Sie sind auf Seite 1von 12

Downloaded 12/06/12 to 192.159.106.200. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.

org/

Angular dependent attenuation in a sandstone


with parallel cracks
Anne Skjcrstein. and ErlingFjcr

ABSTRACT

Laboratorystudieson acousticwavepropagationin syntheticrockswith parallelcracks


showthatthe attenuationis stronglydependenton the incidenceangle.The scatteringwhich
causesthisattenuationalsodistortsthe waveformsandchangesthe frequencycontentof the
signal.The angulardependenceof the attenuationthereforedependssomewhaton the fea-
ture usedas a measureof the signalstrength.
The studyshowsthat underthe conditionsstudiedhere, the attenuationanisotropyis
much more visible than the velocity anisotropy.Most of the measurementsuseda 100 kHz
excitationpulse,which corresponds to a wavelength-to-crack-diameterratio in the range
1.7 - 3.0. In the dry sample,the attenuationanisotropymatchesfairly well theoreticalex-
pectations,and is qualitativelysimilar to the velocityanisotropy.However,the attenuation
anisotropyis muchmorevisible:for the P-wave,the velocityratio Vm/Vm=0.80, while the
amplituderatioAm/Am= 0.01. At higherfrequenciesthe P-wave velocityanisotropyvan-
ishes,while the attenuationanisotropyremainshigh andclearly visible.The effectof water
saturationis a significantreductionin the velocity anisotropyfor the P-wave, while the
attenuationanisotropyremainshigh. The S-wavesare lessaffectedby saturation.

INTRODUCTION

Studieson seismicanisotropyare normally focusedon velocity anisotropy,while at-


tenuationanisotropyattractsmuch less attention.However, the attenuationmay contain
informationaboutthe formations'internalstructurethat is similar and partly complemen-
tary to the velocities,and thereforerepresentsan interestingsourceof information.Liu et
al. (1993) suggestedfor instancethat the main differencein the seismicresponsesof large
and small fractures is the level of attenuation and the waveforms, not the velocities. Further-
more, in AVO processingattenuationanisotropymay have a significantimpacton the re-
sults, and should be corrected for.
Acousticanisotropyis often causedby parallel cracksor fractures.The mechanism
which inducesthis anisotropyis scattering:Acousticwavestravellingthrougha cracked
solidlooseenergyby meansof scattering,andthe magnitudeof thisenergylossis strongly
dependenton the directionof the wave relativeto the orientationof the cracks.

SINTEFPetroleum Research, N-7465 Trondheim, Norway


2NorskHydro ASA, N-0246 Oslo, Norway

205
206 Angular dependent attenuation

We presenta laboratorystudyon attenuationanisotropycausedby parallelcracks.The


experimentalmethodis the sameas usedby Rathoreet al. (1995) for studieson velocity
anisotropy:Acousticwavesare transmittedin variousdirectionsthrougha syntheticsand-
stonesamplecontaininga set of parallel cracks.The cracksare madeduringthe manufac-
Downloaded 12/06/12 to 192.159.106.200. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

turingof the sample,and the crackgeometryis well defined.


To defineattenuationanisotropy,we needa suitableparameteras a measureof signal
strength.However,a single,scalarparameterwhich uniquelydefinessignalstrengthfor an
acousticpulsein a dispersivemediumis hardto find, asthe pulsecontainsmanyfrequency
components,andthe attenuationof eachcomponentmay be different.A parameterwhich is
easilyobtainedfrom the data is the peak-to-peakamplitudeof the initial pulse.We usethis
parameterto obtaina quick-lookestimateof the attenuation.We alsoconsiderthe ampli-
tude at a fixed frequency,derivedby FourierTransformof the signal,to obtaina more well
defined measure of the attenuation.
For the definition of velocities,it is also a problem that there may be a difference
betweenphaseandgroupvelocities.This occursin non-symmetrydirectionsin anisotropic
media (discussedby Dellinger and Vernik, 1994, who concludedthat the velocity deter-
mined from the first break arrival time representsthe phasevelocity), and in dispersive
media(discussed by Rathoreet al., 1993,who proposedthat the firstbreakvelocityrepre-
sentsthe phasevelocityof the fastestfrequencycomponents,while the firstzero cross-over
velocity representsthe phasevelocityof the dominantfrequencycomponentpropagating
throughthe rock).
Velocity anisotropyis often defined as (Vlll{lA..-Vllllll)/Vl,.,f,.,
where V,,,,.,.
and V,,,,,
are the ve-
locitiesin the fastestand slowestdirections,respectively(Crampin, 1989). We shall usea
differentdefinitionof velocityanisotropy,which will allow consistencywith the definition
of velocity and attenuationanisotropy.We define velocity anisotropyAv as

