You are on page 1of 7

ISSN 1062-7391, Journal of Mining Science, 2020, Vol. 56, No. 1, pp. 29–35. © Pleiades Publishing, Ltd., 2020.

Russian Text © The Author(s), 2020, published in Fiziko-Tekhnicheskie Problemy Razrabotki Poleznykh Iskopaemykh, 2020, No. 1, pp. 33–40.

___________________________________
____________________________________________________________________________________________________________________________________________
ROCK __________________________________
________________________________________________________________________________________________________________________________________

FAILURE
Modeling Seismic Vibrations under Massive Blasting
in Underground Mines
A. N. Kholodilova* and A. P. Gospodarikovb**
a
Saint-Petersburg State University of Airspace Instrument Engineering, Saint-Petersburg, 190000 Russia
*e-mail: kholodilov@mail.ru
b
Saint-Petersburg Mining University, Saint-Petersburg, 199106 Russia
**e-mail: Gospodarikov_AP@pers.spmi.ru
Received October 28, 2019
Revised February 3, 2020
Accepted February 4, 2020

Abstract—The theoretical model is proposed, which allows tracing continuation of seismic vibrations after
cessation of blast load by the analysis of velocigrams recording on ground surface during underground
explosions. The comparison of the model and experimental velocigrams of massive blasts shows validity of
elliptical filters for second-order low frequencies in modeling waveforms of the velocigrams. The model
efficiency in the detection of time-delay errors is proved. The conditions of predicting peak particle
velocities based on explosive weight per blast are determined. The pre-conditions for the resonance
excitation in the rock mass–guarded object system are discussed, and the blast-induced load is predicted.
Keywords: Blasting, velocigram, seismic safety, blast-induced seismic vibrations, underground mine,
installations above ground, seismic load prediction, elliptical filter.
DOI: 10.1134/S1062739120016454

INTRODUCTION
The necessity for studying the blast-induced load on the installations above ground at the Novo-
Shirokinskoye Mine is caused by the brickwork collapse immediately after the massive blast.
Preliminary analysis showed that the collapse occurred at an estimated value of peak particle
velocity in the order of 1.3 cm/s [1]. More detailed studies based on peak velocity module
enveloping method led to the conclusion about an additional source of seismic vibrations in the
installation–ground system acting after cessation of blast-induced seismic vibrations [2].
A similar problem is encountered in mining rockburst-hazardous deposits [3]. However, Novo-
Shirokinskoe deposit of polymetallic ores at the initial stage of mining cannot be related to such
deposits. The influence of rock mass structure on particle velocity of seismic vibrations is considered
in [4]. The ground conditions of resonant amplification of seismic vibrations in predicting peak
particle velocity were studied in [5]. Recently, in addition to traditional formulas, models based on
artificial neural networks, fuzzy logic and genetic algorithms are widely used to predict peak particle
velocity [6]. Moreover, the calculation model of massive blast wavefield should be noted, which is
implemented as a representation of seismic vibrations in the form of the Puzyrev pulse [7]. The above
studies do not guarantee the possibility of analyzing additional sources of seismic vibrations, which
appear after cessation of blast-induced seismic vibrations. We consider an approach based on
modeling the waveform of a velocigram.
1. BASIC THEORETICAL PROVISION OF THE MODEL
Assume that medium vibrations are elastically damped at a distance of 60 blasthole radii in hard
rock. In a first approximation, the basis of rock mass disturbance description is the approximation for
overpressure in the shock air wave depending on time at a given point, according to [8]:

