FFSSM seismic source upper region of the arched tunnel.
location and parameters set in Scheme #4, the PPV distribution has The above evidence shows that the tunnel shape influences the
a good symmetry in circular and rectangular tunnels, approximate site effect of seismic waves, which can dominate the location of
symmetry in gate arch-shaped tunnels, and the asymmetry in ground motion amplification and weakened zones and affect the
arched tunnels. For example, the PPVs at the floor of arched tunnels degree of ground motion amplification to some extent. However, in
are higher compared to those at the left wall. Compared with the Fig. 12, the effect of tunnel shape on the ground motions in the
other three tunnels, the PPVm of the circular tunnel is slightly larger, near-tunnel rock masses is relatively minor in terms of the change
approaching 3.4 cm/s; whilst the other three values are closer, of PPVs and amplification factors. Fig. 13 shows the maximum
about 3 cm/s. This difference is more sound in the amplification amplification factors of tunnel-induced ground motions for four
factor distribution shown in Fig. 12b and c. The amplification factor tunnel shapes with FFSSM seismic sub-source lengths of 4 m (Set 1)
distributions of the circular and rectangular tunnels are similarly and 6 m (Set 2). For four tunnel shapes, the maximum amplification
symmetric about the direction of seismic wave incidence. Those of factors, am , are 1.87 and 2.04 for Sets 1 and 2 of circular tunnels, and
 Q. Wang et al. / Journal of Rock Mechanics and Geotechnical Engineering 14 (2022) 1761—1781 1773

Fig. 9. PPV contours and amplification factor distributions for D = 1.5 m, 3 m, 6 m, 10 m, 16 m: (a) PPV contours for the excavation model; (b) Amplification factor distributions
around the tunnel (25 m × 25 m); and (c) Amplification factor distributions for the entire computation domain (100 m × 100 m), in which Zone-A, Zone-B, and Zone-D are the high-
value zones of PPV. Zone-C is the low-value zone of PPV. A1, A2, and A3 are PPV amplified zones.
1774 Q. Wang et al. / Journal of Rock Mechanics and Geotechnical Engineering 14 (2022) 1761—1781

various diffractions, refractions, and reflections of the incident waves

at the EDZ, EFZ, and tunnel borders make this site effect more
noticeable on both sides of the left arch shoulder and the right floor
of the tunnel. The interaction of these multiple seismic sub-waves
changes the location and extent of the appearance of strong
ground motion and the degree of its amplification. Additionally, it
can be found from Fig. 14 that the change of k can result in different
tunnel shielding effects. With the increase of k, the ground motion on
the right arch shoulder side of the tunnel tends to weaken, implying
the gradual enhancement of the shielding effect.
Fig. 15 illustrates the distribution of PPV contours and amplifi-
cation factors corresponding to the conditions of Fig. 14. As the k
increases, PPVm in the Zone-A increases from 3 cm/s to 6.2 cm/s, and
the extent of PPVm gradually extends from the left floor to the left
sidewall of the tunnel. Additionally, with the increase of k, several
small high-PPV regions also appear gradually in the right arch
shoulder and right wall areas of the tunnel, where their PPVm values
also increase progressively and can reach 5 cm/s when k is 1. In
contrast to Zone-A, the range of Zone-B and Zone-D decreases as k
increases. For the PPV amplification factors as shown in Fig. 15b and
c, with the increase of k from 0 to 1, the am in the zone A1 increases
gradually from 1.7 to 3.7. The range of A1 extends from the left floor
region to the left sidewall region. Meanwhile, the am increases
progressively to 3.3 in both the tunnel’s right shoulder and sidewall
regions. However, in the other two PPV amplification zones, A2 and
A3, the am decreases from 4.2 to 2.1 as k increases, whose range also
decreases. Moreover, with the increase of k, the am is reduced from
0.4 to 0.2 in the PPV weakened zone. In short, combining the results
of Figs. 14 and 15, the increase of the extents of EDZ and EFZ can
increase ground motions in the near-tunnel areas, which is man-
ifested by the increase in PPVm and am . On the contrary, it also de-
creases ground motions in the far-tunnel and tunnel shielding areas,
where the shielding effect of the tunnel increases significantly, as
illustrated in the PPV weakened zones in Fig. 15b and c.
Considering four ranges of damage zones and two conditions of
the FFSSM seismic sub-source length in Scheme #5, Fig. 16 shows the
am in the near-tunnel areas. Taking the excavation model with k = 0 as
a reference, when k is 0.5, 0.8, and 1, the am values are 1.4, 1.8, and 2
times that of k = 0, respectively. These relative times are determined
based on the average amplification times for the FFSSM seismic source
Fig. 10. The relationship curves between the near-field ground motion in the near-
lengths of 4 m (Set 1 in Fig. 16) and 6 m (Set 2 in Fig. 16).
tunnel area and the tunnel span: (a) The variation of maximum PPVs with the tun-
nel span; and (b) The variation of maximum amplification factors with the tunnel span.
5. Empirical formula of the amplification factor in the near-
tunnel area
1.84 and 1.99 for rectangular tunnels, 1.77 and 1.94 for gate arch-
Based on the above results of the control effects of the four
shaped tunnels, and 1.75 and 1.79 for arched tunnels, respec-
factors (e.g. the main seismic source wavelength, tunnel span,
tively. Taking the amplification factor of circular tunnels as the base,
tunnel shape, and range of damage zones) on the amplification of
the amplification factor of the rectangular tunnel, gate arch-shaped
near-field ground motion in the near-tunnel area, an empirical
tunnel, and the arched tunnel is 0.98, 0.96, and 0.9 times of the am
formula for am can be proposed, as a product similar to that used in
of the circular tunnel, respectively, which is calculated by the ratios
the Q-system for classifying rock masses:
of the average amplification factor of Sets 1 and 2.
am = Fλ FD Fs Fd (1)
4.4. Influence of the range of damage zones

