applied

sciences
Article
Seismic b-Value for Foreshock AE Events Preceding
Repeated Stick-Slips of Pre-Cut Faults in Granite
Xinglin Lei 1, *, Shinian Li 2 and Liqiang Liu 2
1 Geological Survey of Japan, National Institute of Advanced Industrial Science and Technology,
AIST Central #7, Higashi 1-1-1, Tsukuba, Ibaraki 305-8567, Japan
2 State Key Laboratory of Earthquake Dynamics, Institute of Geology, China Earthquake Administration,
Beijing 100029, China; lishinian2007@126.com (S.L.); liulq48@hotmail.com (L.L.)
* Correspondence: xinglin-lei@aist.go.jp; Tel.: +81-29-861-2468




Received: 11 October 2018; Accepted: 20 November 2018; Published: 23 November 2018


Abstract: In this study, the b-values for acoustic emission (AE) events during stick-slip cycles of
pre-cut faults in granite (as an analogue of unfavorably oriented immature faults) under triaxial
compression (confining pressure: 40 MPa) are investigated. Using a multi-channel AE waveform
recording system and two peak detectors, we recorded AE waveforms at 16 bits and at a sampling
rate of 25 MHz, as well as the maximum amplitude of AE events with a dynamic range of 55 dB.
For stick-slip events, the b-value decreases from 1.2 to 1.5 to approximately 0.6 as the shear stress
increases, and then quickly jumps back to 1.0 to 1.3 immediately prior to the dynamic stress drop.
The minimum b-value coincides with the maximum event rate and a stress level of 70 to 95% of the
shear strength. It is also observed that the AE activity during each cycle was linked with the pre-failure
fault slip, which accounts for 30% of the dynamic slip. Our results on b-value evaluation preceding
repeated stick-slips can be used as an indicator of the degree of fault maturity and shear stress acting
on the fault, which is important in seismic hazard assessment and earthquake prediction, especially
for the injection-induced seismicity for fields in which reactivated shear rupture of unfavorable and
immature faults or tensile fractures is important.

Keywords: acoustic emission; granite; stick-slip; seismic b-value

1. Introduction
Reactivation of pre-existing immature and/or unfavorably oriented faults is an important issue
in earthquake seismology, particularly in cases of seismicity induced by fluid injection or extraction
in applications such as enhanced geothermal systems, shale gas fracking, nonconventional oil/gas
production, and sequestration of CO2 and other fluid wastes into deep formations [1,2]. The conditions
for shear rupturing along favorably oriented (easiest to be reactivated under the present-day regional
stress environment) and well-developed faults (weakened by fault gauges and smooth fault surfaces)
provides the lower limiting bound to overpressures, which is a key parameter in the design and
management of injection applications. In geothermally active areas, earthquakes that occurred on
unfavorably oriented faults are considered to be evidence of over-pressurized fluids and have attracted
increasing attention (e.g., [3–6]). At the same time, immature and/or unfavorably oriented faults,
especially those distributed in sedimentary formations, are also important because such faults are
locked and show very low levels of background seismicity, but might be reactivated due to man-made
local pore pressure and stress changes [2]. Much more attention should be paid to such faults in
geomechanical analyses and risk assessments of related applications.
Acoustic emission (AE) events have been observed during frictional sliding on pre-cut faults in
laboratory rock specimens. In addition to rock blocks, the roughness of the fault surface, geometric

Appl. Sci. 2018, 8, 2361; doi:10.3390/app8122361 www.mdpi.com/journal/applsci

Appl. Sci. 2018, 8, 2361 2 of 10

features (bend and orientation) of the fault, gouge, and loading speed have been investigated [7–9].
Variations in b-value during stress buildup and release were observed in laboratory-created fault
zones [10]. In a very recent study, the temporal evolution of the seismic b-value during stable and
unstable frictional sliding experiments on layers of glass beads in a double-direct shear configuration
was investigated [11]. Thus far, AE activity related to the stick-slip of unfavorably oriented faults
has not been systematically investigated in the laboratory. The authors, therefore, carried out AE
studies using rock specimens containing artificially pre-cut faults as well as naturally created faults in
well-controlled laboratory conditions [12]. In the present study, stick-slip tests were performed on a
rough fault pre-cut at 45◦ in granitic specimens. Such a fault can be considered to be an analogue of
unfavorably orientated immature faults.

