See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/262489881

Numerical study on tunnel damage subject to blast-induced shock wave in

jointed rock masses

Article  in  Tunnelling and Underground Space Technology · July 2014

DOI: 10.1016/j.tust.2014.04.004

CITATIONS READS

97 609

6 authors, including:

Jianbo Zhu Shyi-Gen Chen

Shenzhen University Tri-Service General Hospital
90 PUBLICATIONS   1,727 CITATIONS    156 PUBLICATIONS   2,564 CITATIONS   

SEE PROFILE SEE PROFILE

Zhiye Zhao Yingxin Zhou

Nanyang Technological University Defence Science and Technology Agency, Singapore
176 PUBLICATIONS   2,942 CITATIONS    63 PUBLICATIONS   2,018 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Dynamic behavior and response of rock, discontinuity and underground opening View project

Breast reconstruction View project

All content following this page was uploaded by Jianbo Zhu on 30 April 2020.

The user has requested enhancement of the downloaded file.

1 Numerical Study on Tunnel Damage Subject
2 to Blast-Induced Shock Wave in Jointed Rock
3 Masses
4

5 X.F. Deng1, 3, J.B. Zhu2, *, S.G. Chen3, Z.Y. Zhao4, Y.X. Zhou4, 5, J. Zhao6
1
6 China Tiesiju Civil Engineering Group Company, LTD., Hefei 230023, China
2
7 Department of Civil and Environmental Engineering, The Hong Kong
8 Polytechnic University, Hung Hom, Kowloon, Hong Kong
3
9 MOE Key Laboratory of Transportation Tunnel Engineering, Southwest Jiaotong
10 University, Chengdu 610031, China
4
11 School of Civil and Environmental Engineering, Nanyang Technological
12 University (NTU) Nanyang Avenue, Singapore 639798, Singapore
5
13 Defence Science and Technology Agency, 1 Depot Road # 03-01J, Singapore
14 109679, Singapore
6
15 Department of Civil Engineering, Monash University, Building 60, Clayton,
16 Melbourne, VIC 3800, Australia
17
18 *
Corresponding author. Tel.: +852 34008447
19 E-mail address: jianbo.zhu@polyu.edu.hk (J.B. Zhu).

1
20 Abstract: In this study, numerical modeling on the damage of existing circular
21 tunnel subject to blast-induced shock wave was carried out with DEM-based code
22 UDEC. The disturbed zones including failure zones, open zones and shear zones
23 around circular tunnel and peak particle velocities (PPVs) at tunnel surface are
24 employed to analyze the damage of tunnel. The effects of joint spatial and
25 mechanical properties, initial stress of rock mass, and magnitude of shock wave
26 amplitude to damage of tunnel were evaluated in this study. The difference of
27 damage between non-supported circular tunnel and bolt-supported circular tunnel
28 subject to the same blast-induced shock wave was also studied. It is found that the
29 orientations of joints in rock mass around the tunnel have great effects on tunnel
30 damage. The initial stress around tunnel has relatively small influence on tunnel
31 damage. The bolt support could greatly increase the stability of tunnel by
32 changing the vibration form of particle velocity rather than the decreasing of PPV.

33

34 Keywords: Tunnel damage, Disturbed zones, Peak particle velocities, Joints,

35 Initial stress, Scaled distance, Bolts

2
36 1 Introduction

37 An explosion, such as accidental explosion, drill and blast excavation or weapon

38 attack etc., has large effect to adjacent underground structures. Generally, the
39 effect includes overpressure, thermal effects, energized projectiles (fragments,
40 debris, and missiles), ground shock, and caterings (Ronald et al., 2010).
41 Especially, the ground shocks are of great interests to engineers concerning the
42 design of underground and surface structures. Kutter et al. (1988) noted that the
43 direct loading by a shock wave created due to explosion is the principal
44 mechanism to cause damage to underground structures. Therefore, research on
45 effects of blast-induced shock wave to underground tunnel damage is both
46 significant and practical (Zhao et al., 1999).
47
48 Studies have shown that the peak particle velocity (PPV) is the most
49 representative parameter to describe the ground motion and tunnel response
50 (Dowding, 1984). Extensive studies on damage of non-supported underground
51 tunnel in terms of definition of tunnel damage and the threshold values of PPVs
52 have been performed (Persson, 1997; Li and Huang, 1994; Hendron, 1977;
53 Coates, 1981; Kartuzov, et al., 1975; Oriad, 1972; Philips et al., 1992; Siskind,
54 1997). For damage of rock tunnel with support, studies carried out by Stjern and
55 Myrvang (1998) and Ortlepp and Stacey (1998) have shown that PPVs up to 1 m/s
56 will not cause any measurable damage to the tunnel. For lined tunnels, Dowding
57 (1984) suggested that the threshold value of PPV was roughly double that for
58 unlined tunnels.
59
60 However, most of the definitions are not well defined and various terms,
61 describing of damage, often have significant differences in definition and practical
62 meaning (Zhou, 2011). Little considerations of effects of discontinuities on tunnel
63 damage is included in these experiment-based studies. Generally, rock mass
64 contains various discontinuities such as bedding planes, foliation, faults, and
65 joints. The behavior (deformation characteristics, stress development, etc.) of rock
66 mass around the tunnel is mainly controlled by the spatial and mechanical
67 properties of the discontinuities (Tülin, 2009).
68

