International Journal of Rock Mechanics & Mining Sciences 79 (2015) 241–249

Contents lists available at ScienceDirect

International Journal of
Rock Mechanics & Mining Sciences
journal homepage: www.elsevier.com/locate/ijrmms

Technical Note

Fractal analysis of acoustic emission during uniaxial and triaxial

loading of rock
R. Zhang, F. Dai n, M.Z. Gao, N.W. Xu, C.P. Zhang
State Key Laboratory of Hydraulics and Mountain River Engineering, College of Water Resources and Hydropower, Sichuan University, Chengdu, Sichuan
610065, China

art ic l e i nf o

Article history:
Received 9 May 2014
Received in revised form
9 August 2015
Accepted 23 August 2015

Keywords:
Acoustic emission
Triaxial tests
Temporal distribution
Spatial distribution
Fractal mechanism

1. Introduction two different precursor phenomena for predicting the macro-

failure of rocks based on AE records, an accelerated release of
The acoustic emission (AE) technology has been widely used energy and a sharp increase in the loading and unloading response
for site monitoring and forecasting geotechnical hazards or in- ratio14. Yang et al. conducted AE tests on coal under compression
stabilities, such as rockbursts1–11. In a broad sense, AE is a phe- and proposed a method for determining the time to failure15.
nomenon of transient elastic wave emission due to rapid local Vilhelm et al. studied the application of self-correlation analysis to
energy release of materials. It is a course of stress relaxation interpret and analyze the AE signals of rocks16. Moradian et al.
during the transition from an unstable high-energy state to a analyzed the acoustic characteristics of three different material
stable low-energy state caused by an uneven stress distribution in combinations, rock–concrete, rock–rock, and concrete–concrete,
the material. Compared with conventional monitoring methods, during a direct shear tests. The results indicate that the AE tech-
such as a multipoint displacement meters, AE technology can nique is sufficiently accurate to monitor the shear behavior of
monitor the internal cracking of surrounding rocks before the joints and can thus be used for site monitoring17. He et al. in-
failure manifestson the surface of a structure. vestigated the effect of the bedding plane orientation on the rock
Many studies have been carried out on the AE characteristics of burst behavior of sandstone using AE events and AE energy to
rocks. For example, Lockner reviewed the successes and limita- show the damage level and accumulated energy, respectively18.
Previous studies mainly focused on the AE temporal distribu-
tions of AE studies as applied to the fracture process in rock with
tion characteristics but seldom on the three-dimensional AE dis-
emphasis on the ability to predict rock failure12. Application of
tribution representing micro-rupture in space due to the un-
laboratory AE studies to larger scale problems related to the un-
certainty of location of AE source space caused by reflection and
derstanding of earthquake processes was also discussed. Osamu
refraction of stress waves as well as the limited sampling fre-
et al. investigated the stochastic process of AE occurrence, event
quency of testing equipment. Recently, by virtue of the rapid de-
times of AE were measured during steady creep phases under
velopment of AE equipment and computation techniques, an in-
different stress levels13. Zhang et al. conducted AE tests on a large
creasing number of studies on AE spatial distribution tests have
gneiss deposit with a sample dimension of up to 1.05 m, studying been reported19–30. For example, Lockner et al. developed a fast-
acting axial control system that adjusts the load applied to the
n
Corresponding author. sample to maintain a constant acoustic-emission rate, and the
E-mail address: [email protected] (F. Dai). post-peak-stress region, in which a macroscopic fault formed in

http://dx.doi.org/10.1016/j.ijrmms.2015.08.020
1365-1609/& 2015 Elsevier Ltd. All rights reserved.
242 R. Zhang et al. / International Journal of Rock Mechanics & Mining Sciences 79 (2015) 241–249

