Characteristics of Microseismicity during Breakthrough

in Deep Tunnels: Case Study of Jinping-II

Hydropower Station in China
Guang-Liang Feng 1; Xia-Ting Feng 2; Bing-Rui Chen 3; Ya-Xun Xiao 4;
Guo-Feng Liu 5; Wei Zhang 6; and Lei Hu 7
Downloaded from ascelibrary.org by University of Western Ontario on 12/11/19. Copyright ASCE. For personal use only; all rights reserved.

Abstract: Rockbursts are a common form of disaster that occur during the construction of deep tunnels in hard rock. This is especially the
case in the breakthrough stage of excavation, when even more attention should be paid to the risk of rockburst in order to ensure construction
safety. This work studied the characteristics of the microseismicity associated with 10 breakthrough cases in the deep tunnels (maximal depth
2,525 m) of the Jinping-II Hydropower Station in China. The results showed that the microseismicity was relatively more active in the
breakthrough period (compared with that in adjacent sections) due to the effect of working two faces in tandem and was concentrated
in the breakthrough section. Furthermore, the characteristic c-value associated with the seismic energy–potency relationship was larger,
indicating that the apparent stress was greater in the breakthrough section. Spatiotemporal changes in microseismicity that are associated
with rockburst development were found in the breakthrough section which can be used to qualitatively warn of the risk of rockburst.
Rockburst risk can be quantitatively assessed during tunnel breakthrough based on the monitored microseismicity and a quantitative method
of rockburst warning. Based on results thus obtained, the rockburst risk in the breakthrough section was found to increase continuously as the
distance between the two working faces decreased. When both working faces are excavated in the breakthrough section, the quantitative risk
of rockburst increases significantly. However, when only one working face is in action, the quantitative risk of rockburst increases only
slightly. The results of this work will be helpful in warning of impending rockbursts, and thus improving the safety of the construction
process, when breakthrough is carried out in deep tunnels excavated in hard rock. DOI: 10.1061/(ASCE)GM.1943-5622.0001574.
© 2019 American Society of Civil Engineers.
Author keywords: Tunnel; Breakthrough; Rockburst; Microseismicity; Jinping-II Hydropower Station.

Introduction
1
Associate Professor, State Key Laboratory of Geomechanics and
Geotechnical Engineering, Institute of Rock and Soil Mechanics, Chinese
To speed the construction process, a multiface construction strategy
Academy of Sciences, Wuhan 430071, PR China (corresponding author). often is adopted in the excavation of long tunnels. Therefore, sit-
ORCID: https://orcid.org/0000-0001-9231-0732. Email: glfeng@whrsm uations will be encountered in which two opposing working faces
.ac.cn are simultaneously excavated to the point at which breakthrough is
2
Professor, State Key Laboratory of Geomechanics and Geotechnical achieved. When this happens in deep tunnels that pass through hard
Engineering, Institute of Rock and Soil Mechanics, Chinese Academy rock, the potential risk of a rockburst occurring is particularly sig-
of Sciences, Wuhan 430071, PR China; Professor, Key Laboratory of nificant. A rockburst is a sudden, and usually violent, collapse of
Ministry of Education on Safe Mining of Deep Metal Mines, Northeastern
rock under stress, and such events also lead to the occurrence of
Univ., Shenyang, Liaoning 110819, PR China. Email: fengxiating@
mail.neu.edu.cn accompanying seismic events (Hedley 1992; Kaiser et al. 1996).
3
Professor, State Key Laboratory of Geomechanics and Geotechnical Close to the point of breakthrough in deep, hard tunnels, the rock
Engineering, Institute of Rock and Soil Mechanics, Chinese Academy between the two working faces is simultaneously affected by the
of Sciences, Wuhan 430071, PR China. Email: [email protected] work being carried out on the two opposing working faces. Energy
4
Associate Professor, State Key Laboratory of Geomechanics and changes in the rockmass of the breakthrough section therefore is
Geotechnical Engineering, Institute of Rock and Soil Mechanics, Chinese more intense. In addition, because there are two free surfaces
Academy of Sciences, Wuhan 430071, PR China. Email: yxxiao@ (i.e., the two working faces) to contend with in the breakthrough
whrsm.ac.cn
5
Lecturer, School of Highway, Chang’an Univ., Xi’an, Shanxi 710064,
section, energy release and deformation can occur in more direc-
PR China. Email: [email protected] tions than when a single working face is involved. As a result, the
6
Ph.D. Student, Key Laboratory of Ministry of Education on Safe stability of the rockmass can be expected to be poorer and the risk
Mining of Deep Metal Mines, Northeastern Univ., Shenyang, Liaoning of rockburst higher. Therefore, even more attention should be paid
110819, PR China. Email: [email protected] to the risk of a rockburst occurring during the breakthrough period
7
Ph.D. Student, Key Laboratory of Ministry of Education on Safe in deep tunnels passing through hard rock.
Mining of Deep Metal Mines, Northeastern Univ., Shenyang, Liaoning Microseismic (MS) monitoring techniques involving the three-
110819, PR China. Email: [email protected]
dimensional monitoring of MS events produced by the microcrack-
Note. This manuscript was submitted on March 16, 2019; approved
on July 11, 2019; published online on December 6, 2019. Discussion ing of rocks have been widely used around the world for many
period open until May 6, 2020; separate discussions must be submitted years to monitor and investigate rockbursts, with different degrees
for individual papers. This paper is part of the International Journal of of success (Mendecki 1993, 1997; Poplawski 1997; Becka and
Geomechanics, © ASCE, ISSN 1532-3641. Brady 2002; Li et al. 2007; Kaiser 2009; Trifu and Suorineni 2009;

© ASCE 04019163-1 Int. J. Geomech.