3
 69 Model tests, analytical methods, and numerical methods are generally employed
70 to evaluate the behavior of rock mass around tunnel in jointed rock masses. Few
71 model tests were performed in jointed rock masses because of the limitation in
72 joints setting. Analytical methods, considering block behavior in tunneling, are
73 mainly based on the block theory (Goodman and Shi, 1985). Although the
74 capability of the analysis is increased with improvement of the block theory with
75 respect to stress conditions, decrease of forces acting on the key block (Brady and
76 Brown, 2004) and use of sophisticated joint models (Pötsch et al., 2002), they are
77 not applicable to cases where ground shows stress induced failure or more
78 complicated problems are involved. Compared with theoretical and experimental
79 studies, numerical modeling provides a convenient, economical approach to study
80 underground explosions, especially for complicated cases where experiments are
81 difficult and expensive to conduct and theoretical solutions are impossible to
82 derive (Zhu et al., 2011).
83
84 In this study, numerical modeling on the damage of existing circular tunnel
85 subject to blast-induced shock wave in jointed rock mass was performed with
86 DEM-based code UDEC. The disturbed zones around tunnel and PPVs at tunnel
87 surface were employed to analysis the damage of tunnel. The aim of this study is
88 to evaluate the effects of joint spatial and mechanical properties, initial stress of
89 rock mass, magnitude of shock wave amplitude, and bolt supports to damage of
90 tunnel.

91 2 UDEC Model

92 In this study, before performing UDEC modeling, an AUTODYN-2D modeling

93 was carried out firstly to generate blast-induced shock waves, which would be
94 applied as the velocity boundary conditions in UDEC model. The shock waves
95 were generated by detonation of high explosive TNT. Fig.1 shows the
96 configuration of AUTODYN-2D model, the width and height of this model are
97 both 30 m. The radius of TNT material r, as shown in Eq. (1), depends on the
98 scaled distance (SD) (Zhou, 2011) with assumption that the actual distance R is
99 fixed at 25 m.
100 SD = R / (m)1/3 = R / (4 r 3  / 3)1/3 (1)

4
101 where SD is the scaled distance, m/kg1/3, R is the actual distance from the
102 explosive center, m is the weight of explosive,  = 1,630 kg/m3 is the density of
103 TNT material. In this study, the scaled distance is assumed to be 0.5, 0.75, 1.0,
104 1.5, 2.5, and 5.0 m/kg1/3, respectively. The actual distance R (25 m) used for shock
105 wave generating is identical to the distance from upper boundary to the circular
106 tunnel in UDEC model as will be shown in Fig.4 (b). The gauge A, recording
107 time-velocity history, is located at bottom boundary 5m far from detonating point.
108 As will be shown in Fig.4 (c), the gauge distance (5 m) is identical to the radius of
109 circular boundary, upon which the shock wave is incident in UDEC model. The
110 rock material is assumed to be elastic with Young’s modulus E = 94.75 GPa,
111 Possion’s ratio  = 0.27 and density  = 2,620 kg/m3.
112
113 The equation of state (EOS) of TNT conform to the JWL state expressed as Eq.
114 (2) (AUTODYN, 2005),
 − R1V  − R2V 
115 p = A(1 − )e + B(1 − )e + E (2)
R1V R2V V

116 where p is pressure and V represents relative volume. The properties of TNT
117 material are taken from the AUTODYN built-in material library and listed in
118 Table 1 (AUTODYN, 2005).
119
120 Fig.2 shows particle velocity-time histories obtained from AUTODYN-2D
121 modeling at gauge A in terms of different scaled distances. For verifying the
122 availability of these wave forms, the empirical formula (Dowding, 1985; Zhou
123 and Ong, 1996) for prediction of PPV in coupled explosion condition is
124 employed:
0.5C 2.17 R − n
125 PPV = ( ) (3)
 C Q1/3
126 where  is the rock mass density, kg/m3, n is the constant for a certain
127 geology and explosion set up and assumed to be 1.77 (Zhou, 2011) in this study,
128 and C is propagation velocity of seismic wave in rock mass and can be
129 estimated by following equation:
130 C = ( K + 4G / 3)  (4)

131 where K and G are rock bulk and shear modulus, respectively. Fig.3 shows
132 the comparison of PPVs between AUTODYN-2D modeling results and
5
133 predictions from empirical formula. It can be seen that the PPVs obtained from
134 AUTODYN-2D modeling agree well with those predicted by empirical formula.
135
136 Fig.4 shows the configuration of UDEC models. The width and height of the
137 model are both 50 m. The center of the circular tunnel is located in the middle in
138 the horizontal direction and 17m in height. There exist two joint sets. The process
139 of UDEC modeling contains three sequenced stages. In the first stage, as shown in
140 Fig.4 (a), initial stresses are generated, where zero-velocity boundary conditions
141 are applied at the bottom and lateral boundaries, while the top boundary is
142 assumed as stress boundary. The redistribution of stress was numerical modeled in
143 the second stage as shown in Fig.4 (b) after excavation of circular tunnel was
144 conducted. In the third stage, modeling on tunnel damage subject to blast-induced
145 shock wave is performed. As shown in Fig.4 (c), except the boundary condition at
146 upper circular boundary, other boundary conditions are changed to non-reflection
147 conditions, which are needed for wave propagation modeling. The particle
148 velocity-time histories, obtained from AUTODYN-2D modeling, are applied as
149 the velocity boundary condition at upper circular boundary. The joints are
150 assumed to behave Mohr-Coulomb criteria ones, while the blocks are treated to be
151 elastic. The properties of the rock blocks and joints are listed in Tab. 2.
152
153 In this study, the damping of rock material is neglected considering wave
154 attenuation in jointed rock masses are mainly caused by rock joints, compared
155 with rock material (King et al. 1986; Cai and Zhao 2000, Pyrak-Notel 1990).
156
157 Disturbed zones around tunnel and PPVs at circular tunnel surface are the two
158 major aspects considered in this numerical study. The disturbed zones are regions
159 where the original state of the in situ rock mass, such as stress, strain, rock
160 stability, water flow, etc., has been affected. However, the definition of the
161 disturbed zone depends on the nature or the purpose of the tunnel (Shen and
162 Barton, 1997).
163
164 In case of tunnels located in jointed rock mass, the joints play a key role in the
165 development of the disturbed zone. Joints can create loosen blocks near the tunnel
166 profile and cause local instability (Goodman and Shi, 1985). In this study, it is