the sample, can be studied31. Inversion of the arrival-time data 2. Experimental description
provides three-dimensional locations of AE hypocenters, resulting
in detailed imaging of fracture nucleation and propagation. Lock- 2.1. Experimental facilities
ner et al. conducted triaxial compression tests on intact samples of
granite and sandstone at 50 MPa confining pressure32. They aimed The MTS815 Flex Test GT mechanical test system and PCI-2 AE
at investigating the AE events leading up to fault nucleation in an system manufactured by the Physical Acoustic Corporation (PAC)
attempt to observe precursory changes in the locations and am- were used for this study. Real-time monitoring of AE spatial–
plitudes of these events. Lockner et al. also discussed the experi- temporal distribution can be realized simultaneously during the
mental evidence related to nucleation and growth of fractures and loading process. The maximum axial loading capacity of the
these observations were used to develop a conceptual model, MTS815 Flex Test GT mechanical test system is 4600 kN, and the
based on crack interactions, which described the fault nucleation maximum confining pressure that can be applied is 140 MPa. The
phase33. Shah et al. performed unconfined compression experi- range of bandwidth frequencies of the PCI-2 AE system is 1 kHz–
ments on Charcoal granite specimens with the monitoring of 3 MHz. The maximum signal amplitude is 100 dB, and the dy-
acoustic emission34. The analysis of interaction between spatial namic range is more than 85 dB. Real-time AE characteristic
and temporal clustering revealed the size of clusters in both space parameter acquisition, waveform acquisition and analysis can be
and time. Meglis et al. carried out uniaxial compression and bi- carried out simultaneously. Graphical image display and storage of
directional loading tests on large Lac du Bonnet granite the real-time liner location, surface location and spatial location
samples19,20. The micro-cracking and failure mechanism of the during crack development in the sample (i.e. migration track of AE
granite samples were monitored via AE techniques. Lei and Moura source) can be carried out during the test.
et al. conducted researches on rock AE location and rock failure
precursor21–25. Xu et al. performed AE location tests on fine 2.2. Experimental materials
sandstone under a circulating load26, and the AE time-space evo-
lution characteristics and damage evolution of rocks were ana- The testing granite, biotite monzogranite, was collected from
lyzed. Graham et al. studied two methods, the polarity method underground caverns of a hydropower station in Southwest China.
It is composed of coarse grains with varying sizes of quartz (grey),
and moment tensor inversion method, for AE source location
potassium feldspar (flesh pink), anorthose (off-white) and biotite
analysis and classification27. Dresen et al. used AE location tech-
(black). The samples were prepared as standard cylinders with a
nology to study the course of crack nucleation, development and
length of 100 mm and a diameter of 50 mm. The real size and the
evolution by applying three-dimensional uniform pressure on
strength of the samples are shown in Table 1.
Bentheim sandstone samples with different apertures up to
195 MPa28. Ai et al. studied the AE space-time evolution rules and
2.3. Experimental methods
energy releasing characteristics during deformation and the failure
process of coal in triaxial and unloaded tests29,30. Their studies
Both uniaxial compression and triaxial compression tests were
indicated that the acoustic spatial distribution can track the
conducted in conjunction with the real-time monitoring of the AE
cracking evolution of rock samples with pre-fabricated cracks or
spatial–temporal distribution. The uniaxial compression tests were
holes.
conducted on four granite rock samples. The triaxial compression
Although fundamental researches on AE have been performed
tests were conducted on five samples with the confining pressures
in the last decades as described above, AE is one of the few
set to be 10, 20, 40, 60, and 80 MPa.The rock samples were col-
technical disciplines in which theoretical study35,36 lags behind
lected from the same rock block to minimize the sampling
engineering practice, as a theoretical understanding of the AE
variations.
mechanism and testing standard are far from maturity. The In the tests, the longitudinal and transverse deformations of
monitored AE data comprise a massive amount of information rocks were measured by a longitudinal extensometer and a cir-
related to damage evolution of the surrounding rocks, but in- cumferential extensometer, respectively, and the axial load was
vestigation of dynamic disasters according to the AE spatial– measured by an axial force sensor. Six or eight AE sensors were
temporal distribution parameters and migration characteristics used to monitor AE spatial–temporal distribution in real time in
remains difficult. Thus, systematic AE tests and research are ne- uniaxial and triaxial tests, respectively. The type of AE sensors is
cessary to unravel the relationship between AE parameters and micro30 and its frequency is 100–400 kHz. To ensure good cou-
rock failure during the complete rock damage process. pling, vaseline was applied to the contact positions between the
Ever since Mandelbrot initiated the concept of fractals, the rock samples and AE sensors; to reduce boundary effects, Teflon
theory has been received increasing attention from scientists and was applied between the test sample and upper/lower pressure
technicians in various fields37,38. The theory of fractals is based on heads. The AE threshold was set to 40 dB. The schematic drawing
the concept of self-similarity or self-affinity, and it provides a of arrangement of AE sensors is shown in Fig. 1. To check and
powerful nonlinear theory to solve chaotic and irregular problems ensure the location resolution of the AE events, we conducted
that appear to be disordered, unsystematic, scattered and frag- pencil lead fracture tests before experiments. The location re-
mented. Some researchers reported that fractal dimensional re- solution in this test is about 4 mm.
duction of the AE signals can be used to predict rock instability
39–43. However, present research mainly focuses on the temporal
or spatial distribution fractal dimension of AE during the loading 3. Fractal dimension calculation of the AE spatial–temporal
of rocks32,39,42-44, the associated studies of the spatial–temporal distribution
fractal dimension is rarely reported, especially during the triaxial
loading of rocks. 3.1. Fractal dimension of the AE temporal distribution
In this paper, advanced AE testing equipment is employed to
perform both uniaxial and triaxial compression tests on granite The fractal dimension is a key characteristic parameter to de-
samples. The real-time evolutions of AE spatial–temporal dis- scribe fractal structures. It can provide a quantitative description of
tribution were obtained; both fractal characteristics of the spatial the complexity of the internal structure of a substance. The cor-
and temporal distribution of AE are revealed. relation dimension is an important fractal dimension widely
 R. Zhang et al. / International Journal of Rock Mechanics & Mining Sciences 79 (2015) 241–249 243