2. Specimens and Experimental Setup

This paper focus on the results of three experiments conducted on intact (SA) or pre-cut (S1 and
S2) cylindrical specimens 50 mm in diameter and 125 mm in length. These specimens were cored from
a single granite block. The rock block is a homogeneous and fine-grained granite, which shows very
weak anisotropy and low density of pre-existing microcracks [12]. The pre-cut surface had an angle of
45◦ with respect to the axis, which is the direction of the greatest principal stress under the experimental
conditions. All specimens were naturally dried under ambient conditions. The fine-grained granite
had a relatively low density of pre-existing microcracks, as indicated by the high P-wave velocity of
approximately 6.0 km/s. The results for an intact specimen were used for reference and for estimating
fault slip in the pre-cut specimens. The typical range for the frictional coefficient for faults in the
upper crust is 0.6 to 0.75. The early results of the present study, including data for specimens pre-cut
at 50◦ , showed that the friction coefficient for the pre-cut faults in the granite specimens does not
depend on the fault angle, but drops from approximately 0.75 to 0.6 after a few stick-slip iterations [12],
and falls within the typical range in the upper crust. Based on the Mohr–Coulomb failure criteria,
the optimal angle between the fault plane and the maximum principal stress axis is approximately
25◦ –30◦ . The primary results show that the pre-cut faults in our test specimens were deemed to be
unfavorably oriented, and their friction strength is expected to be 50% greater than that for an optimally
oriented fault [12].
The roughness of the pre-cut surfaces was measured using a laser scanner. As shown in Figure 1,
the power spectral density function follows a power law: Sv ( f ) ∝ λ β = f1β . Thus, the generalized
Hurst exponent (H, [0, 1]) and the fractal dimension (D) can be determined, because these parameters
are correlated with each other: β = 2H + 1 and D = d + 1 − H. The estimated fractal dimension along
the dip direction (slip direction during the experiment) was 1.30, and that for strike direction was 1.55.
Since the fractal features of the fault surface are key to the similarity (scale independence) between
faults of different size, the pre-cut specimens used in the present study can be considered to be an
analogue of natural faults.
The loading and AE monitoring systems used in the present study were similar to those used in
previous studies [13]. First, specimens were hydrostatically compressed by gradually increasing the
confining pressure to 40 MPa. The axial load was then increased at a rate of 10 MPa/min, while the
confining pressure was kept constant. In our loading system, the axial force is balanced by a spring,
which is simply compressed by a stepping motor. After a stress drop, the force of the spring does not
change, so the reloading speed is very fast (5.5 MPa/s), and the time interval between two consecutive
stick-slip events is a few tens of seconds. Thus, time-dependent healing can be ignored.
Appl. Sci. 2018, 8, 2361 3 of 10
Appl.
Appl. Sci.
Sci. 2018,
2018, 8,
8, xx 33 of
of 10
10

Figure
Figure1.
Figure 1.Roughness
1. Roughnessof
Roughness ofpre-cut
of pre-cutsurface
pre-cut surface(S1)
surface (S1)measured
(S1) measured
measuredbyby laser
bylaser scanner.
laserscanner. A
scanner.A and
Aand
andBBBin
in(b)
in (b)(rectangle
(b) (rectangleregion
(rectangle region
region
in
in (a)) indicates direction of two artificially selected profiles, which were used for the calculation
in (a)) indicates direction of two artificially selected profiles, which were used for the calculation of
(a)) indicates direction of two artificially selected profiles, which were used for the calculation of
of
power
power spectral
powerspectral density
spectraldensity function
densityfunction (PDF)
function(PDF)
(PDF)of of the
ofthe roughness
theroughness in
roughnessin (c).
in(c).
(c).

3.
3. AE Data
Data and Methods
3. AE
AE Data andand Methods
Methods
In
In order to record ultrasonic signals radiating from microcracking and stick-slip events, and as
In order
order to
to record
record ultrasonic
ultrasonic signals
signals radiating
radiating from
from microcracking
microcracking and and stick-slip
stick-slip events,
events, and
and as
as
a pulse source for performing velocity measurements, 12 piezoelectric transducers (PZT) (diameter:
a pulse source for performing velocity measurements, 12 piezoelectric transducers (PZT) (diameter:
5 mm, resonant frequency: 2 MHz) were mounted on the curvilinear
5 mm, resonant frequency: 2 MHz) were mounted on the curvilinear curvilinear surface
surface of of the
thespecimen
specimen using
specimen using
epoxy agent (Figure
(Figure2).
epoxy agent (Figure 2). The
The specimen
specimen and and two
two end
end pieces
pieces were
were coated
coated with
with silicone
silicone sealant
sealant toto prevent
prevent
confining oil from
confining oil from leaking into
from leaking intothe
leaking into thespecimen.
the specimen.The
specimen. Thespecimens
The specimenscan
specimens canhandle
can handle
handle a total
aa total shear
total shear
shear displacement
displacement
displacement of
of
of approximately
approximately 2 2
mm,mm, which
which is is sufficient
sufficient for
for upup toto approximately
approximately 20
20
approximately 2 mm, which is sufficient for up to approximately 20 stick-slip events. Our new stick-slip
stick-slip events. Our
Our new
waveform
waveform recorder
recorder has 24
has 24 A/D
24 A/D channelswith
A/D channels
channels withaaasampling
with samplingrate
sampling rateof
rate ofofup
upup toto
to 100100
100 MHz,
MHz,
MHz, a dynamic
aa dynamic
dynamic range
range
range of
of
of
16 16 bits,
bits, andand
256 256
mega mega
wordswords
of of onboard
onboard memory
memory for for
each each channel.
channel. In the In the
present
16 bits, and 256 mega words of onboard memory for each channel. In the present study, the AE signal present
study, study,
the AE the AE
signal
signal
from fromsensor
from each
each each sensor
sensor is is pre-amplified
is pre-amplified
pre-amplified by
by 40 by and
40 dB
dB 40
anddB anddigitized
then
then then digitized
digitized at at a sampling
at aa sampling
sampling rate
rate of ofrate
25 of 25 and
25 MHz
MHz MHz
and aa
and a
sample sample
sample length length
length of
of 4096 of
4096 points4096 points
points (approximately(approximately
(approximately 160 160 microseconds),
160 microseconds),
microseconds), including
including 1600 including
1600 points
points of1600 points of
of pre-trigger
pre-trigger
pre-trigger
data.
data. data.

