Tunnelling and Underground Space Technology 51 (2016) 189–201

Contents lists available at ScienceDirect

Tunnelling and Underground Space Technology

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

Frequency comparison of blast-induced vibration per delay for the full-

face millisecond delay blasting in underground opening excavation
J.H. Yang a,⇑, W.B. Lu b, Q.H. Jiang a, C. Yao a, C.B. Zhou a
a
School of Civil Engineering and Architecture, Nanchang University, Nanchang 330031, China
b
State Key Laboratory of Water Resources and Hydropower Engineering Science, Wuhan University, Wuhan 430072, China

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

Article history: Responses and damage to the structures subjected to blast-induced vibration are highly dependent on
Received 15 December 2014 vibration frequency. Understanding the vibration frequency characteristics and its formation mechanism
Received in revised form 19 June 2015 is essential for the safe and economic design of mining and construction blasts. When millisecond delay
Accepted 19 October 2015
blasting is used in excavation, a global frequency value of all of the delays of vibrations fails to present a
Available online 11 November 2015
true picture of the vibration frequency characteristics. In the present study, comparisons of frequency
characteristics for the blast-induced vibration per delay were first conducted via an underground opening
Keywords:
case excavated by the full-face millisecond delay blasting sequence. The results show that if the blasthole
Underground excavation
Millisecond delay blasting
geometry and charge structure are kept the same in each blast delay, the frequencies of single-delay
Vibration frequency vibration signals will decrease as the dimensions of the equivalent blasting vibration sources increase.
Free face However, the cutting blast used for the headmost holes inside is an exception, and its vibration frequency
Frequency spectrum is lower than that of the breaking blast outside. It is the authors’ belief that this counter-intuitive phe-
Wavelet packet decomposition nomenon is caused by the free faces created by the cutting blast for the breaking blast. A small-scale
blasting test and the related numerical simulation were subsequently performed to demonstrate this
belief. It is found that the vibration frequency from the holes in the free-face blast is higher than that
in the confined blast. In terms vibration frequency, therefore, the vibration from the breaking blast is less
harmful to structures compared to the cutting blast for a certain velocity, and decreasing the burden of
breaking blastholes is beneficial to reduce the vibrational damage of structures.
Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction
Rock fragmentation by drilling and blasting is widely used in
mining, quarrying and civil engineering constructions. However,
only 20–30% of the energy released during explosion is utilized
directly for rock breakage, while the rest of the energy is dissipated
in the form of ground vibrations, air blasts, flying rocks and noises
(Hagan, 1977; Trivedi et al., 2014). Among all of the ill effects,
blast-induced ground vibration adversely affects the integrity of
surrounding structures such as buildings, dams, roads, pit slopes,
and underground workings, and has become a major concern in
mining and construction blasts (Khandelwal, 2010; Verma and
Singh, 2013). When explosive charges are detonated in blastholes,
intense dynamic stresses are initiated around the blastholes due to
the instantaneous acceleration of rock masses caused by the deto-
nation gas pressure on hole walls (Singh et al., 2008). The dynamic
stresses spread away from the source into the surrounding rock
⇑ Corresponding author.
E-mail address: whuyjh@163.com (J.H. Yang).
masses in the form of wave motion, and the waves in the elastic
zone are known as ground vibration. It is well known that three
parameters of blast-induced vibration, i.e. peak particle velocity
(PPV), frequency and duration play an active role in the relation-
ship of blasting vibration and structural hazard. Some researchers
emphasized the importance of the frequency content in the blast-
ing vibration-induced structural damage because a resonance
occurs and the resultant vibration amplitude on the structures gets
amplified if the vibration has a frequency in the range of the natu-
ral frequency of the structures. Responses and damage to the struc-
tures subjected to blasting vibration of various frequencies have
been extensively studied by means of wave theory, field tests
and numerical simulation (Siskind, 1997; Ma et al., 2000; Ozer,
2008; Aldas, 2010; Singh and Roy, 2010; Sazid and Singh, 2013).
It is generally accepted that a low-frequency vibration has a
greater possibility of structural damage than a high-frequency
vibration for a certain velocity because the natural frequency of
the structures is generally below 10 Hz (Siskind, 1997; Ozer,
2008; Aldas, 2010; Singh and Roy, 2010). Ma et al. (2000) noted
that although high-rise buildings have a natural frequency far

http://dx.doi.org/10.1016/j.tust.2015.10.036
0886-7798/Ó 2015 Elsevier Ltd. All rights reserved.
190 J.H. Yang et al. / Tunnelling and Underground Space Technology 51 (2016) 189–201
away from the dominant frequency of the blast-induced vibration,
it is still vulnerable because of the concentrated damage at low sto-
ries, which greatly affects the overall structural stability.
Frequency characteristics of blast-induced vibration are influ-
enced by a variety of parameters, such as physico-mechanical
properties of rock masses, explosive characteristics, propagation
distances of waves, blasting geometry and sequence. It is well
known that each type of rock mass and each terrain has a domi-
nant transmission frequency that favors wave propagation for that
frequency (Singh and Roy, 2010; Alvarez-Vigil et al., 2012). The
dominant frequency of ground vibration on a soil surface is lower
than that on a rock surface, and the deeper the soil layer is, the
lower the dominant frequency of vibration is (Ma et al., 2000). In
general, the frequency content of blasting vibration decreases with
the increase in propagation distance. The Earth’s crust is consid-
ered as a low pass filter, hence the high-frequency vibration com-
ponent attenuates more rapidly than the low-frequency vibration
(Singh et al., 2008). The presence of discontinuities in rock masses
also contributes to the propagation of low-frequency vibration by
filtering the higher frequencies (Park et al., 2009; Li et al.,
2011b). Therefore, extra care must be taken for large blasts where
low-frequency energy may exceed vibration limits at far distances.
Explosive charge detonated per delay is one of the most important
parameters related to blast-induced vibration. Ling et al. (2005)
found that the ratio of the high-frequency energy to the total
energy decreases with the increase of the maximum charge per
delay. Ground vibration induced by different types of explosives
conducted experimentally by Leidig et al. (2010) shows that black
powder with relatively long blast duration and opening of long
fractures, accentuates the low-frequency energy of the ground
vibration, while Composition B causes more high-frequency energy
of vibration as the detonation gases are unable to drive the long
fractures. Yilmaz and Unlu (2013) stated that the most efficient
explosive in rock fragmentation is the one with a low frequency
content but a sufficiently high borehole wall pressure. There is a
popular conception in the spectral control of ground vibration that
a smaller delay interval generates a higher-frequency vibration,
and it will be more readily attenuated in the ground, thus reducing
the blasting vibration. However, Blair (2009) found that the delay
time and accuracy can be used to control the frequency content
of vibration just over a restricted distance range. Although some
progress has been made in the propagation and attenuation laws
of blasting vibration frequency, it is still much less compared to
the large amounts of research achievements in the PPV. Therefore,
it is significant to study the frequency characteristics and its mech-
anism of blast-induced vibration.
Because structural responses and damage characteristics are
highly dependent on the vibration level and dominant frequency,
to determine in which level of vibration the surrounding structures
are damaged, many blasting vibration standards and damage crite-
ria have been developed in terms of the PPV limits in combination
with the dominant frequency of vibration (Rai et al., 2005; Singh
and Roy, 2010; Aldas, 2010; Karadogan et al., 2014). According to
these standards and criteria, structural hazard assessments were
made by many authors using measured particle velocities and fre-
quency values in the field (Ozer, 2008; Nateghi, 2011; Caylak et al.,
2014; Yu et al., 2014).
In underground opening excavation, the full-face millisecond
delay blasting technique is generally used to reduce the charge
weight in a single fire for meeting the safety control criterion. It
should be noted that in the current evaluation of ground
vibration-structural hazard, a global frequency value which covers
the vibration waves in all of the delays is always adopted to be
compared with the frequency limits of the structural hazard stan-
dards and criteria. Actually, the global vibration frequency of all of
the delays is unable to represent the actual frequency characteris-
tics of blast-induced vibration very well because blasting geome-
try, charge characteristics and boundary conditions per delay of
blast are not strictly consistent. Most obviously, cutting blastholes
detonated in the first delay are blasted under a confined boundary
condition, while breaking blastholes and contour blastholes deto-
nated in the subsequent delays are blasted under free-face condi-
tions. This study focuses on the frequency comparison of blast-
induced vibration per delay for the full-face millisecond delay
blasting in underground opening excavation. In the present study,
analytical analyses and a case study are first conducted to find the
frequency variation of the blast-induced vibration per delay. Sub-
sequently, a small-scale blasting test and related numerical model-
ing are performed to explain the mechanism lying behind the
frequency variation.
2. Analytical frequency spectrum of blast-induced vibration
The study of a blasting vibration frequency begins with the ana-
lytical frequency spectrum of blast-induced seismic waves to
determine the factors influencing the vibration frequency. Blasting
technology using cylindrical charges has found widespread appli-
cation in underground opening excavation in mining, hydropower
and transportation industries. A cylindrical charge can be divided
into a series of spherical charges when analytically determining
the blast-induced stress field, for example by using the Starfield
superposition method. Thus, the frequency spectrum of blast-
induced vibration just for the spherical charge is introduced in this
section. It is generally accepted that during rock fragmentation by
blasting, a crushed zone and a cracked zone are generated around a
blasthole by explosion shock waves and subsequent explosion gas
pressure where rock masses are broken or damaged to a different
extent. The outer boundary of the cracked zone is located outside
the region of individual far-reaching fissures, and it is also the
beginning of the seismic zone where the propagation of stress
waves is nearly elastic. Blast-induced vibration in the far field
results from the propagation of elastic seismic waves. To analyti-
cally determine seismic wave propagation by the elastic wave the-
ory, the concept of equivalent radius of the blasting vibration
source is introduced here. In this concept, the crushed zone and
the cracked zone are treated as parts of the blasting vibration
source, and the blasting load is applied to the boundary of the
equivalent blasting vibration source. Let an equivalent source of
blasting vibration with radius re, be located in a homogeneous, iso-
tropic and elastic medium, characterized by the elastic constants of
Lame k and l and longitudinal wave velocity cp. At the boundary of
the equivalent source, the applied action pe(t) is (Kuzmenko et al.,
1993)
pe ðtÞ ¼ rðtÞ ð1Þ
Dynamic disturbances arising from this action satisfy the wave
equation
@ 2 u 2 @u 2u 1 @ 2 u
þ  ¼ 2 2 ð2Þ
@r 2 r @r r 2 cp @t
where u is the particle displacement, r is the distance from the
sphere center, and t is time.
The solution to Eq. (2) can be found in
@f ðs=rÞ f ðs=rÞ 1 @f ðs=rÞ
uðtÞ ¼ ¼ 2  ð3Þ
@r r rcp @ s
with s ¼ t  ðr  r e Þ=cp .
f(s/r) is an unknown function which can be established from
boundary conditions
 J.H. Yang et al. / Tunnelling and Underground Space Technology 51 (2016) 189–201 191


