Sunteți pe pagina 1din 4

GEOPHYSICAL

RESEARCH
LETTERS,
VOL.20,NO. 19,PAGES
2119-2122,
OCTOBER
8, 1993

OBSERVATIONSOF THE KAISER EFFECTUNDER MULTIAXIAL STRESSSTATES:IMPLICATIONS FOR


ITS USE IN DETERMINING IN SITU STRESS

David J. Holcomb

SandiaNationalLaboratories,Albuquerque

Abstract.Experimental
testsof theKaisereffect,thestress- promising.Stressis measured
directlyandtheentirestress
history
dependenceofacousticemission production, showthat tensorwould be available. Several efforts to determinein situ
interactions
betweenprincipalstresses
cannotbe ignoredas stressusingtheKaisereffectin thesimplemannerdescribed
is commonly donewhentryingto usetheKaisereffectto herehavebeenreported[e.g. Hughson andCrawford,1986;
determinein situstress.Experimental
resultsobtainedunder Michiro et al. 1985; Holcomband Martin, 1985; Hayashiet
multiaxialstressstates
areexplained
in termsof a qualitative al., 1979;Kanagawa et al., 1981].Oneof themostconvincing
model.Theresultsshowthatthecommonly-used techniqueof studiesisthatof Momayez andHassani (1992)whodetermined
loading
uniaxially alongvarious
directionstodeterminestress theprincipal stresses
andtheirorientation attheAtomicEnergy
historymustbereevaluatedasit cannotbejustified in terms of CanadaUnderground Research Laboratory.Theirresults
of thelaboratory
experiments.Onepossible resolutionof the arein closeagreement with stresses
determined usingmore-
conflictbetween
laboratory
andfieldresults is thattheKaiser accepted methods, suchasover-coring.
effectphenomenonobserved
incores retrieved fromtheearthis Despite itspromise,theKaiser effectmethod fordetermining
notthesamephenomenon asisobserved in rockloaded under in situstressis notwidely-accepted. Althoughtheinterpreta-
laboratoryconditions. tion of the in situ Kaiser effect is basedon observationsof
the laboratory Kaisereffect,therehasnot beena systematic
Introduction investigationof thelaboratory Kaisereffectto seeif thebasic
assumptions arewarranted.
Followingthe discoverythatproduction of acousticemis- Of greatinterestis theeffectof interaction betweenstress
sionswasstress-historydependent [Kaiser,1950],therehave components in determining the locationof the Kaiserstep.
beenmanyattempts to usethe Kaisereffectto determinein Mostrockproperties aredependent on the pressure or more
situstress.For conventional
triaxialcompressiontests,it was generally, meanstress.In particular, ultimatecompressire
foundthat the Kaisereffect markedthe stresspeak attained strength is stronglydependent on confining pressure (Pc)in
in a previousloadcycle[Goodman,1963]. Additionalwork triaxialtests.Typically,strength
increasesseveralMPaforeach
hasbeendoneon the lengthof time that the effectendured MPa increasein Pt.. Sinceacousticemissions
areproducedby
afterinitialloading,
theeffectsof waterandwhattypesof rock a failureprocess,