29
30 KHOLODILOV, GOSPODARIKOV

⎛ t ⎞ ⎛ t ⎞
ΔP (t ) = ΔP0 ⎜ 1 − ⎟ exp ⎜ − a ⎟, (1)
⎝ τ ⎠ ⎝ τ ⎠
where ΔP0 is the pressure at the front of shock air wave; τ is the duration of compression phase;
a is the empirical coefficient.
We take into account that acoustic impedance of rock for the conditions of Novo-Shirokinskoye
deposit is 20–30 thousand times higher than acoustic impedance of air, i.e., the disturbance relaxation
in rock occurs much faster than in air. In the approximation of a plane seismic wave, the stress
(pressure) at the front of a seismic blast wave is directly proportional to the particle velocity. Using
the Sadovsky formula and approximation (1), we obtain the expression for particle velocity at a given
point for a short-delay blast:
N ⎧⎪⎛ Q ⎞ n ⎛ t − t ⎞ ⎡ ⎛ t − t1, k ⎞b ⎤ ⎫⎪
v (t ) = v0 ∑ ⎨⎜ k ⎟ ⎜ 1 − 1, k
⎟ exp ⎢ − a ⎜ ′ ⎟ ⎥ ⎬ . (2)
k =1 ⎪⎝ Qm ⎠ ⎝ τ′ ⎠ ⎢⎣ ⎝ τ ⎠ ⎥⎦ ⎪⎭

Here, Qk is the explosive charge weight attributable to the k-th group of simultaneously blasted
N
charge; N is the number of groups of simultaneously blasted charges; QΣ = ∑ Qk is the explosive
k =1

weight per massive blast; Qm is the maximum weight of simultaneously blasted charge; t1,k is the
operation time of the k-th group of simultaneously blasted charge, counted from the operation time of
the first group of charges (zero delay, t1,k =1 = 0) ; v0 is the normalizing particle velocity; n is the
damping coefficient of seismic vibrations; b is the empirical coefficient taking into account the
difference in acoustic impedances of rock and air. The duration of compression phase τ ′ is chosen
with regard to the maximum height of charge column, explosive detonation velocity and estimated
duration of quasistatic blasting stage.
In the approximation of a spherical source of blast-induced seismic waves, let us consider the
contribution of a longitudinal volume wave and surface Rayleigh wave to blasting concussion. Given
that the waves originate from the same source, and the particle vibrations in the Rayleigh wave have
radial and vertical components, it is reasonable to assume the additivity of wave radial components.
For particle velocity of the resulting vibration, we obtain:
v′(t ) = η p v (t ) + η R v(t − τ 2 ) , (3)
where η p , η R is the contribution of longitudinal and surface waves, respectively, to the velocity
amplitude of resulting vibrations (η p + η R = 1) ; τ 2 is the delay time of arrival of surface waves
relative to the moment of arrival of longitudinal waves at a given point.
The actual form of model components of a velocigram (horizontal, vertical) can be obtained only if
the transformation of function (3) is applied, which in its characteristics repeats the medium behavior
under dynamic loading. As such a transformation, we take the low frequency filter (LFF). Let us
formulate the requirements for it.
One of the rock mass model representations is Hooke’s body, so the LFF order must be even.
According to Kotelnikov (sampling) theorem, we choose the LFF cut-off frequency as f ≈ 1/ (2τ ′) .
Seismic signal damping in the LFF stop band is taken as D1 = 40 dB. Seismic signal damping within
the LFF pass band D2 is a characteristic of velocigram registration point remoteness from the blast
source, and therefore must satisfy the condition D2 < D1 .