@u u blast has the same blasthole geometry and charge structure,
ðk þ 2lÞ þ 2k ¼ rðtÞ ð4Þ
@r r r¼re according to Eq. (8), the vibration frequencies per delay induced
by the full-face millisecond delay blasting sequence will decrease
Substitution of Eq. (3) into the boundary condition Eq. (4) yields gradually from the cutting blasts to the contour blasts. Admittedly,
the geometric parameters associated with blasthole layout such as
k þ 2l @ 2 f ðs=rÞ 4l @f ðs=rÞ 4l
rðtÞ ¼  þ  2  3 f ðs=rÞ ð5Þ spacing, burden and S/B ratio cannot be considered specifically in
r e c2p @ s2 r e cp @ s re
the theoretical analyses due to the simplification of equivalent
The complex spectra of the displacement function u(t), the func- blasting vibration source.
tion f(s/r) and the action function r(t) are defined as Su(jx), Sf(jx)
and Sr(jx), respectively. Using the Fourier transformation of Eqs. 3. Frequency spectrum analysis of measured blasting vibration
(3) and (5), the complex spectrum of the displacement, Su(jx), during underground blasting
can be found out as
 
r e 1=r2 þ jx=ðrcp Þ In this section, a case study of an underground opening, which
Su ðjxÞ ¼ Sr ðjxÞ ð6Þ is excavated with the full-face millisecond delay blasting technol-
4l=r 2e  ðk þ 2lÞx2 =c2p þ 4ljx=ðr e cp Þ
ogy, is conducted to investigate whether the vibration frequencies
where j is the imaginary unit and x is the circular frequency. for a single blast delay from inside to outside will vary according to
As the recording instruments used for blasting vibration moni- the rule predicted above.
toring usually record velocities, it is advisable to determine the
spectrum of velocity Sv(jx). To obtain the spectrum Sv(jx), let us 3.1. Project background
multiply the LHS and RHS of Eq. (6) by jx. Then,
The Pubugou Hydropower Station is located in the midstream of
Sv ðjxÞ ¼ jxSu ðjxÞ the Tatu River, a branch of the Yangtze River. It has a total installed
ðjxc2p  x2 rcp Þre capacity of 3300 MW. Its underground powerhouse system mainly
¼ Sr ðjxÞ ð7Þ consists of a main powerhouse, a main transformer chamber, a
4lr 2 ½ðk þ 2lÞx2 =ð4lÞ  ðcp =re Þ2  jxðcp =r e Þ
gate chamber, diversion tunnels and tailrace tunnels, as shown in
From Eq. (7), the amplitude–frequency spectrum of velocity is Fig. 2. The main transformer chamber arranged at the downstream
given by side of the main powerhouse in parallel measures
250.3  18.3  26.0 m (length  width  height). The under-

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
r e cp x r2 x2 þ c2p
F v ðxÞ ¼ jSv ðjxÞj ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi jSr ðjxÞj ð8Þ
4lr 2 ðcp =re Þ4 þ ½1  ðk þ 2lÞ=ð2lÞðcp =r e Þ2 x2 þ ½ðk þ 2lÞ=4l2 x4

From an analysis of Eq. (8), it can be found that the amplitude–
frequency spectrum of velocity Fv(x) depends on the parameters of
blasting effect re, r(t), the rock properties k, l, cp, and the distance
from the blasting source r. The study of Lu et al. (2013) shows that
the mean frequency (centroid frequency) of the amplitude–fre-
quency spectrum is linearly proportional to the ratio of cp to re,
and is related to the ratio of re to r by the power function.
In underground opening excavation by blasting, it is a common
occurrence to detonate an array of blastholes simultaneously in the
same delay. Under the condition of multiple blastholes, the envel-
ope of the cracked zones around each hole can be regarded as the
boundary of the equivalent blasting vibration source, as shown in
Fig. 1. For the underground opening excavation by the full-face
millisecond delay blasting sequence, the equivalent blasting vibra-
tion source of the headmost cutting blasts inside is smaller than
that of the breaking hole blasts and the contour hole blasts outside.
The larger the equivalent blasting vibration source is, the lower the
vibration frequency is (Lu et al., 2013). Therefore, if each delay of
Equivalent blasting
vibration source
Cracked zone
boundary
ground powerhouse is located in granite of Class II–III in the Chi-
nese rock classification system, characterized by a longitudinal
wave velocity of more than 4500 m/s.
The main transformer chamber is excavated in three horizontal
layers with heights of 8, 9.4 and 8.6 m, respectively, as shown in
Fig. 2. The upper layer uses an excavation procedure where an
8.3 m wide middle pilot tunnel is blasted prior to the two sides
by 20–30 m. The blast of the middle pilot tunnel is divided into
nine delays, and non-electric millisecond (MS) delayed detonators
from 1 to 13 are adopted to fire various rounds of blastholes. The
blast uses blastholes of 50 mm in diameter and the explosive
adopted is the 2# rock emulsion explosive. The cutting holes,
breaking holes and bottom holes use the same charge of 35 mm
in diameter. To prevent an overbreak, smooth blasting is used in
the roof with a charge of 25 mm in diameter in contour blastholes.
Detailed drilling and blasting parameters for different types of
blastholes are listed in Table 1.
During a blast of the middle pilot tunnel, a digital data logging
and analysis system was employed to monitor the blast-induced
vibration in the main transformer chamber itself. In the survey,
velocity sensors were arranged on the floor of the upper layer at
different distances from the blasting work face to measure vertical
and transverse velocity–time histories, as shown in Fig. 2.
Recorded velocity–time histories are plotted in Fig. 3 (only the
3# measurement point is given in the figure).