Table 1 between this point and the other N − 1 points. In this manner, the
Geometrical parameters and failure stress of the cylindrical specimens tested under number of points falling in the volume element with point Xi as
different confining pressures. the center and a small scalar r as the radius can be obtained, and
Sample Confining pres- Diameter Height Peak stress
the “correlation function” Cm(r) can be subsequently determined
number sure (MPa) (mm) (mm) (MPa)
N N ⎡ ⎡ m−1 ⎤0.5⎤
∑ ∑ H ⎢⎢ r − ⎢ ∑ ⎥
1 2
G-6-1 0 50.02 98.36 96.25 Cm(r ) =
N2 ⎢⎣
( xi + k − xj + k) ⎥⎥ ⎥
i=1 j=1 ⎣ k=0 ⎦ ⎦
G-30-2 0 50.01 94.68 91.56 (1)
G-60-1 0 50.01 98.43 79.37
G-60-3 0 50.03 95.18 99.22 when yardstick r adopts different values ri: r1, r2, … , rT , the cor-
G-C10 10 49.97 95.26 249.18 responding correlation function values Cm(ri ) : Cm(r1) , Cm(r2) , … , Cm(rT )
G-C20 20 49.27 97.59 342.91
(where T is the number of r values) can be determined based on Eq.
G-C40 40 49.39 99.39 480.28
G-C60 60 49.15 100.35 594.57 (1). If the temporal distribution possesses a fractal structure, certain
G-C80 80 49.17 100.45 742.70 D can be found to obtain C (r ) ∝ r D for Cm(r ) with r varying over a
certain value range. The slope of the linear segment of the
lg Cm(r )~ lg r curve is defined as the correlation dimension
D = lg C (r )/ lg r . (2)

Generally, the studied fractal systems have no strict self-simi-

larity but possess a similar or statistical self-similarity in a certain
range of scale. Once out of the range, self-similarity no longer
exists. The scale range is the fractal scale-free area. Therefore, the
definition of the scale-free area is crucial. To avoid dispersibility, a
yardstick r = kr0 is typically adopted, where k is a scale factor
( k = 1, 2, 3, …). The value of k significantly affects the calcula-
tion result. If the scale-free area fails to be obtained, the k value
can be adjusted, for example, by setting k = 0.1, 0.2, 0.3, …. r0 is
the mean value of distance between all pairs of points in the set.
N N
1
r0 = ∑∑ Xi − Xj
N2 i=1 j=1 (3)