Figure
Figure2. 2.Schematic
Schematic of experimental configuration and distributions of
ofpiezoelectric transducers (solid
Figure 2. Schematicofofexperimental
experimental configuration andand
configuration distributions
distributionspiezoelectric transducers
of piezoelectric (solid
transducers
circles)
circles) and
and strain
strain gages
gages (crosses).
(crosses). The
The maximum
maximum amplitudes
amplitudes of
of AE
AE signals
signals from
from
(solid circles) and strain gages (crosses). The maximum amplitudes of AE signals from k1 and k1
k1 and
and k2
k2 were
were
were
recorded
recordedby
recorded by the
bythe peak
thepeak detectors
peakdetectors and
detectorsand used
andused for
usedfor the
forthe calculation
thecalculation of
calculationof seismic
ofseismic b-value.
seismicb-value.
b-value.

The
The maximum
The maximum amplitudes
amplitudesof
amplitudes ofAE
of AEsignals
AE signalsfrom
signals fromtwo
from two selected
two selected
selected sensors,
sensors,
sensors, which
which
which were mounted
were
were mounted
mounted on
on the
on
the
central
the
central surface
central surface
surface of
of the specimen
of the
the specimen
specimen (indicated
(indicated
(indicated as k1
as as and
k1k1 and
and k2
k2k2in
ininFigure
Figure2),
Figure 2),were
2), wererecorded
were recorded by
recorded by the
the peak
the peak
peak
detectors
detectors after
after aaa pre-amplifier
detectors after pre-amplifier of
pre-amplifier of 20
of 20 dB.
20 dB. The
dB. The noise
The noise level
noise level for
level for AE
for AE signals
AE signals is
signals is approximately
is approximately
approximately 40 40 dB
40 dB or
dB or less
or less
less
after
after applying
after applying
applying thethe 20
the 20 dB
20 dB pre-amplifiers.
pre-amplifiers. Thus, the threshold for the peak detectors was set at 45 dB,
dB pre-amplifiers. Thus, the threshold for the peak detectors was set at 45 dB,
yielding
yielding an
an effective
effective dynamic
dynamic range
range ofof 55
55 dB
dB (=100
(=100 −− 45
45 dB).
dB). Based
Based on on the
the definition
definition of
of the
the body-wave
body-wave
earthquake
earthquake magnitude,
magnitude, the the magnitude
magnitude (M) (M) of
of an
an AE
AE event
event isis determined
determined according
according toto the
the log
log of
of the
the
Appl. Sci. 2018, 8, 2361 4 of 10

yielding an effective dynamic range of 55 dB (=100 − 45 dB). Based on the definition of the body-wave
earthquake magnitude, the magnitude (M) of an AE event is determined according to the log of the
maximum amplitude (Amax ) of the AE signal (it is somewhat reflecting the vibration velocity of the
elastic wave [13]), as follows:
M = log( Amax /A0 ) + M0 (1)

where A0 is the threshold for detection, and M0 is a constant (approximately −10, in our system) for
converting the obtained value to a magnitude that is roughly comparable with the magnitude of an
earthquake. Normally, M0 = 0 is used, and Equation (1) yields a relative magnitude that can be used
for calculation of seismic b-values in the Gutenberg-Richter magnitude-frequency relation.