To understand the influence of the range of damage zones, the 5.1. The wavelength control factor, F λ
simulations of Schemes #1 and #5 in Table 2 are conducted. Four
ranges of damage zones are selected as shown in cases 1—4 of Fig. 4, According to the relation between am and λs illustrated in Fig. 7b
set by a damage range parameter, k. in Section 4.1, a best-fit curve of a Gaussian function between am
For four cases with k = 0, 0.5, 0.8, and 1, the vertical velocity and λs is proposed using the least square method, as shown in
distributions at 16.8 ms in Set 1 of Scheme #5 are shown in Fig. 14. As Fig. 17a. The wavelength control factor, Fλ , is established through a
the range of EDZ and EFZ of the rock masses around the tunnel in- formula expressed by Eq. (2). The coefficients in Eq. (2) are finally
creases, that is, the k value increases, the site effect caused by the obtained by the least-squares iteration, and the minimum fit con-
interaction of seismic waves is enhanced accordingly. In Fig. 14, verges error optimization.
 Q. Wang et al. / Journal of Rock Mechanics and Geotechnical Engineering 14 (2022) 1761—1781 1775

( )2 amplification factor and its control factors (λs and D), the Fs is
λs -19 1583 quantified using a rating scale, as listed in Table 5. In Table 5, only
-0 5 11 9305
Fλ = 0 8798 + 0 2594e (2) the four most used tunnel shapes are considered and given sepa-
rate scores, with circular tunnels selected as a baseline for scoring,
i.e. Fs = 1 for a circular tunnel. The scoring values for the other three
5.2. The tunnel span factor, F D tunnel shapes are given based on the laws obtained in Section 4.3.
As listed in Table 5, the values of Fs for four tunnel shapes ranged
The tunnel span factor (FD ) is introduced to quantify the from 0.9 to 1, which illustrates that the tunnel shape has a relatively
contribution of the tunnel span to the increase in near-tunnel small effect on the near-tunnel ground motion amplification effect.
ground motion. It can be found that Eq. (3) has a typical logistic
functional form, and the best-fit curve between am and D given by 5.4. The excavation damage factor, F d
Eq. (3) is illustrated in Fig. 17b. According to the analysis in Section
4.2, am shows a tendency to increase with increase of D, and the The excavation damage factor, Fd is shown in Table 6. From the
increasing trend decreases gradually. simulation results in Section 4.4, the range of the damage zones has
a significant impact on am . As the ratio parameter k increases, that
1 is, the range of EDZ and EFZ increase, the amplification effect of
FD = 1 958 - (3)
1 + (0 9225D)0 9322 ground motion in the near-tunnel area is significantly enhanced.
For the convenience of the Fd values, the tunnel damage is classified
into five levels according to the maximum excavation damage
5.3. The tunnel shape factor, F s depth (the maximum depth of the EDZ, dm ): no, weak, medium,
strong, and extremely strong. The am values for different damage
Similarly, the tunnel shape factor, Fs , is introduced to quantify levels are interpolated based on the simulation results in Section
the contribution of the tunnel shape to the increase in near-tunnel 4.4. In Table 6, for very strong excavation damage tunnels, the
ground motion. However, unlike the Fλ and FD that are established recommended value of Fd is 2.5, but the value can be higher than
by using the best-fit relationship between the maximum 2.5 in special cases. However, for deep hard rock tunnels, the