Int. J. Geomech., 2020, 20(2): 04019163

 Feng et al. 2012, 2013, 2017; Lu et al. 2013, 2015; Feng et al. to be an effective method of rockburst monitoring and warning.
2015a, c, 2019; Chen et al. 2015; Ma et al. 2015; Cao et al. However, although a great deal of research has been carried out
2016; Xu et al. 2016; Dong et al. 2016a, b; Zhang et al. 2016; on the technique, very little has concentrated on the breakthrough
Cheng et al. 2017; Yu et al. 2017, 2018; Wang et al. 2018; Zhang period.
et al. 2018; Li et al. 2019). The technique also has been used to Therefore, this paper studied the characteristics of the microseis-
investigate the stability of underground caverns and rock slopes micity that occurred during breakthrough (10 cases) in the deep,
(Xu et al. 2015; Salvoni and Dight 2016; Dai et al. 2016, 2017; hard tunnels of the Jinping-II Hydropower Station (with a maximal
Ma et al. 2017; Liu et al. 2018). MS sensors generally are laid depth of 2,525 m). The rockburst risk was analyzed quantitatively
out spatially with different azimuths. The MS waves released dur- based on the microseismicity during tunnel breakthrough. Based on
ing rock fracture then can be captured and analyzed so that the time, the quantitative and dynamic warning results, the effect of having
location, intensity, and type of rock fracture occurring can be ob- two opposing working faces on the risk of rockburst in the break-
tained. One can subsequently speculate about the internal stresses through section was investigated. The results of the work will be
Downloaded from ascelibrary.org by University of Western Ontario on 12/11/19. Copyright ASCE. For personal use only; all rights reserved.

and fracture states of the rock masses present. Using this informa- helpful in warning of impending rockbursts in future deep, hard
tion, a warning can be issued if the risk of an impending rockburst rock excavations, and in improving construction safety during
is high. A good example of the application of this technique is breakthrough.
the deep tunnels of the Jinping-II Hydropower Station in China
(Feng et al. 2015c). The excavations required to construct this
particular hydropower station involved seven parallel tunnels with Engineering and MS Monitoring Tests: Overview
a mean length of 16.7 km. They lie at great depth and pass through
hard rockmasses, so rockbursts of varying intensity were frequent The Jinping-II Hydropower Station is located on the Yalong River,
during the construction work. Therefore, MS activity was moni- Sichuan Province, in southwestern China. The project was
tored continuously during a part of the Jinping-II excavation project described in detail by Shan and Yan (2010) and Zhang et al.
and warnings were issued about impending rockbursts, which (2012). The geological cross section of the MS-monitored tunnels
significantly reduced the number of casualties suffered. The con- is illustrated in Fig. 1. There are many branch tunnels, some of
struction schedules of the tunnels thus were ensured and the safety which were excavated collaboratively to accelerate the construc-
of the workers somewhat improved. The technique thus proved tion process. Therefore, there were many situations in which

Fig. 1. Geological cross section along the monitored tunnels and distribution of the 10 breakthrough points (1-1, 2-1, 3-1, 3-2, 3-3, 4-1, 4-2, 4-3, P-1,
and P-2). A-#1, A-#2, and A-#3 are branches that were used to accelerate the construction of the headrace and drainage tunnels.

© ASCE 04019163-2 Int. J. Geomech.

Int. J. Geomech., 2020, 20(2): 04019163

 two opposing working faces were simultaneously excavated to the face. This is the moment when the breakthrough period can be con-
point of breakthrough. sidered to start. The rockmass between the two working faces that
The zones monitored were excavated using the drilling and is affected by breakthrough is called the breakthrough section.
blasting method. Ten breakthrough cases were monitored and The excavation of a working face affects the surrounding rock.
analyzed in this paper (denoted by solid rectangles in Fig. 1). There is a redistribution of the stress in the rock and new fractures
Of these, eight occurred in the headrace tunnels with diameters are generated in the hard rockmass which leads to the release of
of 12.4–13 m (1-1, 2-1, 3-1, 3-2, 3-3, 4-1, 4-2, and 4-3); the other energy. The MS monitoring system is able to capture most of these
two occurred in the drainage tunnel and had a 7.2-m diameter (P-1 events, and the information collected can be used to determine, to
and P-2). The MS monitoring system included a server, smart sen- some extent, the range of influence of the excavation process and
sors, geophysical seismometer, intelligent uninterruptible power the effect of unloading.
system, and communication elements (junction boxes, cables, The spatial distribution of the microseismicity recorded in the
and so forth). The MS monitoring layout was described in detail tunnels of the Jinping-II Hydropower Station was analyzed by
by Feng et al. (2013) and Chen et al. (2015). Two groups of sensors
Downloaded from ascelibrary.org by University of Western Ontario on 12/11/19. Copyright ASCE. For personal use only; all rights reserved.

Zhao et al. (2013). The distribution of MS events recorded near

were set up just behind each working face to ensure that the rock the working face in Headrace tunnels 1–4 was found to obey a nor-
mass between the working faces was near and within the array mal distribution. The mean value (x) corresponded to a separation
of MS sensors. The MS sensors had a natural frequency of 14 Hz of −7.36 m, and the standard deviation (σ) to 22.17 m. The range
and an approximate usable frequency range that varied from 7 to ðx − σ; x þ σÞ can be regarded as the main range of importance of
2,000 Hz. A sectional velocity model (Feng et al. 2015b) was used the distribution (this approximately corresponds to the region from
to locate the MS events. In the model, the velocities from the MS 30 m behind the working face to 15 m in front). The distribution of
source to the MS sensors in any one group are the same but those to MS events in the drainage tunnel also obeyed a normal distribution
different groups of MS sensors are different. The location errors (x ¼ −9.12 m; σ ¼ 17.00 m). In this case, the major part of the
throughout most of the monitoring area were less than 10 m, which distribution, ðx − σ; x þ σÞ, approximately corresponded to the re-
perfectly satisfied the location accuracy required of engineering
gion from 26 m behind the working face to 8 m ahead. Overall, the
monitoring. As might be expected, the further an MS event was
MS events in front of the working face were concentrated mainly in
from the sensor array, the larger was the location error. The location
a range that started 15 and 8 m from the working face in Headrace
accuracy was different in different directions. The location
tunnels 1–4 and the drainage tunnels, respectively.
accuracy was slightly higher along the tunnel. MS monitoring
To investigate the effect of excavation on the distribution of
was conducted 24 h each day and the data recorded continuously.
energy in the rock surrounding in the tunnels, numerical models
Several MS parameters used in this paper are interrelated, as
were established using FLAC3D version 3.0. Fig. 2 shows the
follows (Mendecki 1997):
model mesh created for Headrace tunnel 1 (covering a region
Z t
8 s 150 m in length and 130 m wide and high). The initial geostress
EP;S ¼ πρvP;S R2 u̇2corr ðtÞdt ð1Þ
5 0
conditions used are listed in Table 1 (Feng et al. 2018). The rock-
mass was assumed to behave as an elastic–plastic continuum that
μP2 yielded according to the Mohr–Coulomb yield criterion; the spe-
VA ¼ ð2Þ cific properties of the rockmass employed are listed in Table 2.
2E
The excavation stages used in the simulations were similar to those
P
4μ2 ΔVΔtðt̄Þ tt21 E used in the actual construction processes employed in the headrace
Scsd ¼ − 2 P ð3Þ and drainage tunnels. The distributions of the elastic energy in the
ρðX Þ ð tt21 M ij Þ2