2 2
Vrax- Vrin
Av= 2 (1)
Vrax

By this definition,Av representsthe splittingof the (real part of the) elasticcoefficientp V2


(where p is the density).Correspondingly, we define attenuationanisotropyAmas

AA-ln Amin (2)


wheref is the frequency,x the lengthof the sample,otthe velocity,andA,,,,.and A,,,,,the
amplitudesof the signal in the directionswhere the amplitudeis largestand smallest,re-
spectively.By thisdefinition,Amrepresents the splittingof the imaginarypartof the elastic
coefficientp V2(providedthat the imaginarypart is smallcomparedto the real part).

EXPERIMENTAL CONDITIONS

The syntheticrocksampleusedin thisstudycontains1425circular,parallelcracksper


dm. The diameter is 8.25 mm and the thickness 0.02 mm for all the cracks. The crack
$1<jaerstein
and Fjaer 207

density,definedasthe numberof cracksper unit volumetimesthe cubeof the crackradius,


is 0.1 for this sample.The porosityof the sampleis about35%. The sampleis shapedas a
16-sidedregularprism, with axis parallel to the cracks,to allow for acoustictransmission
measurements in 8 directions,which becauseof symmetryreduceto five different direc-
Downloaded 12/06/12 to 192.159.106.200. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

tionsrelativeto the cracknormaldirection(Figure 1). Eachsideof the prismis 2 cm wide.


A partof the sample(not shownin Figure 1) is keptcrackfree, for referencemeasurements.
Matched pairs of 1 inch diameterbroad band transducers,with a center frequencyat 500
kHz, were usedfor the measurements. The transducers were pressedagainstthe sampleby
meansof a deadweight,which provideda repeatablenormalstressof about0.1 MPa at the
transducer/sample interface.Syrupwas usedas a couplingagentbetweenthe transducers
and the rock samplefor measurements when the samplewas dry, in orderto minimisethe
uncertaintyin the amplitudemeasurements. To reducereflectionsfrom the side-walls,the
samplewas embeddedin a mixture of plasticineand steel spheres,in a ratio adjustedto
makethe acousticimpedancematchthatof the sample.No couplingagent,nor any embed-
ding material(exceptwater) was usedwhen the samplewas saturated.
The studyincludedmeasurements on compressionalwaves(P), shear-wavespolarised
parallelto the sampleaxis and henceparallelto the cracks(S,), and shear-wavespolarised
normalto the sampleaxis and thuspartly normalto the cracks(SO. For excitation,a single
cyclepulsewith a centralfrequencyat 100kHz was usedfor mostof the measurements. For
two minorsetsof measurements, 250 kHz singlecycle and500 kHz five cycle pulseswere
used,respectively.

RESULTS

Dry sample
Examplesof the recordedsignalson the dry sample,usinga 100 kHz excitationpulse,
are shownin Figure 2. The upperpart of the figure showsthe P-wave signalsas recorded
(with true amplitudes),while the lower partof the figureshowsthe samesignalswith indi-
viduallyscaledamplitudessothatthe waveformsareclearlyvisible.The directionof propa-

DISC PLANES

ANSDUCERS

INCIDENCE CRACK-VOIDS
ANGLE, 0

6 SIDED REGULAR
PRISM

Fro. 1. Geometry of the synthetic sandstonesample.