JOURNAL OF MINING SCIENCE Vol. 56 No. 1 2020


MODELING SEISMIC VIBRATIONS UNDER MASSIVE BLASTING IN UNDERGROUND MINES 31

There are different approximations for the frequency characteristics of filters. We choose an
elliptic filter [9], which is symmetrical with respect to the ripples of amplitude-frequency
characteristic in the pass and stop bands and has maximum discrimination in the transition region.
At the same time, the elliptical filter has the worst phase-frequency characteristic and can weakly
suppress the high-frequency components of the signal. The above disadvantages are an advantage of
the elliptical filter in terms of transfer of rock mass dynamic properties.
2. MODELING WAVEFORMS OF EXPERIMENTAL VELOCIGRAMS
For modeling the waveforms of velocigrams obtained from massive blasting, it is required to
determine empirical coefficients a and b included in (2). To do this, we use the velocigram recorded
from blasting of a single explosive charge at a distance of 60 blasthole radii [10]. In this case (2) takes
the following form:
⎛ t ⎞ ⎡ ⎛ t ⎞b ⎤
v (t ) = v0 ⎜ 1 − ⎟ exp ⎢ − a ⎜ ⎟ ⎥ .
⎝ τ′⎠ ⎢⎣ ⎝ τ ′ ⎠ ⎥⎦
The velocigram approximation is obtained at τ ′ = 0.8 ms, a = 0.026 , b = 4 (Fig. 1). A noticeable
oscillation of the experimental curve is observed relative to its approximation. However, this does not
impair the identification of temporary compression and rarefaction regions of the medium.
When modeling waveforms of the velocigrams obtained from massive blasting during ore breaking
by vertical fans, according to [1], the damping coefficient of seismic vibrations n = 2 . Based on the
average length of explosive charge column and average velocity of AS-8 granulite detonation, as
a first approximation we take the compression phase duration τ ′ = 10 ms. Then the upper estimate of
the LFF cut-off frequency will be f = 50 Hz. Other technological parameters of massive blasting are
given in [1].
Modeling begins with the coordination of the initial sections of the model and experimental
velocigrams by selecting the time intervals for short-delay blasting (SDB). Then, the delay time of
surface wave arrival is adjusted and SDB time intervals are corrected. The modeling procedure is
completed by fitting the scaling factor and choosing the shares of surface and longitudinal waves to
the best agreement between the model and experimental velocigrams.
The results of determining the parameters of waveforms of velocigrams for the vertical component
of particle velocity are presented in Table for second-order elliptical LFF. Except for massive blast I,
measurements were made near the guarded object. In case of massive blast I, the velocigram was
recorded approximately in the middle of seismic road between the blast and a guarded object.

Fig. 1. Ore body velocigrams: 1—experiment; 2—approximation.

JOURNAL OF MINING SCIENCE Vol. 56 No. 1 2020


32 KHOLODILOV, GOSPODARIKOV

Velocigram waveform modeling results


Qm , LFF parameters Velocigram parameters
Blast no. QΣ , kg N
kg D2 , dB f , Hz v0 , cm/s ηp ηR τ 2 , ms
I 2986 371 10 16 35 5.3 0.30 0.70 33
IV 2147 360 8 22 38 1.2 0.45 0.55 80
V 1748 471 4 22 38 3.8 0.17 0.83 84
VI 3922 601 8 21 39 5.2 0.50 0.50 66

Figure 2a shows an example of input signal curve for the low frequency filter plotted using (2)
on the basis of tabular parameters for massive blast I, Fig. 2b demonstrates model (after filtration) and
experimental velocigrams for the same blast. Model and experimental velocigrams for massive
blast IV and VI are shown in Figures 2c and 2d. Replacing the second-order LFF with the fourth-
order LFF does not practically change the model parameters of velocigrams. The difference is in the
necessity to adjust the cut-off and damping frequency in the stop band. The second-order LFF
behaves more steadily than the fourth-order filter.
A very significant difference is noted between the model and design times of short-delay blasting
(Fig. 3). The SDB design time was 50 ms. This problem is inherent in the systems of non-electric
blasting initiation, when practically simultaneously initiation of the explosive charge groups
belonging to different delay stages is possible [4, 5, 11]. The seismic load of massive blasts (except
for blast V) is characterized by the combined contribution of volume longitudinal and surface waves
in equal proportions.
Figure 2 shows the reliably recorded abnormally high particle velocities in time domain when the
seismic source ceased its activity. This effect is especially pronounced in the velocigram from
massive blast VI. In the 0.45–0.75 s range, experimental particle velocities exceed model ones up to
3 times. From the Table analysis it follows that the mentioned excess correlates with the explosive
weight per one massive blast reliably.

Fig. 2. Velocigrams from (a, b) I, (c) IV and (d) VI massive blasts: full line—experimental velocigram; dashed
line—model velocigram; vertical lines—responce times of delay stages.

JOURNAL OF MINING SCIENCE Vol. 56 No. 1 2020


MODELING SEISMIC VIBRATIONS UNDER MASSIVE BLASTING IN UNDERGROUND MINES 33

Fig. 3. Velocigrams from (a) IV and (b) V massive blasts: line—experimental velocigram; dashed line—model
velocigram with model and design SDB times (dots); vertical lines—responce times of delay stages.