Fig. 11. Snapshots of the vertical velocity component at 15.2 ms of four cases with different tunnel shapes in Set 1 of Scheme #4: (a) Circular tunnel; (b) Gate arch-shaped tunnel; (c)
Rectangular tunnel; and (d) Arched tunnel. The colors of the seismic waves represent the positive or negative of the vertical velocity components, red represents positive while blue
represents negative; the white dashed ovals and green dashed ovals represent wave superimposed areas and wave weakened areas.
1776 Q. Wang et al. / Journal of Rock Mechanics and Geotechnical Engineering 14 (2022) 1761—1781

Fig. 12. PPV contours and amplification factor distributions for the circular tunnel, gate arch-shaped tunnel, rectangular tunnel, and arched tunnel in Set 1: (a) PPV contours for the
excavation model; (b) Amplification factor distributions around the tunnel (25 m × 25 m); and (c) Amplification factor distributions for the entire computation domain
(100 × 100 m2). Zone-A, Zone-B, and Zone-D are the high-value zones of PPV. Zone-C is the low-value zone of PPV. A1, A2, and A3 are PPV amplified zones.
 Q. Wang et al. / Journal of Rock Mechanics and Geotechnical Engineering 14 (2022) 1761—1781 1777

engineering conditions where the depth of EDZ is greater than 5 m

are relatively uncommon. The Fd value of 2.5 has given sufficient
consideration to the ground motion amplification effect.

5.5. Applications

To verify the reliability of the empirical formula, the am fitting

value obtained through Eq. (1) is compared with the recorded value
obtained by the receiver in Schemes #2—5 (see Table A1). The am
fitting values of the empirical formula are very close to the am
recorded values of the seismic wave propagation simulation based
on the FFSSM. Among the 34 cases in the four schemes, the relative
error (δ) is less than 5% in 21 cases, the δ value is between 5% and
10% in 11 cases, and the δ value exceeds 10% (14.14% and 17.53%) in
only two cases. Among these 34 cases, the proportion of cases with
relative errors within 5% is as high as 61.8%. The proportion of cases
with relative errors within 10% is nearly 94.1%. Therefore, for the
cases in this paper, the proposed empirical formula can estimate
the am in the near-tunnel area very well.
Furthermore, for six published typical deep tunnel projects, the
Fig. 13. The maximum amplification factor of the near-tunnel area around deep tunnel estimated am values from Eq. (1) in this paper and the measured
with different tunnel shapes in Scheme #4. Cases 1 to 4 are for circular tunnel, gate values from the field monitoring under nine cases were further
arch-shaped tunnel, rectangular tunnel, and arched tunnel, respectively.
compared, as shown in Table 7. Those estimated am values in nine

Fig. 14. Snapshots of the vertical velocity component at 16.8 ms of four cases with different k values in Set 1 of Scheme #5: (a) k = 0; (b) k = 0.5; (c) k = 0.8; and (d) k = 1. The colors
of the seismic waves represent the positive or negative of the vertical velocity components, red represents positive while blue represents negative; the white dashed ovals, and
green dashed ovals represent wave superimposed areas and wave weakened areas, respectively; k refers to the ratio of the modeled EDZ and EFZ of the tunnel to the actual EDZ and
EFZ.
1778 Q. Wang et al. / Journal of Rock Mechanics and Geotechnical Engineering 14 (2022) 1761—1781

Fig. 15. PPV contours and amplification factor distributions of four cases with different k values in Set 1 of Scheme #5: (a) PPV contours for the excavation model; (b) Amplification
factor distributions around the tunnel (25 m × 25 m); and (c) Amplification factor distributions for the entire computation domain (100 m × 100 m). Zone-A, Zone-B, and Zone-D
are the high-value zones of PPV. Zone-C is the low-value zone of PPV. A1, A2, and A3 are PPV amplified zones.
 Q. Wang et al. / Journal of Rock Mechanics and Geotechnical Engineering 14 (2022) 1761—1781 1779

Table 5
List of the tunnel shape factor, Fs .