log N = a − bM (2)

where N is the number of events of magnitude M or greater, and a and b are constants. In order to
investigate the temporal evolution of the b-value, we calculated b-values using both the maximum
likelihood method for a running window of an equal number of events (n) with a running √ step of n/4
events (75% overlap). The approximate standard error of the b-value estimation is b/ N. A larger
window size results in smoother evolution of the b-value. In the present study, n = 1000 was used
for a long-term view, while n = 250 was used to explore b-value variation for each stick-slip event.
We performed this analysis using various values of n and found that n = 250 was a good choice for
exploring the details of b-value evolution with sufficiently small errors. Due to attenuation, some
small events could be recorded by only one peak detector close to the AE hypocenter. As a result,
the number of events counted by two detectors could be different. However, we confirmed that the
two datasets yielded consistent results for the temporal variation of the b-value [7].
Waveform data were recorded for the purposes of hypocenter determination and source
mechanism solution, which require clear first motion at more than four sensors. Hence, we designed a
complex triggering logic for the waveform recording system. First, three pairs of sensors were selected
as triggering sensors. Signals from the triggering sensors were fed to a trigger unit. An OR logic was
used within each pair, and an AND logic was used among the three pairs. In order to avoid biased
triggering, the three pairs were required to be homogeneously distributed on the specimen surface.
The triggering threshold level, which is normally approximately 10 times higher than that for the peak
detector, was determined. As a result, the AE hypocenter data constituted a subset of relatively large
events among the AE count data. The higher threshold and complex triggering configuration ensured
that AE events were collected without significant bias throughout the specimen space.
Axial load and strain data were digitized at a sampling rate up to 10 KHz. While confining
pressure, which was kept constant during the experiments, was recorded at 5 Hz. Both shear stress (τ)
and normal stress (σ) can be calculated from axial stress and confining pressure. Hence, an apparent
frictional coefficient (µ = τ/σ) could be estimated from the shear stress and normal stress. The obtained
coefficient was considered as either static or dynamic ones according to the slip velocity of the fault.
The axial shortening of the specimen was estimated from piston displacement and was calibrated
using the elastic portion of the mean axial strain. Since the results for the intact specimen were used
for the calculation of elastic shorting, it is possible to make rough fault slip in the pre-cut specimens.

4. Results
As in previous studies [13], the b-value, event rate, and strain data for the intact specimen (SA)
had three distinct phases: the primary, secondary, and nucleation phases (indicated by P, S, and N,
respectively in Figure 3). The b-values were estimated using the maximum likelihood method for a
running window of 1000 events with a step of 256 events. Acoustic emission rates were calculated for
every 10 s. Axial, circumferential, and volumetric strains were obtained as the average of data recorded
by six pairs of stain gauges, glued directly onto the surface of the specimen. The axial shortening of
the specimen was estimated from piston displacement and was calibrated using the elastic portion
Appl. Sci. 2018, 8, 2361 5 of 10

of the mean axial strain. Acoustic emission activity was initiated when the axial stress reached 75%
Appl. Sci. 2018, 8, x 5 of 10
(approximately 370 MPa) of the failure strength (495 MPa). The AE initiation also coincides with the
start of dilatation,
coincides with the as indicated
start by the as
of dilatation, mean volumetric
indicated by thestrain.
meanInvolumetric
the primary phase,
strain. the b-values
In the primary
increase
phase, the b-values increase with increasing stress from approximately 1.4 to 1.7 and thensecondary
with increasing stress from approximately 1.4 to 1.7 and then decrease during the decrease
phase
duringfrom the approximately
secondary phase 1.7 to
from Finally, the b-values
1.0. approximately 1.7 todrop
1.0.quickly
Finally,tothe
approximately
b-values drop0.5 quickly
during theto
nucleation
approximately phase 0.5and growthe
during quickly during
nucleation the final
phase and failure stage. during the final failure stage.
grow quickly

Figure 3.
Figure 3. Results
Results ofof the
the intact
intact specimen
specimen forfor reference.
reference. AE rates (red
AE rates (red line),
line), b-values
b-values (blue
(blue line),
line), axial
axial
stress (σ ), confining (Pc) pressure, axial/circumferential/volumetric strains (ε /ε
stress (σa ), confining (Pc ) pressure, axial/circumferential/volumetric strains (εa /εc /εv ), and axial
a a c /εv ), and axial
shortening are
shortening areplotted
plotted against
against time.
time. P, S,P,
andS, Nand N indicates
indicates three phases:
three distinct distinct the
phases:
primary, thesecondary,
primary,
secondary,
and nucleation and phases,
nucleation phases, respectively
respectively (see text for(see text for details).
details).

