Upload
others
View
8
Download
0
Embed Size (px)
Citation preview
8th Asian Rock Mechanics Symposium ARMS8 14-16 October 2014, Sapporo, Japan
Insights on Injection-induced Seismicity Gained from Laboratory AE Study—Fracture Behavior of Sedimentary Rocks
X.-L. Leia*, X.-Y. Lib, and Q. Lib
a Geological Survey of Japan, AIST, Ibaraki, 3058567, Japan
bState Key Laboratory of Geomechanics and Geotechnical Engineering, Institute of Rock and Soil Mechanics, Chinese Academy of Science, Wuhan, 43071, China
Abstract
Injection-induced seismicity associated with applications, in which fluids are intensively pressed into the deep formations of the Earth's crust such as Enhanced Geothermal System (EGS), fracking shale gas, geological sequence of CO2, have attracted growing attentions. Motivated by the desire to better understand the mechanism of damaging events so that they can be avoided or mitigated, we have started an integrated study on rock fracturing and fault reactivation in multiscales. In this paper, we present some preliminary results of an ongoing experimental study utilizing acoustic emission technique in laboratory scale. We systematically carried out rock fracture tests using samples of typical sedimentary rocks collected from the Sichuan Basin, China, where a number of injection-induced seismic swarms with sizable earthquakes ranging up to M4~5 have been observed in some gas/oil reservoirs. Since most injection-induced earthquakes are located in sedimentary strata of a wide range of lithology and depth, the fracturing behaviors of such rocks are thus important. Our results indicate that the Pre-Triassic rocks in the Sichuan Basin, including dolomite or dolomitic limestone and shale are strong and demonstrate brittle fracturing behaviors. Such properties are necessary conditions for maintaining high level reservoir stress and resulting seismic fracturing. Keywords: Acoustic Emission, Rock Fracturing, Sedimentary rocks, Injection-Induced Seismicity 1. Introduction
In applications including Enhanced Geothermal System (EGS), fracking shale gas, and geological sequence of CO2, fluids of various viscosity and temperature are intensively pumped under high pressure into the deep formations of the crust. In recent years, associated with the rapid increase of such applications, injection-induced seismicity has attracted growing public and scientific attentions. Earthquakes of moderate sizes (~M5), which are damaging in areas having not experienced with large earthquakes in the history, were occasionally observed. Injection induced seismicity is not required but unavoidable in many cases (Technologies et al., 2012). The impact of induced earthquakes is a controversial issue that has caused delays and threatened cancellation of some projects, such as the recent project at Basel (Kraft et al., 2009). It is thus a scientific challenge to make clear the mechanism and geomechanical conditions of damaging events so that they can either be avoided or be mitigated.
The cases, especially recent ones, of injection-induced earthquake sequences in the southwestern Sichuan Basin, which is a relatively stable region in China showing very low level of background seismicity, provide us a chance of comprehensive studies on which. During the past three decades, a number of seismic sequences have been observed with sizable earthquakes ranging up to M4~5. Their timing, location and occurrence pattern involved in statistical models, convincingly suggest that these sequences were induced by injections of unwanted water in deep wells in depleted gas fields. All sequences were initiated in days to a few weeks after water injection under high pressure begun. Event rate fluctuates following changes of injection rate and injection frequency and tapered quickly after shut down. Detailed studies on some recent and well monitored cases demonstrate that most events, particularly larger ones, mirror the reactivation of pre-existing faults (including joints and fractures) in the sedimentary formations or faults underlying/overlying the reservoir (Lei et al., 2013b; Wang et al., 2012). As a typical example, Fig.1 shows hypocenter distribution of injection-induced earthquakes superimposed on the simplified geological cross section and stratigraphy of a gas
Insights on Injection-induced Seismicity Gained from Laboratory AE Study—Fracture Behavior of Sedimentary RocksX. Lei, X. Li, Q. Li 947
Insights on Injection-induced Seismicity Gained from Laboratory AE Study—Fracture Behavior of Sedimentary RocksX. Lei, X. Li, Q. Li
8th Asian R reservoirwere degas reseand playclouds, migratiofaults oflargest edepletedbehaviorinjectionintegrate
Herefrom typwere perfor many
Fig.