3.2. Frequency spectrum analysis and discussion

Blasthole

Fig. 1. Schematic diagram of the equivalent blasting vibration source for a As the blasting design in Fig. 2 and Table 1 shows, blasthole
multiple-blasthole blast. geometry and charge structure are identical in the blasts of the
192 J.H. Yang et al. / Tunnelling and Underground Space Technology 51 (2016) 189–201

Main power house Main transformer chamber

Main transformer chamber 5.0 8.3 5.0 703.3
Diversion tunnels
Gate chamber
Middle pilot tunnel 695.3

Tail race tunnels 685.9

Generatrix cavern 677.3

0.55 MS13
MS11
MS10
0.7

0.9 MS9 MS3 Upper layer of main transformer chamber Middle pilot tunnel

0.25 0.25 0.25 0.25

0.8
0.8
0.8

MS7

5
0.8

8.3
MS5 MS1 Vibration
MS12 0.7 0.8 0.8 0.8 0.8 MS12 measuring points
3# 2# 1# Blasting site

5
5 5 20.8 20 4
0.7 0.8 0.8

0.25 0.25

0.7 MS13

Fig. 2. Layout and excavation procedure of the main transformer chamber (units: m).

Table 1
Drilling and blasting parameters for the blast of the middle pilot tunnel.

Blasthole type and detonator series Cutting Breaking blasthole Bottom blasthole Contour blasthole
blasthole
MS1 MS3 MS5 MS7 MS9 MS10 MS11 MS12 MS13 MS13
Blasthole diameter (mm) 50 50 50 50 50 50 50 50 50 50
Blasthole length (m) 4.1 4.1 4.1 4.1 4.1 4.1 4.1 4.1 4.1 4.1
Charge diameter (mm) 35 35 35 35 35 35 35 35 35 25
Charge length (m) 2.6 2.6 2.6 2.6 2.6 2.6 2.6 2.6 2.6 3.1
Charge weight per hole (kg) 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 1.4

2.0
Vertical velocity
Velocity (cm/s)

1.0 MS1 MS5 MS7 MS9 MS10 MS11 MS12 MS13

0.0

-1.0 MS3

-2.0
0 100 200 300 400 500 600 700 800 900
Time (ms)
2.0 MS1
MS9 Transverse velocity
Velocity (cm/s)

MS5 MS7 MS10 MS11 MS12 MS13

1.0

0.0

-1.0
MS3
-2.0
0 100 200 300 400 500 600 700 800 900
Time (ms)

Fig. 3. Blast-induced velocity–time histories recorded at the 3# measurement point.

breaking holes of MS7 and MS11 delays, also in the MS9 and MS10 that of the MS7 delay, and also the equivalent source of the MS10
delays, except for the dimensions of the equivalent blasting vibra- delay is larger than that of the MS9 delay. Thus comparing fre-
tion sources. The equivalent source of the MS11 delay is larger than quency spectra of blast-induced vibration between the MS7 and
 J.H. Yang et al. / Tunnelling and Underground Space Technology 51 (2016) 189–201 193

MS11 delays, as well as the MS9 and MS10 delays enables us to Table 2
investigate the effects of the dimensions of equivalent blasting Mean frequencies of single vibration signals.

vibration sources on blasting vibration frequency. Measurement Direction Cutting Breaking hole blast
For the vibration signals f(t) (in our case the vibration velocity), point hole blast
its amplitude–frequency spectrum F(x) is available by the Fourier MS1 and MS7 MS11 MS9 MS10
MS3 delay delay delay delay
transformation. The mathematical expression of the Fourier trans- delays
formation is as follows
1# Vertical 189.6 230.4 211.5 218.7 208.7
Z 1
Transverse 198.8 239.6 221.8 233.5 210.6
SðjxÞ ¼ f ðtÞ expðjxtÞdt ð9Þ 2# Vertical 175.4 206.4 203.4 211.3 193.0
1
Transverse 183.9 230.7 219.7 205.9 196.9
where S(jx) is the complex spectrum of the function f(t), and the 3# Vertical 159.9 201.6 197.1 205.1 183.3
Transverse 147.7 223.0 213.8 201.5 192.5
absolute magnitude of this quantity F(x) = |S(jx)| is simply termed
the amplitude–frequency spectrum of f(t).
Fig. 4 presents comparisons of the amplitude–frequency spec-
tra, where the ratio between the current value of the amplitude
lower. The dominant frequency of vertical vibration for the MS10
of spectrum to its maximum value is plotted on axis y for better
delay blast is also lower than that for the MS9 delay blast. Although
comparison (only the 3# measurement point is given in the figure).
the transverse vibration velocity of the MS10 delay blast has a
The dominant frequency is generally determined from the maxi-
higher dominant frequency, its mean frequency of the ampli-
mum peak in the amplitude–frequency spectrum. However, if
tude–frequency spectrum is still lower than that for the MS9 delay
there exist two or more picks of almost equal amplitudes, then
blast, as listed in Table 2. The same result is also obtained from
the determination of a unique dominant frequency becomes
other measurement points, and the mean frequencies for the
ambiguous. Moreover, the dominant frequencies may change
MS11 delay and the MS10 delay are lower without exceptions. It
abruptly with the increase of the propagation distance, and cannot
follows that the dimensions of the equivalent blasting vibration
give the frequency attenuation effectively (Lu et al., 2013). Thus, in
sources have a significant impact on the frequency composition
addition to the dominant frequency, the mean frequency or also
of blast-induced vibration, and the frequencies become lower with
called centroid frequency is also employed here to better charac-
increasing dimensions of the equivalent sources. This is consistent
terize the frequency composition. The mean frequency of the
with the theoretical analyses above.
amplitude–frequency spectrum is defined as
, The cutting blastholes in the MS1 and MS3 delays also have the
X
m X
m same charge structure with the breaking blastholes. The blast of
fc ¼ ðF i f i Þ Fi ð10Þ the cutting holes has the smallest equivalent blasting vibration
i¼1 i¼1
source. According to the above result, vibration frequencies of
where fc is the mean frequency or centroid frequency, and Fi is the the cutting hole blast are presumably the highest among these
amplitude corresponding to the frequency fi in the amplitude–fre- delays. Fig. 4 also gives the frequency spectra of the cutting blast
quency spectrum. vibrations. Because the cutting hole blasts in the MS1 delay and
It can be seen from Fig. 4 that, compared to the MS7 delay blast, MS3 delay have no clear identification points in the recorded
the frequency spectra of both vertical and transverse single-delay velocity–time histories (see Fig. 3), the vibrations induced by blasts
vibration signals move toward the low-frequency component for of the two delays are analyzed together. From Fig. 4 and Table 2,
the MS11 delay blast, and its dominant frequencies are relatively however, the dominant vibration frequencies and the mean fre-

MS1 and MS3 delays MS7 delay MS11 delay MS1 and MS3 delays MS9 delay MS10 delay
1.0 1.0

0.8 0.8
F( )/Fmax( )

F( )/Fmax( )

0.6 0.6

0.4 0.4

0.2 0.2

0.0 0.0
0 50 100 150 200 250 300 350 400 0 50 100 150 200 250 300 350 400
Frequency (Hz) (a) Frequency (Hz)

MS1 and MS3 delays MS7 delay MS11 delay MS1 and MS3 delays MS9 delay MS10 delay
1.0 1.0

0.8 0.8
F( )/Fmax( )

F( )/Fmax( )