Overall,
Overall, the theresults
resultsforfor
thethepre-cut specimens
pre-cut specimens (S1 and
(S1 S2)
andwereS2) similar. Experiment
were similar. S1 were S1
Experiment stopped
were
after
stopped7 stick-slip events due
after 7 stick-slip to leakage
events due to of confining
leakage oil into the
of confining oil pre-cut
into thefault.pre-cutIn this
fault.paper,
In thiswe thus
paper,
mainly
we thusfocusmainly on focus
resultsonofresults
experiment S2, in which
of experiment S2, in18which
stick-slip events were
18 stick-slip eventsrecorded. As shown
were recorded. As
in Figure
shown 4, AE 4,
in Figure activity was initiated
AE activity at a at
was initiated very lowlow
a very differential
differential stress
stresslevel.
level.InInfact,
fact,some
some events
were observed during during the thehydrostatic
hydrostaticloadingloadingstage.stage.In In addition,
addition, thethe maximum
maximum axialaxial stress
stress waswas 210
210
MPa, MPa,
whichwhichwaswas less less
thanthan
60%60% of theof stress
the stresslevellevel (approximately
(approximately 370 MPa)
370 MPa) at which
at which AE activity
AE activity was
was initiated
initiated in the in intact
the intact specimen,
specimen, as described
as described above.
above. SinceSince
bothboththe the
shearshear
andand normal
normal stress
stress acting
acting on
on
the the fault
fault surface
surface could
could bebecalculated,
calculated,we wetherefore
thereforeestimated
estimatedthe the static
static and
and dynamic frictional
coefficients [12].
coefficients [12].During
Duringthe thedynamic
dynamicstress stress drop
drop associated
associated withwiththethe
firstfirst stick-slip
stick-slip event,event, the static
the static and
and dynamic
dynamic frictional
frictional coefficients
coefficients were were
0.75 and 0.750.5,
andrespectively
0.5, respectively
(Figure (Figure
4b). With4b). increasing
With increasing axial
axial stress,
stress,
the the seismic
seismic b-valueb-value
gradually gradually
decreased decreased from approximately
from approximately 1.8 to 1.01.8 priorto to
1.0the
prior
firsttostick-slip
the first event.
stick-
slipestimated
We event. Wethe estimated
slip alongthethe
slipfault
alongfromthe sample
fault from sample shortening
shortening and elastic and strainelastic
data.strain
Prior data.
to thePrior
first
to the firstthe
stick-slip, stick-slip,
fault had thestatically
fault hadorstatically or quasi-statically
quasi-statically slipped approximately
slipped approximately 0.6 mm (Figure 0.6 mm (Figure
4a). Since
4a). initiation
the Since the of initiation
the faultofslip
thecoincides
fault slip withcoincides with the of
the initiation initiation of the AE
the AE activity, and activity,
the AEand therate
event AE
event ratecontinuously
increases increases continuously
with increasing withfault
increasing
slip, wefault slip, we
concluded concluded
that that the
the AE activity and AEtheactivity and
pre-failure
the pre-failure
fault fault slipwith
slip were coupled wereeachcoupled
other.with each other.
Following the first stick-slip, 17 stick-slip stick-slip events were were observed
observed (Figure
(Figure 4b).4b). The experimental
experimental
results indicate that the static frictional coefficients decreased from the maximum value prior to the
stick-slipevent
first stick-slip eventto toaastable
stablemean
meanvalue
valueofof approximately
approximately 0.60.6 after
after several
several stick-slip
stick-slip events,
events, while
while the
the dynamic
dynamic frictional
frictional coefficient
coefficient fluctuated
fluctuated between between 0.450.52.
0.45 and and After
0.52. After each stick-slip
each stick-slip event,event, there
there was a
was a quiescent
quiescent periodperiodwith awith a relatively
relatively low AE low AE rate.
event eventWhen
rate. When
the shearthe stress
shear reached
stress reached 50 toof90%
50 to 90% the
of the strength,
shear shear strength,
the AEthe AE rate
event event rate increased
increased rapidlyrapidly and reached
and reached a maximuma maximum value immediately
value immediately before
before
the theofonset
onset of the
the next next dynamic
dynamic slip (Figure
slip (Figure 5). In the
5). In total, total, the number
number of AE of AE events
events per stick-slip
per stick-slip cycle
cycle exhibits
exhibits a tendency a tendency to gradually
to gradually decreasedecrease as theofnumber
as the number stick-slipsof events
stick-slips eventsThe
increases. increases.
mean valueThe
mean value of the dynamic fault slip is approximately 0.12 mm, while the pre-failure static slip is
approximately 0.04 mm.
As shown in Figure 5c, for stick-slip events, the b-value decreased from 1.2 to 1.5 to
approximately 0.6 as shear stress increased, and then quickly jumped back up to 1.0 to 1.3 prior to
Appl. Sci. 2018, 8, x 6 of 10

the dynamic stress drop. The time of the minimum b-value corresponds to a stress level of 70 to 95%
of the shear strength.
Appl. As an example,
Sci. 2018, 8, 2361 Figure 6 shows AE hypocenters projected to a vertical profile perpendicular 6 ofto
10
the fault plane for different stick-slip cycles. AE hypocenters were concentrated on the fault surface
and event far from the fault surface must be very limited. This result is in agreement with the
of the dynamic fault
aforementioned facts,slip
theismaximum
approximately
axial 0.12 mm,
stress waswhile the pre-failure
less than 60% of the static slip
stress is approximately
level at which AE
0.04 mm.
activity was initiated in the intact specimen.