3.2. Fractal dimension of the AE spatial distribution

The failure of rock under loading is mainly due to the genera-

tion and expansion of cracks inside the rock. The AE space locating
points dictate the fracture position and amount of energy dis-
sipated. Whether the distribution of the AE space locating points is
fractal and what calculation method for the fractal dimension can
best characterize the spatial distribution of the damage remain
unclear.
The yardstick method was the first method developed for
fractal measurement but is not most widely used37. The box-di-
mension method or box-counting method, i.e., the covering
method, is more widely used. For the covering method, a three-
dimensional cube with length r is used to cover the entire rock
sample. Some of the small cube is hollow without any AE events
inside, and some is not hollow, i.e., with AE events inside. The
number of non-hollow cubes may be marked as N(r), and when
Fig. 1. Schematic diagram of AE sensors arrangement. r → 0, the fractal dimension can be obtained as
lg N(r )
applied to handle radar signals, submarine sound signals, geolo- D = lim
gical exploration etc.45,46. The method most widely used to di- r→0 − lg r (4)
rectly calculate the correlation dimension from a one-dimensional For a natural fractal constituted by discrete elements, such as
temporal distribution utilizes the algorithm proposed by Grass- the distribution of rock fragments, the dimension of the covering
berger and Procaccia47–50 based on the embedding theorem of method can be defined as
Whitney and the phase space reconstruction theory of Packard.
The time-varying AE parameters, such as the AE count rate, can be N(r ) = Cr −D (5)
considered one-dimensional temporal distribution; the correlation
dimension can thus be utilized to analyze the fractal character- where N(r) is the number of discrete elements with a characteristic
istics of the AE temporal distribution. size larger than r, and C is a material constant. To show the re-
The correlation dimension is an extension of the traditional lationship between the number and radius, Eq. (5) can also be
dimension, which reflects the degree of freedom of a fractal expressed as follows:
structure. The temporal distribution of an m-dimensional phase M(r ) = Cr D (6)
space is defined as {Xi}i = 1,2, … , N , select one point
Xi = {x i1, x i2, … , x im} as a reference and then calculate the distance where r is the radius of the circle covering the natural discrete
244 R. Zhang et al. / International Journal of Rock Mechanics & Mining Sciences 79 (2015) 241–249

a b c By drawing the M(r )~r curve in logarithmic coordinates, we can

fit the linear segment by the least-squares method. If the two
variants have an adequate linear correlation, the distribution of AE
space locating points during the damage evolution of the sample
can be considered to have self-similarity, i.e., the spatial distribu-
tion of AE has fractal characteristics. Dimension D obtained from
Eq. (6) is named a cluster dimension.
The theories behind the circle, sphere and column covering
methods are shown in Fig. 2. In the circle covering method, the AE
spatial distribution within a columniform specimen is first pro-
jected onto the bottom surface. The center of the bottom is used as
the base, and the circle diameter r is gradually increased to cover
AE events. Then, the number of AE events M(r) covered in each
circle is counted. The sphere covering method uses the shape
center as the base, and the sphere diameter r is gradually in-
creased to cover the AE events. Then, the number of AE events M(r)
covered in each sphere is counted. The column covering method
Fig. 2. The circle, sphere and column covering fractal methods of AE space dis-
tribution (a) Circlecoveringmethod (b) sphere covering method and (c) column
uses the center of the specimen as the base, and the size of the
covering method. covering column (radius: r, height: h) is gradually increased to
cover the specimen. Then, the number of AE events M(r) covered in
elements, and M(r) indicates the number of discrete elements in- each covering column is counted. The fractal dimension of AE
cluded in the circle with radius r. If the discrete element possesses spatial distributions for the circle, sphere and column covering
a fractal distribution, it meets the requirement of the number– methods can be obtained via Eq. (7).
Compared to the circle covering method and sphere covering
radius relation. Taking the logarithm on both sides of Eq. (6), the
method, the column covering method is particularly advantageous
following equation can be obtained:
for studying columniform samples. The circle covering method can
cover all of the damage positions projected onto a plane but fail to
lg M(r ) = lg C + D lg r . (7) reflect the three-dimensional distribution properties of rock

Fig. 3. AE spatial‐temporal evolution during the process of triaxial compression ( σ3 = 20 MPa) (a) AE temporal distribution parameters of count rate (b) AE temporal
distribution parameters of energy rate and (c) AE spatial distribution.
 R. Zhang et al. / International Journal of Rock Mechanics & Mining Sciences 79 (2015) 241–249 245

Fig. 4. Evolution of fractal dimension of AE spatial–temporal distribution under uniaxial compression (a) G-6-1 sample (b) G-30-2 sample (c) G-60-1 sample and (d) G‐60‐3
sample.
246 R. Zhang et al. / International Journal of Rock Mechanics & Mining Sciences 79 (2015) 241–249

damage. The sphere covering method can partially reflect the compression (20 MPa confining pressure). Adopting correlation
spatial distribution properties of rock damage, but the reflected dimension method for the AE temporal distribution fractal, and
properties are only a part of the AE events and do not necessarily column covering method for the AE spatial distribution fractal, the
include the position of the initial damage of the rock sample. The fractal characteristics of the spatial–temporal evolution of AE are
column covering method possesses the feature of being com- analyzed. Here, the fractal dimension of the AE temporal dis-
pletely similar to the sample. It not only includes all of the damage tribution is expressed as Dt , and the fractal dimension of the AE
points but also truly reflects the three-dimensional distribution spatial distribution is expressed as Ds .
properties of rock damage. According to the testing results, for both the uniaxial and
triaxial compression tests, the yardstick r is linearly proportional
to the correlation function Cm(r ) within a certain range in loga-
4. Fractal characteristics of the AE experimental results rithm coordinates, and the correlation coefficient is relatively high.
This indicates that the temporal distribution of AE throughout the
Fig. 3 shows the typical AE temporal distribution parameter entire rock damage process is self-similar. In addition, the yard-
and spatial distribution evolution of a testing sample under triaxial stick r is linearly proportional to the number M(r) in logarithmic