log E ¼ c þ d log P ð4Þ

where EP;S = P- or S-wave energy; ρ = density of rock mass;

vP;S = velocity of either P- or S-wave; R = distance from MS
source; ts = duration; u̇2corr = radiation pattern corrected for the
square of the far-field velocity pulse; V A = MS apparent volume;
P = seismic potency; μ = shear stiffness; Scsd = seismic Schmidt
number; ΔV = volume of interest; Δt = time increment; t̄ = mean
time between events; t1 = start time; t2 = end time; X̄ = mean dis-
tance between consecutive sources of interacting seismic events;
M ij = seismic moment; E = energy of P- and S-waves; and c and
d = correlation constants.

Microseismicity Characteristics in Tunnel

Breakthrough Section and Period

Breakthrough Period and Section

When the two opposing working faces are far apart, their mutual
interference is negligible and they behave as if they are two separate
working faces. On the other hand, when the distance between them
is small, and the two faces are about to meet, each excavation is
Fig. 2. Numerical model constructed for Headrace tunnel 1.
bound to affect the energy adjustment area of the opposing working

© ASCE 04019163-3 Int. J. Geomech.

Int. J. Geomech., 2020, 20(2): 04019163

 Table 1. Initial geostress conditions used in numerical model and drainage tunnels were set to be 30 and 16 m in length, respec-
Stress components (MPa) Value tively (i.e., twice the ranges of the main concentration ranges of
MS events in front of single working faces). To investigate the ef-
σx −47.86
fect of breakthrough on microseismicity, the breakthrough sections
σy −54.55
σz −62.19
needed to be compared with normal sections. Hence, two other sec-
τ xy −2.52 tions were selected for comparison purposes (one on either side of
τ yz −0.30 the breakthrough section). These sections, referred to hereafter as
τ xz 12.41 the left and right sections, were taken to be of the same size as the
breakthrough section itself.

Microseismicity Characteristics
Table 2. Basic mechanical parameters of rockmass
Downloaded from ascelibrary.org by University of Western Ontario on 12/11/19. Copyright ASCE. For personal use only; all rights reserved.

Parameter Value Microseismicity Distributions

Elastic modulus (GPa) 25.3 The microseismicity in the breakthrough sections and the two sets
Poisson’s ratio 0.22 of adjacent sections was compared. The monitoring and statistical
Tension strength (MPa) 1.5 results thus obtained (for the headrace and drainage tunnels) are
Cohesion strength (MPa) 5.5 shown in Figs. 4 and 5, respectively. The figures present MS data
Internal friction angle (degrees) 32.2 recorded in all 10 cases—for example, Fig. 4(a) pertains to the 8
cases that occurred in the headrace tunnels, and Fig. 4(b) shows the
2 cases in the drainage tunnel.
In both tunnels (headrace and drainage), the microseismicity
headrace and drainage tunnels during excavation were simulated,
was more active in the breakthrough section than in the adjacent
and the results are listed in Fig. 3.
sections spatially (Fig. 4). A clearer way of illustrating the differ-
In both tunnels, the elastic energy was high in front of the work-
ences between the different sections is to create spatial density
ing face. In Headrace tunnel 1, the high elastic energy range
nephograms of the MS events (Fig. 5). The nephograms clearly
spanned 3–12 m in front of the working face, reaching a maximum
show that the microseismicity was more concentrated in the sec-
value of 168.1 kJ=m3 . In the drainage tunnel, the high-energy re- tions influenced by breakthrough. This implies that breakthrough
gion spanned the range 2.5–6 m in front of the working face, and a causes the microseismicity to be comparatively more spatially con-
maximum value of 138.2 kJ=m3 was suggested. The values simu- centrated in these sections of the tunnel.
lated for the drainage tunnel clearly were smaller than those simu- Fig. 6 shows a simulation of the energy distribution in Headrace
lated for Headrace tunnel 1, which is related to their different tunnel 1 when the two working faces were 30 m apart, i.e., the
excavation sizes. This feature of the simulations is in good agree- faces were separated by a distance equal to the length of the break-
ment with the results of microseismicity monitoring experiments. through section. Because the region between the two faces was
The microseismicity distributions revealed the same characteristics: affected by both working faces, the elastic energy distributions be-
the larger the size of the tunnel, the larger was the range of tween the two working faces overlapped one another in the break-
concentration of the MS events. through section. As a result, the maximum energy value increased
From the preceding analysis, it is apparent that the elastic energy (to 176.1 kJ=m3 ) compared with the elastic energy release for a
in front of the working face was concentrated mainly in a range single working face [i.e., the results in Fig. 3(a)]. The range and
that started about 12 m (6 m) from the working face of the headrace maximum value of the elastic energy both increased in the break-
(drainage) tunnel. The corresponding ranges of concentration of through section. This made the microseismicity more active and
MS events in front of the working faces were slightly larger than spatially concentrated in the breakthrough section, as discussed
the ranges of energy concentration; however, they were very close previously.
to each other. The simulations were somewhat simplified because The numbers, energies, and apparent volumes of the MS events
actual rockmasses tend to be rather complex. In addition, a certain occurring in the different sections for each of the 10 cases are
amount of error was associated with the locations of the MS events, plotted in Fig. 7. Each of these MS-related parameters generally
which is inevitable in MS monitoring (Feng et al. 2015b; Li et al. was larger in the breakthrough section than in the corresponding
2017). neighboring sections. The mean values of the number of MS events,
Because breakthrough sections are influenced by both of the MS energies, and MS apparent volumes in the breakthrough sec-
opposing working faces, the breakthrough sections in the headrace tion were 2, 7.3, and 14.4 times those in the adjacent sections,

(a) (b)

Fig. 3. Simulation results showing the elastic energy distribution after excavation for (a) Headrace tunnel 1; and (b) the drainage tunnel.