0.6 0.6

0.4 0.4

0.2 0.2

0.0 0.0
0 50 100 150 200 250 300 350 400 0 50 100 150 200 250 300 350 400
Frequency (Hz) Frequency (Hz)
(b)
Fig. 4. Comparisons of the amplitude–frequency spectra between single vibration signals: (a) spectra of vertical velocities and (b) spectra of transverse velocities.
194 J.H. Yang et al. / Tunnelling and Underground Space Technology 51 (2016) 189–201
quencies of the cutting blast are lower than those of the breaking
blasts, which is completely contrary to our inference from the ana-
lytical frequency spectrum in Eq. (8).
The cutting blastholes are detonated prior to the breaking blast-
holes. In consideration of accumulated rock damage due to
repeated blasts in the millisecond delay blasting sequence, the
integrity of rock masses when the cutting holes are detonated is
better than that when the breaking holes are blasted. It is generally
known that the vibration frequency becomes higher as the integ-
rity of propagation media increases. From this point of view, the
vibration frequencies of the cutting blast should also be higher
than those of the breaking blasts.
It should be noted that when underground openings are exca-
vated with the full-face millisecond delay blasting sequence, the
cutting holes are detonated on the condition of only one free face,
i.e. the blasting work face, and the blast of the former delay creates
another free face for the latter one, while the breaking holes and
the contour holes are all detonated under two free faces, as shown
in Fig. 5. Some studies have shown that free faces have a consider-
able impact on vibration intensity, and thus we believe that they
would also have an influence on vibration frequency. Therefore,
the authors hold the belief that it is the free faces created by blasts
of the former delays that influence the vibration frequencies of the
breaking hole blasts and the contour hole blasts, and result in the
counter-intuitive phenomenon occurring in Fig. 4. To demonstrate
our view, a small-scale blasting test and related numerical model-
ing are performed.
4. Experimental study on the influence of blast-created free
faces on vibration frequency
4.1. Review of the influence of free faces on blasting vibration
Free faces around blastholes are one of the most significant
blasting parameters. The presence of free faces and their amounts,
dimensions and locations directly affect rock fragmentation by
blasting, as well as blast-induced vibration. The influence of free
faces on blast-induced vibration effects was studied extensively
by many researchers. There is a common belief within the blasting
community that the presence of free faces will reduce the magni-
tude of blasting vibration, and decreasing burden (the distance
from the blasting source to the free face) will weaken the blasting
vibration, as the seismic wave energy is more readily released on
the free faces and reflected back as rarefaction waves (Heilig
et al., 1997; Singh, 2002; Nateghi, 2011). Of course there are some
objections to this belief. Blair and Armstrong (2001), and Brent
et al. (2002) performed many experiments to examine the effect
of burden width on vibration. From their results, it is found that
vibration is insensitive to burden, and the vibration from confined
blasts is not higher than that from free-face blasts. Similar work
has also been done by Uysal et al. (2007), who reported that bur-
den width has a significant impact on vibration, but vibration will
Fig. 5. Schema of the free face created by the cutting blast for the breaking blast.
decrease as burden increases. Their explanation of this problem is
the fact that, with increasing burden, more explosion energy is uti-
lized to fracture rock masses and move over rock fragments, result-
ing in a decrease of seismic wave energy. By numerical study of
underground blast induced ground surface motion, Wu and Hao
(2005) found that vibration on the ground surface is larger than
that in the free field at the same distance due to the reflection of
the ground surface.
Conventionally, most studies on the relationship of free faces
and blasting vibration are focused on the PPV to evaluate the vibra-
tion magnitude. However, unfortunately, there is still no unified
understanding for this problem. Furthermore, few open literatures
can be found to study the influence of free faces on vibration fre-
quency. The present study experimentally and numerically investi-
gates the influence of blast-created free faces on vibration
frequency during the full-face millisecond delay blasting sequence.
4.2. Field test and blast-induced vibration monitoring
As mentioned in Section 3.2, during underground opening exca-
vation by the full-face millisecond delay blasting technology, cut-
ting holes are blasted under a confined condition, while breaking
holes are blasted under a free-face condition. To investigate the
influence of blast-created free faces on the vibration frequency, a
side-by-side comparison of the vibration frequency characteristics
is made between the cutting hole blast and the breaking hole blast.
Limited by the field conditions, the blasting test is performed on an
almost vertical slope of the drainage tunnel entrance in a hydro-
power station instead of an underground opening because the drai-
nage tunnel has been reinforced by concrete lining and blasts are
not allowed in the tunnel. Horizontal blastholes, including one
delay of cutting holes (MS3 delay) and two delays of breaking holes
(MS5 and MS11 delays), are drilled in the vertical slope which is
regarded as a blasting work face to model an underground excava-
tion by the full-face millisecond delay blasting sequence, as shown
in Fig. 6. For the project case in Fig. 2, the blasthole number and
layout of the cutting blast is different from that of the breaking
blasts. To prevent other factors from contaminating results as
much as possible in this blasting test, blasthole geometry, charge
structure, blasthole number and blasthole layout are all kept the
same between the cutting blast and the breaking blasts. The test
uses blastholes that are 40 mm in diameter and 1.8 m in length,
with a spacing of 0.5 m and a burden of 0.5 m. The 2# rock emul-
sion explosive is charged continuously in these blastholes, with
32 mm diameter, 1.6 m length and 1.1 kg weight per hole. At dis-
tances of 28–80 m to the blasting site, four triaxial velocity sensors
are arranged on the slope surface along a straight line to measure
transverse, vertical and longitudinal velocity–time histories
induced by blasting, as shown in Fig. 6.
To investigate the effect of the vertical free faces created by the
cutting hole blast for the latter breaking hole blast, the transverse
vibration velocities whose directions are perpendicular to the ver-
 J.H. Yang et al. / Tunnelling and Underground Space Technology 51 (2016) 189–201 195

Vibration
measuring points
MS11
4# 3# 2# 1#
MS3
Blasting site
33m 13m 7m 28m
MS5

Cutting holes
Breaking holes

Fig. 6. Layout of blastholes and blasting vibration monitoring in the blasting test.

0.60 MS5 MS11 0.50 MS11

Velocity (cm/s)

Velocity (cm/s)
MS3 MS3 MS5
0.30 0.25
0.00 0.00
-0.30 -0.25
-0.60 -0.50
0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700
Time (ms) Time (ms)
(a) (b)
0.08 MS3 MS5 0.02
MS3 MS5
Velocity (cm/s)

Velocity (cm/s)
MS11
0.04 MS11 0.01
0.00 0.00
-0.04 -0.01
-0.08 -0.02
0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700
Time (ms) Time (ms)
(c) (d)
Fig. 7. Transverse velocity–time histories recorded in the blasting test: (a) 1# measurement point; (b) 2# measurement point; (c) 3# measurement point and (d) 4#
measurement point.

tical free faces are chosen as analysis objects. Fig. 7 plots the
recorded transverse velocity–time histories. These curves are
divided obviously into three segments, the first one results from
the blast of the cutting holes, and the second and third ones orig-
inate from the blasts of the breaking holes. Slopes are located in
open environments, while underground openings are located in
confined space. It should be noted that influenced by the different
spatial structures, the seismic wave propagation path and associ-
ated vibration frequency will be different for the blasts on slopes
and in underground openings. Some related studies have shown
that when the drilling and blasting parameters are kept the same
for the two cases, the blast-induced vibration on slope surfaces
accentuates lower frequency components at the same distances.
Therefore, the velocity–time histories measured in this blasting
test can not completely reveal the actual frequency characteristics
of blast-induced vibration for the full-face millisecond delay blast-
ing in underground opening excavation. But undoubtedly in this
blasting test, the free faces which are created by the cutting hole
blast for the latter breaking hole blast is really existent. Hence a
side-by-side comparison of the cutting blast vibration with the
breaking blast vibration in this test is able to investigate the
influence of blast-created free faces on vibration frequency. Not
withstanding its limitation, the blasting test performed in this
study does provide some important experimental data for our
research goals.
4.3. Frequency analysis methods
The Fourier transformation is a traditional analysis method pro-
cessing vibration signals, which can capture relatively intact fre-
quency information of stationary signals. However, it is
inherently defective in processing non-stationary random signals
has become a powerful tool for blasting vibration signal processing
due to its characteristics of time shift and multi-scale resolution.
The wavelet packet decomposition originating from the wavelet
transformation has the predominance of analyzing details of
high-frequency components and is a more accurate method than
the wavelet transformation in signal processing. In this study, in
addition to the Fourier transformation, the wavelet packet decom-
position is also used to analyze the blasting vibration signals.
When using the wavelet transformation, signals are first
decomposed into two parts, a low-frequency part and a high-
frequency part, and then the low-frequency part is subdivided into
two parts likewise, while the high-frequency part is not. If the fre-
quency band of signals is in the range of 0–x, it will be divided into
two bands of 0–x/2 and x/2–x after the first layer of decomposi-
tion. The following step is a further decomposition of the low-
frequency band 0–x/2 into 0–x/4 and x/4–x/2, and so on. The
method of wavelet packet is more detailed in the decomposition
of frequency. It subdivides the high-frequency bands, which are
not further decomposed by the wavelet transformation method,
into two bands in each layer, resulting in a much higher resolution
in the high-frequency bands.
After the decomposition of vibration signals for n layers by the
wavelet packet method, we define the reconstructed signal in the
ith frequency band as sn, i and its energy as En, i, then
Z X
N
En;i ¼ jsn;i j2 dt ¼ jxi;k j2 ð11Þ
k¼1
where xi, k represents the vibration magnitude of the discrete sam-
ples in the reconstructed signal sn, i, i = 0, 1, 2, . . ., 2n  1, k = 1, 2, . . .,
N, and N is the amount of discrete samples.
The total energy of vibration signals E0 is available as
2X
n
1

of blasting vibration characterized by short duration and abrupt E0 ¼ En;i ð12Þ

change in signal structure. Recently, the wavelet transformation i¼0
196 J.H. Yang et al. / Tunnelling and Underground Space Technology 51 (2016) 189–201