Tunnel shape Fs

Circular tunnel 1
Rectangular tunnel 0.98
Gate arch-shaped tunnel 0.96
Arched tunnel 0.9

Table 6
List of the excavation damage factor, Fd .

Excavation damage level dm (m) Fd

No dm = 0 1
Weak 0 < dm ≤ 1 1.2
Medium 1 < dm ≤ 3 1.6
Strong 3 < dm ≤ 5 2
Extremely strong 5 < dm 2.5

cases match the measured values, suggesting that our proposed

empirical formula can be applied to practical engineering with high
Fig. 16. The maximum amplification factor in the near-tunnel area with different EDZ
accuracy. Nevertheless, due to few studies on the ground motion
and EFZ ranges in Scheme #5.
amplification effect in the near field of the tunnel, some parameters
in Eq. (1) are adopted from engineering experiences when calcu-
lating the maximum amplification factor in Table 7. This may affect
the accuracy of the proposed empirical formula. However, in any
case, the empirical formula proposed in this paper is very practical
and simple for estimating the amplification effect of near-field
tunnel ground motion.

6. Conclusions

This paper studied the characteristics and distributions of strong

ground motions caused by the interaction between near-field
seismic waves and deep tunnels based on the proposed FFSSM.
The control effects of the main seismic source wavelength, the
tunnel span, the tunnel shape, and the range of damage zones on
the near-field ground motion amplification behaviors are revealed.
Then, an empirical formula in the form of am = Fλ FD Fs Fd that can
be used to estimate the maximum amplification factor of near-field
ground motions in the near-tunnel area is proposed. The conclu-
sions are made as follows.

(1) The main seismic wavelength (λs ) plays a crucial role in

controlling the ground motion behavior around the tunnel,
which affects the interaction time of seismic waves with the
tunnel and changes the distribution of PPVs and their
amplification factors. The interaction time and PPVs increase
with increasing λs , while the maximum amplification factor
am in the near-tunnel area shows a trend of first increasing
and then decreasing. Additionally, several PPV amplification
zones are formed in the interaction regions of the incident
waves and the secondary reflected waves. Conversely, the
PPV weakened zones are formed on the back side of the
tunnel in the direction of the incident seismic waves.
(2) The tunnel span (D) also affects ground motion around the
tunnels. When the tunnel span is changed, even if the source
parameters of the FFSSM are constant, the interaction time of
seismic wave with the tunnels and the induced strong
ground motions in the near-tunnel area is changed. As D
increases, the PPVm , and its corresponding am increase.
(3) The tunnel shape slightly affects the propagation of seismic
waves radiated by the FFSSM seismic source and its strong
Fig. 17. The fitting relationship between the amplification effect in the near-tunnel ground motion distribution in the near-tunnel area. A study
area and λs and D (a) Fitting curve of am and λs (b) Fitting curve of am and D. of four common tunnel shapes (circular, gate arch-shaped,
1780 Q. Wang et al. / Journal of Rock Mechanics and Geotechnical Engineering 14 (2022) 1761—1781

Table 7
Analysis table of relative error between the fitting values and measured values of am in six underground engineering projects.

Underground Case No. λs Fλ D FD Fs Fd The fitting The measured References

engineering value of am value of am

East driefontein No.4 1 20.7 1.7 28.28* 1.3 0.96 2.5 5.31 5.1 Cichowicz et al. (2000); Milev and Spottiswood (2005);
2 50.2 1.17 28.28* 1.3 0.96 2.5 3.66 4* Durrheim (2012); Wesseloo (2018)
3 20.7 1.7 28.28* 1.3 0.96 1.6 3.40 3*
4 50.2 1.17 28.28* 1.3 0.96 1.6 2.34 2.4
Taohuazui mining area 5 53.1* 0.88 12 1.86 0.98 1.2 1.94 1.69 Li et al. (2018, 2020)
Tau Tona gold mine 6 33 1.01 7.67* 1.82 0.96 2.5 4.43 4.2—6.2 Cichowicz et al. (2000); Milev and Spottiswood (2005)
Mponeng gold mine 7 45.1 0.9 11.9* 1.86 0.96 2.5 4.04 4—6 Milev and Spottiswood (2005); Davidsen and Kwiatek (2013);
Share et al. (2013); Kozlowska et al. (2015)
Kloof gold mine 8 24.1 1.1 11.9* 1.86 0.96 2.5 4.91 3.3—5.3 Milev and Spottiswood (2005); Manzi et al. (2014)
Harmony-Orkney 9 58.6 0.88 10.48* 1.85 0.96 2.5 3.9 2.7—4.7 Milev and Spottiswood (2005)