© ASCE 04019163-4 Int. J. Geomech.

Int. J. Geomech., 2020, 20(2): 04019163

 Local Magnitude Local Magnitude

-4.00 -4.00
-3.56 -3.56
-3.11 -3.11
-2.67 -2.67
-2.22 -2.22
-1.78 -1.78
-1.33 -1.33
-0.89 -0.89
-0.44 -0.44
0.00 0.00
Downloaded from ascelibrary.org by University of Western Ontario on 12/11/19. Copyright ASCE. For personal use only; all rights reserved.

(a) (b)

Fig. 4. Spatial distribution of the MS activity in the breakthrough and adjacent sections for (a) the headrace tunnels; and (b) the drainage tunnel.
A sphere’s size represents the energy of the MS event (the larger the sphere, the greater was the energy released).

(a) (b)

Fig. 5. Cloud diagrams (nephograms) showing the spatial distribution density of the MS events in the breakthrough and adjacent sections for (a) the
headrace tunnels; and (b) the drainage tunnel.

respectively. In Fig. 7, most of the curves are higher in the middle However, although the microseismicity in the breakthrough
and lower at both ends. This is a clear indication of the greater level section tends to be more active than in the adjacent sections, it does
of MS activity in the section influenced by breakthrough (compared not always have to be a highly active MS section. For example,
with those that were not). in breakthrough Cases 1-1, 3-4, and 4-3, the microseismicity (num-
ber of MS events, MS energy, and MS apparent volume) was
smaller than that corresponding to the mean levels. At the same
time, there was some variation between cases, so some difference
in the microseismicity occurred in the various cases. This is illus-
trated by the cumulative number of events, energy, and apparent
volume, which ranged from 1 to 31, from 102.6 to 105.9 J, and from
1.1 to 61,660 m3 , respectively. Because microseismicity is related
to rockburst risk, this means that the rockburst risk in different
breakthrough sections is different, and thus these sections are not
always high rockburst risk regions.

MS c-Values: Stress Levels

Apparent stress expresses the amount of radiated MS energy per
unit volume of inelastic coseismic deformation. MS waveforms
do not contain direct information on absolute stress. Rather, they
Fig. 6. Simulation results showing the elastic energy distribution in
merely contain information about the dynamic decrease in stress at
Headrace tunnel 1 with two opposing working faces undergoing
the MS source. A number of seismological studies suggested that a
excavation.
reliable estimate of stress decrease, or preferably apparent stress,

© ASCE 04019163-5 Int. J. Geomech.

Int. J. Geomech., 2020, 20(2): 04019163

Downloaded from ascelibrary.org by University of Western Ontario on 12/11/19. Copyright ASCE. For personal use only; all rights reserved.

(a) (b)

(c)

Fig. 7. Microseismicity in the different sections of the 10 breakthrough cases showing: (a) the number of MS events; (b) MS energy; and
(c) MS apparent volume.

can be made by determining the correlation between energy (E) case (4-2), there was no rockburst occurrence in the breakthrough
and potency (P). This so-called E–P correlation was investigated section.
and described by Mendecki (1993) and Amidzic (2005). Normally, Fig. 9 illustrates the spatiotemporal evolution of the microseis-
for a specific temporal and spatial domain, E and P are correlated in micity associated with the intense rockburst that occurred in
a way best revealed via a log–log plot [i.e., they obey the relation- Case P-1. On this occasion (1) a large number of MS events oc-
ship in Eq. (4) in section “Engineering and MS Monitoring Tests: curred that were spatially clustered together in the breakthrough
Overview”]. The fitting constants derived from such correlation section [Fig. 9(b)]; (2) before the rockburst occurred, the cumula-
plots [specifically, the value of c in Eq. (4)] yield a useful parameter tive number of MS events and amount of MS energy released in-
that can be used to analyze the apparent stress level. That is, a large creased heavily [Fig. 9(c)]; and (3) the Schmidt number decreased
c-value corresponds to a high apparent stress. sharply prior to rockburst and the cumulative apparent volume in-
Fig. 8 presents log–log plots of energy versus potency for the creased abruptly at the same time [Fig. 9(d)]. The Schmidt number
MS events recorded in the different sections studied. There were measures the spatiotemporal complexity of the seismic flow of the
good linear correlations between log E and log P in each section. rock—the lower the Schmidt number, the less stable is the flow and
The equations representing the linear fits to the data also are given the stronger is the microseismicity (Mendecki 1997). This case
in Fig. 8, which therefore specify the corresponding c-values. should be compared with Case 4-2, wherein breakthrough occurred
Clearly, the c-values for the left and right sections both were without rockburst and the microseismicity was relatively much qui-
smaller than that for the breakthrough section. This implies that eter. Fig. 10 presents a similar set of figures to those shown in Fig. 9.
when the two working faces interact (in the breakthrough section), Because no rockburst occurred during breakthrough in Case 4-2,
the apparent stress and stress decrease are enhanced. This variation the resulting tunnel had good integrity and shape [Fig. 10(a)]. Only
is why the microseismicity is more active in the breakthrough a few MS events occurred in the breakthrough section, so the spatial
section. distribution was relatively sparse [Fig. 10(b)]. The cumulative num-
ber of MS events and cumulative MS energy released increased
gently with time and there was only a small jump in these quantities
Microseismicity Changes during Rockburst just before breakthrough [Fig. 10(c)]. The Schmidt number de-
Development in Breakthrough Section creased sharply but the cumulative apparent volume increased only
slightly [Fig. 10(d)].
This section concentrates on two cases. In one case (P-1), there The preceding comparison shows that the microseismicity is
was an intense rockburst in the breakthrough section; in the other correlated with the region in which the rockburst occurs; rockbursts

© ASCE 04019163-6 Int. J. Geomech.

