Anisotropy can be thought of as directional

inequality - the variation of properties with

direction. Anisotropy occurs at all scales from

core plugs to reservoirs. Until recently, anisotropy

was ignored - or treated as noise to be removed

from reservoir models. Recent research has

changed this perception. Modern techniques and

tools can detect and use the small variations in

velocity which are found in sonic and seismic studies.

Anisotropy can make a huge difference to reservoir

performance. Variations between horizontal and vertical

permeability and the presence of low-permeability

barriers can turn a carefully planned and executed

waterflood or gasflood into a production lottery, with early

and unpredictable water or gas breakthrough.

In this article, Bruce Cassell, Mahmood Akbar and Roy Nurmi

examine the origins of anisotropy in the reservoir and describe

the advanced interpretation techniques which are helping to

bring anisotropy into the mainstream of

reservoir characterization.

44 Middle East Well Evaluation Review

Until recently, geoscientists usingseismic methods to survey sedi-mentary sequences assumed that

waves travelled through rock at thesame speed in all directions. However,this assumption is now known to be in-accurate, because rock sequences arevariable. While it might be convenient toassume that a reservoir is basicallyhomogeneous this rarely proves to bethe case.

This phenomenon, where certainphysical properties vary with direction,is called anisotropy. It is caused by analignment of crystals, bedding planes,joints or fractures at a scale which issmaller than the wavelength of the ultra-sonic, sonic or seismic waves passingthrough the rock.

Despite field and laboratory experi-ments confirming the presence ofanisotropy, it has usually been neglectedby exploration and production geophysi-cists. The velocity differences are smalland were usually treated as noise to beremoved from the seismic image signal.However, as a result of recent advancesin acquisition, processing and interpreta-tion some highly anisotropic velocitieshave recently been recorded in ultra-sonic, sonic and seismic data.

Anisotropy, whether on a small orlarge scale, is the rule in oil and gasreservoirs. Only by understanding moreabout the ways in which mechanical andstress-related properties vary betweenand within layers can geoscientistsimprove their reservoir models.

Compression or shear?

There are two main types of wavemotion: compressional waves and shearwaves. The difference between them canbe illustrated by considering thebehaviour of a single rock particle. Incompressional waves, particles moveparallel to the direction of wave propa-gation, whereas in shear waves particlemotion is perpendicular to propagation(figure 3.1). There are two main types ofanisotropy.• TIV (transverse isotropy with a verticalaxis of symmetry), where vertical veloc-ity differs from horizontal velocity. This iscaused by differences between adjacentrock strata or by fine sedimentary layer-ing as found in shales (figure 3.2a).• TIH (transverse isotropy with a hori-zontal axis of symmetry), where verticalvelocities are identical in a vertical planeparallel to the fractures, but the horizon-tal velocities are different. TIH is usuallycaused by stress-induced micro-fracturesor other discontinuities in the formation(figure 3.2b).

Maximumhorizontal stress

Minimumhorizontal stress

Vertical stressFast

Slow

(a) (b) Fig. 3.3: Shear wavespropagating throughan isotropic rockmass (a) do sowithout modification.Shear waves passingthrough a layeredmass (b) will splitinto two separatecomponents - a fastwave and a slowwave. Thesecomponents alignthemselves with themaximum andminimum horizontalstress directions.

Fig. 3.2: There aretwo basic types ofanisotropy. TIV (a) iscaused by the finelayering typicallyfound in shales.Micro-fracturesproduce TIH (b)where the horizontalvelocities aredifferent.

Shale Fractures(a) (b)

Compressional

Fast shear

Slow shear

Wave propagation direction

(a)

(b)

(c)

Fig. 3.1: Particlemotion reveals thedifference betweencompressional andshear waves. Incompressional waves(a) particle motion isparallel to the wavepropagationdirection; in shearwaves (b and c)particles oscillate in adirectionperpendicular to thedirection of wavepropagation.

45Number 18, 1997

waves split into two separate com-ponents (figure 3.3). These propagate atdifferent slownesses (the fast and slowwaves) and have orthogonal particle-motion properties (i.e. the two wavesvibrate the rock in directions which are90° to one another) parallel to the direc-tions of maximum and minimum hori-zontal stress.