Notes: The data marked with * are estimated values or approximate ranges obtained from engineering data and related references, due to lack of data.

rectangular, and arch) shows that circular tunnels have the Beresnev, I.A., Atkinson, G.M., 1999. Generic finite-fault model for ground-motion
prediction in eastern North America. Bull. Seismol. Soc. Am. 89 (3), 608—625.
largest maximum PPV amplification factors.
Berkhout, A.J., 1987. Applied Seismic Wave Theory. Elsevier, Amsterdam, The
(4) The ranges of the EDZ and EFZ of the surrounding rocks Netherlands.
induced by the high-stress unloading of the deep tunnels Carcione, J.M., 2007. Wave Fields in Real Media: Wave Propagation in Anisotropic,
have a significant impact on the near-field seismic wave Anelastic, Porous and Electromagnetic Media. Elsevier, Amsterdam, The
Netherlands.
propagation and its strong ground motion distributions Chaljub, E., Komatitsch, D., Vilotte, J.P., Capdeville, Y., Valette, B., Festa, G., 2007.
around deep tunnels. As the EDZ and EFZ ranges increase, the Spectral-element analysis in seismology. Adv. Geophys. 48, 365—419.
PPV and its amplification factor distribution change, while Chapman, C.H., 2004. Fundamentals of Seismic Wave Propagation. Cambridge
University Press, Cambridge, UK.
the PPVm and am also increase significantly. Chen, Q.J., Yuan, W.Z., Li, Y.C., Cao, L.Y., 2013. Dynamic response characteristics of
super high-rise buildings subjected to long-period ground motions. J. Cent.
The applicability and accuracy of the empirical formula are South. Univ. 20, 1341—1353.
Cichowicz, A., Milev, A.M., Durrheim, R.J., 2000. Rock mass behaviour under seismic
verified. This empirical formula provides an easy approach for loading in a deep mine environment: implications for stope support. J. S. Afr.
accurately estimating the ground motion parameters in seismic Inst. Min. Metall 100 (2), 121—128.
hazard risk evaluation (e.g. fault-slip rockburst and seismic- Davidsen, J., Kwiatek, G., 2013. Earthquake interevent time distribution for induced
micro-, nano-, and picoseismicity. Phys. Rev. Lett. 110 (6), 068501.
induced collapse) and the rock support design of the deep tun-
Diao, T., Chen, S.J., Jiang, Z.F., 2011. Amplification effect of peak ground motion ac-
nels under dynamic load conditions. celeration in class II and III sites over Shandong Province. Earthq. Res. China 25
(4), 498—506.
Durrheim, R.J., Kullmann, D.H., Stewart, R.D., Cichowicz, A., 1996. Seismic excitation
Declaration of competing interest of the rock mass surrounding an excavation in highly stressed ground. In:
Proceedings of the 2nd North American Rock Mechanics Symposium, pp. 389—
The authors declare that they have no known competing 394. Montreal, Canada.
Durrheim, R.J., Spottiswoode, S.M., Roberts, M.K.C., Brink, A.Z., 2005. Comparative
financial interests or personal relationships that could have
seismology of the Witwatersrand Basin and Bushveld Complex and emerging
appeared to influence the work reported in this paper. technologies to manage the risk of rockbursting. J. S. Afr. Inst. Min. Metall 105
(6), 409—416.
Durrheim, R.J., 2012. Functional specifications for in-stope support based on seismic
Acknowledgments
and rockburst observations in South African mines. In: Proceedings of the 6th
International Seminar on Deep and High Stress Mining, pp. 41—55. Perth,