it wouldbeexpected
thatthestress
required
exhibitedtheKaisereffect[YoshikawaandMogi, 1978, 1981; to causeAE wouldalsobe pressure dependent.It hasbeen
KuritaandFujii, 1979;FriedelandThill, 1990;Momayezand shownin severalrocktypes[Holcomb,1984,Holcomband
Hassani,1992]. Experimenters haveexaminedthe basicas- Martin, 1985, Holcomband Costin,1986, Holcomb,1992]
sumption thattheprincipalstressesweredecoupled suchthat that the stressat which a laboratory-induced
Kaiser stepis
the Kaiserstepobservedfor loadingalonga givenaxiswas observed is similarlydependent
on Pc. Thusthefundamental
independentof stress
components alongorthogonal directions assumption thattheKaiserstepisobserved ata stress
equalto
[Murayamaet al. 1984;HolcombandCostin,1986;C. Stuart, theprevious peakstress component alonga givenaxisisshown
personalcommunication, 1992]. to be incorrect.At a minimum,thepressure dependence would
Knowingthe onsetof AE activitywashistory-dependent, require a correction to the uniaxialloading result.
it was naturalto hypothesizethat the stressexperienced by The purposeof this paperis to showthat sucha correction
rockin theearthcouldbedetermined by uniaxiallyrestressing cannotaccountfor the effectsof stressinteraction. Existing
sub-corestaken at various orientationsand monitoring AE. resultsandonenewexperimentareanalyzedto investigatethe
Observation
of a Kaiserstepwouldindicatethestress
thecore effectsof stressinteraction. It will be shownthat the simple
hadbeensubjected to in theearthalongthe loadaxis. To uniaxialloadingmethodfor determiningin situstresscannotbe
clarifythefollowing discussion,
thephenomenon observedas justifiedon the basisof laboratoryresults,if the in situKaiser
a resultof reloadingaftera known,laboratory-imposed
stress effectis the samephenomenon asthelaboratoryKaisereffect.
historywill bereferredtoasthelaboratoryKaisereffect.The
apparently-similarphenomenon observedwhenrockisloaded ExperimentalResults
for thefirsttime in thelaboratory,
freshfromthein situstress
field of the earth will be called the in situ Kaiser effect. Sampleswereparallelepipeds of Tennessee
marble,54 by 54
Becausethe Kaisereffect methodrequiresno referenceto by 25.4 millimeters,jacketedin copperandpolyurethane.Each
an intervening
materialmodel,as do all of the othercore- samplewasinstrumented with two AE transducersgluedto the
basedstressdeterminationmethods,it has always beenvery centerof the largefaces. Loadingwas by a servo-controlled,
truetriaxialsystem[Wawersiket al., 1992]capableof applying
Copyright1993by the AmericanGeophysical
Union. any compressive stressstatewith principalstresses
parallelto
the axesof the parallelepiped.For eachexperiment,a sample
Papernumber93GL01270 was prestressed underextensionconditions(0-• = 0'22> 0'33,
0094-8534/93/93 GL-01270503.00 compression positive)until AE wereinduced,showingthat a