Int. J. Geomech., 2020, 20(2): 04019163

 (a) (b)
Downloaded from ascelibrary.org by University of Western Ontario on 12/11/19. Copyright ASCE. For personal use only; all rights reserved.

(c)

Fig. 8. Log–log plots of energy (E) versus potency (P) for the MS events occurring in three different regions: (a) left sections; (b) breakthrough
sections; and (c) right sections.

Local Magnitude
-4.00
-3.56
-3.11
-2.67
-2.22
-1.78
-1.33
-0.89
-0.44
0.00

(a) (b)

(c) (d)

Fig. 9. Data relating to the intense rockburst that occurred during breakthrough Case P-1: (a) photograph of rockburst area; (b) spatial distribution of
microseismicity in the rockburst area, in which a sphere’s size represents the energy of the MS event (the larger the sphere, the was greater the energy
released); (c) temporal variation in the number of MS events and energy released; and (d) the temporal change in Schmidt number and apparent
volume.

© ASCE 04019163-7 Int. J. Geomech.

Int. J. Geomech., 2020, 20(2): 04019163

 Local Magnitude
-4.00
-3.56
-3.11
-2.67
-2.22
-1.78
-1.33
-0.89
-0.44
0.00

(a) (b)
Downloaded from ascelibrary.org by University of Western Ontario on 12/11/19. Copyright ASCE. For personal use only; all rights reserved.

(c) (d)

Fig. 10. Comparable diagrams to those in Fig. 9 for breakthrough Case 4-2 (no rockburst): (a) photograph of the tunnel highlighting its good physical
condition; (b) spatial distribution of the microseismicity, showing its quiet nature, in which a sphere’s size represents the energy of the MS event
(the larger the sphere, the greater was the energy released); (c) temporal variation in the number of MS events and energy released; and
(d) corresponding change in Schmidt number and apparent volume.

are more likely to occur in regions of high microseismicity. More- the jth MS parameter. The probability value for each rockburst in-
over, the way the microseismicity changes in breakthrough sections tensity type ranges from 0% to 100% (naturally, the greater the
with and without rockbursts is quite different. Therefore, we can probability, the greater is the rockburst risk). The sum of the
qualitatively warn of the potential risk of rockburst based on the probabilities of rockbursts of all intensity types is 100%.
different microseismicity changes outlined previously. During the breakthrough periods, the method was used to quan-
titatively and dynamically warn of the potential occurrence of rock-
burst in the breakthrough sections. Daily risk assessments were made
Quantitative and Dynamic Rockburst Warning in and released to support the safety of the construction process. Figs. 11
Breakthrough Section and 12 present the results obtained for the two breakthrough cases
discussed previously (P-1 and 4-2). The results showed that
Feng et al. (2015c) devised a method to quantitatively and dynami- 1. The calculated rockburst risks changed continually as excava-
cally warn of the risk of a rockburst occurring in deep tunnels tion proceeded in both cases. As the distance between the
based on microseismicity. The method established a quantitative two working faces became smaller, the potential risk of rock-
relationship between rockburst risk and microseismicity, and it burst increased continuously. The risk changes were significant
was eventually used to form a quantitative formula for rockburst and rockburst risk increased rapidly when the two working faces
warning in the deep tunnels of Jinping-II were very close or about to break through.
2. In Case P-1, the rockburst risk in the breakthrough section was
X
n
Pi ¼ wj Pji ð5Þ high when the distance between the two working faces was less
j¼1 than 44 m [Fig. 11(a)]. On January 22, when the distance be-
tween the two faces was 44 m, there was a 54.3% probability of
where i = intensity of rockburst (i.e., extremely intense, intense, a moderate rockburst risk. As the two working faces continued
moderate, slight, or none); and Pi = probability that rockburst to move forward, the rockburst risk increased to an intense level
occurs with intensity i. The summation extends over n MS on January 24 (with 23 m between the two working faces).
parameters which are labeled with the index j. The functional cor- The risk of an intense (moderate) rockburst occurring on that
relation between the occurrence of a rockburst of intensity i and day was 66.2% (33.5%). Two days later, on January 26, the risk
MS parameter j is Pji , and wj is a weighting coefficient for of an intense rockburst occurring increased to 71.2%, and an

© ASCE 04019163-8 Int. J. Geomech.

Int. J. Geomech., 2020, 20(2): 04019163

Downloaded from ascelibrary.org by University of Western Ontario on 12/11/19. Copyright ASCE. For personal use only; all rights reserved.

(a) (b)

Fig. 11. Variation in the probability that a rockburst of given intensity will occur in the days leading up to breakthrough for (a) Case P-1
(intense rockburst); and (b) Case 4-2 (no rockburst).

intense rockburst occurred on that day. The quantitative rock- was halted (so the chainages of the working faces were the same
burst warning result therefore was in good agreement with the on the next day). As a result, on the following day (January 25), the
actual situation in the tunnel. risk of an intense rockburst increased only slightly, to 69.5%. On
3. In Case 4-2, the rockburst risk in the breakthrough section January 25, excavation was resumed on one face only (the left).
remained low throughout the overall construction period That is, the workers changed from two-directional to single-
[Fig. 11(b)]. This reflected well the actual situation in the tunnel, directional excavation in an attempt to reduce the disturbance to
because no rockburst occurred at this site. From September 19 to the surrounding rockmass. Once again, no rockburst occurred. On
25, the warning results suggested a low risk of rockburst (on January 26, the distance between the two working faces was only
these days, the no-rockburst probability ranged from 68.7% to 18 m and the intense rockburst risk had reached 71.2%. The micro-
100%). When the distance between the two working faces was seismicity was abnormally active and concentrated in the left work-
less than 13 m, the rockburst risk increased somewhat. From ing face (Fig. 13). Under such circumstances, single-directional
September 24 to 26, the risks associated with slight and mod- excavation should have been continued to reduce the intense mu-
erate rockbursts increased as excavation continued (and the tual effect of simultaneously excavating and unloading two work-
working faces moved from 13 m apart to 0 m). On September ing faces. Unfortunately, this strategy was not implemented on this
26, the risk of slight rockburst rose to 44.0% and breakthrough occasion (Fig. 12). Instead, both working faces were excavated
was achieved in the tunnel. This probability (that a slight rock- simultaneously. The left and right working faces advanced by ap-
burst might occur) should not be viewed as particularly high, proximately 2 and 4 m on that day, respectively. This resulted in the
because there also was a 30.5% chance that no rockburst would occurrence of an intense rockburst. After that rockburst, measures
happen at all at this time. were again taken to improve the situation. From January 27 on-
The excavation progress and intense rockburst warning results ward, single-directional excavation was again employed. This time,
in Case P-1 are shown in Fig. 12 in greater detail. From January 21 excavation was resumed at the right working face (where the micro-
to 24, the risk of an intense rockburst occurring increased heavily seismicity was relativity low) which continued to advance until the
(from 0% to 66.2%) as both working faces were excavated. On tunnel broke through safely.
January 24 (66.2% chance of intense rockburst) work on both faces The preceding discussion shows that rockburst risk can be
quantitatively assessed, and warnings issued dynamically, based
on microseismicity recorded during breakthrough. Based on such