This work was jointly supported by the National Natural Science Australian.
Faccioli, E., Maggio, F., Paolucci, R., Quarteroni, A., 1997. 2D and 3D elastic wave
Foundation of China (Grant No. 41877256), the Natural Science
propagation by a spseudo-spectral domain decomposition method. J. Seismol. 1
Foundation of Hubei Province (Grant No. ZRQT2020000114), the (3), 237—251.
Key Research Program of the Chinese Academy of Sciences (Grant Feng, X.T., Chen, B.R., Li, S.J., et al., 2012. Studies on the evolution process of rock-
No. KFZD-SW-423). bursts in deep tunnels. J. Rock. Mech. Geotech. Eng. 4 (4), 289—295.
Fichtner, A., 2011. Full Seismic Waveform Modelling and Inversion. Springer, Berlin,
Germany.
Appendix A. Supplementary data Hashash, Y., Hook, J.J., Schmidt, B., Yao, J., 2001. Seismic design and analysis of
underground structures. Tunn. Undergr. Space Technol. 16 (4), 247—293.
He, M.C., Ren, F.Q., Liu, D.Q., 2018. Rockburst mechanism research and its control.
Supplementary data to this article can be found online at Int. J. Min. Sci. Technol. 28 (5), 829—837.
https://doi.org/10.1016/j.jrmge.2021.12.024. Hedley, D.G.F., 1988. Peak particle velocity for rock bursts in some Ontario mines. In:
Proc 2nd International Symposium on Rockbursts and Seismicity in Mines,
pp. 345—348. Minneapolis, USA.
References Hedley, D.G.F., 1992. Rockburst Handbook for Ontario Hardrock Mines. Energy,
Mines and Resources. Canada Centre for Mineral and Energy Technology,
Aki, K., Richards, P.G., 2002. Quantitative Seismology, second ed. Univ Science Ottawa, Australia. CANMET Special Report SP92-1E.
Books, Oxford, UK. Hudyma, M.R., Heal, D., Mikula, P., 2003. Seismic monitoring in mines—old
Almadani, S.A., Abdelrahman, K., Mansour, F.I.B., 2020. Site response assessment technology-new applications. In: Proceedings of the 1st Australasian Ground
and ground conditions at king saud university campus, riyadh city, Saudi arabia. Control in Mining Conference, pp. 201—218. Sydney, Australia.
Arabian J. Geosci. 13, 357. Kaiser, P.K., McCreath, D., Tannant, D., 1996. Rockburst Support Handbook. Geo-
Bao, H., Zhang, K.K., Yan, C.G., Lan, H.X., Wu, F.Q., Zheng, H., 2020. Excavation mechanics Research Centre. Laurentian University, Sudbury, Canada.
damaged zone division and time-dependency deformation prediction: a case Komatitsch, D., 1997. Spectral and Spectral—Element Methods for the 2D and 3D
study of excavated rock mass at Xiaowan Hydropower Station. Eng. Geol. 272, Elastodynamics Equations in Heterogeneous Media. Institut de Physique du
105668. Globe Paris, Paris, France.
Basabe, J.D., Sen, M.K., 2007. Grid dispersion and stability criteria of some common Komatitsch, D., Tromp, J., 1999. Introduction to the spectral element method for
finite-element methods for acoustic and elastic wave equations. Geophysics 72 three-dimensional seismic wave propagation. Geophys. J. Int. 139 (3), 806—822.
(6), T81—T95. Komatitsch, D., Tromp, J., 2003. A perfectly matched layer absorbing boundary
Beresnev, I.A., Atkinson, G.M., 1997. Modeling finite-fault radiation from the ωn condition for the second-order seismic wave equation. Geophys. J. Int. 154 (1),
Spectrum. Bull. Seismol. Soc. Am. 87 (1), 67—84. 146—153.
 Q. Wang et al. / Journal of Rock Mechanics and Geotechnical Engineering 14 (2022) 1761—1781 1781

