ebook img

Acoustic emission characteristics in hydraulic fracturing of stratified rocks: A laboratory study PDF

10 Pages·2020·1.947 MB·English
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Acoustic emission characteristics in hydraulic fracturing of stratified rocks: A laboratory study

PowderTechnology371(2020)267–276 ContentslistsavailableatScienceDirect Powder Technology journal homepage: www.elsevier.com/locate/powtec Acoustic emission characteristics in hydraulic fracturing of stratified rocks: A laboratory study ZhizhongJianga,b,QuanguiLia,b,c,⁎,QiantingHua,b,YunpeiLianga,b,YangchengXua,b,LeLiua,b,XiaobingWua,b, XuelongLia,b,c,XiaoguangWanga,b,LiangpingHua,b,FapingLinga,b aStateKeyLaboratoryofCoalMineDisasterDynamicsandControl,ChongqingUniversity,Chongqing400044,China bSchoolofResourcesandSafetyEngineering,ChongqingUniversity,Chongqing400044,China cStateKeyLaboratoryCultivationBaseforGasGeologyandGasControl(HenanPolytechnicUniversity),China a r t i c l e i n f o a b s t r a c t Articlehistory: Acousticemission(AE)isapopulartechniquetomonitortheprocessofrockfailureduringhydraulicfracturing Received30November2019 forunconventionalnaturalgasproduction.Itcontainsabundantinformationthatwillbeusefultostudyin-depth Receivedinrevisedform1April2020 thenatureofhydraulicfracturing.Inthisstudy,wefocusedontheAEcount,energy,peakfrequency,crackclas- Accepted14May2020 sification,andlocationrecordedfromfourrockspecimenssubjectedtoaspecifictriaxialstresscondition.We Availableonline20May2020 foundthemulti-frequency-responsephenomenonofAE,andproposedthemulti-frequency-responseindexto indicatethemomentofthemacrohydrauliccrackformation.Furthermore,itwasfoundthatthepowerlawdis- Keywords: tributionindexofAEenergyofnon-stratifiedspecimenwasbiggerthanthatofstratifiedspecimensduringhy- Acousticemission Hydraulicfracturing draulicfracturing.Thetensilecrackdominatedinallhydraulicfracturingtests.Ourresultsareofsignificance Coalbedmethane forunderstandinghydraulicfracturinginstratifiedrocks. Peakfrequency ©2020ElsevierB.V.Allrightsreserved. 1.Introduction AE characteristics,such ascount,energy, frequency,location, havebeenextensivelystudied,amongwhichtheAEcountfeature Stratifiedrocksarewidelydistributedinunconventionalnaturalgas isthemoststudied.Ingeneral,themoretheAEcountthegreater productionsources,suchasshalesandcoalbeds[1–3].Inordertoen- theenergyofAEevents,andthelargerthescaleofcoal/rockfrac- hancetheirpermeability,andthusincreasegasproduction,manytech- tures.AEcountcharacteristicsdemonstrategoodKaisereffectina niquesareemployed,amongwhichhydraulicfracturingiscommonly homogeneousmedium[14],andFelicityeffectinaheterogeneous applied[4–8].Hydraulicfracturinginducesfailureandfractureofrock medium[25].AEcountoftengoesthroughseveral“suddenincrease mass,whichresultsinacousticemission(AE)[9–13].AE,whichhas andcalm”stagesduringloadingtestsofcoalandrock(heteroge- beenasubjectofresearchonrockfailuresincethe1950s[14],isbasi- neousmedium)[22].Hydraulicfracturingofcoalandrockisame- callyanelasticwave,whichissimilartomicroseismingeophysics.AE chanicalprocess.AEcountpresentsanobviousincreasewhencoal eventsareassociatedwiththeinitiationandpropagationofhydraulic orrockisdestroyedbywaterinjection[12].Inaddition,inhighstress cracks[15,16].Globally,scientistsandengineersaretryingtounder- regions,AEcountinducedbyfailureofcoalorrockmassismorethan standtheintrinsicmechanismofrockmassfracturefromAEcharacter- thatinlowstressregions[15,26–28].Ishidaetal.[9]foundthatthe istics,aswellastorecognizetheseAEcharacteristicsduringhydraulic AEcountincreaseddramaticallyandthecracksexpandedrapidly fracturing,asrockswithdifferentphysicalproperties[17–20]orunder with the sudden drop in pressure after the injection of high- differentstressconditions[21,22]willproduceAEwaveswithvarying pressurewaterintogranite.Lietal.[29]reportedthattheAEcount characteristicsduringthedamageprocess.Furthermore,theAEcharac- duringthestageofbreakdownwasfarmorethanthatduringother teristicsvaryduringdifferentstagesofhydraulicfracturing[15].There- stages. This phenomenon was also reported in other studies fore, by studying AE characteristics, the failure law of hydraulic [15,30–32].AEwaveformcharacteristicsarestudiedtoidentifyfail- fracturingcanbeindirectlyrevealed,withsufficientinvestigationof uremodes,suchastensionfailure,shearfailure,ortheircombina- AEcharacteristicsinthelaboratoryexpectedtoimprovethemicroseis- tion. There are two methods to identify failure modes: moment micmonitoringofhydraulicfracturingduringfieldapplications[23,24]. tensoranalysis[33]andanalyzingtherelationshipbetweenRAand AF,whichisobtainedfromAEdata[34,35].AEfrequencyisanimpor- tantparameterofAEsignal,whichcanindirectlyreflectthematerial ⁎ Correspondingauthorat:StateKeyLaboratoryofCoalMineDisasterDynamicsand properties.Whilestudyingcompositefiberdamage,AEfrequency Control,ChongqingUniversity,Chongqing400044,China. E-mailaddress:[email protected](Q.Li). presents good zoning characteristics with different materials https://doi.org/10.1016/j.powtec.2020.05.050 0032-5910/©2020ElsevierB.V.Allrightsreserved. 