geologlines in
2. Expe2.1 Rock
In compresthe Sichformatioof Dengare founare one o
The twere dria constafracture confininthese co0~30° w
Rock Mechanic
r in Zigong termined wirvoir. Shale
y a role in arseismic acti
ons. Two M4f relative higevent (M4.4)d reservoir. Itrs under hign-induced eaed research ie, we report spical formatirformed at ay injection ap
1 Hypocentegical cross sen the inlet m
eriments k sample order to in
ssion conditihuan Basin. on (J2x); 2) Lgying formatind in formatiof the major test samples ied under noant stress rat
strength atng pressure wonditions, thewith respect to
cs Symposium
city, locatedithin the Permand mudston
rresting fractivity is clea
4+ events shogh dipping a) was probabt is know thagh pressure.arthquakes inncluding labsome prelimions of gas re
a confining ppplications s
er distributionection and str
map) in the so
nvestigating ions, we coll
The test roLower Triassion (Zd); andions of sandscap rocks. Twere shaped
rmal room cte or at a cot room tempwas maintaine rock samplo the axis wh
d in southwemian dolomne in the ovetures in the darly boundedow a revise fangle, whichbly initiatedat typical mu. It is thus n the Sichuanboratory studyinary results eservoirs in tpressure of 22such as geolo
n of injectionratigraphy ofouthwestern S
the fracturlected variouocks in this ic limestone d 4) Precambstone of very
Tab. 1 lists md into cylindonditions foronstant stresperature (air
ned at a consles are normahich is the di
est of the Sicmitic limeston
erlying and udolomitic limd by dippinfaulting mechh are known
in the Pre-Cudstone, shal
an interestian Basin. Mody on the frac
from triaxiathe Sichuan B2.5 MPa, whogical storag
n-induced eaf the HuangjiSichuan Bas
re behavior us kinds of r
study are of Feixiangubrian shale oy low perme
major mechanders of 125 mr more than
ss (creep tesr-conditione
stant level ofally fractureirection of th
chuan Basinne dominatedunderlying st
mestone formng faults leahanism indic
n from petroCambrian dole, and limesting issue tootivated by scture behavioal compressioBasin. Rock hich correspoe of CO2.
arthquakes suiachang gas rin, China (M
of typical rock blocks f1) Middle Juan formatio
of Zd. In the eability, limenical and phymm length aone month a
st) at a stresd at ~25°Cf 22.5 MPa (d along a co
he maximum
14-16 Octo
. More than d strata surrotrata act as f
mation. At theading to upwcating reactivo-geophysicaolomite formtone pronoun
make clearsuch purposeors of typicalon tests of rofracture testsonds to a dep
uperimposedreservoir (ind
Modified from
reservoir rfrom a field Jurassic sandon (T1f); 3) PSichuan Bas
estone, and dysical propertnd 50 mm d
and then theys level of ~). During th
(or 10 MPa iompression sh
compressive
tober 2014, Sap
90% of hypounding the fluid diffusioe front of hyward and dovation of preal investigatimation undernce ductile fr the mechae, we have sl reservoir roock samples s under dry cpth of ~1 km
d on the simpdicated by th
m Lei et al., 2
rocks undergeological s
dstone of XPrecambrian sin, typical rdolomite. Shrties of these diameter. Ally were comp
~95% of the the deformain some caseshear fault ore stress.
ARMS8 pporo, Japan
pocenters depleted
on barrier ypocenter ownward e-existing ons. The lying the
fracturing anism of tarted an
ocks. collected condition
m, typical
plified he cross 2013b).
r triaxial survey in
Xiaximiao dolomite
reservoirs ale strata rocks .