Local Magnitude

-4.00
-3.56
-3.11
-2.67
-2.22
-1.78
-1.33
-0.89
-0.44
0.00

Fig. 13. Spatial distribution of the microseismicity on January 26 for

Fig. 12. Change in probability that an intense rockburst will occur Case P-1. A sphere’s size represents the energy of the MS event
and the evolution of the construction strategy for Case P-1. (the larger the sphere, the greater was the energy released).

© ASCE 04019163-9 Int. J. Geomech.

Int. J. Geomech., 2020, 20(2): 04019163

 evidence, we found that rockburst risk is strongly influenced by the Notation
distance between the two working faces and the excavation
procedures carried out there. As the interface distance becomes The following symbols are used in this paper:
smaller, the potential risk of rockburst increases almost continu- c, d = correlation constants;
ously. When both working faces are being excavated, the quanti- E = energy of P-waves and S-waves;
tative risk of rockburst can increase rather rapidly. However, under EP;S = energy of P-wave or S-wave;
these circumstances, the rate at which rockburst risk increases can i = rockburst intensity;
be dramatically reduced by carrying out excavation at only one of j = MS parameter;
the working faces. Mij = seismic moment;
n = number of MS parameters;
P = seismic potency;
Conclusions Pi = probability that rockburst occurs with intensity i;
Downloaded from ascelibrary.org by University of Western Ontario on 12/11/19. Copyright ASCE. For personal use only; all rights reserved.

Pji = functional correlation between occurrence of rockburst

Data on 10 breakthrough cases was collected during the construc- of intensity i and MS parameter j;
tion of the deep, hard rock tunnels of the Jinping-II Hydropower R = distance from MS source;
Station in China (with a maximum depth of 2,525 m). The char-
Scsd = seismic Schmidt number;
acteristics of the microseismicity occurring in these breakthrough
t1 = start time;
cases were studied in detail. When two opposing working faces
t2 = end time;
approach one another, the microseismicity in the region between
them becomes more active (compared with when they are far apart) ts = duration;
and the MS events tend to cluster together in this region (forming a t̄ = mean time between events;
breakthrough section). u̇2corr = radiation pattern corrected for square of far-field
Certain properties of the MS events (number of events and their velocity pulse;
energies and apparent volumes) are greater in the breakthrough sec- V A = MS apparent volume;
tion than in the neighboring sections. The c-value associated with vP;S = velocity of either P- or S-wave;
the seismic energy–potency relationship also is larger, showing that wj = weighting coefficient of j-th MS parameter;
the apparent stress is greater in the breakthrough section. The x = mean value;
spatiotemporal evolution of the microseismicity occurring during X̄ = mean distance between consecutive sources of
the development of rockbursts also was studied in the breakthrough interacting seismic events;
section. Before an intense rockburst in the breakthrough section, Δt = time increment;
a large number of concentrated MS events become clustered in ΔV = volume of interest;
the region. Moreover, the cumulative number of MS events and μ = shear stiffness;
MS energy released build up as time goes on. The Schmidt number ρ = density of rock mass; and
decreases sharply and the cumulative apparent volume increases σ = standard deviation.
sharply at the same time. In contrast, when breakthrough is achieved
without rockburst occurrence, the microseismicity is much quieter
and follows a very different pattern. References
Rockburst risk in the breakthrough section was quantitatively
assessed based on the microseismicity and a quantitative method Amidzic, D. 2005. “Energy-moment relation and its application.” In
of rockburst warning. We found that the risk of a rockburst occur- Controlling Seismic Risk-Proc., 6th Int. Symp. on Rockburst and
ring is strongly affected by the distance between the two working Seismicity in Mines, edited by Y. Potvin and M. Hudyma, 509–513.
Crawley, Australia: Australian Centre for Geomechanics.
faces and the excavation activity carried out on them. As the inter-
Becka, D. A., and B. H. G. Brady. 2002. “Evaluation and application of
face distance becomes smaller, the potential risk of rockburst controlling parameters for seismic events in hard-rock mines.” Int. J.
increases almost continuously. When both working faces are exca- Rock Mech. Min. Sci. 39 (5): 633–642. https://doi.org/10.1016/S1365
vated, the quantitative rockburst risk increases rapidly. However, -1609(02)00061-8.
when only one working face is in action, the quantitative rockburst Cao, A. Y., L. M. Dou, C. B. Wang, X. X. Yao, J. Y. Dong, and Y. Gu. 2016.
risk increases much more slowly. “Microseismic precursory characteristics of rock burst hazard in mining
The results of this work clearly will be helpful in warning of im- areas near a large residual coal pillar: A case study from Xuzhuang coal
pending rockburst occurrence and improving the safety of the con- mine, Xuzhou, China.” Rock Mech. Rock Eng. 49 (11): 1–16. https://doi
.org/10.1007/s00603-016-1036-7.
struction process in deep tunnels during the breakthrough stage.
Chen, B. R., X. T. Feng, Q. P. Li, R. Z. Luo, and S. J. Li. 2015.
“Rockburst intensity classification based on the radiated energy
with damage intensity at Jinping II Hydropower Station, China.” Rock
Data Availability Statement Mech. Rock Eng. 48 (1): 289–303. https://doi.org/10.1007/s00603-013
-0524-2.
The MS monitoring data and the rockburst case data during the Cheng, G. W., T. H. Ma, C. A. Tang, H. Y. Liu, and S. J. Wang. 2017.
study are available from the corresponding author by request. “A zoning model for coal mining-induced strata movement based on
microseismic monitoring.” Int. J. Rock Mech. Min. Sci. 94 (Apr):
123–138. https://doi.org/10.1016/j.ijrmms.2017.03.001.
Dai, F., B. Li, N. W. Xu, Y. L. Fan, and C. Q. Zhang. 2016. “Deformation
Acknowledgments forecasting and stability analysis of large-scale underground powerhouse
caverns from microseismic monitoring.” Int. J. Rock Mech. Min. Sci.
The authors gratefully acknowledge financial support from 86 (Jul): 269–281. https://doi.org/10.1016/j.ijrmms.2016.05.001.
the National Natural Science Foundation of China (Grant Dai, F., B. Li, N. W. Xu, and Y. G. Zhu. 2017. “Microseismic early warning of
Nos. 51621006 and 51709256). surrounding rock mass deformation in the underground powerhouse of the