268 Z.Jiangetal./PowderTechnology371(2020)267–276 [36,37].AEfrequencydistributionisregularinrockuniaxialcom- ofthePCI-2workstation(PhysicalAcousticsCorporation,USA)and pressionfailure[37–39].Forexample,thefrequencybandcanreflect RS54Bsensors(BeijingSoftlandTimesScientific&TechnologyCo.Ltd., thefailuretypeofAEsource.DifferentrockmaterialsemitAEsofdif- China).ParameterssetfortheAEacquisitionworkstationandsensors ferentfrequencieswhentheyareloadedtofailure[22].Yaoetal.[40] arelistedinTables1and2.ThePDT,HDT,andHLTinTable1arethe foundthattheenergyofhigh-frequencyAEissmallerthanthatof peakdefinitiontime,hitdefinitiontime,andhitlockouttime,respec- low-frequencyAEwhenrockfailureoccursbyuniaxialcompression. tively.ThecorrectsettingofPDTwillensurecorrectidentificationof Caietal.[41]foundthathigh-frequencyAEcorrespondstosmall- risetimeandpeakamplitudedetection.ThecorrectsettingofHDT scalecracks,whereaslow-frequencyAEcorrespondstolarge-scale willensurethatanAEsignalinthedatastructureisreflectedinthesys- cracksinthenumericalsimulationofundergroundcavernexcava- temastheoneandonlyonehit.ThecorrectsettingofHLTwillavoid tion.Uniaxialcompressionexperimentshaveshownthatfrequency falsedetectionofsignalattenuationandimprovethespeedofdata of AE signals following coal failure is lower than that of rock acquisition. [15,17].Thesestudiesprovideaprimaryreferencefortheinvestiga- ThetriaxialstressenvironmentandplacementofAEsensorsareil- tionofAEsignalfrequencyduringhydraulicfracturing. lustratedinFig.2.Thevalueoftheprincipalstressissetasσ =5.0 v Still, AE characteristic and evolution are rarely investigated in MPa,σ =7.0MPa,andσ =4.5MPa,whichishalftheratioofdiggings H h stratifiedrockssuchascoalbedseams.Mechanismoffractureinitia- inthesouthwestofChina.Sensorswereevenlydistributedonthefour tion and propagation are complex in stratified coalbed reservoirs surfacesofthespecimen. withdifferentproperties.Itisdifficulttoperformhydraulicfracturing inthesestratifiedrocks[42].Dingetal.[43]havestudiedhydraulic fracturinginastratifiedformation,whichwasmadeofsimilarmate- 2.2.Specimenpreparation rials.TheyalsoanalyzedtheAEcountwiththehydraulicfracturing Theexperimentalmaterialsweregypsumpowder,Portlandcement, pressure.However,theydidnotanalyzetheAEfrequencyandenergy, andthecrackclassification,whicharecrucialtounderstandmecha- andcrushedstone(diameter≤8mm).Thematerialproportionswith different uniaxial compressive strengths were obtained as demon- nismsoffracturepropagation.Inthispaper,weaimtoobtainrelation- stratedpreviously[44,45].Theuniaxialcompressivestrengthofrock shipsbetweenmultipleAEcharacteristicsandinjectionpressureof hydraulicfracturing.Wewillanalyzetheinfluenceofstratifiedforma- was14.34MPa,whereasthatofcoalwas1.98MPa. Fourspecimens corresponding to different engineering condi- tiononfracturingpressureandAE.Tothisend,thepiezoelectricce- tionswereprepared(Fig.3a).Thefourspecimenswereallmade ramicAEsensorsareusedtorecordtheAEsignalsgeneratedduring with a 300 × 300 × 300-mm3 mold, and numbered as S1, S2, S3, hydraulicfracturingoffourkindsofspecimensunderatruetriaxial and S4 (Fig. 3 b). Among them, S1 simulates a single coal seam; stressconditions. S2simulatesasinglerockseam;andS3andS4arestratifiedsam- pleswithtwo coalseamsandthreerockseams.Allinterfacesare 2.Materialsandmethod naturallyformedwithoutanyspecialtreatment.Allrockseamsin S3andS4were50-mmthick.BothcoalseamsinS3hadathickness 2.1.Experimentalequipment of 75 mm. The upper coal seam in S4 was 50-mm thick,whereas the lower was 100 mm. A 20-mm diameter borehole was drilled Aschematicdiagramofthetruetriaxialhydraulicfracturingexperi- in themiddleof each specimen.The length of the boreholein S1 mentalsystemusedinthisworkispresentedinFig.1.Thesystem and S2 was 155 mm, whereas in S3 and S4 it was 255 mm. A mainlyconsistsofatruetriaxialstressloadingsubsystem,anAEand metalpipewasinsertedintotheboreholeandawateroutletwas pressuremonitoringsubsystem,andapumpsubsystem.Thetruetriax- arrangedatthebottomofthemetalpipe.Theexternalandinternal ialsubsystemcouldsimulateinsitustressbyloadingindependentlyin diametersofthemetal pipewere15and8mm,respectively.The threemutuallyperpendiculardirections,witheachofthemaximum externalwallofthemetalpipewasmaderoughtoincreasefriction pressurebeingabletoreach70MPa.Themaximumpumpratewas withtheboreholewallandpreventslidingduringfracturing.Epoxy 