l samples pressed at
nominal tion, the e). Under riented at
948
Insights on Injection-induced Seismicity Gained from Laboratory AE Study—Fracture Behavior of Sedimentary RocksX. Lei, X. Li, Q. Li
8th Asian R
SaLiDlim
2.2 Exp
The lby our g(Lei et awith a swords (Mresonantmonitoribefore fsamplingcross-typare digitused to c40 dB pselected
The mod, anevent armost 10during ewe norm~5000 eOnce th
Rock Mechanic
Table 1 B
Rock
andstone imestone olomitic mestone Shale
Fig.2 Blosyst
erimental seloading syste
group (Lei etal., 2000) is sampling rateMW) capacit frequency ing and velofeeding into g rate of 25 pe strain gautized with vacapture the vre-amplifierssensor, in tonew wavefo
nd continuoure directly tr0 events per earlier stagesmally switchevents per sehe onboard
cs Symposium
Basic inform
Geology time J2x T1f Zd
Zd
ock diagram otems used fo
etup and prem and AE mt al., 2000; Lrecently renee up to 100 ity. Twenty-of 2 MHz w
ocity measurthe waveforMHz and a uges were marying sampvalues of the s. An automaotal 18, to theorm recordins mode. In t
ransformed tosecond unti
s of loading, h the system econd since
memory is
mation and ke
Vp (Km/s)
Po
3.98 5.81 6.5
5.66
of the renewr rock fractu
ocedure monitoring s
Lei et al., 201ewed by a neMHz, a dyn
-four piezoelwere mounterement. The rm recordisample lengt
mounted to thpling interval
maximum aatic switchine pulse gener
ng system hahe single evo the hard dil the hard din which AEto the mult
digitized wa full (6553
ey results of t
orosity Pe
6% <- -
-
wed experimeure test and a
system are b11). The 32-cew system (Fnamic rangelectric transded to the cusignal from
ing system. th of 4096 whe surface ol between 0.amplitudes, fng sub-systemrator of 20Vas three wor
vent mode, thdisk of the hdisk array is E rate is lessti-event modaveform data36 events un
the test samp
ermeability
<0.1 mD - -
-
ental apparatucoustic emis
basically simchannel rapidFig.2). The n
e of 16 bit, aducers (PZTurvilinear su
every AE sIn this study
words (~ 160f the sample001 and 1 s
from 2 artificm was used fpp volt outpu
rking modeshe digitized whost PC. As a
full. Thus ts than 10 ev
de, under wha are directlynder aforem
14-16 Octo
ples used in t
Pc (MPa) 22.5 22.5 10
22.5
us and data asion monitor
ilar to that ud AE wavefonew system and onboard ) of 5 mm irface of the
sensor is prey, AE signamicrosecon
e for strain mecond. Two
cially selectefor sequentiaut for velocit: single-evenwaveform daa result the sthe single-events per seco
hich the systy stored in th
mentioned sa
tober 2014, Sap
this paper.
Strength(MPa)
225 249 321
239
acquisition ring.
used in earlieorm recordinhas 24 A/D
d buffer of 2in diameter
e test samplee-amplified bal is digitizends). Furthermmeasurement peak detect
ed sensors, afally connectity measuremnt mode, muata of a triggsystem can
vent mode isond. For latetem can recohe onboard ampling con
ARMS8 pporo, Japan
h
er studies ng system
channels 256 mega having a
e for AE by 40 dB d with a more, six ts, which tors were fter 20 or ing every
ment. ulti-event gered AE record at s applied er stages, ord up to memory. nditions),
949
Insights on Injection-induced Seismicity Gained from Laboratory AE Study—Fracture Behavior of Sedimentary RocksX. Lei, X. Li, Q. Li
8th Asian R waveforthe onbosoftware
Fig.3fracture and hydr
Afterof the Cin the ssample s
3. ExpeFigs.
summeryforeshoc
The An accenumber while afcreated coefficiefaults in
The lslowly fsome sustructurecomplicobservedrecordindistributwithin th
The comparethe peaksignificashort timevents wsample i(Fig.6).
The sCT scansample aby one o
Rock Mechanic
rm data are aoard data at e-TSpro (Fig3 shows typic
tests performro-fracturingr the experimT image is 0ample. We sine we can e
Fig.3 Schemstick-slip
a
erimental re 4-7 show ke
ry. For conveck and afterssandstone selerated incrof aftershock
ftershocks cofault is com
ent of ~1.1, n the crust (Filimestone safrom the peaub events shoe in the limeated fault ged. During th
ng system, ated on the fhe sample (Fdolomite (ded with the lk value of 32ant foreshockme period (~were observeindicate a ve
shale samplen images (Figand thus is oof the beddin
cs Symposium
automaticallyany time thr
g.2). cal loading pmed under dg tests will bements, 3D X-0.2 mm/pixelcan relate theasily identif
matic diagramp or stable slaxial stress w
esults ey results of enience, we shock, respec
sample showrease of AE ks are observoncentrated mplicated bywhich is sigig.4). ample demonak value of 2owing rapid estone formaeometry whi
he fracturing among whicfault plane, bFig.5). dolomitic limlimestone sa21 MPa to 24ks occurred
~3s) after theed during theery complica
e contains stg.7). The bedoptimally oring structures
y transformedrough the in
paths used indry conditione reported in-ray CT scanl, sufficient fhe direction fy the beddin
m showing lliding. In caswhile case B)
rock fracturesimply term
ctively. ws very large
activity is ved. Foreshoalong the finy bends andgnificantly hi
nstrates duct249 MPa to stress drop aation, has a ich can be idstage, a few
ch a few tebut some ev
mestone) samample. The a40 MPa, andbefore the p
e peak stress rapid stress
ated fracture
trong beddindding plane hiented for frs. The fractu
d to the hostnteractive use
n the experimn (Fig.3 A). n later papersnning is carrifor examinin
of formationg planes of
loading pathsse A), the tes) is designed
e test of the fthe AE even
e volumetric observed im
ocks are randnal fault plad rough surigher than th
tile fracturin~50 MPa. H
and thus coudominant ro
dentified fromw hundreds oens event covents are cle
mple demonaxial stress dd thus demonpeak stress. A
and before dropping pegeometry ch
ng structureshas an angle
racturing. Thuring process
t PC. We caner interface o
ment. In this sResults of fr
s. ied out for alg the mesosc
on and loadithese rocks.