Figure
Figure 4.4. Results
Results of
of experiment
experiment S2. S2. (a)
(a) Acoustic
Acoustic emission
emission counts
counts (every
(every 1010 s),
s), seismic
seismic b-value,
b-value, axial
axial
stress (σa ), shortening of specimen (d), and elastic shorting plotted with respect to time. (b)
stress (σ a ), shortening of specimen (d), and elastic shorting plotted with respect to time. (b) Detailed
Detailed
view
view ofof b-value
b-value and
and event
event rate
rate (every
(every second),
second), with
with additional
additional shear
shear stress
stress and
and friction
friction coefficient
coefficient for
for
time
time window
window as as shown
shown in in (a).
(a). The
The seismic
seismic b-value
b-value was
was estimated
estimated using
using aa running
running window
window with
with 1000
1000
or
or 250
250 event
event number.
number. The
The vertical
vertical dashed
dashed lines
lines in
in (b)
(b) indicate
indicate the
the onset
onset of
of stick-slip
stick-slip events.
events. The
The vertical
vertical
and horizontal bars at each b-value point indicate the standard error and time window,
and horizontal bars at each b-value point indicate the standard error and time window, respectively. respectively.

As shown in Figure 5c, for stick-slip events, the b-value decreased from 1.2 to 1.5 to approximately
0.6 as shear stress increased, and then quickly jumped back up to 1.0 to 1.3 prior to the dynamic stress
drop. The time of the minimum b-value corresponds to a stress level of 70 to 95% of the shear strength.
As an example, Figure 6 shows AE hypocenters projected to a vertical profile perpendicular to
the fault plane for different stick-slip cycles. AE hypocenters were concentrated on the fault surface
Appl. Sci. 2018, 8, 2361 7 of 10

and event far from the fault surface must be very limited. This result is in agreement with the
aforementioned facts, the maximum axial stress was less than 60% of the stress level at which AE
activity was initiated in the intact specimen.
Appl. Sci. 2018, 8, x 7 of 10
Appl. Sci. 2018, 8, x 7 of 10

Figure 5.
Figure 5. (a,b) Detailed view of acoustic emission (AE) rate, seismic b-value, shear stress, stress, and
and friction
friction
coefficient
coefficient for
for selected
selected stick-slip
stick-slip events
events in in experiment
experiment S2. (c) S2. (c)
b-value b-value
plotted plotted
with with
respect
Figure 5. (a,b) Detailed view of acoustic emission (AE) rate, seismic b-value, shear stress, and friction to respect
normalized to
normalized
shear stress shear
for all stress for
stick-slip all stick-slip
events. See events.
caption See
of caption
Figure 4 of
for Figure
details. 4 for details.
coefficient for selected stick-slip events in experiment S2. (c) b-value plotted with respect to
normalized shear stress for all stick-slip events. See caption of Figure 4 for details.

Figure 6. AE hypocenters projected to a vertical profile perpendicular to the fault plane for different
stick-slip
Figure 6.
Figure cycles
6. AE in experiment
AE hypocenters
hypocenters S1. Hypocenter
projected
projected to symbols
to aa vertical
vertical profile are
profile color with event
perpendicular
perpendicular thesequential
to the
to number.
fault plane
fault plane for different
for different
stick-slip cycles
stick-slip cycles in
in experiment
experiment S1.
S1. Hypocenter
Hypocenter symbols
symbols are
are color
color with
withevent
eventsequential
sequentialnumber.
number.
5. Discussion
5. Discussion
5.1. Seismic b-Value and AE Events in Pre-Cut Faults
5.1. Seismic b-Value
Previous andon
studies AE Events inoriented
favorably Pre-Cut (oriented
Faults at 30° with respect to the maximum stress axis)
faultsPrevious
[10] and studies
pure shear experiments
on favorably [11] showed
oriented thatatb-values
(oriented 30° withdecrease with
respect to theincreasing
maximumshear
stressstress
axis)
Appl. Sci. 2018, 8, 2361 8 of 10

5. Discussion

5.1. Seismic b-Value and AE Events in Pre-Cut Faults

Previous studies on favorably oriented (oriented at 30◦ with respect to the maximum stress axis)
faults [10] and pure shear experiments [11] showed that b-values decrease with increasing shear stress
until the stress drops. The results for unfavorably oriented (45◦ ) faults show a similar decreasing
b-value trend, followed by a quick jump prior to a stress drop for most stick-slip events. Note that
AEs were concentrated in a narrow zone from the fault surface. The authors also conducted similar
experiments using specimens having pre-cut faults at 50◦ . Detailed hypocenter data suggest that
there are two competing mechanisms that govern the evolution of the frictional properties and the
damage zone characteristics of such unfavorably oriented faults [12]. On the one hand, the fault plane
is smoothed by fault slippage because of asperities failing on the fault plane, and a fault gauge is
created. On the other hand, the fault plane is roughened by new damage. In particular, for a 50◦ fault,
a number of sub-faults were created along the favorable direction [12]. The quick jump in the b-value
corresponds to a decreasing event rate, or, in other words, a quiescence period preceding failure. Such
short-term quiescence basically agrees with the damage model for a fractal population of cracks [14].
The model predicts that higher b-values are associated with either lower stress intensity or greater
material heterogeneity.