Fig. 5. Evolution of fractal dimension of AE spatial–temporal distribution under triaxial compression (a) G-C10 sample ( σ3 ¼ 10 MPa), (b) G-C20 sample ( σ3 ¼ 20 MPa), (c) G-
C40 sample ( σ3 ¼40 MPa), (d) G-C60 sample ( σ3 ¼60 MPa) and (e) G-C80 sample ( σ3 ¼ 80 MPa).
 R. Zhang et al. / International Journal of Rock Mechanics & Mining Sciences 79 (2015) 241–249 247

Fig. 5. (continued)

coordinates for locating the points of AE in the cylinder within a to orderliness and finally to macroscopic fracture and catastrophic
certain range, and the correlation coefficient is also relatively high, failure51. The fluctuation of the thermodynamic state of the rock is
indicating that the spatial distribution of AE is self-similar as well. mainly reflected in the evolution of various micro-defects. As
Therefore, both temporal and spatial distributions of AE are fractal shown in Fig. 6, the entire rock deformation process up to failure
distributions during the deformation and damage evolution of involves four states: the initial equilibrium state, sub-stable state,
rocks under both uniaxial and triaxial tests. The evolutions of Dt critical state and new stable state. In the sub-stable state, micro-
and Ds under uniaxial compression and triaxial compression are defects may appear at any position inside the rock and are irre-
illustrated in Figs. 4 and 5, respectively, where the horizontal or- gularly distributed. At this stage, the micro-defects are small in
dinate represents the ratio of the load to the peak stress, σ /σc . both dimension and quantity, and the fluctuation never results in a
It should be noted that at a lower stress level during the early sudden change of state. At the critical state, micro-defects occur
loading in the triaxial tests, the confining pressure prevents crack and mainly concentrate in certain areas. Some fluctuations are
formation and extension and thus causes fewer AE signals. The gradually enlarged to cause macroscopic fracture, indicating the
fractal distribution of the spatial–temporal evolution of AE cannot
approach of rock instability and the final formation of a new stable
be calculated at this stage. Thus, in Fig. 5, the fractal dimension is
state. Therefore, rock damage depicts a self-organization phe-
deficient at a lower stress level. For triaxial tests on granite, this
nomenon from orderliness at the initial equilibrium state to the
threshold stress is approximately 50% of the strength.
disordered distribution of micro-defects and orderly macroscopic
fracture.
4.1. Fractal characteristics of the AE temporal distribution
The AE activity of rock reflects the evolution of internal defects
and can be studied statistically. When the stress level is low, the
The dimension of the AE temporal distribution Dt appears to be
corresponding correlation dimension is low, reflecting the order-
in a “rising-reduction mode”, whereas the turning point from
liness of the rock deformation and damage approaching the initial
rising to reduction varies under different loading conditions. For
the uniaxial compression tests, Dt rises rapidly from the initial equilibrium state. With increasing load, the correlation dimension
loading to the region with a σ /σc of approximately 30% and then increases continuously to a maximum, reflecting the disorderli-
reduces with further increases of loading. For the triaxial com- ness of rock deformation and damage approaching the sub-stable
pression tests, Dt rises from initial loading to the region with a σ /σc state. At this moment, the micro-damage in rocks is of different
of approximately 50–80% and decreases with further increases of sizes, disorganized and randomly distributed. With the continuous
loading. increase in the stress level, the correlation dimension drops from
The rock deformation and damage is a process of continuous the maximum to the trough. This indicates that the critical state of
evolution of micro-cracking from randomness and disorderliness rock deformation and damage will transit to a new stable state.
248 R. Zhang et al. / International Journal of Rock Mechanics & Mining Sciences 79 (2015) 241–249

Fig. 6. Variation of thermodynamic state in the process of rock deformation and damage (reproduced from [51]).