6
167 assumed that the tunnel damage is caused by the large displacements of joints
168 around the tunnel, so the damage of intact rock is neglected. The joint
169 displacements contain normal and shear displacements. In this study it is assumed
170 that the zones, in which the normal displacement and shear displacement are
171 larger than 0.003 m and 0.01 m respectively, are disturbed zones. In order to
172 comprehensively investigate the distribution laws of joint normal and shear
173 displacements around the tunnel in terms of different mechanical and geometrical
174 parameters, the disturbed zone is divided into three sub-zones in this study: (1) the
175 failure zone, where loose rock blocks fall into the tunnel; (2) the open zone, where
176 joints open up with the separation displacement larger than 0.003 m; and (3) the
177 shear zone, where joints experience a shear displacement larger than 0.01 m and
178 the separation displacement smaller than 0.003 m. The division of disturbed zone
179 refers to the study by Shen and Barton (1997).
180
181 In this numerical study, UDEC modeling for non-supported circular tunnel were
182 performed in terms of different combinations of joint dip angle, joint stiffness,
183 joint spacing, tunnel depth, lateral stress coefficient and scaled distance (SD). A
184 numerical modeling for bolt-supported circular tunnel was also performed to
185 investigate the difference of damage between non-supported and bolt-supported
186 tunnels subject to the same blast-induced shock waves.

187 3 Effect of joint properties on the damage of

188 circular tunnel

189 3.1 Joint orientation

190 In this numerical modeling on effects of joint orientation on damage of existing

191 circular tunnel subject to blast-induced shock wave, two intersected persistent
192 joint sets are included in the model. One joint set is fixed in horizontal direction,
193 while the orientation of the other joint set varies between 15o and 90o from
194 horizontal direction with an interval of 15o. Thus, the expression of “joint dip
195 angle” adopted in the following only represents the orientation of second joint set.
196 The configurations of different combinations of these two joint sets can be seen in
197 Fig.5. Joint spacing is 0.5 m. The normal and shear stiffness of the joints are equal
198 to 50 GPa and 25 GPa, respectively. The depth of tunnel and lateral stress

7
199 coefficient are assumed to be 25 m and 0.8, respectively. The scaled distance of
200 blast-induced shock wave employed in this numerical modeling is 0.75 m/kg1/3.
201
202 Fig.5 and Fig.6 show disturbed zones around circular tunnel and PPVs at tunnel
203 surface in terms of different joint dip angles, respectively, after 20000 numerical
204 steps (about 75 ms). The open zones are marked by red color, while the color of
205 shear zones is blue in this study. It can be seen that in the first and second cases
206 (joint dip angle is 15o and 30o, respectively), no visible failure zones occur. When
207 joint dip angle reaches 45o or larger as shown in Fig.5 (c, d, e and f), the failure
208 zones are distinct and the size of failure zones increases with increasing joint dip
209 angle. The open zones and shear zones both increase obviously when joint dip
210 angle changes from 15o to 30o. However, in cases of joint dip angle equal to 75o
211 and 90o, as shown in Figs. 5(e) and 5(6), the size of shear zones decreases
212 obviously because of the existence of larger failure and open zones around
213 circular tunnel. As Fig.6 shows, the maximum PPV occurs at the tunnel roof when
214 dip angles are equal to 15o, 30o, 45o and 90o, while the left roof (225o in Fig.6)
215 experiences the maximum PPV when joint dip angle is 60o or 75o. It can be seen
216 from Fig.5 that the locations, where maximum PPV occurs, experience serious
217 damage. Also it can be seen that the PPV at tunnel roof decreases when joint dip
218 angle increases from 15o to 45o.

219 3.2 Joint stiffness

220 Effect of joint stiffness on damage of existing circular tunnel subject to blast-
221 induced shock wave is also numerically studied with UDEC. The joint normal
222 stiffness kn (kn = 2ks) was assumed to be 5, 10, 20, 50 and 100 GPa for both joint
223 sets, while the joint dip angle and joint spacing are equal to 60o and 0.5 m,
224 respectively. The parameters of tunnel depth, lateral initial stress coefficient and
225 scaled distance of blast-induced shock wave employed in this numerical modeling
226 are the same as those used in Section 3.1. In this study, the value of 100 GPa/m is
227 higher than the elastic modulus. However, in the real rock tunnel projects, the
228 joints are filled with stuffing, which could not be modeled directly with UDEC
229 code. The different stuffing may cause large difference of wave transmission, so it
230 is reasonable that the joint stiffness is higher than the elastic modulus of the rock
231 block when considering the effect of stuffing in real rock tunnel project.

8
232
233 Fig.7 shows the disturbed zones around circular tunnel in terms of different joint
234 stiffness after 20000 numerical steps. It can be seen that all the failure zones, open
235 zones and shear zones increases with increasing joint stiffness. Fig.8 shows the
236 PPVs at tunnel left roof (225o in Fig.6) in terms of different joint normal stiffness,
237 where the maximum PPV occurs. It can be seen that the PPV is increased with
238 increasing of joint stiffness, which could be a powerful explanation of the
239 difference of disturbed zones around tunnel as shown in Fig.7.