The origins of anisotropy

Anisotropy in oil and gas reservoirsstarts to develop during the depositionand diagenesis of the sediments.Sandstones can develop anisotropic fea-tures during and after depositionwhereas in carbonate rocks anisotropy isgenerally caused by post-depositionalevents such as fracturing or diagenesis.Sandstone anisotropy generally devel-ops where there is an ‘ordering’ of sedi-ment grains. This typically occurs in oneof two ways:• an alternation of sediment typesdeposited sequentially as bedding(figure 3.4)• alignment of individual grains causedby directional deposition (this generallyreflects the predominant water or winddirection at time of deposition) or dia-genetic effects (figure 3.5) .

Fracturing forces

Fracturing is a major source ofanisotropy, particularly in carbonatesequences. Whether open or filled withporous breccia, the properties of fracturezones are very different to those of thesurrounding rock. This fact, and thedirectional nature of fracture sets, makesthem important sources of anisotropy atevery scale from core plug to reservoir.

Mineralized fractures have the samedirectional controls, but may be quitesimilar to the surrounding rock and willproduce less obvious anisotropic effectsthan open fractures.

An additional complication withanisotropy is that it varies with scale aswell as direction. For example, a singlecrystal may have an atomic structurethat is anisotropic for electric currentflow or acoustic propagation, while apiece of rock formed from a randomlypacked group of the same crystals mightbe isotropic for the same properties atthe larger scale.

So the detection of anisotropydepends on the technique being usedand the scale of investigation.

Fig. 3.5: The clay grains in this sediment have a random distribution at deposition. However,as the weight of accumulating sediment increases in the early stages of diagenesis, thegrains are rotated to develop a rock fabric which exhibits elastic anisotropy.

Fig. 3.4: Normalsedimentary layeringand associatedprocesses lead toanisotropy. Theirregular layering inthe siltstone (a) andthe more regularlayers in thesandstone (b) reflectvariations indepositionalenvironment, originalmineralogy and grainsize or diageneticevents. All of thesefactors will combineto give these rocksvery different degreesof anisotropy.

In a borehole, compressional wavesalways propagate along the length of thehole (i.e. in the direction of the longaxis), whether the borehole is horizontalor vertical. For this reason a sonic log-ging method designed to measureanisotropy using compressional wavesalone would require a number of bore-holes with different orientations in thezone of interest. In a field containing only

vertical wells compressional acousticanisotropy cannot be assessed as mea-surements are in the vertical directiononly. Compressional waves do not splitin anisotropic formations, so for anygiven well deviation there is only one logvalue for a compressional wave.

Shear waves are different. For anysub-vertical direction of propagationthrough an anisotropic layer, shear

(a)

(b)

46 Middle East Well Evaluation Review

Rotation angle Offline energyof dipole source and receiver

0°Minimum

90°Minimum

45°Maximum

Fast

Slo

w

Fig. 3.7: The orientation of the dipole transmitters and receivers, with respect to thefast and slow shear planes, controls the energy level recorded at the offline sensor.When the transmitters and receivers are aligned with the fast and slow planes theenergy recorded in the offlines is at a minimum. When they are at 45° to theseplanes offline energy reaches its maximum value.

Fig. 3.6: The DSI toolfires its shear sonicpulse alternately fromtwo perpendiculartransmitters to anarray of receivers.The pulse splits intotwo components andthe shear wavefield isrecorded. Usingamplitude and travel-time difference data,the operator canidentify the fast shearwave direction whichcorresponds to theorientation of anyaligned fractures orthe maximumhorizontal stress.

FastS wave

SlowS wave

Sourcepulse

Dipole source

Dipole receivers

Fast, slow or in between?

The Dipole Shear Sonic Imager (DSI*)tool makes dipole measurements inorthogonal directions, detecting theshear wave and allowing the measure-ment of sonic-scale anisotropy in fastand slow formations (figure 3.6).

If one of the DSI tool's dipole trans-ducers is aligned with the fast-sheardirection, fast-shear waves are logged inthat plane and slow-shear waves in theorthogonal plane. When, as is usually thecase, the dipole transducers are alignedsomewhere between fast and slow direc-tions, both sets of shear waves will splitinto fast and slow components and willbe recorded by the inline and offlinereceivers. Inline implies that the trans-ducer and receivers are in the sameplane; offline means that the receiversare measuring in the plane orthogonal tothe transducer.