2119
2120 Holcomb:Experimental
Testof KaiserEffect

2OO 0.5
Test1' ' ' ,' ' Table 1. StresshistoryandAE onsetsfor thetestsshownin
-o Figures1 and2
-- x
0.4

Test l
0-11 0'22 0'33 PathType
lOO Stress Peak 127 127 10 Extensional
0.2 rc AE onset 80 10 10 Triaxial
Stress Peak 110 10 10 Triaxial
0.1 AE onset 10 89 10 Triaxial
Test 2
o 0.0 Stress Peak 149 149 10 Extensional
2000 3000 4000 5000 6000 7000
AE onset 10 56 10 Triaxial
Time (sec) Stress Peak 10 71 10 Triaxial
AE onset 89 10 10 Triaxial
Fig. 1. AE rate(thinjaggedline)andstress
history(heavy
andmediumlines)for an extensionalstresspeakof 120 MPa,
hypothesesisthatindividualstress
components wouldbefound
followedbyconventional triaxialloading(Test1). Dottedlines for eachcore orientation. Certainlythis is correctfor the
separate thethreeloadphases. AE onsetfor theconventional conventional triaxiallaboratorytestwhenthe prestress and
triaxialphasesis shownby the arrows. reloadpathsareidentical.It hasbeenassumed thatthesame
techniquecouldbeapplied tocoreretrieved fromtheearthto
determinethecompletein situstresstensor.
Kaiserstephadbeencreated.Thenthestress wasreturned to If the in situKaisereffectis relatedto the laboratoryKaiser
hydrostaticandtwo conventional triaxialtests(0-• > 0'22=
effect( an untested assumption), severallinesof reasoning
0'33)weredonewiththemaximumcompressire loadaxisfirst
indicatethatthissimpleapproach wouldnotwork. First,as
paralleltotheoriginal0'• axisandthenthe0'22 axis.Duringthe
discussedearlier,thelaboratory Kaisereffectobservedin con-
triaxialphase,stress
wasraiseduntilAE haddefinitelybegun.
ventionaltriaxialcyclictestsis pressure-dependent.Thusthe
Thesamplewasthenunloaded toavoidshiftingtheKaiserstep
stress
required to generateAE cannot uniquely determine the
any morethannecessary.
stresshistoryevenin thesimplecaseof conventional triaxial
Figures1 and2 showtheAE ratesandstresshistories
for
loading.
testson differentsamples.ArrowsindicatetheAE onsetstress
A second argument is basedonthenon-damaging character
for the conventionaltriaxial loadingphases. To emphasize
thedifferencein onsetstresses, the stresspeakfor the second of hydrostatic loading. Untilporecollapse isinitiated, usually
sample(Test2) underextensional loadingwasincreased to 149 atquite high stresses,
hydrostatic stress
does not induce aKaiser
MPa. Table 1 summarizes the stresspeaksandthe stressstate step.Yet, thereisnothing in thecommonly-used approachthat
at whichAE wasfirstobserved.For all testing,0'53washeld suggests thatthe hydrostaticcase shouldbe treated differently.
constant at 10 MPa. As usuallydescribed, it shouldbepossible to determine the
magnitude of a previoushydrostaticstressbyuniaxialloading
Discussion of threemutually-perpendicular coresuntila Kaiserstepisob-
served.Implicitin theconventional method is theassumption
thatit isonlyapplicable tostressstateswitha shearcomponent.
Currentattemptsto usethe Kaisereffectfor in situ stress
Butthatin turnimpliesthattheprincipalstresses arecoupled
determination involvetakingsubcores at variousorientations
and applyinguniaxialstressalongeachsub-coreaxis. The in determining thestressstateswhere a Kaiser stepisobserved.
All experimentsto dateagreethereis no interactionif uni-
axial stresses
areappliedsequentiallyalongorthogonal axes.
200 However,the experiments do not settlethe questionof stress
Test 2
o•
.-
' • 0.9 interactionwhen stresses
are appliedsimultaneously,
as they
'• .• --• 0.8 are in the earth. Experimental
resultspresentedearlierad-
c ! o• -0.7 dressthe questionof interactionwhen non-triaxialstresses
are appliedsimultaneously.An extensionalstressstatewas
._j
• 0.5 usedto induceacousticemissionsand a Kaiser stepat a stress
significantly
higherthanthe initial, virgin,Kaiserstep. For
/ i\ •
0.4
0.3 an extensionalstresspath, the Kaiser step was moved to
0.2 (0'11,peak
= 0'22,peak,
0'33)-The conventional
approach
would
!
• 0.1
then arguethat subsequent loadingalongthe 1- and 2- axes
o • • [• -• 0.0 wouldfind a Kaiserstepat 0'1•= 0'11.p,akand0'22--
lOOO 2000 3000 4000 5000 6000
respectively.As Figures1 and 2 and Table 1 show,this is
Time (sec) not the case. Instead,for loadingalongthe 1- and 2- axes,
a Kaiserstepwasfoundat a muchlower stress.Clearly,the
Fig. 2. AE rate (thinjaggedline) andstresshistory(heavy conventionalapproach incorrectlypredictstheonsetof AE for
andmediumlines)for an extensionalstresspeakof 149 MPa, the laboratoryKaisereffect.
followedby conventional
triaxialloading(Test2). AE onset A qualitativeargumentthatmakesfew assumptions can be
for theconventionaltriaxialphasesis shownby the arrows. usedto explaintheseresults.The argumentis basedon the
Holcomb: ExperimentalTestof KaiserEffect 2121