Therefore, the variation trend of the “rising-reduction mode” of AE 5. Conclusion

fractal characteristics described in the paper, i.e., the “low value-
high value-low value” of the correlation dimension, reflects the In this paper, uniaxial and triaxial tests on granite samples are
self-organization phenomenon of “orderliness-disorderliness-or- conducted in conjunction with real-time AE temporal-spatial
derliness” during the process of rock deformation and damage. monitoring. The fractal characteristics of AE temporal-spatial dis-
tribution evolution were jointly studied. The AE temporal dis-
4.2. Fractal characteristics of the AE spatial distribution tribution fractal dimension is calculated through the algorithm
proposed by Grassberger and Procaccia, and the AE spatial dis-
For both uniaxial and triaxial tests, the dimension of AE spatial tribution fractal dimension is obtained through the column cov-
distribution appears to be in a “reduction mode” and continually ering method.
declines from the initial loading to the peak strength. At the early For both uniaxial and triaxial tests, the temporal distribution
loading stage, when the applied load is up to approximately 30% of fractal dimension presents a “rising-reducing” mode. The temporal
the peak strength, the dimension of AE spatial distribution Ds is distribution fractal dimension increases dramatically during the
relatively high at approximately 2.85–2.65. In the middle loading initial loading period; as the applied load increases, it shifts from
stage, the load is approximately 30–60% of the peak strength, and an increasing trend to a decreasing trend. The spatial fractal di-
the dimension of the AE spatial distribution Ds continues to de- mension of AE shows a “reducing mode”, decreasing throughout
crease to approximately 2.65–2.45. In the mid- to late loading the loading process up to peak. In the initial loading period, the
stages, the load is approximately 60–100% of the peak strength, spatial fractal dimension tends to be 3, and the micro-failure
and the dimension of the AE spatial distribution Ds decreases ra- spatial distribution of the superficial characteristic of AE tends to
pidly to approximately 2.45–2.1. fill the entire space inside the rock;in the middle and late loading
Dimension is a key factor for a geometric figure. In traditional periods, the spatial fractal dimension continues to decrease and
Euclidean geometrical space, a point is zero-dimensional, a line is tends to be 2, and the micro-failure spatial distribution of the
one-dimensional, a plane is two-dimensional, and a body is three- superficial characteristic of AE tends to converge.
dimensional. Fractal dimension breaks through the traditional
viewpoint of Euclidean integral dimensions. The fractal dimension
of a fractal curve is typically between 1 and 2. A larger fractal di- Acknowledgments
mension results in a more complex curve and leads to a larger
amount of the plane being filled. The fractal dimension of a fractal The authors are grateful for the financial support from the
surface is between 2 and 3. A larger fractal dimension results in a Major State Fundamental Research Project of China (Nos.
more complex surface and leads to a larger amount of the space 2011CB201201 and 2015CB057903), the National Natural Science
being filled. Therefore, at early loading, the dimension of the AE Foundation of China (Nos. 51204113, 51134018) and the New
spatial distribution Ds is relatively high (approximately 3). The AE Century Excellent Talents in University (NCET-13-0382).
space locating points are inclined to fill the entire space inside the
rock. The AE monitoring points for the test were relatively dis-
persed. At the mid and late loading stages, the dimension of the AE References
spatial distribution decreases to 2. The AE events incline to
converge. 1 Tang CA, Xu XH. Evolution and propagation of material defects and Kaiser effect
 R. Zhang et al. / International Journal of Rock Mechanics & Mining Sciences 79 (2015) 241–249 249