Offline energy changes as the orient-ation of the tool changes. It is at a mini-mum when the transducers are alignedwith the fast or slow shear directions andreaches a maximum when the transduc-ers are arranged at 45° to these fast andslow shear planes (figure 3.7)

There are two major applications ofsonic anisotropy measurements:• assessment of stress directions whichreveal the orientation of faults and frac-tures (prior to drilling horizontal wells).• determination of mechanical rockproperties by using the two horizontalPoisson’s ratios.

On the right line

At present, dipole shear logging is usedmainly to obtain shear logs in slow for-mations where conventional monopoleshear measurements are not possible.Dipole logging provides the directionalmeasurements which monopole loggingcannot. Two dipole transducers aremounted orthogonally around the tooland can excite shear waves in these dif-ferent directions.

Inline measurements are made at thereceiver whose polarization directioncorresponds to that in the transmitter(i.e. the receiver operates in the sameplane as the transmitter).

Conversely, offline measurements aremade at receivers polarized at 90° to thesignal. If there is no azimuthal variationin stress, shear waves generated in oneplane will not split and will, therefore,not be recorded in the offline.

47Number 18, 1997

3

Vertical velocity (km/sec)

2

1

-1

-2

-3

-4 -2 2 4

qP

qS1

Horizontal velocity (km/sec)

0

0

Fig. 3.8: Velocityvariation plottedaround an axis ofsymmetry for ashale, showingwavefront velocitiesfor compressional(qP) and shear (qS1)waves. The ellipticalmodel of anelasticanisotropy had to bemodified becausethe relationshipbetween velocities atdifferent angles forshear waves is notelliptical but‘anelliptical’. Tickmarks on theparticles indicate thedirections of particlemotion.

Fig. 3.9: Non-uniform compressive stress operating on microcracks. In unstressed rockswith randomly oriented microcracks (a) the cracks may be open whatever theirorientations. However, when stress is applied (b) cracks perpendicular to the direction ofmaximum compressional stress will close, while cracks parallel to it will remain open.

WAVEFRONTS AND VELOCITY VARIATIONSIf there is an orderly arrangement ofcrystals, fractures, grains, joints or bed-ding planes on a scale smaller than thelength of the incident wave, then wavevelocity will vary with direction.

Laboratory and field experiments inthe 1950s detected velocity anisotropy:vertically and horizontally propagatingwaves were found to have differentvelocities.

However, the directional velocityvariations produced by anisotropiceffects were so small (typically less than5% in standard surface seismic measure-ment techniques) that they were usu-ally ignored. The idea that wavespropagated through a rock at equalrates in all directions was a standard oil-field simplification for many years.

Early assessments of anisotropy pre-dicted an elliptical relationship forwaves travelling at angles between the

horizontal and vertical. Ellipses wereapplicable because, once the horizontaland vertical velocities were known,velocities at intermediate angles couldbe computed easily. Later laboratoryand field experiments aimed at quantify-ing anisotropy continued to measurevelocities parallel and perpendicular toperceived alignments and many olderpublications list anisotropies of differentrock types in terms of the percentage ofdifference between the fast and slowvelocities or the ellipticity.

Unfortunately, an ellipse does notreflect the true complexity of anisotropicrocks. Experiments have established thatthe relationship for shear waves is notelliptical but a squarish non-ellipse (fig-ure 3.8).

The elastic properties of a rock can beused to correlate with other propertiessuch as lithology or porosity. Most geo-

scientists would say that when the den-sity and P-wave and S-wave velocitieshave been established the rock is com-pletely described; but this is correctonly for isotropic rocks where veloci-ties do not vary with direction.

Because most oil industry researchhas focused on reservoir rocks (whichare usually relatively isotropic sand-stones or carbonates) anisotropy hasnot had a major influence on reservoircharacterization.

However, most of the rocks around areservoir are anisotropic shales. In themost extreme case 21 numbers wouldbe required to characterize a rock, andeven in the simple anisotropic rocksdescribed in figure 3.2, five velocities(two transversely polarized S, verticalP, horizontal P, P at 45°) plus densitywould be necessary.

Change of a stress

Taking cap-rock variation into accountcan be very important when explo-rationists are trying to assess target for-mations. If an Amplitude Versus Offset(AVO) survey does not take account ofthe transverse isotropy (TIV) in a shalecap rock, the underlying gas-bearingsandstone may be overlooked becausethe modelled AVO curve (for an oilsand overlain by an anisotropic shale)would not fit the observed AVOresponse from the survey.