© ASCE 04019163-10 Int. J. Geomech.

Int. J. Geomech., 2020, 20(2): 04019163

 Houziyan hydropower station, China.” Tunnelling Underground Space Mech. Min. Sci. 76 (Jun): 18–32. https://doi.org/10.1016/j.ijrmms
Technol. 62 (Feb): 64–74. https://doi.org/10.1016/j.tust.2016.11.009. .2015.02.005.
Dong, L. J., J. Wesseloo, Y. Potvin, and X. B. Li. 2016a. “Discriminant Ma, K., C. A. Tang, Z. Z. Liang, D. Y. Zhuang, and Q. B. Zhang. 2017.
models of blasts and seismic events in mine seismology.” Int. J. Rock “Stability analysis and reinforcement evaluation of high-steep rock
Mech. Min. Sci. 86 (Jul): 282–291. https://doi.org/10.1016/j.ijrmms slope by microseismic monitoring.” Eng. Geol. 218 (Feb): 22–38.
.2016.04.021. https://doi.org/10.1016/j.enggeo.2016.12.020.
Dong, L. J., J. Wesseloo, Y. Potvin, and X. B. Li. 2016b. “Discrimination Ma, K., C. A. Tang, L. X. Wang, W. D. Zhang, and L. Wang. 2015.
of mine seismic events and blasts using the Fisher classifier, naive Baye- “Rockburst characteristics and microseismic monitoring of deep-buried
sian classifier and logistic regression.” Rock Mech. Rock Eng. 49 (1): tunnels for Jinping II Hydropower Station.” Tunnelling Underground
183–211. https://doi.org/10.1007/s00603-015-0733-y. Space Technol. 49 (Jun): 345–368. https://doi.org/10.1016/j.tust.2015
Feng, G. L., X. T. Feng, B. R. Chen, and Y. X. Xiao. 2015a. “Microseismic .04.016.
sequences associated with rockbursts in the tunnels of the Jinping II Mendecki, A. J. 1993. “Real time quantitative seismology in mines.”
hydropower station.” Int. J. Rock. Mech. Min. Sci. 80 (Dec): 89–100. In Proc., 3rd Int. Symp. on Rock-bursts and Seismicity in Mines,
https://doi.org/10.1016/j.ijrmms.2015.06.011. 287–295. Rotterdam, Netherlands: A.A. Balkema.
Downloaded from ascelibrary.org by University of Western Ontario on 12/11/19. Copyright ASCE. For personal use only; all rights reserved.