240 3.3 Joint spacing

241 In this numerical modeling of investigating the effect of joint spacing on damage
242 of existing circular tunnel subject to blast-induced shock wave, the joint spacing
243 varies from 0.25 m to 1.0 m with interval of 0.25 m for both two joint sets. The
244 joint dip angle and joint normal stiffness kn (kn = 2ks) are equal to 60o and 50 GPa,
245 respectively. The parameters of tunnel depth, lateral initial stress coefficient and
246 scaled distance of blast-induced shock wave employed in this numerical modeling
247 are also the same as those used in Section 3.1.
248
249 Fig.9 shows the disturbed zones around circular tunnel in terms of different joint
250 spacing after 20000 numerical steps. It can be seen that in the first case where
251 joint spacing is assumed to be 0.25 m, the size of all failure zones, open zones and
252 shear zones are the smallest among all four cases. The open zones and shear zones
253 are increase with increasing joint spacing. The size of failure zones is almost the
254 same when joint spacing changes from 0.5 m to 1.0 m. Fig.10 shows the PPVs at
255 tunnel left crown in terms of different joint spacing. It can be seen that the PPV
256 increases with increasing of joint spacing when joint spacing is less than 0.5 m.
257 However, the PPV tend to be decreased when joint spacing is larger than 0.5 m.

258 4 Effect of initial stress distribution on tunnel

259 damage

260 4.1 Tunnel depth

261 The depth of tunnel is assumed to be 25, 50, 100, 200 m, respectively, and the
262 lateral initial stress coefficient is fixed to be 0.8. The joint spacing is 0.5 m for

9
263 both two joint sets, while the joint dip angle and joint normal stiffness kn (kn = 2ks)
264 are equal to 60o and 50 GPa, respectively.
265
266 Figs.11 and 12 show the disturbed zones around tunnel and PPVs at tunnel
267 surface, respectively, in terms of different tunnel depth, where the scaled distance
268 is assumed to be 2.5 m/kg1/3 after 20000 numerical steps. It can be seen that the
269 size of failure zones is largest when tunnel depth is equal to 25 m, and it keeps
270 almost the same when the tunnel depth is 50 m or larger. However, different
271 depths result in different distributions of open zones and shear zones. The
272 maximum PPVs occur at the tunnel roof and decrease with increasing tunnel
273 depth as shown in Fig.12. Fig.13 shows the PPVs at tunnel surface in terms of
274 different tunnel depths, where all parameters adopted are the same as those in
275 Fig.12 except the scaled distance is 0.75 m/kg1/3. It can be seen that the influence
276 of depth to PPVs at tunnel surface can be ignored. It indicates that the influence of
277 tunnel depth on PPVs decreases or even disappears when the scaled distance
278 decreases.

279 4.2 Lateral initial stress coefficient

280 The lateral initial stress coefficient (K) is assumed to be 0.2, 0.5, 0.8, 1.0, 1.5, and
281 2.0, respectively, and the tunnel depth is 25 m. The joint spacing is 0.5 m for both
282 two joint sets, while the joint dip angle and joint normal stiffness kn (kn = 2ks) are
283 60o and 50 GPa, respectively.
284
285 Fig.14 shows the PPVs at tunnel surface in terms of different lateral initial stress
286 coefficient when the scaled distance was assumed to be 5.0 m/kg1/3 after 20000
287 numerical steps. The maximum PPVs occur at the tunnel left roof and decreases
288 with the increasing lateral initial stress coefficient. However, this difference of
289 PPVs does not cause any difference of tunnel disturbed zones when magnitude of
290 PPVs less than 1.0m/s as shown in Fig.15. Similar to tunnel depth, the influence
291 of lateral initial stress coefficient on PPVs at tunnel surface decreases or is
292 negligible with decreasing of scaled distance. Fig.16 shows the PPVs at tunnel
293 surface in terms of different lateral initial stress coefficient where the scaled
294 distance is assumed to be 0.75 m/kg1/3. It can be seen that the influence of lateral
295 initial stress coefficient to PPVs at tunnel surface can be ignored.

10
296 5 Effect of scaled distance on the tunnel damage

297 This numerical modeling is performed to investigate the damage of existing

298 circular tunnel subject to blast-induced shock wave with different scaled
299 distances. The joint spacing, joint dip angle and joint normal stiffness kn (kn = 2ks)
300 are equal to be 0.5 m, 60o and 50 GPa, respectively. The tunnel depth and lateral
301 initial stress coefficient employed in this numerical modeling are assumed to be
302 25 m and 0.8, respectively.
303
304 Fig.17 shows the disturbed zones around circular tunnel in terms of different
305 scaled distance after 20000 numerical steps. It can be seen that the size of failure
306 zones, open zones or shear zones decreases with increasing scaled distance. Fig.18
307 shows the PPVs at tunnel roof versus scaled distances obtained from UDEC
308 modeling and empirical formula (Dowding, 1985; Zhou and Ong, 1996) at same
309 locations. It can be seen that the numerical results of PPVs agree well with those
310 predicted from empirical formula. The PPVs at tunnel roof increase
311 approximately 2.0 times after excavation of tunnel is conducted. All these two
312 groups of results present a tendency that the PPVs decrease with increasing scaled
313 distance, which could be a powerful explanation to the difference of disturbed
314 zones around tunnel as shown in Fig.17.