208 Angular dependent attenuation

gation is defined by the incidenceangle 0, which is the angle betweenthe direction of


propagationand the cracknormal(Figure 1).
The velocitiesfor the P-, S,- and S_-waves
versusthe incidenceangle- measuredon
the dry sample- are shownin Figure 3. For eachanglethere are two equivalentdirections
Downloaded 12/06/12 to 192.159.106.200. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

(0 and -0, seeFigure 1), which are both shownin orderto give an indicationof the scatterin
the data. First zero cross-over(FZC) was usedas the referencepoint in the waveformsto
definethe travel-times.As mentionedabove,the FZC velocityis assumedto representthe
phasevelocityof the dominantfrequencyin the receivedsignal(Rathoreet al., 1993). The
velocityanisotropycoefficientAv for the threewaves,calculatedaccordingto equation(1),
is presentedin Table 1.

67.5

45

22.5

! I I I
0.00 0.05 o. 10 o. 15 0.20

Time [ms]

0.00 0.05 0.10 0.15 0.20

Time [ms]

Fro. 2. ReceivedP-wave signalsin the dry sample. Upper part' true amplitudes, lower part:
scaledamplitudes. The incidenceanglesare shown on the left.
$t<jaersteinand Fjaer 209

The peak-to-peak amplitudeof thefirstcycleis shownin Figure4 for all threewaves.


The attenuationanisotropyis seento be pronouncedboth for the P-wave and the S,-wave,
while theanisotropyfor the S,-wave is muchless.Qualitatively,theangulardependence of
theamplitudes resembles closelytheangulardependence of thevelocities(Figure3). The
Downloaded 12/06/12 to 192.159.106.200. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

anisotropycoefficient,A, calculatedaccordingto equation(2) is shownin Table 1. The


centerfrequencyof theemittedpulseis usedforf in thesecalculations, whilethevelocity
measuredin thecrack-freepartof the sampleis usedto representor. The signalamplitude
at 100 kHz, obtainedby FourierTransformanalysisof the receivedsignal,is shownin
Figure5. The anisotropy is seento bequalitativelysimilarto theanisotropy for thepeak-to-
peakamplitude(Figure4). The anisotropy coefficientAnis presented in Table1.

3000

25OO

2000

1500

1 ooo i i i i i i
15 30 45 60 75 90

Incidence Angle []

Fro. 3. Velocitiesversusincidenceangle for the dry sample.The linesare drawn only as guides
for the eye. - P-wave, El = &-wave,A = Sz-wave.

Table 1. The anisotropycoefficients,obtained from measurementswith 100 kHz excitation


pulse.

Dry Saturated

P Sii S _ P Sii S _

A4 (peak-to-peak) 0.36 0.12 0.10 0.20 0.10 0.03

AA( 100 kHz) 0.37 0.15 0.05 0.35 0.22 0.07


AA(Theory) 0.31 0.09 0.03 0.03 0.09 0.09
Av 0.36 0.23 0.06 0.06 0.23 0.09
210 Angular dependent attenuation
1.4

1.2
Downloaded 12/06/12 to 192.159.106.200. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

0.8

0.6

E 0.4

0.2

I I
0 20 40 60 80 100

Incidence Angle []

Fro. 4. Peak-to-peak amplitudes versusincidenceangle for the dry sample.The lines are drawn
only as guidesfor the eye. - P-wave, FI - S,-wave, 6 - Sz-wave.

800 --

600 -

400 -

200 -

I I T I I I

0 15 30 45 60 75 90

Incidence Angle []

Fro. 5. Amplitudes at 100 kHz (obtained by Fourier Transform analyses) versus incidence
angle for the dry sample. The lines are drawn only as guidesfor the eye.
- P-wave, FI- S,-wave, A- S-wave.
$1<larsteinand Fjaer 211

Saturated sample
The samesamplewas saturatedwith water and the measurementswere repeated.Fig-
ure 6 showsthe recordedP-wave signalsin the saturatedsample.The lower part of the
figureshowsthatthewaveformsandarrivaltimesarenowquitesimilarfor all directions,as
Downloaded 12/06/12 to 192.159.106.200. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

comparedto the dry case(Figure2), hencethe velocityanisotropyhasalmostdisappeared.