Feng, G. L., X. T. Feng, B. R. Chen, Y. X. Xiao, and Q. Jiang. 2015b. Mendecki, A. J. 1997. Seismic monitoring in mines. London: Chapman &
“Sectional velocity model for microseismic source location in tunnels.” Hall.
Tunnelling Underground Space Technol. 45 (Jan): 73–83. https://doi Poplawski, R. F. 1997. “Seismic parameters and rockburst hazard at
.org/10.1016/j.tust.2014.09.007. Mt Charlotte mine.” Int. J. Rock Mech. Min. Sci. 34 (8): 1213–1228.
Feng, G. L., X. T. Feng, B. R. Chen, Y. X. Xiao, and Y. Yu. 2015c. https://doi.org/10.1016/S1365-1609(97)80072-X.
“A microseismic method for dynamic warning of rockburst Salvoni, M., and P. M. Dight. 2016. “Rock damage assessment in a large
development processes in tunnels.” Rock Mech. Rock. Eng. 48 (5): unstable slope from microseismic monitoring—MMG Century mine
2061–2076. https://doi.org/10.1007/s00603-014-0689-3. (Queensland, Australia) case study.” Eng. Geol. 210 (Aug): 45–56.
Feng, G. L., X. T. Feng, B. R. Chen, Y. X. Xiao, and Z. N. Zhao. 2019. https://doi.org/10.1016/j.enggeo.2016.06.002.
“Effects of structural planes on the microseismicity associated with Shan, Z., and P. Yan. 2010. “Management of rock bursts during excavation
rockburst development processes in deep tunnels of the Jinping-II of the deep tunnels in Jinping II Hydropower Station.” Bull. Eng. Geol.
Hydropower Station, China.” Tunnelling Underground Space Technol. Environ. 69 (3): 353–363. https://doi.org/10.1007/s10064-010-0266-2.
84 (Feb): 273–280. https://doi.org/10.1016/j.tust.2018.11.008.
Trifu, C. I., and F. T. Suorineni. 2009. “Use of MS monitoring for rockburst
Feng, X. T., et al. 2012. “Study on the evolution process of rockbursts in
management at VALE INCO mines.” In Proc., 7th Int. Symp. on Rock
deep tunnels.” J. Rock. Mech. Geo. Eng. 4 (4): 289–295. https://doi.org
burst and Seismicity in Mines, edited by C. A. Tang, 1105–1114.
/10.3724/SP.J.1235.2012.00289.
New York: Renton.
Feng, X. T. 2017. Rockburst: Mechanism, monitoring, warning and
Wang, P., L. S. Jiang, J. Q. Jiang, and P. Q. Zheng. 2018. “Strata behaviors
mitigation Elsevier-Health Sciences.
and rock burst–inducing mechanism under the coupling effect of a
Feng, X. T., B. R. Chen, C. Q. Zhang, S. J. Li, and S. Y. Wu. 2013. Mecha-
hard, thick stratum and a normal fault.” Int. J. Geomech. 18 (2):
nism, warning, and dynamic control of rockburst development proc-
04017135. https://doi.org/10.1061/(ASCE)GM.1943-5622.0001044.
esses. [In Chinese.] Beijing: Science.
Xu, N. W., T. B. Li, F. Dai, B. Li, Y. G. Zhu, and D. S. Yang. 2015.
Feng, X. T., Z. B. Yao, S. J. Li, S. Y. Wu, C. X. Yang, H. S. Guo, and
“Microseismic monitoring and stability evaluation for the large scale
S. Zhong. 2018. “In situ observation of hard surrounding rock displace-
underground caverns at the Houziyan hydropower station in Southwest
ment at 2400-m-deep tunnels.” Rock Mech. Rock. Eng. 51 (3): 873–892.
https://doi.org/10.1007/s00603-017-1371-3. China.” Eng. Geol. 188 (Apr): 48–67. https://doi.org/10.1016/j.enggeo
Hedley, D. G. F. 1992. Rockburst handbook for Ontario hardrock .2015.01.020.
mines: CANMET SP92-1E. Ottawa: Energy, Mines and Resources Canada. Xu, N. W., T. B. Li, F. Dai, R. Zhang, C. A. Tang, and L. X. Tang. 2016.
Kaiser, P. K. 2009. “Seismic hazard evaluation in underground construc- “Microseismic monitoring of strainburst activities in deep tunnels at the
tion.” In Proc., 7th Int. Symp. on Rock Burst and Seismicity in Mines, Jinping II hydropower station, China.” Rock Mech. Rock Eng. 49 (3):
1–26. New York: Renton. 981–1000. https://doi.org/10.1007/s00603-015-0784-0.
Kaiser, P. K., D. D. Tannant, and D. R. McCreath. 1996. Canadian rock- Yu, Y., B. R. Chen, C. J. Xu, X. H. Diao, L. H. Tong, and Y. F. Shi. 2017.
burst support handbook. Sudbury, Canada: Geomechanics Research “Analysis for microseismic energy of immediate rockbursts in deep
Centre, Laurentian Univ. tunnels with different excavation methods.” Int. J. Geomech. 17 (5):
Li, A., F. Dai, N. V. Xu, G. K. Gu, and Z. H. Hu. 2019. “Analysis of a 04016119. https://doi.org/10.1061/(ASCE)GM.1943-5622.0000805.
complex flexural toppling failure of large underground caverns in lay- Yu, Y., D. X. Geng, L. H. Tong, X. S. Zhao, X. H. Diao, and L. H. Huang.
ered rock masses.” Rock Mech. Rock. Eng. 52 (9): 3157–3181. https:// 2018. “Time fractal behavior of microseismic events for different inten-
doi.org/10.1007/s00603-019-01760-5. sities of immediate rock bursts.” Int. J. Geomech. 18 (7): 06018016.
Li, T., M. F. Cai, and M. Cai. 2007. “A review of mining-induced seismicity https://doi.org/10.1061/(ASCE)GM.1943-5622.0001221.
in China.” Int. J. Rock. Mech. Min. Sci. 44 (8): 1149–1171. https://doi Zhang, C. Q., X. T. Feng, and H. Zhou. 2012. “Estimation of in situ stress
.org/10.1016/j.ijrmms.2007.06.002. along deep tunnels buried in complex geological conditions.” Int. J.
Li, X. B., F. Q. Gong, M. Tao, L. J. Dong, K. Du, C. D. Ma, Z. L. Zhou, Rock. Mech. Min. Sci. 52 (Jun): 139–162. https://doi.org/10.1016/j
andT.B.Yin.2017.“Failuremechanismandcoupledstatic-dynamicloading .ijrmms.2012.03.016.
theory in deep hard rock mining: A review.” J. Rock Mech. Geotech. Eng. Zhang, J. F., F. X. Jiang, S. T. Zhu, and L. Zhang. 2016. “Width design
9 (4): 767–782. https://doi.org/10.1016/j.jrmge.2017.04.004. for gobs and isolated coal pillars based on overall burst-instability
Liu, J. P., Y. H. Li, and S. D. Xu. 2018. “Relationship between microseis- prevention in coal mines.” J. Rock Mech. Geotech. Eng. 8 (4):
mic activities and mining parameters during deep mining process.” 551–558. https://doi.org/10.1016/j.jrmge.2015.12.006.
J. Appl. Geophys. 159 (Dec): 814–823. https://doi.org/10.1016/j Zhang, K., T. H. Yang, H. B. Bai, and R. P. Gamage. 2018. “Longwall
.jappgeo.2018.10.018. mining–induced damage and fractures: Field measurements and
Lu, C. P., L. M. Dou, N. Zhang, J. H. Xue, X. N. Wang, H. Liu, and J. W. simulation using FDM and DEM coupled method.” Int. J. Geomech.
Zhang. 2013. “Microseismic frequency-spectrum evolutionary rule of 18 (1): 04017127. https://doi.org/10.1061/(ASCE)GM.1943-5622
rockburst triggered by roof fall.” Int. J. Rock. Mech. Min. Sci. 64 (6): .0001040.
6–16. https://doi.org/10.1016/j.ijrmms.2013.08.022. Zhao, Z. N., X. T. Feng, B. R. Chen, G. L. Feng, and T. Y. Chen. 2013.
Lu, C. P., G. J. Liu, Y. Liu, N. Zhang, J. H. Xue, and L. Zhang. 2015. “Study of relativity between rockburst and MS activity zone in deep
“Microseismic multi-parameter characteristics of rockburst hazard in- tunnel.” [In Chinese.] Rock Soil. Mech. 34 (2): 491–497. https://doi
duced by hard roof fall and high stress concentration.” Int. J. Rock. .org/10.16285/j.rsm.2013.02.037.

© ASCE 04019163-11 Int. J. Geomech.

Int. J. Geomech., 2020, 20(2): 04019163