100ml/min. resinwasusedtosealtheholetoensurethemetalpipeistightly Aneight-channelhigh-frequencyAEinstrumentwasusedtocollect attached to the hole wall. The sealing lengths of S1 and S2 were theAEsignalsduringhydraulicfracturing.TheAEinstrumentconsisted 145mm,andthoseofS3andS4were50mm. Loading frame AE sensor Water pump e ol h HF fier Oil pump Jack Specimen mpli a e AE sensor r P Oil pump AE collec(cid:2)ng instrument Jack Fig.1.Truetriaxialhydraulicfracturingandacousticemissionmonitoringsystem.Thereisthetwo-dimensiondiagram,threejacksareactuallyemployedfortruetriaxialstress. Z.Jiangetal./PowderTechnology371(2020)267–276 269 Table1 Table3 ParametersofthePCI-2AEsystem. Sensorscoordinates. Preamp Frequency Sampling PDT HDT HLT Signal Sensors x y z Sensors x y z gain(dB) range(kHz) frequency (μs) (μs) (μs) threshold No. (mm) (mm) (mm) No. (mm) (mm) (mm) (103s−1) (dB) 1 0 75 225 5 300 75 75 40 100–400 10,000 50 200 300 45 2 0 225 75 6 300 225 225 3 75 75 0 7 225 75 300 4 225 225 0 8 75 225 300 Table2 Parametersofthesensor. simulatingthestressconditionsexistinginthefieldspecifiedinFig.2. Scale Working Hit Peaksensitivity Frequencyrange Enoughred-dyedwaterwaspreparedforinjectiontoinvestigatethe (mm) temperature(°C) limit (10−6Pa) (kHz) cracks. 8×21.9 −20–130 10,000 −650 100–400 Finally,thewaterpumpandAEsystemwereturnedonsimulta- neouslytoinjectwaterintotheboreholeandcollectAEsignals.The pumpratewassetas100ml/min.Aftertheinjectedwaterflowedout ofthespecimensurface,thewaterpumpwasclosed,markingthecom- 2.3.Experimentalprocedure pletionoftheAEsignalsacquisitionprogram. Firstly,thepreparedspecimenwasputintotheloadingframe.AE 3.Resultsanddiscussion sensorswereinstalledasillustratedinFig.2.Theceramiccontactsur- faceoftheAEsensorwasinclosecontactwiththespecimensurface Eachspecimen wassuccessfullytreatedbyhydraulicfracturing, throughVaseline(couplingagent).AfterconnectingthelineofAEin- whichresultedinobviouscracks.AsillustratedinFig.4,therewere strument,AEparametersspecifiedinTable1andTable3wereset, markedtracksoffluid(red-dyedwater)flowingoutofthefractured andthestabilityofAEsystemwastestedtoensuretheexperimentef- specimensandcracksinthesectionsacrossthespecimens.Thedirec- fects.Meanwhile,thefracturingpipelinewasconnected,withitsgas tionsofthecracksvariedbetweenthefourspecimens.Thedirection tightnesstestedinadvance. ofcrackscanbecategorizedaseitherhorizontalorvertical.Thehorizon- Secondly,stresseswereindependentlyappliedtothespecimenin taltrendwasseeninS1andS4,whereastheverticaltrendwasseenin threedirectionsusingflatjackswith0.01MPapersecondincreasing, S2andS3.Thesedifferencesmightbecausedbytheirregularityofthe Fig.2.TriaxialstressconditionandAEsensorslayoutdiagram. Fig.3.Specimensfortest.(a)specimenimage,(b)profilemapofinteriorstructureofspecimens,brownpresentsepoxyresin(sealinglength),redpresentsinjectionfluid,arrowspresents interfacebetweeninjectionfluidandspecimen.(Forinterpretationofthereferencestocolourinthisfigurelegend,thereaderisreferredtothewebversionofthisarticle.) 270 Z.Jiangetal./PowderTechnology371(2020)267–276 Fig.4.Crackformationafterhydraulicfracturing. specimensurface,whichledtoasymmetricalloadingthatresultedin thespecimencube.Thespecimenbrokedownat475s,atwhichpoint thedeviationofprincipalstressorientation.Inaddition,thesediffer- the injection pressure dropped again, beyond which it never rose encesmightbecausedbytheloadstressdifferenceamongthethreeor- again. The breakdown pressure of S2 was 6.39 MPa, and pressure thogonaldirection.Smallstressdifferenceinthistestsisapttoincrease remainedatabout5.71MPa(89%ofbreakdownpressure)intheend. theuncertaintyofdirectionofcrack'spropagation.Somecrackbranches ThemaximumAEcumulativecountswere1963.Itisworthnoting werealsonotedinthestratifiedspecimensS3andS4,indicatingthat thatthepressurehadfluctuatedforover385sfrom90sto475s, morecracknetworksformedinthestratifiedspecimenthaninthe whichwasextremelylongerthanthatobservedinS1.Thislongerpe- non-stratifiedspecimen. riodofpressurefluctuationsuggeststhatthecrackproducedpropa- gated gradually rather than instantaneously. This result is also 3.1.AEcount supportedbytheoveralltrendofgradualincreaseintheAEcount. In thetest involving S3,the breakdown pressurewas4.39MPa DamageandfractureofrockmassinduceAEactivity.Inthewave- (Fig.5).Theinjectionpressuredecreasedlinearlyforabout35safter formofAEsignal,thenumberoftimestheamplitudecrossesthethresh- themomentofbreakdown.Therateofdescentwithinthe35swasrel- oldvalueiscalled“ringingcount”,whichisalsocalled“AEcount”.The ativelylessthanthoseofS1andS2,whichpresentedasharpdrop.These moretheAEcount,themoreseverethedamageinsidetherockmass. resultsindicatethatalthoughsomecracksmighthaveformed,causing InthetestinvolvingS1,theinjectionpressurereacheditsmaximum