The effect of short-delay blasting parameter spread on seismic load of blasting is shown in Fig. 3.
The deviation of SDB model parameters from design ones ambiguously affects the velocigram. For
massive blast V, this deviation does not change peak particle velocities. In case of blast IV, a three-
fold increase in peak particle velocity is observed. In this regard, we note the influence of blasting
time-delay error on seismic load of blasting [12].
Let us estimate the equivalent charge of a massive blast IV for model delay intervals using the
formula Q ′ = Qm (v1 / v2 )3/ n ( v1 / v2 is the ratio of peak experimental particle velocity to peak model
particle velocity obtained for the SDB design parameters). Assuming Qm = 360 kg, v1 / v2 = 3 , n = 2 ,
we obtain Q′ = 1871 kg. If QΣ = 2147 kg, then an unexpected conclusion follows: when the SDB
parameters deviate from the design values, the seismic load of a massive blast becomes similar to
a simultaneously blasted charge. To estimate the seismic load of a massive blast in the mining
practice, there is an approach when not the maximum weight of simultaneously blasted explosive
charge is used, but the weight of an explosive blown up per massive blast [13].
3. RESULTS AND DISCUSSION
Let us analyze the anomaly of particle velocity in the time domain, in which the primary source of
seismic vibrations is not active. Assume that physical explanation for this phenomenon is the
temporary accumulation of the energy of seismic vibrations by the rock mass. Then, the rock mass
should be considered as a secondary source of seismic vibrations. In a first approximation, the
primary source can be considered quasiperiodic with a frequency of seismic vibrations of about
25 Hz. Since seismic vibrations of the secondary source occur at the same frequency, the prerequisites
for resonant excitation of the rock mass near the above-ground installations are created.
Assume that the amplitude changes according to the exponential law [14]. To estimate the ratio
of peak experimental particle velocity to peak model particle velocity from the secondary source, we
take the following regression equation:
v′
ln 1 = k1QΣ + k 2 , (4)
v2′
where k1 , k2 are the regression coefficients. For the weight of QΣ expressed in tons, k1 = 0.38 ,
k2 = −0.38 . The condition v1′ / v2′ = 1 means the absence of a secondary source, from which QΣ = 1 t.
Specify the seismic prediction in the guarded object area before the brickwork collapse
considered in [1]. Massive blast parameters are QΣ = 5052 kg, Qm = 1946 kg. The distance from the
massive blast to emergency section of the building was R = 277 m.

JOURNAL OF MINING SCIENCE Vol. 56 No. 1 2020


34 KHOLODILOV, GOSPODARIKOV

Assuming deviation of the SDB parameters from design values, we determine the value of the
reduced distance RΣ = R / QΣ1/3 = 16.1 m/kg1/3. Based on the maximum explosive weight in the group
of simultaneously blasted charges, the value of the reduced distance RΣ = R / Qm1/3 = 22.2 m/kg1/3,
which corresponds to peak vertical component of particle velocity vv = 1.3 cm/s. In this case, the
prediction is vv′ = ( Rm / RΣ ) n = 2.5 cm/s.
In an approximation of resonant excitation in the rock mass-guarded object system, according
to (4), we obtain v1′ / v2′ = 4.7 . From the analysis of massive blast VI velocigram (Fig. 2c) it follows
that the relation of peak vertical components of particle velocities from the secondary and primary
sources of seismic vibrations is approximately η = 1.3 . Then, the prediction includes the values from
vv′′ = vv ⋅ v1′ / v2′ ⋅η = 2 cm/s to vv′′′ = vv′ ⋅ v1′ / v2′ ⋅η = 3.9 cm/s. From the above analysis it follows that,
in order to ensure the seismic safety of guarded object, it is required that peak particle velocity from
the secondary vibration source should not exceed peak particle velocity from the primary (blasting)
vibration source.
In accordance with [1], maximum admissible value 0.7 cm/s was adopted for the peak vertical
component of particle velocity, which turned out to be exceeded by more than 5 times in case of the
most unfavorable prediction.
CONCLUSIONS
From the analysis of velocigrams recorded on the Earth’s surface during underground blasting, the
theoretical model for finding seismic loads not associated with blast-induced seismic vibrations
allows detecting seismic activity in the rock mass-guarded object system and deviation of the SDB
time intervals from nominal values.
The condition for the existence of a secondary source of seismic vibrations is determined and
relative contribution of longitudinal and surface blast-induced seismic waves to the seismic load on
the guarded object is specified. The correct use of even-order elliptic low frequency filters
in modeling velocigram waveforms is shown by comparing model and experimental velocigrams
from massive blasts. In some cases, when the parameters of short-delay blasting deviate from design
values, the seismic load of a massive blast becomes similar to a simultaneously blasted charge. The
limiting value of seismic load at which the guarded object remains secured corresponds to the equality
of peak values of particle velocities of the rock mass-guarded object seismic response and blast-
induced seismic waves.