5.2. Implication for Seismic Hazard

The results of the present study indicate that seismic b-values could be an indicator of fault
slip history and shear stress acted on the fault. The present study is thus helpful in seismic hazard
assessment and earthquake prediction, especially for injection-induced seismicity in fields having
unfavorably oriented faults, immature faults, or tensile fractures created by hydrofracturing. We must
pay more attention to such faults because shear fracturing contributes a high level of injection-induced
seismicity. Figure 7 shows the magnitude-frequency distributions and seismic b-values for earthquakes
induced by injection during shale gas fracking at two shale gas sites, Weiyuan and Shangluo, in Sichuan
China [2]. The Weiyuan site is located in the central uplift of the Sichuan basin, and has a lower
density of pre-existing faults. Induced seismicity is probably associated with immature fractures
created by hydraulic fracturing operations, and thus has a higher b-value (1.52). The Shangluo site,
which is located in the south fold zone of the Sichuan basin, has a high density of pre-existing faults
(most of which are unfavorable in the present-day stress field). Since the mean seismic b-value is 0.91,
pre-existing and well-developed faults play a significant role in injection-induced seismicity at the
Shangluo site. These interpretations can also explain the maximum earthquake magnitudes observed
thus far, which are approximately 3.5 and 5, at Weiyuan and Shangluo, respectively. At the shale
sites in the Sichuan basin, many well pads exhibited induced seismicity, with a sizable magnitude
of up to approximately 3.5. Events with a magnitude greater than 4 are limited to well pads close to
known or unknown faults. In order to produce an earthquake of magnitude 5, a fault having a size
greater than several kilometers is required [2,15], which is larger than fractures that can be created by
shale fracturing.
earthquake magnitudes observed thus far, which are approximately 3.5 and 5, at Weiyuan and
Shangluo, respectively. At the shale sites in the Sichuan basin, many well pads exhibited induced
seismicity, with a sizable magnitude of up to approximately 3.5. Events with a magnitude greater
than 4 are limited to well pads close to known or unknown faults. In order to produce an earthquake
of magnitude
Appl. 5, a fault having a size greater than several kilometers is required [2,15], which is larger
Sci. 2018, 8, 2361 9 of 10
than fractures that can be created by shale fracturing.

Figure 7.
Figure 7. Magnitude-frequency
Magnitude-frequency distributions
distributions and
and seismic
seismic b-values
b-values for
for induced
induced earthquakes
earthquakes at
at two
two
shale gas sites, Weiyuan and Shangluo, in Sichuan China. (The plot for Shangluo was modified
shale gas sites, Weiyuan and Shangluo, in Sichuan China. (The plot for Shangluo was modified from from
Lei et
Lei et al.
al. 2017
2017 [2].)
[2].) In
In the figure, Mc
the figure, Mc indicates
indicates the
the lower
lower cut
cut magnitude
magnitude for
for completeness,
completeness, and
and SESE
indicates the
indicates the standard
standard error
error of
of the
theb-value.
b-value.

6. Conclusions
In conclusion, the authors investigated the seismic b-value for AE events during repeated stick-slip
cycles of pre-cut faults in granite under triaxial compression conditions (confining pressure: 40 MPa).
The pre-cut fault had an angle of 45◦ with the axial stress and thus is unfavorably oriented. Using
a multi-channel AE waveform recording system, AE waveforms were recorded at 16 bits and at a
sampling rate of 25 MHz. At the same time, two peak detectors were used to record the maximum
amplitude of AE events. Acoustic emission amplitude data, ranging over three orders of magnitude,
show the frequency-magnitude relation of Gutenberg and Richter.
At first, the b-value, event rate, and surface strain for the intact specimen demonstrate three
distinct phases: the primary, secondary, and nucleation phases. This pattern is similar to that observed
for other fine-grained granites, such as Westerly granite and Oshima granite [7].
The results of the present study indicate that the static frictional coefficient decreased from
the maximum value prior to the first stick-slip event to a stable mean value of approximately 0.6
after several stick-slip events, while the dynamic frictional coefficient fluctuated between 0.45 and
0.52. The pre-failure static slip accounts for approximately 30% of the dynamic slip. AEs during the
pre-failure static slip stage were concentrated on the fault surface.
Prior to the first stick-slip event, the seismic b-value decreased gradually from approximately
1.8 to 1.0 with a fault slip of approximately 0.6 mm. For later stick-slip events, the seismic b-value
decreased from 1.2 to 1.5 to approximately 0.6 as the shear stress increased, and then quickly jumped
back up to 1.0 to 1.3 before the stress decreased. The time of the minimum b-value, which coincided
with the maximum event rate, corresponded to a stress level of 70–95% of the shear strength.
Our results on b-value evaluation preceding repeated stick-slips demonstrate that the seismic
b-value can be used as an indicator of fault slip history or the degree of fault maturity and shear
stress acting on the fault, which is important in seismic hazard assessment and earthquake prediction,
especially for injection-induced seismicity in fields in which reactivated shear rupture of unfavorable
and immature faults or tensile fractures is important. For individual earthquakes, there are no
significant differences, in a wide aspect of earthquake seismology, between induced-earthquakes
by fluid injection and earthquakes related to tectonic activity [2]. A combination of multi pieces of
statistical behaviors is required making reliable discrimination. This study indicates b-value evaluation
is an important parameter.