After signal decomposition for n layers, the ratio of the energy
in each frequency band to the total energy is given by
pn;i ¼ En;i =E0
Thus, energy distribution of blasting vibration signals in
different frequency bands can be obtained by the wavelet packet
analysis in Eqs. (11)–(13). When signals are decomposed by the
wavelet packet method, the decomposition layers are decided by
the signal frequency band and the working frequency band of the
seismic instrument. Because frequencies of blasting vibration sig-
nals on structures are commonly less than 200 Hz, the sampling
rate of signals in this study is set as 2000 Hz. According to the
Nyquist sampling theorem, the highest frequency that the instru-
ment is able to record is 1000 Hz. So the blasting vibration signals
recorded in this test can be decomposed for six layers, and its low-
est frequency band has a range of 0–15.625 Hz. A crucial step in the
application of the wavelet packet decomposition method is choos-
ing an appropriate wavelet basis due to its non-uniqueness. The
choice of a wavelet basis depends on the objectives of analysis
and the characteristics of the signals analyzed. In general, the wave-
let basis that attenuates speedily and has a similar wave shape to
the analyzed signals should be chosen. The Daubechies’ wavelets
(Db), which are orthogonal, nearly symmetric, as well as relatively
smooth and also allow for compact representation of the original
signals and fast algorithm implementation, have been successfully
applied to process non-stationary signals of blasting vibration (Ling
et al., 2005; Amiri and Asadi, 2009). According to the requirement of
wavelet basis choice and the characteristics of analyzed signals, the
wavelet function of the 8th-order Daubechies (Db8) is employed in
the present study.
4.4. Analysis results and discussion
Fig. 8 gives the amplitude–frequency spectra of each delay
vibration signal by the Fourier transformation. The amplitudes of
the spectrum on axis y are also normalized here for convenience
of comparison. Take vibration signals at the 1# measurement point
for instance, the dominant vibration frequency for the cutting blast
ð13Þ
is 71.9 Hz, while it is 94.2 Hz and 106.5 Hz for the latter two delays
of the breaking blasts. Compared to the cutting hole blast, the
amplitude–frequency spectra of the vibrations by the breaking
blasts move toward a high-frequency component, leading to higher
dominant frequencies and mean frequencies. The same conclusion
can also be drawn from the other measurement points.
The two delays of the breaking hole blasts have the same blast-
ing parameters, as well as the same boundary dimensions of the
equivalent vibration sources in the vertical direction. In theory at
least, their frequencies in transverse vibrations should be coinci-
dent. However, due to a combination of some accidental factors,
such as delay errors of detonators at the same delay and local rock
looseness or compaction by vibration, their amplitude–frequency
spectra are not completely consistent, but they share a very similar
spectrum structure, as shown in Fig. 8.
The energy distribution of each delay vibration in different fre-
quency bands is shown in Fig. 9 by wavelet packet decomposition.
Considering the 1# measurement point again, the energy of blast-
ing vibration signals lies in the frequency band of 0.00–265.00 Hz
for both the cutting hole blast and the breaking hole blasts. For
the cutting hole blast, the ratio of the energy in the frequency band
of 0–78.13 Hz to the total energy is 86%, and only 14% of the vibra-
tional energy lies in the frequency band of 78.13–265.63 Hz. While
for the two delays of the breaking hole blasts, the ratios of the rel-
atively high-frequency energy in the band of 78.13–265.63 Hz are
60% and 76%, respectively. It is concluded that the ratio of the high-
frequency vibrational energy for the breaking hole blasts is signif-
icantly higher than that for the cutting hole blast, which is also
confirmed at the other measurement points.
Although the dominant frequency band of vibration signals
obtained by the Fourier transformation is not exactly the same as
that by the wavelet packet decomposition, the results from the

1.0 1.0
Cutting blast Cutting blast
MS3 delay MS3 delay
0.8 0.8
Breaking blast Breaking blast
F( )/Fmax( )

F( )/Fmax( )

MS5 delay MS5 delay

0.6 0.6
Breaking blast Breaking blast
MS11 dealy MS11 delay
0.4 0.4

0.2 0.2

0.0 0.0
0 50 100 15 0 200 250 300 0 50 100 15 0 200 250 300
Frequency (Hz) Frequency (Hz)

(a) (b)
1.0 1.0
Cutting blast Cutting blast
MS3 delay MS3 delay
0.8 0.8
Breaking blast Breaking blast
F( )/Fmax( )
F( )/Fmax( )

MS5 delay MS5 delay

0.6 0.6
Breaking blast Breaking blast
MS11 dealy MS11 delay
0.4 0.4

0.2 0.2

0.0 0.0
0 50 100 150 200 250 300 350 400 0 50 100 150 200 250 300
Frequency (Hz) Frequency (Hz)
(c) (d)
Fig. 8. Comparisons of the amplitude–frequency spectra of vibration signals between the cutting blast and the breaking blasts: (a) 1# measurement point; (b) 2#
measurement point; (c) 3# measurement point and (d) 4# measurement point.
 J.H. Yang et al. / Tunnelling and Underground Space Technology 51 (2016) 189–201 197

80 50
Cutting blast Cutting blast
MS3 delay 40 MS3 delay
60 Breaking blast Breaking blast
MS5 delay MS5 delay
30
p (%)

p (%)
Breaking blast Breaking blast
40
MS11 delay MS11 delay
20
20
10

0 0
0 50 100 150 200 250 300 0 50 100 150 200 250 300
Frequency (Hz) Frequency (Hz)
(a) (b)
50 70
Cutting blast Cutting blast
MS3 delay 60 MS3 delay
40
Breaking blast 50 Breaking blast
MS5 delay MS5 delay
30
p (%)

p (%)
Breaking blast 40 Breaking blast
MS11 delay 30 MS11 delay
20
20
10
10
0 0
0 50 100 150 200 250 300 0 50 100 150 200 250 300
Frequency (Hz) Frequency (Hz)
(c) (d)
Fig. 9. Comparisons of the energy distribution of vibration signals between the cutting blast and the breaking blasts: (a) 1# measurement point; (b) 2# measurement point;
(c) 3# measurement point and (d) 4# measurement point.

two signal processing methods both show that, the vibration fre-
quency from the breaking hole blasts is higher and its ratio of
higher-frequency vibration is larger than that from the cutting hole
blast. As stated above, the biggest difference between the cutting
blast and the breaking blasts in the test is that the blasting itself
creates free faces for the breaking blast. The analysis results from MS11
the blasting test prove that the blast-created free faces indeed have Blast-created
a significant impact on blasting vibration frequency, leading to an Observation
free face
points
increase of the frequency. MS3
MS5