thefluidtoleak,thesewerestilltoosmalltoallowamassiveflow.Inad- valueof9.32MPaat270sandthendroppedimmediately(Fig.5),indi- dition,followingbreakdown,AEactivitywasdetectedbyonlyC2and catingthatthespecimenwasbrokendown.Attheprecisemomentof C6.At710s,alltheAEcountofallchannelsincreasedenormously(C5 breakdown,AEcountsofC1,C5,andC6rosesteeply,especiallyforC6 wasbroken),withaconsiderablefallinpressure.Whencameto710s, whichincreasedfrom56to836;bycontrast,allremainingchannels injectionfluidof1183.3ml(equalto4.4%volumeofS3specimen) respondedpoorly.Thisdifferencebetweenthechannelsmightbedue hadbeeninjectedintoS3specimen.Whiletheporosityofallspecimens topriorloadingoftruetriaxialstress,whichcouldchangetheinterface waslessthan1%.Basedontheseresults,weinferredthatasnaplikely connectingthespecimenwiththeAEsensor.AfterbreakingdownS1, occurredinthespecimen,whichwouldcausethefluidtoleakoutof theinjectionpressuredroppedandthenroseagain;however,itcontin- the specimen. The injection pressure finally remained around 3.14 uedtofluctuateforover200s.Duringthepressurefluctuationperiod, MPauntilthepumpwasturnedoff.AmongallAEchannels,theevolu- theAEcountofC1increasedsimultaneously,suggestingthatcrack tionofC6agreedwellwiththeinjectionpressurehistory. wasproducedandcontinuedtopropagateandenlargethroughthe InthetestinvolvingS4,breakdowndidnotoccuruntil1276s(peak specimencube.Meanwhile,theAEcountofC5andC6,whichwassen- pressure,8.04MPa;Fig.5).Beforebreakdown,perturbationofinjection sitiveatthemomentofbreakdown,hardlyincreased,implyingthat pressurewasverydrasticandremainedthesameforabout236s.Dur- thecontactofthesechannelswiththespecimenwasswitchedoffdue ing the period, the AE cumulative counts simultaneously grew the tothedisplacementandmovementofS1duringbreakdown.Atthe fastest.Thisresultsuggestedthat236swasrequiredformacrocrackini- endoftheinjectionprocessofS1,thepressureremainedatabout4.19 tiationandpenetrationthroughS4.Beforedrasticperturbationofinjec- MPa (45% of breakdown pressure) and the AE cumulative counts tionpressure,anS-stylechangeinpressurewasnoted.Initially,thefluid reached1255. wasinjectedtofillthevolumeoftheinjectionhole,whichcausedthe InthetestinvolvingS2,bysynthesizinginjectionpressurecurveand pressuretogrowslowly.Oncetheinjectionholewasfull,thepressure AEcumulativecountsevolution,S2wasfoundtohaveexperienceda grewapproximatelylinearity.Whenthepressuregrowthtrendshifted relativelyintensefailureat348s(marketedinFig.5).Atthispoint, tononlinearity,fluidleakagebegan,andtheslopeofthepressurecurve thepressurereached5.46MPaanddroppedimmediatelyto4.48MPa, decreased,indicatingmicrocrackswereformingbeforebreakdown, andtheAEcountofC1andC4presentedanobviousincrease.However, andthesemicrocrackscouldcontinuouslyletfluiddiffusioninthe thepressurecontinuedtoincreaseandbecamehigherbeyondthis specimen. point;meanwhile,theAEactivityhadnotcalmeddown.Thissuggested SimilartoIshidaetal.[29,34],theAEcountsbeforeandaftera thatalthoughacrackmighthaveformedinside,itdidnotyetpenetrate sharpdropinpressureduringhydraulicfracturingwerefarmore Z.Jiangetal./PowderTechnology371(2020)267–276 271 Fig.5.InjectionpressurehistoryandAEcountaccumulationofallchannels. than at other times in our experiments, especially in S1 and S3 theinjectionpressurehistorythroughoutexceptC1andC4.Thus,in (Fig.5).Bycontrast,wefindthattheinjectionpressuresometimes thefollowingsections,datafromC1andC4weremainlyanalyzed. experiences a long, slow fluctuation process (Fig. 5), and the AE count increases rapidly. This is due to the fact that microcracks 3.2.AEenergy havebeenproducedinthespecimenallthetime;however,they didnotpenetratetheoutersideofthespecimen.Oncethecrack TheAEenergystatisticsofthefourtestaresummarizedinTable4,in propagationbreaksthroughtheoutersurfaceofthespecimen,the whichS2hasthehighestenergyvalueofmean,sum,andmaximum, pressurewilldecreasedramatically.Itisworthaffirmingthatthe whereasS1hasthelowestvalues,whichagreedwellwiththestrength moreintensethepressurefluctuation,thegreaterthescaleandthe ofthespecimens.ThestrengthofS3andS4wascalculatedbyanequiv- longerthelengthofthecracksformed,althoughthecracksdidnot alentapproach[46].Inaddition,bycomparingthefourspecimens,it penetrate the specimen. Therefore, the faster the increase in AE wasfoundthattheAEenergywaslargerifrockstratumwasbroken. count, the larger the crack size. The sudden rapid growth of AE Therefore,themagnitudeofAEenergyislikelytoindicatethetypeof countsindicatesthatlarge-scalecracksarebeinggenerated. rupturedstratum. Unexpectedly,theresponseofeachAEsensorisdifferentinthefour TheAEenergyofbrittlematerialsisknowntoobeythepower tests.TheAEcumulativecountswerenotequalateachtimepoint;in law distribution [47]. To investigate the power law distribution