s and possiblst rock is fauld for hydro-fr
four rocks lisnts before an
strain indicmmediate befdomly locatedane. X-Ray Crface. The phe normal ra
ng behavior iHowever, theuld be seismiole on the frm the X-ray
of AE eventsould be precearly correla
nstrates quit dropped quicnstrating moAbout 2000 the rapid str
eriod. The 3Dharacterized
which are ce of ~25 degrhe finally cres is very bri
14-16 Octo
n also force tof the wavef
study, we focfracture tests
ll broken samcale structureng axis for
le post failurelted due to inracturing test
sted in Tab. nd after the d
ating signififore the pead in the centrCT image shpost slip datange (0.6~0.8
in total. The e fracturing pc. Stylolite, racturing proy CT image. were recordcisely locateated with the
different frackly (within re brittle fraAE events w
ress drop begD X-ray CT i
by bends an
clearly reconree with respeated fracturettle, and the
tober 2014, Sap
the system toform data ac
cus only on runder wet c
mples. The rees and fault gshale and s
re processes, ncreasing t.
1. Follows adynamic frac
icant compreak stress andral part of thhows that thta show a f8) of well-d
e axial stress process also which is the
ocess leadingNo foresho
ded by the wed. Most eve Stylolite s
acturing behaa few secon
acturing behawere recordegun. More thimages of thnd complex
nstructed in tpect to the axe is clearly g
e axial stress
ARMS8 pporo, Japan
o transfer cquisition
results of condition
esolution geometry sandstone
re a brief cturing as
essibility. d a large he sample he finally frictional eveloped
dropped contains
e weakest g to very cks were
waveform vents are structures
aviors as nds) from avior. No ed in the han 1000 he broken branches
the X-ray xis of the governed dropped
950
Insights on Injection-induced Seismicity Gained from Laboratory AE Study—Fracture Behavior of Sedimentary RocksX. Lei, X. Li, Q. Li
8th Asian R very quipeak strThe fauencountewere ob
Fig di
fria
of
Rock Mechanic
ickly from thress. More thult surface iered the harserved (Fig.7
g.4 Basic ressplacement, ictional coefand post rupf the fracture
Fig.5 Baaxial dbetwee
cs Symposium
he peak (239han 1000 aftis quite simd end pieces7-P2).
sults of triaxistrains, AE n
fficient actedture are plott
ed sample and
asic results oisplacement,
en AE hypoc
9 MPa) to ~tershocks occ
mple as coms at both sid
ial compressinumber coun
d on the final ted against tid AE hypoce
of triaxial com, strains, andenters and X
~100 MPa. Ncurred imme
mpared with des, additiona
ion test of santed by wave fracture planime. A compenters during
mpression ted AE numberX-ray CT scan
No significanediately follo
that of all al AE event
andstone sameform recordne during theparison betwg different sta
est of limestors against timn image of th
14-16 Octo
nt foreshocksowing the str
other samps associated
mple (J2x). Axding system, e periods of feen X-ray CTages (F, P) a
one show axime and a comhe fractured
tober 2014, Sap
s occurred btress drop (Fples. Since
with over d
xial stress, axshear stress, fault formatiT scan imag
are also plotte
ial stress,
mparison sample.
ARMS8 pporo, Japan
efore the Fig.7-P1). the fault
damaging
xial and
ion e ed.
951
Insights on Injection-induced Seismicity Gained from Laboratory AE Study—Fracture Behavior of Sedimentary RocksX. Lei, X. Li, Q. Li
8th Asian R
Rock Mechanic
Fig.6 Basicstrains, an
Fig.7 Basaxial dbetwee
cs Symposium
c results of trnd AE numbe
and X-
sic results of isplacement,
en AE hypoc
iaxial comprers against ti-ray CT scan
triaxial com, strains, andenters and X
ression test oime and a con image of th
mpression testd AE numberX-ray CT scan
of dolomite smparison be
he fractured s
t of shale samrs against timn image of th
14-16 Octo
ample show tween AE hy
sample.
mple show axme and a comhe fractured
tober 2014, Sap
axial stress, ypocenters
xial stress, mparison
sample.