315 6 Numerical modeling on the damage of existing

316 circular tunnel with bolts support

317 The support for tunnel could restrict tunnel from damage. In this numerical
318 modeling, the tunnel will be strengthened by bolts with different bolt length and
319 bolt number. To facilitate the comparison, the parameters adopted in this section,
320 such as joint dip angle, joint spacing, joint stiffness, tunnel overburden and lateral
321 initial stress coefficient etc., are the same as those used in Section 5, except that
322 the scaled distance is fixed to be 0.75 m/kg1/3. The parameters of bolts are listed in
323 Table 3.
324
325 Fig.19 shows the disturbed zones around bolt-supported circular tunnel in terms of
326 different bolt length, while Fig.20 presents the disturbed zones around bolt-
327 supported circular tunnel in terms of different bolt number. It can be seen in
11
328 Fig.19 that the bolt length must be longer than the dimension of failure zones in
329 non-supported condition (approximate 2 m) to prevent the blocks from collapse
330 except dropping of several discrete blocks. The size of open zones and shear
331 zones decrease when bolt length increases from 2m to 4m. This is because the
332 increasing of bolt length decreases the displacement of rock at tunnel roof in
333 downward vertical direction. To investigate the influence of bolt number on
334 tunnel damage subject to blast-induced shock wave, the bolt length is fixed to be 2
335 m as shown in Fig.20. It can be seen that the size of failure zones, open zones or
336 shear zones obviously decrease with increasing bolt number. When bolt number is
337 as large as 17 as shown in Fig.20(c), no block dropping appears at the tunnel
338 crown. As shown in Fig.21, the PPVs at tunnel surface are also investigated. It can
339 be seen that the support of bolts almost has no effect to the magnitude of PPV at
340 tunnel surface. However, the existing of bolts changes the vibration form of
341 particle velocities. Fig.22 and Fig.23 show the wave forms of particle velocity at
342 left roof of bolt-supported tunnel with considering of different bolt length and bolt
343 number, respectively. It can be concluded that the decreasing of bolt length or bolt
344 number results in the vibration of particle at a higher velocity after the first peak
345 (PPV).

346 7 Discussion and Conclusions

347 In this numerical study, the damage of tunnel is mainly caused by the slipping and
348 opening of rock joints. This is consistent with field observations that initial tunnel
349 damage is a result of falling loose rocks, rather than damage of rock material
350 created by the shock wave (Zhou, 2011).
351
352 It is extremely important to have a common understanding of the definition of
353 dynamic damage before any meaningful discussion of tunnel damage can be
354 made. Unfortunately, there currently exist no established damage criteria for rock
355 and rock support (Zhou, 2011). PPV has been shown to be the most representative
356 parameter when describing the ground motion and tunnel response (Dowding,
357 1984). Some researches on this issue give definitions of tunnel damage and the
358 threshold value of PPVs. Persson (1997) gives damage criteria for Swedish hard
359 rock as listed in Table 3. Li and Huang (1994) discussed the damage criteria for
360 rock tunnel (Table 4). Additionally, Zhou (2011) suggested that visible tunnel

12
361 damage in competent rock will not occur until the incipient PPV reaches about 1-
362 2 m/s, while Hendron (1977) stated that serious spalling (or damage) does not
363 occur until the PPV reaches 4 m/s. These different damage criteria have been
364 employed for further discussion of the numerical modeling results in this study.
365
366 For numerical modeling of effect of scaled distance (SD) on tunnel damage
367 subject to blast-induced shock wave, when the scaled distance is assumed to be
368 5.0, where the PPV is approximate 0.45 m/s, almost no damage occurs. This is
369 consistent with the definitions of “no damage” by Persson (1997) and Li and
370 Huang (1994). When the scaled distance is decreased to 2.5 (PPV = 2.65 m/s),
371 although there are no blocks falling down into the tunnel, the obvious open zones
372 with height of 1.5m appear, as shown in Fig.17. The tunnel is seriously damaged
373 with appearing of failure zones when the PPV reaches 6.5 m/s or larger as shown
374 in Fig. 18 (a, b, c, and d). The present findings with respect to tunnel damage
375 agree better with conclusions by Persson (1997), Zhou (2011) and Hendron
376 (1977) than the suggestions by Li and Huang (1994).
377
378 However, the above numerical results are obtained by fixing the joint dip angle to
379 be 60o. The joint orientation also has great effects on the damage of tunnel, as
380 shown in Section 3.1. Some of the numerical results in Section 3.1 could not agree
381 well with the suggestions by Persson (1997), Zhou (2011), Hendron (1977) or Li
382 and Huang (1994). When joint dip angle is equal to 15o or 30o, no obvious
383 damage occurs even though the PPV reaches 22.4 m/s where scaled distance is
384 assumed to be 0.75 m/kg1/3. This is because the slipping of joints around tunnel
385 becomes difficult when the angle between wave transmission direction and
386 normal direction of joint plane is smaller than joint friction angle (35o in this
387 study). The damage zones increases with joint dip angle is because wave
388 attenuation generally decreases with increasing joint dip angle (as shown in Fig.
389 6). This is coincident with the conclusion of in situ tests that waves propagating
390 along the path across joints have much smaller amplitudes than those propagating
391 along the direction parallel to the joints (King et al. 1986; Hao et al. 2001).
392 Therefore, when giving the prediction of rock tunnel damage, the joint orientation
393 must not be ignored.
394

13
395 In this study, larger joint stiffness causes larger PPV at tunnel surface and
396 disturbed zones around tunnel as shown in Fig.7 and Fig.8. This is mainly because
397 that the larger joint stiffness results in larger shock wave transmission coefficient
398 if other parameters are fixed. However, the extent of increasing of size of
399 disturbed zone is much smaller than that of increasing of PPV at tunnel surface.
400 This phenomenon is mainly because that higher stiffness limits the joints from
401 opening up. In this special case, stiffer joints are more vulnerable to damage.
402 However, other parameters such as explosive weight, the distance between
403 explosion source and tunnel, numbers of joints and joint spacing also control the
404 tunnel damage. For example, if the explosive weight is sufficiently small, stiffer
405 joints may be more secure for tunnel.
406
407 Different from the joint stiffness, the increasing of joint spacing does not result in
408 increasing or decreasing tendency of the size of disturbed zones around tunnel and
409 PPV at tunnel surface. This is because when incident wave propagation through a
410 rock mass with fixed distance, the smaller joint spacing means larger joint number
411 in this rock mass. As shown in Fig.25, both the joint spacing and joint number
412 have influence to wave propagation in jointed rock mass.
413
414 The tunnel initial stress around tunnel has relatively smaller influence on tunnel
415 damage. This is mainly because that if the amplitude of stress wave arriving at the
416 tunnel is smaller or slightly larger than initial stress around tunnel (SD=5.0 in this
417 study), even though the difference of PPVs at tunnel surface are large, the tunnel
418 experiences no different obvious damage. When the amplitude of stress wave
419 arriving at the tunnel is much larger than initial stress around tunnel (SD=0.75 in
420 this study), from the numerical modeling results as shown in Fig.13 and Fig.16, it
421 can be found that the initial stress around tunnel has little influence to PPV at
422 tunnel surface.
423
424 The bolt-supported tunnel experiences relatively small damage compared to the
425 non-supported tunnel. This phenomenon, caused by the different vibration forms
426 of particle velocities as shown in Fig.22 and Fig.23, can be explained by the fact
427 that in dynamic rock support system, bolts are also able to absorb a certain
428 amount of the released energy through their deformations (Li, 2011).