However, the attenuationanisotropystill remains,as the upper part of Figure 6 shows.
These observationsare reflected in the anisotropycoefficients,which are presentedin
Table 1. For the S-wavesthe differencesbetweenthe dry andsaturatedcasesare smallerand
less consistent.

90

67.5

45

22.5

! I I I
0.00 0.05 o. 10 o. 15 0.20

Time [ms]

! I I I
0.00 0.05 0.10 0.15 0.20

Time [ms]

Fro. 6. Received P-wave signals in the saturated sample. Upper part' true amplitudes, lower
part' scaledamplitudes. The incidenceanglesare shown on the left.
212 Angular dependent attenuation

Higher frequencies
The resultspresentedso far were obtainedfrom measurements performedwith a 100
kHz excitationpulse.AdditionalP-wave measurements were performedon the dry sample
with 250 kHz and500 kHz excitationpulses,respectively.For the measurements
performed
Downloaded 12/06/12 to 192.159.106.200. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

at 500 kHz, a five cyclepulsewasusedto forcemorehigh frequencies throughthe sample.


Examplesof the receivedsignalsare presentedin Figure7. All the signalsin the plot are
scaled.The resultsat 100 kHz are the sameas presentedin Figure2; they are includedfor
comparison.The receivedsignalsin the crack parallel directionare almostidenticalto the
respectiveexcitationpulses.The signalsin the cracknormaldirectionareon the otherhand
nearlyidentical,regardlessthe excitationpulse.The anisotropycoefficientAn, asobtained
by FourierTransformanalysis,is presentedin Table2 for the threefrequencies.

J 9o
o 500kHz

0o
J 4s & / \ /"-'x

I i
0.00 0.05 0.10 0.15 0.20

Time [ms]

Fro. 7. Received P-wave signals using three different excitation frequencies.Scaled ampli-
tudes. The incidence angles are shown on the left.

Table 2. Attenuation coefficient A, for the P-wave


measured on dry sample, derived from Fourier
Transform amplitudes.

100 kHz 250 kHz 500 kHz

A 0.37 0.25 0.12


$1<jaersteinand Fjaer 213

DISCUSSION

The observedanisotropyin the dry sampleis seento be qualitativelyconsistentwith


what we would expectin a materialwith parallel cracks(see for instanceHudson, 1981).
Downloaded 12/06/12 to 192.159.106.200. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

The P- and the S,-wavesare slowestand mostattenuatedin the crack-normaldirection,and


fastestand leastattenuatedin the crack-paralleldirection.The S_-wave anisotropyis much
smaller,as expected,but is not properlyresolvablein thesemeasurements. The attenuation
anisotropythuscarriesmuchof the sameinformationaboutthe crackgeometryas the ve-
locity anisotropy.However,the attenuationanisotropyis muchmore visible:while the ratio
V,,,ff,,,,,c
is 0.80 for the P-wave, the ratio A,,,ffA,,,,for the samewave is 0.01.
Notice that for intermediateincidenceangles(in particulararound45) there is an
offsetin thedirectionof energypropagationdueto theanisotropy, sothata largepartof the
wave energywill not reachthe receivingtransducer(nicely illustratedby Dellinger and
Vernik, 1994). Hencethe recordedamplitudesfor thesedirectionsare probablysomewhat
too low.

The sourceof the attenuationanisotropyis Rayleighscatteringon the cracks.Hudson


( 1981)presenteda theory['orattenuationcausedby scatteringon parallelcracks.Although
Hudsonhasquestionedthe validity of this low order theory in later works (Hudson, 1986,
1990), we wish to testour resultson this theorysincewe do not havea betteralternativeat
hand.The crack geometryis fully known in our sample,hencethe theoreticalattenuation
can be calculatedwithoutany adjustableparameters.Hudsonpresentedhis resultsin terms
of the attenuationcoefficient(0),which relatesto our anisotropycoefficientAAas

(3)