fact,theyvariedsignificantly.Besides,fewchannelsagreedwellwith index, we analyzed the energy data in double logarithmic Table4 AEenergystatistics. Specimen Strength Mean StandardDeviation Sum Minimum Median Maximum (Mpa) (aJ) (aJ) (aJ) (aJ) (aJ) (aJ) S1 1.98 690.61 1013.31 147,099.55 14.61 318.16 6937.00 S2 14.34 2120.58 18,084.43 897,007.20 6.59 420.66 367,434.00 S3 5.61 1017.20 1639.33 545,219.76 23.42 472.41 16,399.00 S4 2.78 935.02 1313.84 476,860.84 6.55 415.35 9017.00 272 Z.Jiangetal./PowderTechnology371(2020)267–276 Theprobabilityplot[48]isagraphicaltechniquetoassesswhether ornotadatasetfollowsagivendistributionsuchasthelog-normal. Thedataareplottedagainstatheoreticaldistributioninsuchaway thatthepointsshouldformapproximatelyastraightline(cf.therefer- encelineinFig.8).Departuresfromthereferencelineindicatedepar- turesfromthespecifieddistribution.AsFig.8illustrates,thefirstfew pointsandthelastfewpointsshowsomedeparturefromthereference line(identifiedbelowthereferenceline).Nevertheless,thetrendofper- centilecoincidedwiththereferencelineonthewhole,whichindicated thattheAEenergywasinaccordancewiththelog-normaldistribution inthefourspecimenstested. 3.3.AEfrequency Thepeakfrequencyisthekeyparameterwhichissubsequentlyac- quiredfromtheAEdetection,indicatingthevibrationfrequencyofma- terial damage and fracture. The peak frequency is calculated correspondingtothemaximumofFFT(fastFouriertransform)ofAE amplitude.Sometimesthepeakfrequencyisalsocalledthemainfre- quency, which refers to the main frequency of the AE sensor. The mainfrequenciesofthesensorsusedinthisexperimentwere100– Fig.6.Fitofpowerlawdistributionindex. 400 kHz; therefore, the AE peak frequency was filtered out to this range(Fig.9).Toinvestigatetheevolutionofpeakfrequencyassociated withtheinjectionpressure,peakfrequencies,pressures,andenergies coordinates and fitted these with linearity as shown in Fig. 6. wereplotted(Fig.9). Theselinesfittedwellwithenergyvalue,whichrangedfromhun- TheAEeventswithhigherenergywereconsideredtoidentifythe dreds to thousands. The opposite number of slope of the fitting failurestagesofhydraulicfractureinthisexperiment.Thepeakfrequen- line was equivalent to the power law distribution index. As ciesoftheseeventswithhigherenergywerecatalogedandclassified shown in Fig. 6, the power law distribution index of non- intotreesbands(Fig.10).Whencombinedthesepeakfrequenciescor- stratifiedspecimen(S1and S2) was biggerthan thatofstratified respondtothefailurestagesofhydraulicfracture.TheeventsinbandA specimens(S3andS4).Thissuggestedthatthestratifiedspecimen werecorrespondtothestageofintensepressurefluctuation,eventsin was more unstable than the non- stratified specimen. Further- bandBtothestageofpressurefluctuationafterthepeakpressure, more,theslopeofS1 wasequaltothatofS2,which indicatesno andeventsinbandCtoallstages. influenceofthestrengthonthepowerlawdistributionindex. Furthermore,asshowninFig.9,arelativelywiderangeofpeakfre- Toinvestigatethedistributionofallenergydata,wecalculatedthe quencywasnotedbetween100and300Hzwhentheinjectionpressure probabilityofenergyandplottedtheprobabilityhistogram(Fig.7). fluctuatedsharplyandthelow-frequencyeventsincreasedatthesame Wefoundthatthedistributionobeyedlog-normaldistributionfitted time.Forthetimebeing,thisphenomenonofpeakfrequencyrange curvesinFig.7. broadening will be termed “multifrequency response (MFR).” MFR Fig.7.Energyprobabilitydistribution.Thethincurveabovebarsisthefittedstandardlog-normaldistributioncurve. Z.Jiangetal./PowderTechnology371(2020)267–276 273 Fig.8.Log-normalprobabilityplotofenergy.Thereferencelinerepresentspercentofabsolutelog-normalprobability.Thecloserdatapoint(percent)istothereferenceline,themoreit agreeswiththelog-normaldistribution,viceversa. characteristicsareassociatedwiththesizeoffrequencyrange,AEnum- vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi bmeorr,eanthdeAAEEdeevnesnittysainndtimtheesscmalaell.eTrhtehelafrrgeeqrutehnecyfredqifufeernecnycerabnegtew,ethene StdevðlÞ¼uutPni¼−11(cid:3)li−li(cid:4)2 ð2Þ them,themoresufficienttheMFR.TheMFRindex(MFRI)wasassumed i n−2 torepresentthevaluepositivelycorrelatedwiththedegreeofMFR. ItisassumedthattherearenAEeventsint~t+Δtduringhydraulic whereStdev(li)denotesthestandarddeviationofli.Thesmallerthe fracturing,andtheirfrequenciesaref (i=1,2,3,…,n).Thefrequency Stdev(li),themoreevenlytheAEeventsaredistributedinLf,andthe i largertheL,themoresufficientthedegreeofMFR. bandwidthofMFRisthedifferencebetweenmaxf andminf int~t+ f Δt.Therefore,thatthefrequencybandwidthcanbei writtenais Therefore,theMFRIcanbeexpressedas vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u Lf ¼ThmeafrxeðqfuiÞe−ncmyidniðfffeiÞrencebetweenadjacenteventsisl =f −ð1fÞ, MFRIðtÞ¼uutðn−2Þ½Pmni¼a−1x1ð(cid:3)fliiÞ−−lim(cid:4)2inðfiÞ(cid:2)2 ð3Þ i i+1 i (i=1,2,3,…,n−1),andthestandarddeviationofl is i Fig.9.AEpeakfrequencycharacteristicsduringthehydraulicfracturingprocess. 