14
429

430 Acknowledgments

431 Deng XF receives financial support from the China Scholarship Council and the
432 Chinese Central University Fundamental Research Funds (SWJTU11ZT33) for
433 the work performed at EPFL Switzerland and NTU Singapore.

434 References
435 AUTODYN, 2005. Revision 4.3 Century Dynamics.
436
437 Brady, B., Brown, H.G., 2004. Rock mechanics for underground mining. Kluvier Academic
438 Publishers, Netherland 255-261.
439
440 Cai JG, Zhao J (2000) Effects of multiple parallel fractures on apparent wave attenuation in
441 rock masses. Int J Rock Mech Min Sci 37: 661-682
442
443 Coates, D.F., 1981. “Rock Mechanics Principles,” Monograph 874, Energy, Mines, and
444 Resources Canada.
445
446 Dowding, C.H., 1984. Estimating earthquake damage from explosion testing of full-scale
447 tunnels. Adv. Tunnel Technology and Subsurface Use 4(3), 113-117.
448
449 Dowding, C.H., 1985. Blasting vibration monitoring and control. Prentice-Hall, NJ, USA.
450
451 Goodman, R.E., Shi, G.H., 1985. Block theory and its application to rock engineering.
452 Prentice-Hall International, London.
453
454 Hao, H., Wu, Y.K., Ma, G.W., Zhou, Y.X., 2001. Characteristics of surface ground motions
455 induced by blasts in jointed rock mass. Soil Dyn Earthq Eng 21:85–98.
456
457 Hendron, A.J., 1977. Engineering rock blasting in civil projects. In: Structural and
458 Geotechnical Mechanics: A Volume Honouring Hathan M. Neumark (ed. W J Hall), Prentice
459 Hall 242-277.
460
461 Kartuzov, M.I., Abramov, N.A., Pazdnikov, N.V., Sinel'nikov, L.M., Kaimashnikov, G.S.,
462 1975. Stability of underground mine workings with various types of supports when subjected
463 to blasting. J Min Sci 11(2), 99-103.

15
464
465 King, M.S., Myer, L.R., Rezowalli, J.J., 1986. Experimental studies of elastic-wave
466 transmission in a columnar-jointed rock mass. Geophys Prospect 34:1185–1199.
467
468 Kutter, B.L., O’Leary, L.M., Thompson, P.Y., Lather, R., 1988. Gravityscaled Tests on Blast-
469 induced Soil-structure. Interaction Journal of Geotechnical Engineering, American Society of
470 Civil Engineers 114(4), 431-447.
471
472 Li, C.C., 2011. Rock support for underground excavations subjected to dynamic loads and
473 failure. In: Zhou, Y. X., Zhao, J. editors. Advances in rock dynamics and applications.
474 London, Taylor and Francis Group 483-506.
475
476 Li, Z., Huang, H., 1994. The calculation of stability of tunnels under the effects of seismic
477 wave of explosions. Proceedings, 26th Department of Defence Explosives Safety Seminar,
478 Department of Defence Explosives Safety Board.
479
480 Oriad, L.L., 1972. Blasting effects and their control in open pit mining. In Geotechnical
481 Practice for Stability in Open Pit Mining, Society of Mining Engineers, New York 197-222.
482
483 Ortlepp, W.D., Stacey T.R., 1998. Performance of Tunnel Support underground Large
484 Deformation Static and Dynamic Loading. Tunn. Undergr. Sp. Tech 13(1), 15-21.
485
486 Persson, P.A., 1997. The relationship between strain energy, rock damage, fragmentation, and
487 throw in rock blasting. International Journal of Blasting and Fragmentation 1, 99-110.
488
489 Phillips, J.S., Luke, B.A., Long, J.W., Lee, J.G., 1992. “MISTY ECHO Tunel Dynamics
490 Experiment – Data Report, Volume 1” Sandia National Laboratories Report SAND89-0972/1,
491 April 1992.
492
493 Pötsch, M., 2002. Influence of the three dimensional stress condition at the tunnel face on the
494 stability of removable block, Diploma Thesis. Graz University of Technology.
495
496 Pyrak-Nolte LJ, Myer LR, Cook NGW (1990) Anisotropy in seismic velocities and
497 amplitudes from multiple parallel fractures. J Geophys Res 95:11345-11358
498
499 Reed, M.B., Grasso, P., Rizzi, D., Rabajoli, G., 1993. Improvement of rock properties by
500 bolting in the plastic zone around a tunnel: a uumerical study. Int J Rock Mech Min Sci &
501 Geomech. Abstr 30(5): 567-571.
502