Kozlowska, M., Orlecka-Sikora, B., Kwiatek, G., Boettcher, M.S., Dresen, G., 2015. Seriani, G., Priolo, E., Pregarz, A., 1995. Modelling waves in anisotropic media by a
Nanoseismicity and picoseismicity rate changes from static stress triggering spectral element method. In: Proceeding of Third International Conference on
caused by a Mw 2.2 earthquake in Mponeng gold mine, South Africa. J. Geophys. Mathematical and Numerical Aspects of Wave Propagation, pp. 289—298.
Res. Solid Earth 120 (1), 290—307. Seriani, G., Oliveira, S.P., 2008. Dispersion analysis of spectral element methods for
Li, D.Z., Helmberger, D., Clayton, R.W., Sun, D.Y., 2014. Global synthetic seismograms elastic wave propagation. Wave Motion 45 (6), 729—744.
using a 2-D finite-difference method. Geophys. J. Int. 197 (2), 1166—1183. Share, P., Milev, A., Durrheim, R., Kuijpers, J., Ogasawaraet, H., 2013. Relating high-
Li, X.P., Bian, X., Luo, Y., Lv, J.L., Ren, G.F., 2020. Study on attenuation law of blasting resolution tilt measurements to the source displacement of an M2.2 event
vibration propagation of side wall of underground cavern. Rock Soil Mech. 41 located at Mponeng gold mine. J. S. Afr. Inst. Min. Metall 113 (10), 787—793.
(6), 2063—2069 (in Chinese). Siren, T., Kantia, P., Rinne, M., 2015. Considerations and observations of stress-
Li, X.P., Lv, J.L., Luo, Y., Liu, T.T., 2018. Mechanism study on elevation effect of blast induced and construction-induced excavation damage zone in crystalline
wave propagation in high side wall of deep underground powerhouse. Shock rock. Int. J. Rock. Mech. Min. 73, 165—174.
Vib. 11, 1—15. Somerville, P., Irikula, K., Graves, R., et al., 1999. Characterizing crustal earthquake
Lin, Z., Qi, Z., Milkereit, B., Grasselli, G., Liu, Q.Y., 2014. Spectral-element simulations slip models for the prediction of strong ground motions. Seismol Res. Lett. 70
of elastic wave propagation in exploration and geotechnical applications. (1), 59—80.
Earthq. Sci. 27 (2), 179—187. Spottiswoode, S.M., 1984. Underground seismic networks and safety. In: Pro-
Linkov, A.M., 2000. Instability, fracture acceleration and wave amplification. Int. J. ceedings of the Monitoring for Safety in Geotechnical Engineering, pp. 39—46.
Rock. Mech. Min. 37 (1), 31—37. South Africa.
Ma, T.H., Tang, C.A., Tang, L.X., Zhang, W.D., Wang, L., 2015. Rockburst characteristics Stacey, T.R., Rojas, E., 2013. A potential method of containing rockburst damage and
and microseismic monitoring of deep-buried tunnels for Jinping II Hydropower enhancing safety using a sacrificial layer. J. S. Afr. Inst. Min. Metall 113 (7), 565—
Station. Tunn. Undergr. Space Technol. 49, 345—368. 573.
Manzi, M.S.D., Hein, K.A.A., Durrheim, R.J., King, N., 2014. The Ventersdorp Contact St John, C.M., Zahrah, T.F., 1987. Aseismic design of underground structures. Tunn.
Reef model in the Kloof Gold Mine as derived from 3D seismics, geological Undergr. Space Technol. 2 (2), 165—197.
mapping and exploration borehole datasets. Int. J. Rock. Mech. Min. 66, 97—113. Sun, Q.H., Ma, F.S., Guo, J., et al., 2021. Excavation-induced deformation and damage
McGarr, A., Green, R.W., Spottiswoode, S.M., 1981. Strong ground motion of mine evolution of deep tunnels based on a realistic stress path. Comput. Geotech. 129,
tremors: implications for near source ground motion parameters. Bull. Seismol. 103843.
Soc. Am. 71, 295—319. Tang, L.Z., Xia, K.W., 2010. Seismological method for prediction of areal rockbursts
McGarr, A., 1984. Scaling of ground motion parameters, state of stress, and focal in deep mine with seismic source mechanism and unstable failure theory.
depth. Geophys. Res. 89 (B8), 6969—6979. J. Cent. South. Univ. 17 (5), 947—953.
Mikula, P.A., 2013. Progress with empirical performance charting for confident se- Vazaios, I., Vlachopoulos, N., Diederichs, M.S., 2019. Assessing fracturing mecha-
lection of ground support in seismic conditions. Min. Technol. 121 (4), 192—203. nisms and evolution of excavation damaged zone of tunnels in interlocked rock
Milev, A.M., Spottiswoode, S.M., Stewart, R.D., 1999. Dynamic response of the rock masses at high stresses using a finite-discrete element approach. J. Rock. Mech.