Crack
Distribution
AfterExtensional
Loading
Crack
Normal a
3

Stress

Crack Parameters and Growth Model

Fig. 3. Crackparameters
andrelationof stressorientationand
crackgrowth.Crackgrowthastheresultof stressin thecrack
planeisindicated
by thedarkershading
andtheincrease
in lal.
Fig. 5. Evolutionof initially-isotropiccrackdistributionunder
observations
thatcracksin rock undercompressive
stresstend extensional stress
appliedin the1-2plane.Crackswithnormals
to growparallelto themaximumcompressivestress,aredriven roughlyin the 3 directiongrowandtheirnormalsplot as the
bythedeviatoric stress
andsuppressed bycompressive stress spikealongthe 3 axis.
normalto thecrackplane.Thus,a cracklying in the 1-2 plane
will be extendedby compressive
stresses in the 1-2 planeand observed thistypeof distributionfor rocksubjected to conven-
suppressed by compressive
stressin the3 direction.In Figure tionaltriaxialtesting.
3, thecracklateraldimension is represented
by [aI, thelength An extensionalloadingpath (increasing 0.• - 0.22) 0'33,
of thevectora. In responseto a compressive
stressin thecrack constant)producesa differentpicture(Figure5). For exten-
plane,theinitially-circular
crackextendsin its ownplaneand sionalloadingin the 1-2plane,onlycrackswith normalsclose
[oI increases.The experimental resultsshownin Figures1 to the3 axisgrow.Thecompressive stresses in the 1-2 plane
and2 andthequalitativeargument agreewithpredictionsof a suppress growthof othercracks.Theresultis thatthespherical
micromechanical modelof crackgrowthundergeneralstress distribution becomespointeddue to growthof crackswhose
states(HolcombandCosfin,1986seeFigure7b). planesare in the 1-2 plane. Crackswith normalsin the 1 and
An ensembleof cracksis requiredto representthe crack 2 directionareundisturbed (Figure5). If thesamplewasthen
population in a rock. For simplicity,the crackpopulationis loadeduniaxiallyalongthe 1 axis,crackswith normalsin the
assumed to be initiallyisotropicwith eachcrackrepresented 2 directionwouldgrowandproduceAE at the sameuniaxial
by thenormalto its plane.Plottingall of the normalvectorsa stressthat would have causedcrack growthin the original,
produces a spherewithradiuslal. never-loaded sample.ThusAE will be producedfor uniaxial
For this very idealizedpicture,the responseof the crack stressalongthe1 axisthatis unrelatedto thestress alongthe1
populationto extensionaland uniaxial stressstatescan be vi-
axisduringextensional loading.Forloadingalongthe2 axis,
sualizedas in Figures4 and 5. In both figures,the crack the situationis identical,involvingcrackswhosenormalslie
distribution
prior to loadingis shownas a sphere.Applying alongtl/e 1 axis. The resultsshownin Figure1 and2 arise
a uniaxialstressalongthe 3 axiscausesgrowth(extension) from exactlythissituation.
of crackswhosenormalslie roughlyin the 1-2 plane. The Fromthissimplified model,it followsthatuniaxialloading
initiallyspherical
distribution
develops
a girdleasseenin Fig- cannot beexpected todetect theextensional
stresshistory
along
ure 4. WawersikandBrace(1971) andKranz(1979, 1980) theloadingaxis. In fact,for anextensional loadinghistory,
uniaxialloadingis blindto theprevious stressandgivesthe
resultthatnoloadhadbeenapplied.If thein situKaisereffect
CrackDistribution
AfterUniaxial
Loading that is measuredon core retrieved from the earth is the same
phenomenon asthelaboratory Kaisereffect,thentheuniaxial-
r 3axis loadingapproach cannot,evenin principle, determine
thein
situstressfor generalstressstates.

Conclusion

TheKaiser effecthasbeenreported
forrockcyclically
loaded
1 2 axes in thelaboratory(laboratory
Kaisereffect)andforrockloaded
for thefirsttimeafterbeingretrievedfromtheearth(in situ
Kaiser
effect).Analysis
ofpreviously-reported
andnewexper-
imental resultshas shownthat becauseof interactionsbetween
simultaneously-applied
stresses,the stresshistorycannotbe
determined by uniaxialloadingfor thelaboratory Kaiseref-
Fig. 4. Evolutionof an initially-isotropic
crackdistribution fect. A qualitativemodelincorporating theknownbehavior
(thesphere)underuniaxialstress alongthe3-axis.Crackswith of microcracksgrowingin response to compressivedeviatoric
normalsin the 1-2planegrowandtheirnormalsareplottedas stressesprovidesan explanationfor theexperimentalresults,
a diskaroundtheinitial sphere. particularly
theextensionalloadingtests.Thusdetermination
2122 Holcomb:Experimental
Testof KaiserEffect

of the in situ stressby the commonly-used uniaxialloading TensileTestsor a Studyof AcousticPhenomena