The theoryis basedon the assumptionthat the wavelength-to-crack-diameter ratio is


large and that the crack densityis low. In our experimentsat 100 kHz, this ratio is in the
range2.4 - 3.0 for the P-wave, and in the range 1.7 - 1.8 for the S-waves,which is really on
the vergeof the validityrangefor the theory.The theoreticalresultsare shownin Table 1, in
comparisonwith the experimentalresults.Our resultsshowthatthe theoreticalpredictions
andtheexperimentalresultsfor theattenuationanisotropyagreesquitewell for thedry rock
at this frequency.However,other studieshavereporteddisagreementbetweenthis theory
and observations(Hudson, 1990; Ass'ad and Ebrom, 1994).
Attenuationcausedby Rayleighscatteringis stronglydependenton the wavelength-to-
crack-sizeratio, and hencestronglydependenton the frequencyfor a given crack size.
Accordingto Hudson'sformalism,is proportionaltof4 and thusAA ocf. We theretore
expectthatthe attenuationanisotropywill decreaserapidlywith decreasingfrequency.Un-
fortunately,we did nothavethe meansto testlower frequenciesin thisstudy,becauseof the
samplesize. The velocity anisotropyis not equally dependenton frequencyin the high
wavelength-to-crack-size regime,thusit is to be expectedthatthe largedifferencebetween
the Vmi,?max
andtheA,,,,,/A,,,,
ratiosreportedabovewill decreasewith decreasing frequency.
At the higher frequencies(250 kHz and 500 kHz) the wavelength-to-crack-diameter
ratio becomesequalto or lessthan 1, and the theoryis definitivelynot valid. The experi-
mental results(Table 2) indicatehoweverthat theA,,ffA,,, ratio remainslow and moreor
214 Angular dependent attenuation

lessconstantat thesefrequencies.It is more difficult to quantifyhow the velocity anisot-


ropychangeswith the frequency,becausethe signalsin the cracknormaldirectionare more
or less independentof the excitationpulse, and hence the FZC method for arrival time
pickingdoesnot give new informationat the higherfrequencies.Determinationof arrival
Downloaded 12/06/12 to 192.159.106.200. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

timesfrom the firstbreak is assumedto give velocitiesat higherfrequencies(Rathoreet al.,


1993), but the actualfrequenciesfor which thesevelocitiesare valid are not well defined.
However,accordingto the first breakarrival times,the velocityanisotropyvanishesat the
higherfrequencies,as Figure7 shows.This indicatesa differencebetweenvelocityanisot-
ropy and attenuationanisotropyat the higherfrequencies,which may be explainedas fol-
lows: At sufficiently high frequencies,the wavelengthis shortcomparedto the distance
betweenthe cracks,and partsof the wavefrontmay passbetweenthe crackswithoutbeing
affectedby them.Thus the fastestpropagatingpart of the waveenergywill be lessaffected
by the cracks,and the anisotropyvanishes.Other partsof the wavefrontwill on the other
handstill be scatteredby the cracks- moreor lessdependingon the orientationof the cracks
relativeto the directionof the propagatingwave - thusthe attenuationanisotropystill exists
at the higher frequencies.
Table I showsthat saturationlargely reducesthe velocity anisotropyfor the P-wave.
This resultdiffers from previouslypresentedresults(Rathoreet al., 1995), where a signifi-
cantly lower reductionin velocity anisotropydue to saturationwasseen,in accordancewith
the theoryof Thomsen(1995). The discrepancyis mostlikely due to the differencein crack
size:the resultspresentedby Rathoreet al. (199.5)wereperformedon a samplewith smaller
cracks,wherethe conditionsfor pore pressureequalisationare different.Unlike the veloc-
ity anisotropy,the attenuationanisotropyis not stronglyaffectedby saturation(Table 1).
This result is not consistentwith theoreticalexpectationsbasedon Hudson'stheory, but
appearsas a reasonableresultif pore pressureequalisationis assumed.Furtherstudiesare
neededto confirm whetherthe resultis generallyvalid, or is a specialresultdue to marginal
conditions.