function. J Seismol Res. 1990;13(2)203–213. complex critical exponents. J Mech Phys Solids. 2006;54:2544–2553.
2 Poplawski RF. Seismic parameters and rockburst hazard at Mt Charlotte mine. 26 Xu J, Tang XJ, Li SC, Tao Y, Jiang Y. Space-time evolution rules study of acoustic
Int J Rock Mech Min Sci. 1997;34(8)1213–1228. emission locations in rock under cyclic loading. J Chongqing Univ. 2008;31(6)
3 Fujii Y, Ishijima Y, Deguchi G. Prediction of coal face rockbursts and micro- 672–676.
seismicity in deep longwall coal mining. Int J Rock Mech Min Sci. 1997;34(1)85– 27 Graham CC, Stanchits S, Main IG, Dresen G. Comparison of polarity and moment
96. tensor inversion methods for source analysis of acoustic emission data. Int J
4 Srinivasan C, Arora SK, Benady S. Precursory monitoring of impending rock- Rock Mech Min Sci. 2010;47:161–169.
bursts in Kolar gold mines from microseismic emissions at deeper levels. Int J 28 Dresen G, Stanchits S, Rybacki E. Borehole breakout evolution through acoustic
Rock Mech Min Sci. 1999;36:941–948. emission location analysis. Int J Rock Mech Min Sci. 2010;47:426–435.
5 Mansurov VA. Prediction of rockbursts by analysis of induced seismictiy data. 29 Ai T, Zhang R, Liu JF, Zhao XP, Ren L. Space-time evolution rules of acoustic
Int J Rock Mech Min Sci. 2001;38:893–901. emission locations under triaxial compression. J China Coal Soc. 2011;36(12)
6 Moriya H, Fujita T, Niitsuma H, Eisenblätter J, Manthei G. Analysis of fracture 2048–2057.
propagation behavior using hydraulically induced acoustic emissions in the 30 Ai T, Zhang R, Liu JF, Ren L. Space-time evolution rules of Acoustic Emission
Bernburg salt mine, Germany. Int J Rock Mech Min Sci. 2006;43:49–57. location of unloaded coal sample at different loading rates. Int J Min Sci Tech.
7 Kaiser P. Seismic hazard evaluation in undergound construction: theory and 2012;22(6)847–854.
practice. In: Proceedings of the Seventh International Symposium on Rockburst 31 Lockner D, Byerlee J, Kuksenko V, Ponomarev A, Sidorin A. Quasi-static fault
and Seismicity in Mines Proceedings 2009:p.55–58. growth and shear fracture energy in granite. Nature. 1991;350(7)39–42.
8 Lasocki S. Probabilistic seismic hazard analysis for mining-induced seismicity. 32 Lockner D, Byerlee J. Precursory AE patterns leading to rock fracture. In Pro-
In: Proceedimgs of the Seventh International Symposium on Rockburst and ceedings. In: Hardy HR, editor. Fifth Conference on Acoustic Emission/Micro-
Seismicity in Mines Proceedings 2009:p.59–72. seismic Activity in Geological Structures and Materials. Clausthal-Zellerfeld, Ger-
9 Tirfu C.I., Suorineni F.T. Use of micoseismic monitoring for rockburst manage- many: Trans-Tech Publications; 1993.
ment at Vale Inco mines. In: Proceedings of the Seventh International Sympo- 33 Lockner D.,Moore D., Reches Z. Microcrack interaction leading to shear fracture.
sium on Rockburst and Seismicity in Mines 2009:p.1105–1114. In: Tillerson JR and Wawersik WR, eds. Proceedings of the thirty thired U.S. Rock
10 Mutke G., Lurka A., Dubinski J.Seismic monitoring and rock burst hazard as- Mechanics Symposium, Balkema: Rotterdam; 1992:807–816.
sessment in deep polish coal mines-case study of rock best wishes, urst on april 34 Shah K, Joseph F. Damage mechanisms in stressed rock from acoustic emission. J
16, 2008 in wujek-slask coal mine. In: Proceedings of the Seventh International Geophys Res. 1995;100(B8)15527–15539.
Symposium on Rockburst and Seismicity in Mines pp. 1413–1424. 35 McLaskey G, Glaser S. Acoustic Emission Sensor Calibration for Absolute Source
11 Paul Young R, Martin CD. Potential role of acoustic emission/microseismicity
Measurements. 2012, http://dx.doi.org/10.1007/s10921-012-0131-2.
investigations in the site characterization and performance monitoring of nu-
36 Duca S, Occhiena C, Sambuelli LA. Procedure to determine the optimal sensor
clear waste repositories. Int J Rock Mech Min Sci GeomechAbstr. 1993;30(7)797–
positions for locating ae sources in rock samples. Rock Mech Rock Eng. 2014,
803.
http://dx.doi.org/10.1007/s00603-014-0582-0.
12 Lockner D. The role of acoustic emission in the study of rock fracture. Int J Rock
37 Xie H.P. Fractals in rock mechanics. A.A. Balkema, P.O. Box 1675, 3000BR Rot-
Mech Min Sci GeomechAbstr. 1993;30(7)883–899.
terdam, Neterlands. A.A. Balkema Publishers, Old Post Road, Brookfield, VT