274 Z.Jiangetal./PowderTechnology371(2020)267–276 Fig.10.AEpeakfrequencyrangedeviationaccordingtoAEeventdata. Usingformula(3),MFRIateachtimefromttot+Δtcanbecalcu- inferthatmacrofracturewasproducedonceinS1,thriceinS2,and lated.Mathematically,ittakesΔt→0togettheMFRIoftimet.AsΔt twiceinS3andS4.Therefore,theMFRImethodcanbeanefficient isassociatedwiththeAEeventrate,bysettingt=1s,theresults approachtoidentifytheoccurrenceofmacrofracture. werecalculated(Fig.11). FromanMFRIviewpoint(Fig.11),itisclearthatarelationship 3.4.Crackclassification existsbetweenAEpeakfrequencyandinjectionpressureevolution. MFRIisprominentwithadropininjectionpressure.Withtheap- Crackmechanismisanimportantfeatureofmaterialfailure.InAE pearance of large MFRI, AE activities increase extensively and testing,theAF–RAmethodiswidelyusedtofastclassifycrackmodelin- macrofracturesareformed.Thisresultisconsistentwiththeprinci- cludingtensilecrackandshearcrack[50],whereAFisAEcountdivided pleofvibrationmechanics[49],becausetheformationofcracksis byduration,andRAisrisetimedividedbyamplitude.Thus,itisimpor- the result of the convergence and penetration of multiple tantthatthethresholdsetisaccurateandappropriatetoobtainthetrue microcracks.Beforethemicrocrackspenetratedthespecimen,the risetimeandduration.IgnoringthedeviationfromAEsystem,wecould rock mass yielded locally in many places, forming microvoids. findthattensilecrackcomprisedover94%ofthecracksinalltests,with Priortothepenetrationofvoids,therockmasswasrelativelystable, theremainingbeingshearcracks(Fig.12).Thedominationoftensile thestrengthwasgoodandconsistenteverywhere,andthepeakfre- cracksdemonstratesthatmostfactureswhichcontributedtoinjection quencyofAEwasrelativelysingle.However,whenthevoidsstartto pressureactedasthetensilestressinsidethespecimens.Thisresult penetrate,cracksstarttoform,andnumerousmicrovoidwallsare was well in accordance with the tensile theory [51] of hydraulic generatedinstantaneously.Therockmassdeformationincreased fracture. subsequently,whichcausedrockmassinstabilityanddeteriorated Unfortunately,theresultofAElocationinthisfourtestswasnotvery itsstrength.Therefore,theAEfrequencybecameunstable,low- encouraging.AsshowninFig.13,whatlocationacquiredintheprofile frequency events occurred, and MFRI increased. Thus, we could depart invariably from the crack marked red induced by hydraulic Fig.11.MFRIduringhydraulicfracturing. Z.Jiangetal./PowderTechnology371(2020)267–276 275 Fig.12.CrackclassificationbytheAF–RAmethod. summarizedasfollows:1)low-frequencyAEeventsincreasedandthe phenomenon of MFR occurred when the pressure curve fluctuated sharply. The MFRI was positively correlated with the formation of macrocracksinhydraulicfracturing.2)asuddenrapidgrowthofAE countsindicatedthegenerationoflarge-scalecracks.TheAEcountin- creasedsharplywhenthepressurecurvefluctuatedsharply;bycon- trast, the AE count increased steadily when the pressure curve fluctuatedslightly.3)AEenergywasmoresensitivetoformationcondi- tions. AE energy tend to be larger if the rock stratum was broken. Concerningthemagnitudes,AEenergywasinaccordancewithlog- normaldistributioninthisfour-sampletest.TheAEenergyvaluevaried fromhundredstothousandsandfittedwellwithpowerlawdistribu- tion;thepowerlawdistributionindexofnon-stratifiedspecimenwas biggerthanthatofstratifiedspecimensduringhydraulicfracturing,sug- gestingthatthestratifiedspecimenwasmoreunstablethanthenon- stratified specimen. 4) The tensile crack, classified by the AF–RA method,dominatedinalltests,indicatingthatmostfactureswhichcon- tributed to injection pressure acted as the tensile stress inside specimens. Overall,hydraulicfracturingofcoalbedmethaneisacomplexpro- cess,someencouragedperspectivesweresummarizedwithafewtypi- Fig.13.AElocation,withS2asanexample.Tofitwiththecrack,theeventslocatedinthe calexperimentsprovidedinthispaper.Theonlydrawbackisthatnotall crosssectionwereextracted. AEsensorscouldcompletelyrecordthecriticalsignalsoffracturing,and theAElocationhaderrorsduetoheterogeneityofspecimens.Toobtain moreinformationaboutAEsignalcharacteristicsduringhydraulicfrac- fracturingtreatment.Thisinconsistencymaybeduetotheanisotropyof turingincoalseamgasreservoirs,moreexperimentalstudiesshouldbe thespecimens[52],namely,AEvelocitywasnotaconstantineverydi- carriedoutusingareliablelocationmethod. rectionandalsoineverymomentoftheinjectionperiod.Furthermore, first-arrivaldetections,whichrequiremanualcorrection,wereallde- pendentonautomationoftheAEsystem,resultingininevitablelocation DeclarationofCompetingInterest errors.Thisdiscrepancypresentedsomechallengesforevaluationofhy- draulicfracturing.Therefore,itisnecessarytocomeupwithanewand Theauthorsdeclarethattheyhavenoknowncompetingfinancial reliablelocationmethod. interestsorpersonalrelationshipsthatcouldhaveappearedtoinflu- encetheworkreportedinthispaper. 