5. Numerical study on the influence mechanism of blast-created

free faces on vibration frequency
The study results described above clearly display the effects of
blast-created free faces on vibration frequency. Limited by experi-
ment conditions, the underlying influence mechanism is not
revealed above. It is studied below via numerically modeling the
blasting test with the LS-DYNA program.
5.1. Numerical model and its parameters of the blasting test
According to the blasting site in Fig. 6, a three-dimensional
numerical model measuring 160  8  8 m (length  width
 height) is developed to simulate the vibration induced by the
cutting hole and breaking hole blasts, as illustrated in Fig. 10. In
the model, the blasthole dimension and layout are in accordance
with reality. The model is divided into a mesh of hexahedron-
shaped brick elements with eight nodes. A fine grid of elements
is used in the vicinity of the blastholes to increase the simulation
accuracy. The grid increases in size away from the blastholes, but
it should be shorter than 1/8–1/10 of the wave length in general
to properly reduce any wave distortion. Convergence tests are con-
ducted to determine how many elements would be needed to
achieve a reliable estimation. The sizes of elements are reduced
Fig. 10. Local view of the finite element model used for numerical modeling of
blasting vibration.
until the difference of the modeling results between two consecu-
tive element sizes is less than 5%. According to the above require-
ments, the largest element size is set as 2.0 m at the border, and
the model mesh has a total of 251808 discretized elements with
271830 nodes. In accordance with the actual situation, free bound-
aries are applied to the front surface of the model, and non-
reflecting boundaries are enforced on the other five surfaces to pre-
vent artificial wave reflections generated at the model boundaries
from reentering the model and contaminating results, as infinite
domains are usually not modeled. Internally, LS-DYNA computers
an impedance matching function for all non-reflecting boundary
segments based on an assumption of linear material behavior. In
the numerical modeling of the blasting test, only blasts of the cut-
ting holes and the first delay of breaking holes are simulated.
When modeling the blast of breaking holes, rock elements exca-
vated by the cutting hole blast are first removed from the numer-
ical model to create free faces for the breaking hole blast.
The LS-DYNA software is one of the few codes that allow an
accurate description of explosive detonation and explosive-rock
198 J.H. Yang et al. / Tunnelling and Underground Space Technology 51 (2016) 189–201
interaction. In LS-DYNA, the explosion is implemented by an equa-
tion of state (EOS). The Jones-Wilkins-Lee (JWL) EOS describes the
relationship between volume, energy and pressure of detonation
products and is widely used in the simulation of blasts. The JWL
EOS is defined as
   
x x xE0
Pd ¼ A 1  expðR1 VÞ þ B 1  expðR2 VÞ þ
R1 V R2 V V
where Pd is the detonation pressure, V is the relative volume, E0 is
the internal energy per initial volume, and A, B, R1, R2, and x are
parameters related to charging explosives. The 2# rock emulsion
explosive used in the field test is characterized by density of 950–
1300 kg/m3 and velocity of detonation of 3500–4500 m/s according
to the explosive instructions of the manufacturer. Hence in the
numerical simulation the explosive density is assigned to
1300 kg/m3 and the velocity of detonation is assigned to 4000 m/
s. The parameters associated with the JWL equation should be
obtained by experiments, but direct measurements of these param-
eters were not carried out due to the lacking of feasible methods
and instruments. Instead, these parameters in our simulation are
given A = 214.4 GPa, B = 0.182 GPa, R1 = 4.2, R2 = 0.9, x = 0.15, and
E0 = 4.192 GPa by referring to the work of Li et al. (2011a), in which
the emulsion explosive is also used and the simulated results of
rock damage and vibration agree well with those obtained by site
investigations.
After explosive detonation, rock masses in the immediate vicin-
ity of the charge will suffer high pressure and large strain instanta-
neously. It is well known that rock mass properties are strain rate
dependent. Thus, a plastic kinematic constitutive model is adopted
in the present simulation to consider the effect of strain rate and
the plastic hardening of rock material. It is generally implemented
by the Cowper–Symonds model, in which the current yield
strength of rock material associated with the strain rate is given by
Table 3
Parameters of the rock mass used for numerical simulation.
Properties Value
Density, q (kg/m ) 3
1700–2500
Initial Young’s modulus, E (GPa) 3–30
Initial tangential modulus, Etan (GPa) 0.8–8
Initial yield strength, r0 (MPa) 3–30
ð14Þ Poisson’s ratio, m 0.22–0.32
Cowper–Symonds constant, C (s1) 2.5
Cowper–Symonds constant, P 4.0
Constant, b 1.0
masses inside. In order to represent the weathering effects in the
numerical calculation for a better simulation as much as possible,
it is assumed that within 2.0 m inside the slope surface, the rock
parameters such as density and Young’s modulus gradually
increase from outside to inside in a linear fashion, and beyond
the range the rock parameters remain the same. Thus the lower
limits of the density, modulus and strength in Table 3 are assigned
to the rock elements on the slope surface, and the upper limits are
assigned to the rock elements beyond 2.0 m inside the slope sur-
face. While the contrary limits are assigned for the Poisson’s ratio.
5.2. Demonstration of numerical modeling
Fig. 11 shows comparisons of the velocity–time histories
obtained from the numerical simulation and from the site monitor-
ing at the 1# measurement point. It can be seen that, the simulated
waves including peak velocities, frequencies and duration show a
satisfactory agreement with the measured ones. It demonstrates
that the numerical modeling adopted in this study is feasible,
and allows predicting the blasting vibration and studying the influ-
ence mechanism of blast-created free faces on vibration frequency.
h 9 According to the arrangement of measurement points at the
e_
1=P i
ry ¼ 1 þ r0 þ bEp eeff >
> blasting test, some observation points aligning a straight line on
C p
>>
= the slope surface are selected to analyze the vibration frequency
Ep ¼ EE
Etan E
ð15Þ attenuation with distance for the cutting hole blast and the break-
tan
>
>
R t 2 p p 1=2 >
> ing hole blast. Fig. 12a gives the attenuation curves of the domi-
ep ¼ 0 3 e_ ij e_ ij dt
eff ;
where ry and r0 are the current and the initial yield strengths,
respectively, e_ is the strain rate, C and P are Cowper–Symonds con-
stants, b is a parameter that ranges from 0 to 1 denoting kinematic
hardening and isotropic hardening, respectively, E, Ep, and Etan are
the Young’s modulus, the plastic hardening modulus and the tan-
gential modulus, respectively, eeff p is the effective plastic strain, eij
_p
is the plastic strain rate, and t is the accumulative duration after ini-
tiation of the plastic strain. Details of the Cowper–Symonds model
can be found in the LS-DYNA theoretical manual (Hallquist, 1998).
Discontinuities of different sizes ranging from microcracks to
faults in rock masses have great effects on the propagation and
reflection of the blast-induced seismic waves through the rock
masses, but unfortunately also create some complication to
numerical modeling. For the purpose of understanding simply
the influence mechanism of blast-created free faces on vibration
frequency, it is assumed that the rock mass is continuous and iso-
tropic; however, this is generally not in realistic case. The rock
nant frequencies. It is found that variation of the dominant
frequencies is not regular. On the whole, the dominant frequencies
for the breaking hole blast are higher than ones for the cutting hole
blast, but they decrease dramatically at the distance of 3.0 m, lead-
ing to lower dominant frequencies locally in the range of 3–20 m
distance. This is because multiple peaks of almost equal ampli-
tudes exist in the amplitude–frequency spectra, and the high-
frequency bands attenuate with increasing distance at a different
rate from the low-frequency ones (Lu et al., 2013). As an example,
vibrations at the typical distances of 10 and 50 m are considered, as
their amplitude–frequency spectra and energy distribution in dif-
ferent frequency bands show in Fig. 13. At 10 m distance, the dom-
inant vibration frequency for the cutting hole blast is 153.9 Hz,
while the vibration from the breaking hole blast has a lower dom-
inant frequency of 90.5 Hz. However, from the energy distribution
figures by wavelet packet decomposition, for the cutting blast, the
ratio of the vibrational energy in the frequency band of 0–265.63 is
Velocity (cm/s)

mass in the field test is basalt. Due to the lack of experimental data 0.70 MS3 Measured signals
MS5
for the rock parameters, numerical tests are carried out by compar- 0.35 Simulated wave
ing simulated vibration with the measured vibration waves. The 0.00
rock parameters that make the simulated waves agree well with -0.35
the measured data are finally adopted, as listed in Table 3. The -0.70
80 100 120 140 160 180 200 220 240
selected rock parameters may be different from the reality, but this
Time (ms)
would have no effect on revealing the internal mechanism. Because
of weathering effects occurring on the slope surface, the rock Fig. 11. Comparison of the velocity–time histories between numerical simulation
parameters of the slope surface are different from those of the rock and site monitoring at 1# measurement point.
 J.H. Yang et al. / Tunnelling and Underground Space Technology 51 (2016) 189–201 199

Dominant frequency (Hz)

400 600

Mean frequency (Hz)

Cutting hole blast Cutting hole blast
500
300 Breaking hole blast Breaking hole blast
400
200 300
200
100
100
0 0
0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80
Distance (m) Distance (m)
(a) (b)
Fig. 12. Comparisons of the vibration frequencies between the cutting blast and the breaking blast: (a) dominant frequency and (b) mean frequency.