surrounding deep level mining excavations tunnels. In: Proceedings of the 9th Geotech. 11 (4), 701—722.
ISRM. Isrm Congress, Paris, France, pp. 220—227. Wang, T., Shang, Q.X., Li, J.C., 2021. Seismic force demands on acceleration-sensitive
Milev, A.M., Spottiswoode, S.M., Noble, B.R., et al., 2001. The meaningful use of peak nonstructural components: a state-of-the-art review. Earthq. Eng. Eng. Vib. 20
particle velocities at excavation surfaces for the optimisation of the rockburst (1), 39—62.
criteria for tunnels and stopes. SIMRAC Final Project Report GAP 709, 2002, 305. Wang, X., Cai, M., 2014. Wave propagation simulation in underground mines by
Milev, A.M., Spottiswood, S.M., 2005. Strong ground motion and site response in SPECFEM2D. In: Proceedings of the seventh international conference on deep
deep South African mines. J. S. Afr. Inst. Min. Metall 105 (7), 515—524. and high stress mining, Sudbury, Canada, pp. 723—738.
Morissette, P., Hadjigeorgiou, J., 2019. Ground support design for dynamic loading Wang, X., Cai, M., 2015. Influence of wavelength-to-excavation span ratio on ground
conditions: a quantitative data-driven approach based on rockburst case motion around deep underground excavations. Tunn. Undergr. Space Technol.
studies. J. Rock. Mech. Geotech. Eng. 11, 909—919, 05. 49, 438—453.
Motazedian, D., Atkinson, G.M., 2005. Stochastic finite-fault modeling based on a Wesseloo, J., 2018. The spatial assessment of the current seismic hazard state for
dynamic corner frequency. Bull. Seismol. Soc. Am. 95 (3), 995—1010. hard rock underground mine. Rock Mech. Rock Eng. 51 (6), 1839—1862.
Mutke, G.Z., 2008. Stability of the underground mine workings in the near-field Zhang, C.Q., Feng, X.T., Zhou, H., Qiu, S.L., Wu, W.P., 2012. Case histories of four
zone of seismic events. In: Proceedings of the 21st World Mining Congress extremely intense rockbursts in deep tunnels. Rock Mech. Rock Eng. 45, 275—
2008—New Challenges and Vision for Mining, pp. 89—97. Cracow, Poland. 288.
Ortlepp, W.D., 1993. High ground displacement velocities associated with rockburst Zhang, P., Swan, G., Nordlund, E., 2015. 1D numerical simulation of velocity
damage. In: Proceedings of the 3rd International Symposium on Rockbursts and amplification of P-waves travelling through fractured rock near a free surface.
Seismicity in Mines, pp. 101—106. Kingston, Canada. J. S. Afr. Inst. Min. Metall 115 (11), 1121—1126.
Pan, W.Y., Kristopher, A., Wang, Y.F., 2020. SeisElastic2D: an open-source package
for multiparameter full-waveform inversion in isotropic-, anisotropic- and
visco-elastic media. Comput. Geosci. 145, 104586.
Patera, A.T., 1984. A spectral element method for fluid dynamics: laminar flow in a Shili Qiu is an Associate Professor at the Institute of Rock
channel expansion. J. Comput. Phys. 54 (3), 468—488. and Soil Mechanics (IRSM), Chinese Academy of Sciences.
Potvin, Y., Jarufe, J., Wesseloo, J., 2010. Interpretation of seismic data and numerical He received his PhD degree in Geotechnical Engineering,
modelling of fault reactivation at El Teniente, Reservas Norte sector. Min. in IRSM, in 2011. His research interests cover rockburst
Technol. 119 (3), 175—181. mechanism and its risk evaluation in deep hard rock en-
Potvin, Y., Wesseloo, J., 2013. Towards an understanding of dynamic demand on gineering. He is interesting in unloading rock testing, true-
ground support. J. S. Afr. Inst. Min. Metall 113 (12), 913—922. triaxial strength criterion and constitutive model of hard
Qiu, S.L., Feng, X.T., Xiao, J.Q., Zhang, C.Q., 2014. An experimental study on the pre- rock, and empirical and numerical methods for rockburst
peak unloading damage evolution of marble. Rock Mech. Rock Eng. 47 (2), 401— vulnerability assessment. In the past decade, Dr. Qiu is
419. responsible for three projects sponsored by the National
Rohnacher, A., Rietbrock, A., Gottschmmer, E., et al., 2021. Source mechanism of Science Foundation of China. He has published over 50 sci-
seismic explosion signals at Santiaguito Volcano, Guatemala: new insights from entific papers.
seismic analysis and numerical modeling. Front. Earth Sci. 8, 603441.