4.Conclusion Acknowledgement Hydraulicfracturingunderfourcoalseamconditionswassimulated inatruetriaxialstressenvironment,andtheirAEdatawerecollected, ThispaperwassupportedbytheNationalMajorScienceandTech- analyzed, and discussed in depth. Some meaningful insights are nology Projects of China (No. 2016ZX05045004, No. 276 Z.Jiangetal./PowderTechnology371(2020)267–276 2016ZX05043005),theNationalNaturalScienceFoundationofChina [24] Z.Z.Jiang,Q.G.Li,Q.T.Hu,J.F.Chen,X.L.Li,X.G.Wang,etal.,Undergroundmicroseis- micmonitoringofahydraulicfracturingoperationforCBMreservoirsinacoal (GrantNo.51604045),theStateKeyLaboratoryCultivationBasefor mine,EnergySci.Eng.7(3)(2019)986–999. GasGeologyandGasControl(HenanPolytechnicUniversity),Natural [25] M.Fuwa,A.R.Bunsell,B.Harris,AcousticEmissionandFatigueofReinforcedPlastics, ScienceFoundationofChongqing,China(cstc2019jcyj-bsh0041).We 197477–79. [26] M.Shimada,A.Cho,H.Yukutake,Fracturestrengthofdrysilicaterocksathighcon- thankSWANEditorialServices(www.swaneditorial.com)forediting finingpressuresandactivityofacoustic-emission,Tectonophysics96(1–2)(1983) thispaper. 159–172. [27] E.C.Donaldson,W.Alam,N.Begum,Rockmechanicsoffracturing-hydraulicfrac- References turingexplained-Chapter3,HydraulicFracturingExplained2013,pp.47–76. [28] D.Tiab,E.C.Donaldson,Chapter9-effectofstressonreservoirrockproperties,in:D. [1] hGi.gDhulyvestarua,tiJfi.Fe.dShroacok,sA,Imnto.dJ.ifiReodcksiMngelcehp.Mlanine.o3f5w(6e)ak(n19es9s8)th8e0o7r–y8f1o3r.thefailureof Tiniagb,,BEu.rCli.nDgotonnal2d0so0n4,(pEpd.s5.)5,4P–e6tr7o0p.hysics,SecondeditionGulfProfessionalPublish- [29] N.Li,S.C.Zhang,Y.S.Zou,X.F.Ma,S.Wu,Y.N.Zhang,Experimentalanalysisofhy- [2] P.Wang,M.Cai,F.Ren,C.Li,T.Yang,Theoreticalinvestigationofdeformationchar- draulicfracturegrowthandacousticemissionresponseinalayeredformation, aGcetoesrciis.tJi.c2s1of(2s)tr(a2t0ifi1e7d)2ro1c3k–s22co2n.sideringgeometricandmechanicalvariability, RockMech.Rock.Eng.51(4)(2018)1047–1062. [30] S.Stanchits,A.Surdi,P.Gathogo,E.Edelman,R.Suarez-Rivera,Onsetofhydraulic [3] N.A.Do,D.Dias,V.D.Dinh,T.T.Tran,V.C.Dao,V.D.Dao,etal.,Behaviorofnoncircular fractureinitiationmonitoredbyacousticemissionandvolumetricdeformation tunnelsexcavatedinstratifiedrockmasses–caseofundergroundcoalmines,J. measurements,RockMech.Rock.Eng.47(5)(2014)1521–1532. RockMech.Geophys.Eng.11(1)(2019)99–110. [31] G.Kwiatek,K.Plenkers,P.Martinez-Garzon,M.Leonhardt,A.Zang,G.Dresen,New [4] Y.Tan,Z.Pan,X.-T.Feng,D.Zhang,L.D.Connell,S.Li,Laboratorycharacterisationof insightsintofractureprocessthroughin-situacousticemissionmonitoringduring fracturecompressibilityforcoalandshalegasreservoirrocks:areview,Int.J.Coal FatiguehydraulicfractureexperimentinAspohardrocklaboratory,Process.Eng. Geol.204(2019)1–17. 191(2017)618–622. [5] A.Rezaei,B.Dindoruk,M.Y.Soliman,Onparametersaffectingthepropagationofhy- [32] X.F.Ma,N.Li,C.B.Yin,Y.C.Li,Y.S.Zou,S.Wu,etal.,Hydraulicfracturepropagation draulicfracturesfrominfillwells,J.Pet.Sci.Eng.182(2019)106255. geometryandacousticemissioninterpretation:AcasestudyofSilurianLongmaxi [6] J. Cai, Z. Zhang, W. Wei, D. Guo, S. Li, P. Zhao, The critical factors for FormationshaleinSichuanBasin,SWChina,Pet.Explor.Dev.44(6)(2017) permeability-formationfactorrelationinreservoirrocks:pore-throatratio,tortuos- 1030–1037. ityandconnectivity,Energy188(2019)116051. [33] K.Ohno,M.Ohtsu,Crackclassificationinconcretebasedonacousticemission, [7] W.Yan,H.Ge,J.Wang,D.Wang,F.Meng,J.Chen,etal.,Experimentalstudyofthe Constr.Build.Mater.24(12)(2010)2339–2346. frictionpropertiesandcompressiveshearfailurebehaviorsofgasshaleunderthe [34] T.Ishida,Acousticemissionmonitoringofhydraulicfracturinginlaboratoryand influenceoffluids,J.Nat.GasSci.Eng.33(2016)153–161. field,Constr.Build.Mater.15(5–6)(2001)283–295. [8] S.Sharma,V.Agrawal,R.N.Akondi,Roleofbiogeochemistryinefficientshaleoiland [35] D.G.Aggelis,Classificationofcrackingmodeinconcretebyacousticemission gasproduction,Fuel259(2020)15. parameters,Mech.Res.Commun.38(3)(2011)153–157. [9] T.Ishida,Q.Chen,Y.Mizuta,Effectofinjectedwateronhydraulicfracturingdeduced [36] Q.Q.Ni,M.Iwamoto,Wavelettransformofacousticemissionsignalsinfailureof fromacousticemissionmonitoring,PureAppl.Geophys.150(3–4)(1997)627–646. modelcomposites,Eng.Fract.Mech.69(6)(2002)717–728. [10] S.Kramadibrata,M.A.Rai,B.Sulistianto,R.K.Wattimena,P.N.Hartami,K.Matsui,De- [37] R.Gutkin,C.J.Green,S.Vangrattanachai,S.T.Pinho,P.Robinson,P.T.Curtis,On terminationofthreedimensionalinsitustressregimeusinghydraulicfracturing acousticemissionforfailureinvestigationinCFRP:patternrecognitionandpeakfre- andacousticemissionmethods,Contrib.RockMech.NewCent.1and2(2004) quencyanalyses,Mech.Syst.Signal.Pr.25(4)(2011)1393–1407. 