Comparingthe estimatesof attenuationanisotropyin Table 1, it is seenthat the peak-


to-peakamplitudeappearsto give reasonableestimatesof the attenuationanisotropyin the
centralfrequencyrange,comparedto the moreaccurateamplitudesderivedby FourierTrans-
forms.However,due to the uncertaintiesin frequency,and couplingto dispersioneffects,
the peak-to-peakamplitudecan only be usedas a quick-lookmethod.

CONCLUSIONS

Measurementsperformedon the dry sampleat a frequencywhere the wavelength-to-


crack-sizeratio is in the range 1.7 - 3.0, showthat parallelcracksproducea strongattenua-
tionanisotropy.This attenuationanisotropyisqualitativelycomparableto thevelocityanisot-
ropy in the samefrequencyrange,but the attenuationanisotropyis muchmore visible:for
the P-wave, the velocity ratio V,,,,,,,,,.
= 0.80, while the amplituderatioA,,ffA,,,,= 0.0 l.
The attenuationanisotropyis found to be consistentwith the theoreticalformulationsof
Hudson(1981), which describeattenuationdue to scatteringfrom parallelcracks.
At higherfrequencies,wherethe wavelength-to-crack-diameter ratio is about 1 or less,
the velocityanisotropyvanishes,while the attenuationanisotropyremainshigh andclearly
visible.This vanishingvelocityanisotropymay be explainedin termsof the possibilityfor
$kjaerstein and Fiaer 21.5

partsof the wavefrontto passin betweenthe crackswithoutbeingaffectedby them,when


the wavelengthis sufficientlysmall. Sinceother partsof the wavefrontare still being scat-
teredby the cracks,the attenuationanisotropyis not equally affected.
The attenuationanisotropyfor the P-wavesis observedto remainhigh after saturation,
Downloaded 12/06/12 to 192.159.106.200. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

while the velocity anisotropyis largely reduced.The S-wavesare lessaffectedby satura-


tion, both with respectto velocity anisotropyand attenuationanisotropy.

AC KNOWL EDG EMENTS

This work wassupportedfinanciallyby Amoco Norway Oil Co., ConocoNorway Inc.


and Elf PetroleumNorge A/S.

REFERENCES

Ass'ad, J. M. and Ebrom, D. A., 1994, Apparentattenuationin azimutally anisotropicme-


dium: a comparisonbetweenexperimentand theory:Sixth InternationalWorkshopon
SeismicAnisotropy(6IWSA), Trondheim,Abstracts,paperP 1-7.
Crampin,S., 1989,Suggestions for a consistent
terminologyfor seismicanisotropy: Geophys.
Prosp.,37, 753-770.
Dellinger, J. and Vernik, L., 1994, Do traveltimesin pulse-transmission experimentsyield
anisotropicgroupor phasevelocities?:Geophysics,59, 1774-1779.
Hudson,J. A., 1981,Wave speedsand attenuationof elasticwavesin materialcontaining
cracks:Geophys.J. R. Astr. Soc., 64, 133-150.
1986,A higherorderapproximationto thewavepropagationconstantsfor a cracked
solid:Geophys.J. R. Astr. Soc.,87, 265-274.
1990, Attenuation due to second-orderscatteringin material containing cracks:
Geophys.J. Int., 102, 485-490.
Liu, E., Crampin,S., Queen,J. H., and Rizer, W. D., 1993, Velocityand attenuationanisot-
ropy causedby microcracksand macrofracturesin a multiazimuthreverseVSP: Can. J.
Exp. Geophys.,29, 177-188.
Rathore,J. S., Fjaer, E., Renlie, L., and Nysaeter,J., 1993, Estimationof phasevelocitiesin
crackedrocks' 55 Conf. Eur. Assn. Expl. Geophys.,Stavanger,ExpandedAbstracts,
paper P 124.
Rathore,J. S., Fjaer, E., Holt, R. M., and Renlie, L., 1995, P-and S-wave anisotropyof a
syntheticsandstonewith controlledcrack geometry:Geophys.Prosp.,43, 711-728.
Thomsen,L., 1995,Elasticanisotropydueto alignedcracksin porousrock:Geophys.Prosp.,
43, 805-829.
Downloaded 12/06/12 to 192.159.106.200. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Das könnte Ihnen auch gefallen