1.0 25
Cutting hole blast Cutting hole blast
0.8 Breaking hole blast 20 Breaking hole blast
F( )/Fmax( )

0.6 15

p (%)
0.4 10

0.2 5

0.0 0
0 200 400 600 800 1000 0 200 400 600 800 1000

Frequency (Hz) Frequency (Hz)

(a) (b)

1.0 50
Cutting hole blast Cutting hole blast
0.8 Breaking hole blast 40 Breaking hole blast
F( )/Fmax( )

0.6 30
p (%)

0.4 20

0.2 10

0.0 0
0 50 100 150 200 250 300 0 50 100 150 200 250 300

Frequency (Hz) Frequency (Hz)

(c) (d)
Fig. 13. Amplitude–frequency spectra and energy distribution of vibrations at different distances: (a) amplitude–frequency spectrum at 10 m distance; (b) energy
distribution at 10 m distance; (c) amplitude–frequency spectrum at 50 m distance and (d) energy distribution at 50 m distance.

91%, and only 9% of vibrational energy lies in the relatively high- 5.3. Influence mechanism analyses
frequency band of 265.63–1000 Hz, while for the breaking blast,
the energy ratio in the band of 265.63–1000 Hz is up to 52%. In In the field test and the numerical modeling, the blast of the
sum, compared to the cutting blast, the breaking blast has a higher breaking holes has the same blasthole geometry and charge struc-
ratio of the high-frequency vibrational energy, but lower dominant ture as that of the cutting holes. These conditions make it very easy
frequency at the distance of 10 m. However, at a distance of 50 m to get into a misunderstanding that the breaking blast has the
(see Fig. 13), consistent results are obtained from the two signal same vibration frequency as the cutting blast at the same distance.
processing methods where both the ratio of the high-frequency Actually, the breaking holes are blasted under another boundary
vibrational energy and the dominant frequency are higher for the condition which the cutting blast does not have, i.e. the free faces
breaking blast than the cutting blast. created by the blast of the former delay. When explosion-induced
As is clear from the above analyses, the dominant frequency of compressive stress waves travel to the free faces, they are reflected
the amplitude–frequency spectrum cannot describe the frequency as back-propagating rarefaction waves. The reflection waves are
composition and its attenuation law very well. The mean fre- superimposed onto the initial compressive stress waves in the pro-
quency attenuation with increasing distance is shown in Fig. 12b. cess of propagation, and it will influence far-field compressive
Unlike the dominant frequencies, the mean frequencies have a reg- stress waves. Fig. 14a shows the blast loading pressures at a dis-
ular attenuation law with increase in distance. Because of the pres- tance of 10 m calculated from the numerical modeling of the cut-
ence of the free faces created by the cutting blast for the breaking ting blast and the breaking blast. It can be seen that the loading
blast, the mean frequencies of the breaking blast vibrations are pressure of the breaking blast has a similar form to the one of
higher throughout the analysis region, which coincides with the the cutting blast, but with a smaller peak, shorter rising time and
results from the wavelet packet analysis. shorter duration. The rising time and the duration of load have a
200 J.H. Yang et al. / Tunnelling and Underground Space Technology 51 (2016) 189–201

0.4 Initial stress wave Cutting hole blast 1.0 Cutting hole blast
0.3
Breaking hole blast

Pressure (MPa)
Observed 0.8 Breaking hole blast

F( )/Fmax( )
0.2 stress wave
0.6
0.1
0.0 0.4
0 1 2 3 4 5 6
-0.1 0.2
-0.2 Reflected stress wave 0.0
-0.3 0 200 400 600 800 1000
Time (ms) Frequency (Hz)
(a) (b)
Fig. 14. Blast loading pressures and their frequency spectra at a distance of 10 m: (a) blast loading pressures and (b) amplitude–frequency spectra of the pressures.

800
significant impact on the frequency composition of blasting vibra- Burden = 0.25 m

Mean frequency (Hz)