Recognizing that rocks areanisotropic, or may become so understress, is important when evaluating therelationships between the velocity atwhich seismic waves propagate througha rock and its reservoir properties.

While most reservoirs are composedof relatively isotropic sandstones orcarbonates, their properties may bemodified by stress (figure 3.9). Non-uni-form compressive stress will have amajor affect on randomly distributedmicrocracks in a reservoir. When therock is unstressed all of the cracks maybe open, However, compressionalstresses will close cracks oriented per-pendicular to the direction of maximumcompressive stress, while cracks para-llel to the stress direction will remainopen. Elastic waves passing through thestressed rock will travel faster acrossthe closed cracks (parallel to maximumstress) than across the open ones.

(a) (b)

48 Middle East Well Evaluation Review

Waveform processingprinciples

The DSI tool gathers a mass of sonic datathat can be processed and presentedusing IMPACT* (Integrated MechanicalProperties Analysis ComputationTechnique). Having completed a statisti-cal analysis of the acoustic anisotropydataset, the operator can generate theo-retical results for any orientation of thetool within the borehole. This avoids thepotential difficulties of turning the toolwith sufficient precision to measure theoffline minimum value.

When the dipole source is aligned withthe fast- or slow-shear polarization planesof the rock layer the offline energy (orcross-components) becomes zero.Consequently, one way to determine thefast- and slow-shear directions at eachdepth is to rotate the data mathematicallyto find the angle which minimizes energyin the offline readings.

The logs generated by the DSI toolshow which parts of the section areanisotropic and give an indication of dataquality throughout the logged interval.

The basis of waveform processinginvolves geometrical (mathematical)rotation of the inline and offline wave-forms. Seven of these overlap at eachrecording depth so there is consider-able data redundancy. After the theoret-ical rotation the offline is minimizedand, i f there is indeed azimuthalanisotropy and no other noise present,the offline energy will have zero ampli-tude. The amount of rotation requiredto minimize the offline energy indicateshow much the DSI tool would have tobe rotated in order to align thereceivers parallel to the anisotropic orfractured system. As the absolute orien-tation of the tool is known, it is possibleto display the anisotropy direction withrespect to true north.

Anisotropy is often observed in indi-vidual layers so it is very important todetermine whether or not it is actuallypresent before proceeding to interpretthe results. Figures 3.10 and 3.11 are fromthe central Arabian Gulf while figure 3.12is from an Egyptian field where the wellpenetrates fractured basement rock.

The green area in the left track of eachfigure is the most important information.It allows the interpreter to determinewhether or not the other results arebased on the observation of anisotropy.The left edge of the green shaded areashows the effect of offline amplitude mini-mization. If the offline energy is not pro-duced by anisotropy it will not becomezero. This indicates that other effects (e.g.borehole rugosity or altered zones) havegenerated the amplitude anomaly andfurther indicators in these zones aremeaningless.

x160

x170

x180

x190

x200

x210

x220

1:200(ft)

100

5025090-90

502500

1000

MinEne%

MaxEne%

Off Ene

Fast shear azimuth

Azimuth uncertainty

(deg) (µs/ft) 30001000 (µs)

30001000 (µs)(µs/ft)

DT fast shear

1000

DT based anisotropyProcessing window

Fast shear waveforms

Slow shear waveformsDT slow shear

x180

x190

x200

x210

x220

x230

x240

1:200(ft)

100

5025090-90

502500

1000

MinEne%

MaxEne%

Off Ene

Fast shear azimuth

Azimuth uncertainty

(deg) (µs/ft) 30001000 (µs)

30001000 (µs)(µs/ft)

DT fast shear

Processing window

Fast shear waveforms

Slow shear waveformsDT slow shear

Fig. 3.10: The left track (green area) is the most important part of the log, indicating whether or notthe results are due to anisotropy. This example indicates a general ENE stress direction withazimuthal anisotropy values up to 15%.

Fig. 3.11: This log shows one zone of major anisotropy between x180 and x200.