Author Contributions: All authors designed the project and participated in the experiments and the interpretation
of the results. X.L. processed and analyzed the data and wrote the manuscript.
Appl. Sci. 2018, 8, 2361 10 of 10

Acknowledgments: The present study was partly supported by the National Science Foundation of China
(NSFC 41174046).
Conflicts of Interest: The authors wish to confirm that there are no known conflicts of interest associated with
this publication and there has been no significant financial support for this work that could have influenced
its outcome.

References
1. Grigoli, F.; Cesca, S.; Rinaldi, A.; Manconi, A.; López-Comino, J.; Clinton, J.; Westaway, R.; Cauzzi, C.;
Dahm, T.; Wiemer, S. The November 2017 Mw 5.5 Pohang earthquake: A possible case of induced seismicity
in South Korea. Science 2018, 360, 1003–1006. [CrossRef] [PubMed]
2. Lei, X.; Huang, D.; Su, J.; Jiang, G.; Wang, X.; Wang, H.; Guo, X.; Fu, H. Fault reactivation and earthquakes
with magnitudes of up to Mw4.7 induced by shale-gas hydraulic fracturing in Sichuan Basin, China. Sci. Rep.
2017, 7, 7971. [CrossRef] [PubMed]
3. Terakawa, T.; Hashimoto, C.; Matsu’ura, M. Changes in seismic activity following the 2011 Tohoku-oki
earthquake: Effects of pore fluid pressure. Earth Planet. Sci. Lett. 2013, 365, 17–24. [CrossRef]
4. Sibson, R.H.; Robert, F.; Poulsen, K.H. High-angle reverse faults, fluid-pressure cycling, and mesothermal
gold-quartz deposits. Geology 1988, 16, 551. [CrossRef]
5. Terakawa, T.; Zoporowski, A.; Galvan, B.; Miller, S.A. High-pressure fluid at hypocentral depths in the
L’Aquila region inferred from earthquake focal mechanisms. Geology 2010, 38, 995–998. [CrossRef]
6. Sibson, R.H. Brittle-failure controls on maximum sustainable overpressure in different tectonic regimes.
AAPG Bull. 2003, 87, 901–908. [CrossRef]
7. Lei, X.; Ma, S. Laboratory acoustic emission study for earthquake generation process. Earthq. Sci. 2014, 27,
627–646. [CrossRef]
8. Smith, W.D. The b-value as an earthquake precursor. Nature 1981, 289, 136–139. [CrossRef]
9. Sammonds, P.R.; Meredith, P.G.; Main, I.G. Role of pore fluids in the generation of seismic precursors to
shear fracture. Nature 1992, 359, 228–230. [CrossRef]
10. Goebel, T.; Schorlemmer, D.; Becker, T.; Dresen, G.; Sammis, C. Acoustic emissions document stress changes
over many seismic cycles in stick-slip experiments. Geophys. Res. Lett. 2013, 40, 2049–2054. [CrossRef]
11. Rivière, J.; Lv, Z.; Johnson, P.; Marone, C. Evolution of b-value during the seismic cycle: Insights from
laboratory experiments on simulated faults. Earth Planet. Sci. Lett. 2018, 482, 407–413. [CrossRef]
12. Lei, X.; Li, S.; Liu, L. Fracturing behaviors of unfavorably oriented faults investigated using an acoustic
emission monitor. In Advances in Acoustic Emission Technology; Springer: Basel, Switzerland, 2017; pp. 229–240.
13. Lei, X. How do asperities fracture? An experimental study of unbroken asperities. Earth Planet. Sci. Lett.
2003, 213, 347–359. [CrossRef]
14. Main, I.G.; Meredith, P.G.; Sammonds, P.R. Temporal variations in seismic event rate and b-values from
stress corrosion constitutive laws. Tectonophysics 1992, 211, 233–246. [CrossRef]
15. Wells, D.L.; Coppersmith, K.J. New empirical relationships among magnitude, rupture length, rupture
width, rupture area, and surface displacement. Bull. Seismol. Soc. Am. 1994, 84, 974–1002.

© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).