Burden = 0.50 m
tion, and it follows that with reduction in the rising time and dura- 600
Burden = 1.00 m
tion, the spectrum of the load moves into the region of higher
frequencies, as shown in Fig. 14b. The high-frequency load during 400
the breaking blast, of course, will naturally induce high-frequency
blasting vibration and a higher ratio of the high-frequency vibra- 200
tional energy to the total energy.
To prove the effects of the back-propagating reflection waves, 0
0 10 20 30 40 50 60 70 80
we attempt to separate the initial explosion stress waves and the
Distance (m)
reflected waves when the breaking holes are blasted. The initial
waves and the reflected waves have no clear time-domain identifi- Fig. 15. Comparison of vibration frequency attenuation at different burdens.
cation points in the observed stress-time histories, and it presents
a considerable challenge to distinguishing them visually. It is
assumed that the observed stress waves in the elastic seismic wave
6. Conclusions
zone are linear superposition of the initial waves and the reflected
waves, and then the reflected waves can be obtained by subtract-
Blast-induced vibration is a major concern for the designers of
ing the initial waves from the observed stress waves. The observed
mining and construction excavations. Responses and damage of
stress wave at the 10 m distance has been calculated as the dashed
structures subjected to the blasting vibration depend not only on
line in Fig. 14a shows, in which the rock elements excavated by the
PPV but also on vibration frequency contents. Therefore, it is quite
cutting hole blast have been removed from the numerical model
significant to investigate the vibration frequency characteristics
and the free faces created by blast are existent. While the initial
and its formation mechanism for hazard assessment of blasting
wave is available when the free faces created by blast are absent,
vibration. In the present study, the frequency characteristics for
and its waveform is almost the same as the solid line in Fig. 14a.
the full-face millisecond delay blasting sequence used in under-
By implementing the above calculating procedure, the back-
ground excavation are investigated by methods of project case
propagating reflection wave is separated as shown with the dot
study, blasting test and numerical simulation. The frequency com-
line in Fig. 14a. It can be seen that the reflected wave is in tension
position of blasting vibration depends on drilling and blasting
phase, and its peak is smaller than that of the initial stress wave
parameters, rock properties and distance to blasting sources. As
due to the effects of geometric spreading and non-elastic proper-
the dimensions of the equivalent blasting vibration sources
ties of rock mass such as damping. It is the tensile reflection waves
increase, the frequency of blast-induced vibration will decrease.
that cause the breaking blast to have smaller peak pressure at far
Therefore, if the blasthole geometry and charge structure are kept
distances.
the same in each delay of the full-face millisecond delay blasting
Characteristics of the reflected back-propagating rarefaction
sequence, the vibration frequencies per delay will decrease gradu-
waves are closely related to rock properties and propagation dis-
ally in the order from inside blasts to outside blasts. This is con-
tance. For the rock parameters determined by the numerical tests
firmed by field vibration monitoring at an underground opening
(see Table 3), ,when the burden of the breaking blastholes varies
project where the full-face millisecond delay blasting technology
from 0.25 to 1.0 m, attenuation curves of the mean frequencies
is used. However, the headmost cutting blast inside is an excep-
for the breaking blast vibrations are shown in Fig. 15. It is found
tion, and its vibration frequency is lower than that of the breaking
that the vibration frequencies will become lower as the burden
blast outside. The authors hold the belief that the counter-intuitive
increases. This is because amplitudes of stress waves decrease con-
phenomena is caused by the free faces created by the cutting blast
tinuously in the process of propagation, especially in the vicinity of
for the breaking blast. When the breaking blastholes are blasted,
the blastholes. Therefore, the magnitudes of the reflected rarefac-
the back-propagating rarefaction waves reflected from the free
tion waves will be considerably reduced as the burden increases,
faces are superimposed onto the initial explosion stress waves in
thus reducing the superposition effects onto the initial compres-
the process of propagation, and it causes the rising time and dura-
sive stress waves. It is well known that low-frequency vibration
tion of the far-field blast loading pressure to become shorter, thus
has greater potential of damage to structures than high-
increasing the loading frequency and the vibration frequency of the
frequency vibration for a certain velocity as the natural frequencies
breaking hole blast. The small-scale blasting test and the corre-
of buildings are below 10 Hz in general. Therefore, from the view of
sponding numerical simulation, conducted especially to study the
vibration frequency, decreasing the burden of breaking holes in
influence of blast-created free faces on vibration frequency, have
underground blasts is beneficial to reduce the structural damage
demonstrated the reliability of the analysis results above. The
due to blasting vibration.
vibration frequency of the breaking hole blast increases as the
 J.H. Yang et al. / Tunnelling and Underground Space Technology 51 (2016) 189–201
burden decreases. Low-frequency vibration has greater potential to
cause structure damage than does high-frequency vibration for a
certain velocity. Therefore, from the view of vibration frequency,
the vibration induced by the breaking hole blast is less harmful
to structures than the cutting blast vibration, and decreasing the
burden can reduce the vibrational damage to structures due to
the breaking blast vibration.
The frequency of blast-induced vibration is influenced by a vari-
ety of factors. The work presented in this paper focuses only on the
dimensions of the equivalent vibration sources and the blast-
created free faces during underground excavation with the full-
face millisecond delay blasting technology. Further studies are
required for other influence factors, such as drilling parameters,
charge structure, interaction of multiple-blasthole blasts, and det-
onator delay time and its accuracy.
Acknowledgments
This work is supported by Chinese National Natural Science
Foundation (51509126 and 51409138), Jiangxi Provincial Natural
Science Foundation (20151BAB213026) and Opening fund of State
Key Laboratory of Water Resources and Hydropower Engineering
Science (Wuhan University). The authors wish to express their
thanks to all supports.
References
Aldas, G.G.U., 2010. Explosive charge mass and peak particle velocity (PPV)-
frequency relation in mining blast. J. Geophys. Eng. 7 (3), 223–231.
Alvarez-Vigil, A.E., Gonzalez-Nicieza, C., Lopez Gayarre, F., Alvarez-Fernandez, M.I.,
2012. Predicting blasting propagation velocity and vibration frequency using
artificial neural networks. Int. J. Rock Mech. Min. Sci. 55, 108–116.
Amiri, G.G., Asadi, A., 2009. Comparison of different methods of wavelet and
wavelet packet transform in processing ground motion records. Int. J. Civ. Eng. 7
(4), 248–257.
Blair, D.P., 2009. Limitations of electronic delays for the control of blast vibration
and fragmentation. In: Proceedings of the 9th International Symposium on Rock
Fragmentation by Blasting. Granada, Spain, pp. 171–184.
Blair, D.P., Armstrong, L.W., 2001. The influence of burden on blast vibration.
Fragblast 5 (1–2), 108–129.
Brent, G.F., Smith, G.E., Lye, G.N., 2002. Studies on the effect of burden on blast
damage and the implementation of new blasting practices to improve
productivity at KCGMs Fimiston Mine. Fragblast 6 (2), 189–206.
Caylak, C., Kocaslan, A., Gorgulu, K., Buyuksarac, A., Arpaz, E., 2014. Importance of
ground properties in the relationship of ground vibration–structural hazard and
land application. J. Appl. Geophys. 104, 6–16.
Hagan, T.N., 1977. Rock breakage by explosives. Acta Astronaut. 6 (3–4), 329–340.
Hallquist, J.O., 1998. LS-DYNA Theoretical Manual. Livermore Software Technology
Corporation, Livermore.
201
Heilig, J., Zoitsas, A., Cox, N., 1997. Free face blasting: is it the best for quarry. In:
Proceedings of the 41st Annual Conference Institute of Quarrying, Australia, pp.
23–29.
Karadogan, A., Kahriman, A., Ozer, U., 2014. A new damage criteria norm for blast-
induced ground vibrations in Turkey. Arab. J. Geosci. 7 (4), 1617–1626.
Khandelwal, M., 2010. Evaluation and prediction of blast-induced ground vibration
using support vector machine. Int. J. Rock Mech. Min. Sci. 47 (3), 509–516.
Kuzmenko, A.A., Vorobev, V.D., Denisyuk, I.I., Dauetas, A.A., 1993. Seismic Effects of
Blasting in Rock. A A Balkema Publishers, Brookfield.
Leidig, M., Bonner, J.L., Rath, T., Murray, D., 2010. Quantification of ground vibration
differences from well-confined single-hole explosions with variable velocity of
detonation. Int. J. Rock Mech. Min. Sci. 47 (1), 42–49.
Li, H.B., Xia, X., Li, J.C., Zhao, J., Liu, B., Liu, Y.Q., 2011a. Rock damage control in
bedrock blasting excavation for a nuclear power plant. Int. J. Rock Mech. Min.
Sci. 48 (2), 210–218.
Li, J.C., Ma, G.W., Zhao, J., 2011b. Analysis of stochastic seismic wave interaction
with a slippery rock fault. Rock Mech. Rock Eng. 44 (1), 85–92.
Ling, T.H., Li, X.B., Dai, T.G., Peng, Z.B., 2005. Features of energy distribution for blast
vibration signals based on wavelet packet decomposition. J. Cent. South Univ.
Technol. 12 (Sup 1), 135–140.
Lu, W.B., Zhang, L., Zhou, J.R., 2013. Theoretical analysis on decay mechanism and
law of blasting vibration frequency. Blasting 30 (2), 1–11 (in Chinese, Abstract
in English).
Ma, G.W., Hao, H., Zhou, Y.X., 2000. Assessment of structure damage to blasting
induced ground motions. Eng. Struct. 22 (10), 1378–1389.
Nateghi, R., 2011. Prediction of ground vibration level induced by blasting at
different rock units. Int. J. Rock Mech. Min. Sci. 48 (6), 899–908.
Ozer, U., 2008. Environmental impacts of ground vibration induced by blasting at
different rock units on the Kadikoy–Kartal metro tunnel. Eng. Geol. 100 (1–2),
82–90.
Park, D., Jeon, B., Jeon, S., 2009. A Numerical study on the screening of blast-induced
waves for reducing ground vibration. Rock Mech. Rock Eng. 42 (3), 449–473.
Rai, R., Shrivastva, B.K., Singh, T.N., 2005. Prediction of maximum safe charge per
delay in surface mining. Min. Technol. 114 (4), 227–231.
Sazid, M., Singh, T.N., 2013. Two dimensional finite element simulation of rock
blasting. Arab. J. Geosci. 6 (10), 3703–3708.
Singh, P.K., 2002. Blast vibration damage to underground coal mines from adjacent
open-pit blasting. Int. J. Rock Mech. Min. Sci. 39 (8), 959–973.
Singh, P.K., Roy, M.P., 2010. Damage to surface structures due to blast vibration. Int.
J. Rock Mech. Min. Sci. 47 (6), 949–961.
Singh, T.N., Dontha, L.K., Bhardwaj, V., 2008. Study into blast vibration and
frequency using ANFIS and MVRA. Min. Technol. 117 (3), 116–121.
Siskind, D.E., 1997. Blast vibration damage to underground mines and unlined
tunnels – a survey and summary review. DESA Consultants DR-36, 1–10.
Trivedi, R., Singh, T.N., Raina, A.K., 2014. Prediction of blast-induced flyrock in
Indian limestone mines using neural networks. J. Rock Mech. Geotech. Eng. 6
(5), 447–454.
Uysal, O., Arpaz, E., Berber, M., 2007. Studies on the effect of burden width on blast-
induced vibration in open-pit mines. Environ. Geol. 53 (3), 643–650.
Verma, A.K., Singh, T.N., 2013. Comparative study of cognitive systems for ground
vibration measurements. Neural Comput. Appl. 22 (Suppl. 1), 341–350.
Wu, C.Q., Hao, H., 2005. Numerical study of characteristics of underground blast
induced surface ground motion and their effect on above-ground structures.
Part I. Ground motion characteristics. Soil Dyn. Earthq. Eng. 25 (1), 27–38.
Yilmaz, O., Unlu, T., 2013. Three dimensional numerical rock damage analysis under
blasting load. Tunn. Undergr. Space Technol. 38, 266–278.
Yu, H.T., Yuan, Y., Yu, G.X., Liu, X., 2014. Evaluation of influence of vibrations
generated by blasting construction on an existing tunnel in soft soils. Tunn.
Undergr. Space Technol. 43, 59–66.