REFERENCES
1. Artemov, V.А., Vinogradov, Yu. I., Kholodilov, А. N., Gustov, S.V., and Shcherbakov, N.Ya., Seismic
Safety of Massive Blasting at the Novo-Shirokinsky Mine, Vzryvnoe delo, 2011, no. 105/62, pp. 239–252.
2. Kholodilov, А.N. and Gospodarikov, A.P., Methodology of the Assessment of Seismic Safety of
Explosions Executed in Underground Mines near Heapsteads, GIAB, 2016, no. 2, pp. 320–328.
3. Eremenko, А.А., Mashukov, I.V., and Eremenko, V.А., Geodynamic and Seismic Events under Rockburst-
Hazardous Block Caving in Gornaya Shoria, J. Min. Sci., 2017, vol. 53, no. 1, pp. 65–70.
4. Oparin, V.N., Yushkin, V.F., Porokhovsky, N.N., Grishin, А.N., Kulinich, N.А., Rublev, D.Е.,
and Yushkin, A.V., Effect of Large-Scale Blasting on Spectrum of Seismic Waves in a Stone Quarry, J.
Min. Sci., 2014, vol. 50, no. 5, pp. 865–877.

JOURNAL OF MINING SCIENCE Vol. 56 No. 1 2020


MODELING SEISMIC VIBRATIONS UNDER MASSIVE BLASTING IN UNDERGROUND MINES 35

5. Verkholantsev, А.V., Dyagilev, R.А., Shulakov, D.Yu., and Shkurko, А.V., Monitoring of Earthquake
Loads from Blasting in the Shakhtau Open Pit Mine, J. Min. Sci., 2019, vol. 55, no. 2, pp. 229–238.
6. Prashanth, R. and Nimaje, D.S. Estimation of Peak Particle Velocity Using Soft Computing Technique
Approaches, Noise and Vibration Worldwide, 2018, vol. 49, nos. 9–10, pp. 302–310.
7. Kishkina, S.B., Seismic Load of Large-Scale Chemical Blasting in the Mines of Kursk Magnetic
Anomaly, Cand. Phys. Math. Sci. Thesis, Moscow, 2002.
8. Orlenko, L.P. (Ed.), Fizika vzryva (Physics of Explosion), Moscow: Fizmatlit, 2004.
9. Oppenheim, А.V and Schafer, R.W, Digital Signal Processing, Pearson, 1975.
10. Kholodilov, А.N., Artemov, V.А., and Vinogradov, Yu. I., Estimation Procedure of a Line of Least
Resistance with Acceleration Information for Rock Blasting Technique, GIAB, 2013, no. 5, pp. 314–318.
11. Ekvist, B.V., Increasing the Safety of Short-Delay Blasting, GIAB, 2017, no. 5, pp. 389–394.
12. Han, L., Li, H., Liu, D., Ling, T., Li, C., and Liang, S., Probability Analysis for Influence of Time-Delay
Error of Detonators on Superposed Seismic Wave Vibration Reduction, J. of Vibration and Shock, 2019,
vol. 38, no. 3, pp. 96–101.
13. Pandula, B. and Jelšovská, K., New Criterion for Estimate of Ground Vibrations during Blasting Operations
in Quarries, Acta Geodynamica et Geomaterialia, 2008, vol. 5, no. 2 (150), pp. 147–152.
14. Yakubovich, V.A. and Starzhinsky, V.M., Lineinye differentsia’lnye uravneniya s periodicheskimi
koeffitsientami i ikh prilozheniya (Linear Differential Equations with Periodic Coefficients and
Applications), Moscow: Nauka, 1972.

JOURNAL OF MINING SCIENCE Vol. 56 No. 1 2020

You might also like