in Tensile
methodcannotbe justifiedon the basisof the laboratoryex- Tests, Ph.D. thesis,Tech. Hosch. Munchen,Munich,
periments.That leavesthequestionof whatis beingobserved Germany,1950.
by thosewhomeasurethein situKaisereffect?Furtherstudy Kanagawa,T, Hayashi,M. andKitahara,Y., Acoustic
Emis-
shouldnow be directedto the questionof what is being ob- sion and Over CoringMethodsfor MeasuringTectonic
servedin field applications
of the Kaisereffect. If in situstress Stresses, in Proc. lnt. Symp. on WeakRock,pp. 1205-
canindeedbe determinedvia the standardapproach,thenthere 1210,Tokyo, 1981.
mustbe anotherphenomenon responsible for the relationship Kranz, R. L., The EffectsOf ConfiningPressureand Stress
betweenonsetof AE andstresshistoryof rock stressed by the DifferenceOn StaticFatigueOf Granite,J. Geophys.Res.,
earthover geologicaltime. 85, 1854-1866, 1980.
Kranz, R. L., CrackGrowthandDevelopmentDuring Creep
Acknowledgments.
Thankyou to L. Costin,L. Teufel, D. of Barre Granite,Int. J. RockMech. Min. Sci., 16, 23-35,
Zeuch and W. Wawersik for careful reviews and enlighten- 1979.
ing discussions.LawrenceCarlsonwas responsible for the Kurita,K. andN. Fujii, StressMemoryof Crystalline Rocks
extensiontests.Thisworkperformedat SandiaNationalLab- In AcousticEmission,Geophys. Res.Lett.,6, 9-12, 1979.
oratoriessupportedby theU.S. Departmentof EnergyOffice Michihiro,K., T. FujiwaraandH. Yoshioka,Studyon Esti-
of BasicEnergySciences, undercontractnumberDE-AC04- matingGeostresses By The KaiserEffectof AE, in Proc.
76DP00789. 26th U.S. Symp.onRockMech.,editedby E. Ashworth,pp.
557-564, A. A. Balkema, 1985.
References Momayez,M. andHassani, F. P., Application
of KaiserEffect
to MeasureIn-situStresses
in Underground Mines,in Proc.
Friedel, M. J. and R. E. Thill, StressDeterminationin Rock of the33rdU.S. Symposium onRockMechanics, editedby
Using the Kaiser Effect, Rept. of invest. R1-9286,U.S. J. R. Tillerson and W. R. Wawersik, A. A. Balkema, 1992.
Dept. of the Interior,Bureauof Mines, 1990. Murayama,
S.,Michihiro,K., Saito,J.,Fujiwara,T, Yoshioka,
Goodman,R.E., SubaudibleNoise During Compressionof H. andHata, K., The KaiserEffectof a GraniteUnderVarious
Rocks, Geol. Soc. Amer. Bull., 74, 487-490, 1963. Loadings,in Proc. 7thInt. Acoustic
EmissionSymp.,edited
Hayashi, M., T Kanagawa,S. Hibino, M. Matojima and by M. Onoe,K. Yamaguchi, andH. Takahashi,
TheJapanese
Y. Kitahara, Detectionof AnisotropicGeostresses
Trying Soc. for Non-DestructiveInspections,
1984.
by AcousticEmission,and Non-linearRock Mechanicson Wawersik,W. A. and Brace,W. F., Post-failureBehaviorof a
LargeExcavatingCaverns,Proc. 4th Cong. Int. Soc.for Granite and Diabase,Rock Mechanics,3, 61-85, 1971.
Rock Mechanics, 2, 212-218, 1979. Wawersik,W. R., J. W. Rudnicki,W. A. Olsson,D. J. Hol-
Holcomb,D. J., Using AcousticEmissionsto DetermineIn- combandK. T. Chau,Localizationof Deformationin Brittle
situ Stress:Problemsand Promise,in Geomechanics,edited Rock: Theoreticaland LaboratoryInvestigations,in Mi-
by S. Nemat-Nasser,
pp. 11-21,ASME, New York, 1984. cromechanics
of Failureof Quasi-BrittleMaterials,edited
Holcomb, D. J., Localization Studies Under Triaxial Condi- by S. P.Shah,S. E. Swartz,andM. L. Wang,pp. 115-124,
tions, in Proc. 33rd U.S. Symposiumon Rock Mechanics, ElsevierAppliedScience,1990.
editedby J. R. TillersonandW. R. Wawersik,pp. 661-670, Yoshikawa,
S. andK. Mogi,A New Methodfor Estimation
of
A. A. Balkema, 1992. the CrustalStressFrom Cored Rock Samples:Laboratory
Holcomb, D. J. and R. J. Martin, Determining Peak Stress Studyin theCaseof UniaxialCompression,
Tectonophysics,
HistoryUsingAcousticEmissions,
in Proc.26th U.S. Sym- 74, 323-339, 1981.
posium On Rock Mechanics,edited by E. Ashworth,pp. Yoshikawa,S.andK. Mogi, KaiserEffectof AcousticEmission
715-722, A. A. Balkema, 1985. in Rocks-Influenceof WaterandTemperature Disturbances,
Holcomb,D. J. andL. S. Costin,DetectingDamageSurfacesin Proc. 4thAcousticEmissionSymposium, (Tokyo,1978),Ses-
BrittleMaterialsUsingAcousticEmissions,J. App. Mech., sion7, pp. 21-40.
108, 536-544, 1986.
Hughson,D. R. andCrawford,A.M., Kaisereffectgauging: David J. Holcomb, Geomechanics6117, Sandia National
a newmethodfor determining thepre-existing
in situstress Laboratories,
Albuquerque,
NM 87185.
from an extractedcoreby acousticemissions,
in Proc. Int.
Symp. on Rock Stressand Rock StressMeasurement,pp.
359-367, Lulea: Centek, Stockholm, Sweden, 1986. Received:April 5, 1993;
Kaiser,J., An Investigation
intotheOccurrence
of Noisesin Accepted:May 7, 1993

S-ar putea să vă placă și