16
503 Ronald, P., Kim, R., Mniszewski, A.L., 2010. Explosion Phenomena and Effects of
504 Explosions on Structures. I: Phenomena and Effects. Pract. Period. Struct. Des. Constr 15,
505 135-140.
506
507 Shen, B. Barton, N., 1997. The disturbed zone around tunnels in jointed rock masses. Int J
508 Rock Mech Min Sci 34(1), 117-125.
509
510 Siskind, D.E., 1997. Vibration damages in industrial situations. Journal of Explosives
511 Engineering, Nov/Dec/1997:40.
512
513 Stjern, G., Myrvang, A., 1998. The influence of blasting on grouted rock bolts. Tunn Undergr
514 Sp Tech 3(1), 65-70.
515
516 Tülin, S., 2009. Ground behavior evaluation for tunnels in blocky rock masses. Tunn Undergr
517 Sp Tech 24, 323-330.
518
519 Zhao, J., Zhou, Y.X., Hefny, A.M., Cai, J.G., Chen, S.G., Li, H.B., Liu, J.F., Jain, M., Foo,
520 S.T., Seah, C.C., 1999. Rock dynamics research related to cavern development for
521 ammunition storage. Tunn Undergr Sp Tech 14:513-526.
522
523 Zhou, Y., Ong, Y.H., 1996. Ground shock prediciton methods-a critical appraisal.
524 Proceedings of the 1st Asia-Pacific Conference on Shock and Impact Loads on Structures
525 477-483.
526
527 Zhou, Y.X., 2011. Explosion loading and tunnel response. In: Zhou, Y. X., Zhao, J. editors.
528 Advances in rock dynamics and applications. London, Taylor and Francis Group 457-481.
529
530 Zhu, J.B., Deng, X.F., Zhao, X.B., Zhao, J., 2012. A numerical study on wave transmission
531 across multiple intersecting joint sets in rock masses with UDEC. Rock Mech. Rock Eng.
532 DOI 10.1007/s00603-012-0352-9.
533
534 Zhu, J.B., Zhao, G.F., Zhao, X.B., Zhao, J., 2011. Validation study of the distinct lattice
535 spring model (DLSM) on P-wave propagation across multiple parallel joints. Comput
536 Geotech 38:298-304.
537

17
 Y

rock

30.0m

Detonating point

TNT

Gauge A
X
r
5.0m
30.0m

Fig. 1 Configuration of AUTODYN-2D model for blast-induced wave

generating

18
 400 Scaled distance=0.5 Scaled distance=0.75
Scaled distance=1.0 Scaled distance=1.5
Scaled distance=2.5 Scaled distance=5.0
30
300
20
Scaled distance=2.5

Velocity (m/s)
Velocity (m/s) 10 Scaled distance=5.0
200 0

-10

-20
100
-30
0.6 0.8 1.0 1.2 1.4
Time (ms)
0

-100
0 1 2 3 4 5
Time (ms)
Fig. 2 Particle velocity-time histories obtained from AUTODYN-2D modeling
at gauge A in terms of different scaled distances

19
 1000

Empirical formula

Peak particle velocity (m/s)

AUTODYN modelling
100

R
PPV = 5.74( ) −1.77
10 Q1/3

1
0.1 1 10
1/3
Scaled distance from center of explosive (m/kg )

Fig. 3 Comparison of PPV between numerical modeling results and empirical

formula

20
 Stress boundary

Jointed rock mass

50m

Zero-velocity boundary
50m

(a)

25m
50m

Existing circular tunnel

8m

17m

(b)
10m
Non-reflection boundary

Incidence of shock wave

(c)
Fig. 4 Configuration of UDEC models for sequenced numerical modeling
stages: (a) Generating initial stress; (b) Calculating of secondary stress
distribution after excavation; (c) Modeling on the stability tunnel subject to blast-
induced shock wave

21
 (a) (b)

(c) (d)

(e) (f)

Fig. 5 Disturbed zones around tunnel in terms of different joint dip angles: (a)
15o; (b) 30o; (c) 45o; (d) 60o; (e) 75o; (f) 90o

22
 90
o
180o joint dip angle=15
80 Tunnel o
joint dip angle=30
o
joint dip angle=45
70 270o 90o joint dip angle=60
o

Peak particle velocity (m/s)

o
joint dip angle=75
60 joint dip angle=90
o
0o
50

40

30

20

10

0
0 45 90 135 180 225 270 315 360
o
Angle from downward vertical ( )
Fig. 6 Peak particle velocities at circular tunnel surface in terms of different
joint dip angles

23
 (a) (b)

(c) (d)

(e)

Fig. 7 Disturbed zones around circular tunnel in terms of different joint normal
stiffness (kn = 2ks): (a) 5 GPa/m; (b) 10 GPa/m; (c) 20 GPa/m; (d) 50 GPa/m; (e)
100 GPa/m

24
 40

Peak particle velocity at tunnel crown (m/s)

35

30

25

20

15

10

0
0 20 40 60 80 100
Joint normal stiffness (GPa/m)
Fig. 8 Peak particle velocities at tunnel left roof (225o in Fig.6) in terms of
different joint normal stiffness

25
 (a) (b)

(c) (d)

Fig. 9 Disturbed zones around tunnel in terms of different joint spacing: (a)
0.25m; (b) 0.5m; (c) 0.75m; (d) 1.0m

26
 25

Peak particle velocity at tunnel crown (m/s)

20

15

10

0
0.2 0.4 0.6 0.8 1.0
Joint spacing (m)

Fig. 10 Peak particle velocities at tunnel left crown (225o in Fig.6) in terms of
different joint spacing

27
 (a) (b)

(c) (d)

Fig.11 Disturbed zones around circular tunnel in terms of different tunnel depth:
(a) 25m; (b) 50m; (c) 100m; (d) 200m

28
 3.5
Depth of tunnel=25m
Depth of tunnel=50m
3.0
Depth of tunnel=100m
Depth of tunnel=200m

Peak particle velocity (m/s)

2.5

2.0

1.5

1.0

0.5

0.0
0 45 90 135 180 225 270 315 360
o
Angle from downward vertical ( )
Fig. 12 Peak particle velocities at circular tunnel surface in terms of different
tunnel depth (SD = 2.5 m/kg1/3)

29
 35

Depth of tunnel=25m
30 Depth of tunnel=50m
Depth of tunnel=100m

Peak particle velocity (m/s)