ARMS8 pporo, Japan
952
Insights on Injection-induced Seismicity Gained from Laboratory AE Study—Fracture Behavior of Sedimentary RocksX. Lei, X. Li, Q. Li
8th Asian Rock Mechanics Symposium ARMS8 14-16 October 2014, Sapporo, Japan 4. Discussion and conclusions
Our results indicate that the Pre-Cambrian formations in the Sichuan Basin, including dolomite or dolomitic limestone, and shale show brittle fracturing behaviors, characterized by no significant foreshocks, rapid dynamic fracturing with large stress drop, and a large number of aftershocks. The well cemented sandstone, having a porosity of ~6% and an extremely low permeability (less than 0.1 mD), also shows brittle fracturing behaviors. Unlike dolomite and brittle shale, sandstone shows increasing foreshock activity preceding the final failure.
The finally created faults in dolomite, limestone, and sandstone samples demonstrate complicated geometry, showing bends, irregular and rough fault surface, and branches. Fault in the shale sample (notes the bedding plan is optimally oriented for fracture) shows relatively simple fault plane. It is worthy to note that sample of badly oriented bedding planes normally demonstrate very irregular shear fault patterns (Lei et al., 2013a). New fault in brittle rocks shows relatively high frictional coefficient (1.0~1.2), expected to decrease to the typical range (0.6~0.8) after certain amount of slip and worth further study.
Our experimental results are helpful for understanding the question has raised in the introduction section—why injection-induced seismicity is so significant in the Sichuan Basin. Major Pre-Triassic sedimentary rocks, including dolomite, shale, and dolomitic limestone are strong and demonstrating brittle fracturing behaviors. Such properties are necessary conditions for maintaining high level reservoir stress and leading to seismic fracturing. Insights gained from this study may shed some lights to the general earthquake seismology and provide a better understanding of why damaging injection-induced earthquakes occur so that they can either be avoided or be mitigated. In general, critically or sub-critically stressed fault of a dimension of a few kms is a necessary condition for resulting M~5 class earthquakes. On the same time, AE, or in other words, micro-seismicity monitoring is also useful in risk assessment and injection management and should be fully utilized in injection applications.
Acknowledgements
We thank Yingxiang Cui, Wenbin Fei, and Miao Jing for their help in the field work for collecting rock samples. References Kraft, T., et al. 2009, Enhanced geothermal systems: Mitigating risk in urban areas, Eos Trans. AGU,
90(32), 273, doi:10.1029/2009EO320001. Lei, X., Kusunose, K., Rao, M.V.M.S., Nishizawa, O., Satoh, T., 2000, Quasi-static fault growth and
cracking in homogeneous brittle rock under triaxial compression using acoustic emission monitoring, Journal of Geophysical Research, 105, 6127.
Lei, X., Funatsu, T., Villaescusa, E., 2013a, Fault formation in foliated rock – insights gained from a laboratory study, In Proc 8th Int Symp on Rockbursts and Seismicity in Mines (RaSim8) (ed. Malovichko A, Malovichko D), Russia Saint-Peterburg-Moscow, 41-49.
Lei, X., Ma, S., Chen, W., Pang, C., Zeng, J., Jiang, B., 2013b. A detailed view of the injection-induced seismicity in a natural gas reservoir in Zigong, southwestern Sichuan Basin, China. Journal of Geophysical Research: Solid Earth 118, 4296-4311.
Lei, X., Tamagawa, T., Tezuka, K., Takahashi, M., 2011. Role of drainage conditions in deformation and fracture of porous rocks under triaxial compression in the laboratory, Geophysical Research Letters, 38, L24310, doi:10.1029/2011GL049888.
Technologies, C.o.I.S.P.i.E., Resources, C.o.E., Engineering, C.o.G.a.G., Geodynamics, C.o.S.a., (BESR), B.o.E.S.a.R., (DELS), D.o.E.a.L.S., Council, N.R., 2012, Induced Seismicity Potential in Energy Technologies,The National Academies Press.
Wang, X., Ma, S., et al., 2012, Fine velocity structure and relocation of the 2010 Ml5.1 earthquake sequence in the Rongchang gas field, Earthquake Research in China, 20, 467-478.
953