1147–1152. [38] P.J.Groot,P.A.M.Wijnen,R.B.F.Janssen,Real-timefrequencydeterminationof [11] Q.G.Li,B.Q.Lin,C.Zhai,Theeffectofpulsefrequencyonthefractureextensiondur- acousticemissionfordifferentfracturemechanismsincarbon/epoxycomposites, inghydraulicfracturing,J.Nat.GasSci.Eng.21(2014)296–303. Compos.Sci.Technol.55(4)(1995)405–412. [12] Y.P.Liang,Y.H.Cheng,Q.L.Zou,W.D.Wang,Y.K.Ma,Q.G.Li,Responsecharacteristics [39] C.R.Ramirez-Jimenez,N.Papadakis,N.Reynolds,T.H.Gan,P.Purnell,M.Pharaoh, ofcoalsubjectedtohydraulicfracturing:anevaluationbasedonreal-timemonitor- Identificationoffailuremodesinglass/polypropylenecompositesbymeansofthe ingofboreholestrainandacousticemission,J.Nat.GasSci.Eng.38(2017)402–411. primaryfrequencycontentoftheacousticemissionevent,Compos.Sci.Technol. [13] P.G.Ranjith,W.A.M.Wanniarachchi,M.S.A.Perera,T.D.Rathnaweera,Investigation 64(12)(2004)1819–1827. oftheeffectoffoamflowrateonfoam-basedhydraulicfracturingofshalereservoir [40] X.-L.Yao,Y.-B.Zhang,X.-X.Liu,P.Liang,L.Sun,Optimizationmethodforkeychar- rockswithnaturalfractures:anexperimentalstudy,J.Pet.Sci.Eng.169(2018) acteristicsignalofacousticemissioninrockfracture,RockSoilMech.39(1)(2018) 518–531. 375–384. [14] J.Kaiser,AStudyofAcousticPhenomenainTensileTest,PhDThesisTechnicalUni- [41] M.Cai,P.K.Kaiser,H.Morioka,M.Minami,T.Maejima,Y.Tasaka,etal.,FLAC/PFC versity,1950. couplednumericalsimulationofAEinlarge-scaleundergroundexcavations,Int.J. [15] S.Stanchits,J.Burghardt,A.Surdi,Hydraulicfracturingofheterogeneousrockmon- RockMech.Min.44(4)(2007)550–564. itoredbyacousticemission,RockMech.Rock.Eng.48(6)(2015)2513–2527. [42] T.T.Jiang,J.H.Zhang,G.Huang,S.X.Song,H.Wu,Effectsofbeddingonhydraulic [16] J.N.Albright,C.F.Pearson,Acousticemissionsasatoolforhydraulicfracturelocation fracturingincoalbedmethanereservoirs,Curr.Sci.India113(6)(2017)1153–1159. -experienceattheFentonHillhotdryrocksite,Soc.Petrol.Eng.J.22(4)(1982) [43] D.Hong,LawofHydraulicFracturingCrackInitiationintheMulti-LayeredCoal 523–530. Seams,MasterChongqingUniversity,2014. [17] M.Seto,T.Kiyama,T.Narita,M.Kouno,K.Shiota,H.Nabeya,etal.,Acousticemission [44] Q.T.Hu,S.T.Zhang,G.C.Wen,L.C.Dai,B.Wang,Coal-likematerialforcoalandgas inhydraulicfracturingofcoalmeasurerock,J.MMIJ105(9)(1989)661–666. outburstsimulationtests,Int.J.RockMech.Min.74(2015)151–156. [18] V.Arumugam,R.N.Shankar,B.T.N.Sridhar,A.J.Stanley,Ultimatestrengthprediction [45] Y.H.Guo,R.J.Cao,L.H.Zhu,Researchonsimilarmaterialinphysicalspecimenpe- ofcarbon/epoxytensilespecimensfromacousticemissiondata,J.Mater.Sci. trographyofrock,Adv.Mater.Res.616–618(2013)346–349. Technol.26(8)(2010)725–729. [46] G.Yin,X.Li,J.Lu,Z.Song,Afailurecriterionforlayeredcompositerockundertrue [19] J.W.Liu,X.Z.Wu,X.X.Liu,Y.Y.Yu,W.Hu,L.B.Yin,Time-frequencycharacteristicand triaxialstressconditions,Chin.J.RockMech.Eng.36(2)(2017)261–269. signalrecognitionofacousticemissiongeneratedfromdifferentrockbrittlefailure, [47] B.Jordi,C.Álvaro,I.Xavier,P.Antoni,SaljeEKH,S.Wilfried,etal.,Statisticalsimilar- Nonferr.MetalsSci.Eng.4(6)(2013)73–77. itybetweenthecompressionofaporousmaterialandearthquakes,Phys.Rev.Lett. [20] X.Jingna,N.Guanhua,X.Hongchao,L.Shang,S.Qian,D.Kai,Theeffectofaddingsur- 110(8)(2013)(088702). factanttothetreatingacidonthechemicalpropertiesofanacid-treatedcoal,Pow- [48] J.M.Chambers,W.S.Cleveland,B.Kleiner,P.A.Tukey,GraphicalMethodsforData derTechnol.356(2019)263–272. Analysis,Wadsworth,1983. [21] S.G.Cao,Y.B.Liu,L.I.Yong,L.Q.Zhang,Experimentalstudyonacousticemission [49] W.T.Thomson,TheoryofVibrationwithApplications,Prentice-Hall,1981. characteristicsofcoalrockatdifferentconfiningpressure,J.Chong.Univ.32(11) [50] R.K.Miller,P.Mcintire,AcousticEmissionTesting,2008. (2009)1321–1327. [51] Y.Aimene,C.Hammerquist,A.Ouenes,Anisotropicdamagemechanicsforasym- [22] X.Wang,E.Wang,X.Liu,X.Li,H.Wang,D.Li,Macro-crackpropagationprocessand metrichydraulicfractureheightpropagationinalayeredunconventionalgasreser- correspondingAEbehaviorsoffracturedsandstoneunderdifferentloadingrates, voir,J.Nat.GasSci.Eng.67(2019)1–13. Chin.J.RockMech.Eng.37(6)(2018)1446–1458. [52] L.Qin,S.Li,C.Zhai,H.Lin,P.Zhao,M.Yan,etal.,Jointanalysisofporesinlow,inter- [23] X.Li,L.Zhonghui,W.Enyuan,L.Yunpei,L.Baolin,C.Peng,etal.,Patternrecognition mediate,andhighrankcoalsusingmercuryintrusion,nitrogenadsorption,andnu- ofminemicroseismic(MS)andblastingeventsbasedonwavefractalfeatures,Frac- clearmagneticresonance,PowderTechnol.362(2020)615–627. tals26(3)(2018)(1850029-1–18).

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.