13 Nishizawa O, Noro H. A self-exciting process of acoustic emission occurrence in
05036,USA; 1993(Geomechanics research series editor-in –chief: Marek A.
steady creep of granite under uniaxial stress. Geophys Res Lett. 1990;17:1521–
Kwasniewski).
1524.
38 Katsuhiko (author), XT (translator) Feng. Application of Acoustic Emission (AE)
14 Zhang HH, Yan YD, Yu HZ, Yin XC. Acoustic emission experimental research on
Technology. Beijing: Metallurgical Industry Press; 1996.
large-scaled rock failure under cycling load—fracture precursor of rock. Chin J
39 Xie HP, Pariseau WG. The fractal characteristics and mechanism of rock burst.
Rock Mech Eng. 2004;23(21)3621–3628.
15 Yang YJ, Chen SJ, Han GD. Experimental on acoustic emission during com- Chin J Rock Mech Eng. 1993;12(1)28–37.
pression rupture procedure of coal sample. J China Coal Soc. 2006;31(5)562– 40 Xie HP, Liu JF, Ju Y, Li J, Xie LZ. Fractal property of spatial distribution of acoustic
565. emissions during the failure process of bedded rock salt. Int J Rock Mech Min Sci.
16 Vilhelm J, Rudajev V, Lokajíček T, Živor R. Application of autocorrelation ana- 2011;48:1344–1351.
lysis for interpreting acoustic emission in rock. Int J Rock Mech Min Sci. 41 Lu C, Mai YW, Xie H. A sudden drop of fractal dimension: a likely precursor of
2008;45:1068–1081. catastrophic failure in disordered media. Philos Mag Lett. 2005;85(1)33–40.
17 Moradian ZA, Ballivy G, Rivard P, Gravel C, Rousseau B. Evaluating damage 42 Hirata T, Satoh T, Ito K. Fractal structure of spatial distribution of micro-
during shear tests of rock joints using acoustic emissions. Int J Rock Mech Min fracturing in rock. Geophys J R Astr Soc. 1987;90:369–374.
Sci. 2010;47:590–598. 43 Gao F, Li JJ, Li XY, Chen J. Fractal analysis of the characteristics of acoustic
18 He MC, Nie W, Zhao ZY, Guo W. Experimental investigation of bedding plane emission of rock. J Wuhan Univ Technol. 2005;27(7)67–69.
orientation on the rockburst behavior of sandstone. Rock Mech Rock Eng. 44 Yin XG, Li SL, Tang HY. Study on strength fractal features of acoustic emission in
2012;45:311–326. process of rock failure. Chin J Rock Mech Eng.. 2005;24(19)3512–3516.
19 Meglis IL, Chow TM, Young RP. Progressive microcrack development in tests on 45 Tiwari RK, Rao KNN. Phase space structure, attractor dimension, Lyapunov ex-
Lac du Bonnet Granite- I. Acoustic emission source location and velocity mea- ponent and nonlinear prediction from earth’s atmospheric angular momentum
surements. Int J Rock Mech Min Sci GeomechAbstr. 1995;32(8)741–750. time series. Pure Appl Geophys. 1999;156:719–736.
20 Meglis IL, Chow TM, Young RP. Progressive microcrack development in tests on 46 Sivakumar B, Jayawardena AW, Fernando TMK. River flow forcasting: ues of
Lac du Bonnet Granite- II. Ultrasonic tomographic imaging. Int J Rock Mech Min phase-space reconstruction and artificial neural networks approaches. J Hydrol.
Sci GeomechAbstr. 1995;32(8)751–761. 2002;265(1)225–245.
21 Lei XL, Nishizawa O, Kusunose K, Cho A, Satoh T, Nishizawa O. Compressive 47 Grassberger P, Procaccia I. Dimensions and entropies of strange attractors from
failure of mudstone samples containing quartz veins using rapid AE monitoring: a fluctuating dynamics approach. Physica D: Nonlinear Phenom. 1984;13(1)34–
the role of asperities. Tectonophysics. 2000;328:329–340. 54.
22 Lei X, Masuda K, Nishizawa O, Jouniaux L, Liu L, Ma W, Kusunose K. Detailed 48 Grassberger P, Procaccia I. Characterization of strange attractors. Phys Rev Lett.
analysis of acoustic emission activity during catastrophic fracture of faults in 1983;50(5)346.
rock. J Struct Geol. 2004;26:247–258. 49 Grassberger P, Procaccia I. Measuring the strangeness of strange attractors.
23 Lei X, Satoh T. Indicators of critical point behavior prior to rock failure inferred Physica D. 1983;9:189–208.
from pre-failure damage. Tectonophysics. 2007;431:97–111. 50 Grassberger P, Procaccia I. Estimation of the Kolmogorov entropy from a chaotic
24 Moura A, Lei X, Nishisawa O. Prediction scheme for the catastrophic failure of signal. Phys Rev A. 1983;28(4)2591–2593.
highly loaded brittle materials or rocks. J Mech Phys Solids. 2005;53:2435–2455. 51 Peng RD. Research on rock damage and strength based on energy dissipation
25 Moura A, Lei X, Nishisawa O. Self-similarity in rock cracking and related and release. Ph.D. Dissertation; 2005[in Chinese].