Depth of tunnel=200m
25

20

15

10

0
0 45 90 135 180 225 270 315 360
o
Angle from downward vertical ( )
Fig. 13 Peak particle velocities at circular tunnel surface in terms of different
tunnel depth (SD = 0.75 m/kg1/3)

30
 1.5

K=0.2
K=0.5
K=0.8

Peak particle velocity (m/s)

K=1.0
K=1.5
1.0
K=2.0

0.5

0.0
0 45 90 135 180 225 270 315 360
o
Angle from downward vertical ( )

Fig. 14 Peak particle velocities at circular tunnel surface in different conditions

of lateral stress coefficient K (SD = 5.0 m/kg1/3)

31
Fig.15 Disturbed zones around circular tunnel in the situation of combination of
K = 0.5 and SD = 5.0 m/kg1/3

32
 30

K=0.2
25 K=0.5
K=0.8

Peak particle velocity (m/s)

K=1.0
K=1.5
20
K=2.0

15

10

0
0 45 90 135 180 225 270 315 360
o
Angle from downward vertical ( )

Fig. 16 Peak particle velocities at circular tunnel surface in different conditions

of lateral stress coefficient K (SD = 0.75 m/kg1/3)

33
 (a) (b)

(c) (d)

(e) (f)

Fig.17 Disturbed zones around circular tunnel in terms of different SDs

(m/kg1/3): (a) 0.5; (b) 0.75; (c) 1.0; (d) 1.5; (e) 2.5; (f) 5.0

34
 100

Peak particle velocity at tunnel crown (m/s)

10

UDEC
Empirical formula (Dowding, 1984)

0.1
0.1 1 10
1/3
Scale distance (m/(kg) )

Fig. 18 Peak particle velocities at circular tunnel crown in terms of different

scaled distances with comparison to those in situation of no excavation of tunnel
and results from empirical formula (Dowding, 1984)

35
 (a) (b)

bolt

(c) (d)

Fig. 19 Disturbed zones around bolt supported circular tunnel in terms of

different bolt length: (a) 1 m; (b) 2 m; (c) 3 m; (d) 4 m

36
 bolt

(a) (b)

(c)

Fig. 20 Disturbed zones around bolt supported circular tunnel in terms of

different bolt number: (a) 5; (b) 9; (c) 17

37
 30

L=1.0m, num=9
L=2.0m, num=9
25
L=3.0m, num=9
L=4.0m, num=9

Peak particle velocity (m/s)

L=2.0m, num=5
20 L=2.0m, num=17

15

10

0
0 45 90 135 180 225 270 315 360
o
Angle from downward vertical ( )

Fig. 21 Peak particle velocities at bolt supported circular tunnel surface in

different conditions of bolt length and number

38
 25
L=1.0m, num=9
L=2.0m, num=9
L=3.0m, num=9
20 L=4.0m, num=9

Velocity (m/s)
15

10

0 5000 10000 15000 20000

Step

Fig. 22 Wave forms at bolt supported tunnel left crown (225o in Fig.6) in terms
of different bolt length

39
 25
L=2.0m, num=5
L=2.0m, num=9
20 L=2.0m, num=17

Velocity (m/s)
15

10

0 5000 10000 15000 20000

Step

Fig. 23 Wave forms at bolt supported tunnel left crown (225o in Fig.6) in terms
of different bolt number

40
 0.7

0.6

0.5
lTl
0.4

0.3

0.2
0.0 0.5 1.0 1.5 2.0


Fig. 24 TN versus  for normally incident P wave transmission across two

joint sets with UDEC (Zhu et al., 2012)

41
 Table 1 Properties of TNT material used in AUTODYN-2D modeling
(AUTODYN, 2005)
Parameters Value
Density (g/mm3) 1.6
A (kPa) 3.7377e8
B (kPa) 3.7471e6
R1 4.15

R2 0.9

 0.35
Detonation velocity (m/s) 6930
Energy/unit volume (kJ/m3) 6.0e6
CJ pressure (kPa/m3) 2.1e7

42
 Table 2 Properties of rock blocks and joints
Property Value
Density (kg/m3) 2,620
Young’s Modulus (GPa) 94.75
Passion’s Ratio 0.27
Joint normal stiffness (GPa/m) 5, 10, 20, 50, 100
Joint shear stiffness (GPa/m) 2.5, 5, 10, 25, 50
Joint friction angle (o) 35
Joint cohesive (MPa) 6

43
 Table 3 Properties of bolt (Reed et al., 1993)
Property Value
Elastic modulus (MPa) 2.1 e +5
Shear modulus (MPa) 0.77 e +5
Grout shear modulus (MPa) 860

44
Table 4 Damage criteria for hard Scandinavian bed rocks (Persson, 1997)
Tensile Stress, Strain Energy,
PPV, m/s Typical effect
MPa J/kg
0.7 8.7 0.25 Incipient swelling
1.0 12.5 0.5 Incipient swelling
2.5 31.2 3.1 Fragmentation
5.0 62.4 12.5 Good fragmentation
15.0 187 112.5 Crushing

45
 Table 5 Tunnel damage criteria for unlined rock tunnels (Li and Huang, 1994)
Rock parameters Peak particle velocity, m/s
Rock Unit Compress Tensile No Slight Medium Serious
type weight strength strength damage damage damage damage
(g/cm3) (MPa) (MPa)
Hard 2.6-2.7 75-110 2.1-3.4 0.27 0.54 0.82 1.53
rock 2.7-2.9 110-180 3.4-5.1 0.31 0.62 0.96 1.78
2.7-2.9 180-200 5.1-5.7 0.36 0.72 1.11 2.09
Soft 2.0-2.5 40-100 1.1-3.1 0.29 0.58 0.90 1.67
rock 2.0-2.5 100-160 3.4-4.5 0.35 0.70 1.07 1.99

46

View publication stats