49Number 18, 1997

x430

x440

x450

x460

x470

x480

x490

x300

x310

x320

x330

1:200(ft)

100

5025090-90

502500

1000

MinEne%

MaxEne%

Off Ene

Fast shear azimuth

Azimuth uncertainty

(deg) (µs/ft) 39001100 (µs)

39001100 (µs)(µs/ft)

DT fast shear

Processing window

Fast shear waveforms

Slow shear waveformsDT slow shear

1000

DT based anisotropy

Similarly, the right edge of the greentrack shows how much anisotropy ispresent. It maximizes in intervals ofmeasurable anisotropy. The secondtrack shows the direction of the fastshear (stress direction). This is usuallydisplayed from west (-90) to east (90).The third track contains the fast andslow shear slownesses which havebeen determined by processing (redand blue curves). When there is

anisotropy these curves separate andthe amount of slowness difference islinked to the percentage of anisotropy.The curve on the left edge of the thirdtrack indicates the quantitative time-based anisotropy. Other indicators canbe displayed as an interpretation aid.The fourth track contains the fast andslow waveforms after processing. In theanisotropic zones, i.e. where the leftedge of the green area in the first track

Fig. 3.12: Fracturedbasement rock inEgypt. This examplecontains two zones ofdifferent stressdirections or fracturealignment - northwestand due north. Theanisotropy in the topzone is small butmeasurable.

minimizes and the right edge maxi-mizes, the waveforms separate. The firsttwo examples show a general ENEstress direction, with azimuthalanisotropy values up to 15%. The frac-tured basement example (figure 3.12)contains two zones of different stressdirections or fracture alignment, north-west and north. The degree ofanisotropy in the top zone, althoughmeasurable, is small.

50 Middle East Well Evaluation Review

Surface shear

The shear anisotropy technique is notconfined to the borehole. A similarapproach has been attempted usingborehole seismic methods (figure 3.13).The surface system has severe limita-tions and complications in that thedowngoing seismic wave can beaffected by one or more shallowanisotropic layers which split the signalinto fast and slow components before itreaches the target layer. To counteractthis problem, the effects of the shallowlayers have to be ‘stripped’ from thesignal. This technique is called ‘AlfordRotation’. A shear wave movingthrough the first anisotropic layer willsplit into fast and slow waves. Each ofthese waves will, in turn, act as an inde-pendent source, splitting into fast andslow waves at the next interface, and soon (figure 3.14). A layer-stripping proce-dure is necessary to remove the timedelay between the two ‘sources’ whenthey reach the interface, effectively pre-dicting the effect the interface wouldhave if it were at the surface. DipoleShear Logging techniques, using the DSItool downhole, eliminate this process-ing problem because they consist of in-situ measurements within layers.

Assessing stress

In-situ stress has a major influence onthe permeability and morphology offractures. Fractures which strike per-pendicular to the maximum horizontalstress are likely to be closed and,clearly, have a lower permeability thanopen fractures which are oriented par-allel to the maximum stress.

Accurate information about in-situstress is extremely helpful in predictinginjectivity of natural fractures, particu-larly for acidizing, waterflooding and gas-injection projects. Advance modelling for‘frac’ jobs in vertical and horizontal wellsrequires a prior knowledge of in-situstresses to obtain the best results.

Borehole failures in horizontal wellsare costly and time-consuming but theycan be avoided if the in-situ stressesare known and the orientation of theborehole adjusted accordingly.Horizontal boreholes drilled parallel tothe maximum horizontal stress direc-tion are unlikely to be distorted.

Fig. 3.13: Boreholeseismic surveyscan use shearwave splitting toinvestigatereservoir layers.However,anisotropic rocklayers close to thesurface distort theshear waves beforethey reach the targetlayer. Geophysicistsmust take account ofthis effect andcorrect for it by usinga layer-strippingtechnique.

Fig. 3.14: The complexity of wave splitting increases with depth. Theseparate fast and slow waves produced in the shallow anisotropic layereach split within the next layer, giving a total of four waves from a singlesource. These effects must be countered by layer stripping.

Surface

Lag fromupper layer

S2

S1

S'2

S'2

S'1

S'1

Upperanisotropicrock

Loweranisotropicrock

Waves sourced by S2 Waves sourced by S1

Upper naturalcoordinate frame

Lower naturalcoordinate frame

S2 S1

S'2S'1

Source

H1

S2

S1

H2

σH

Well

Source

Surface

Anisotropicrock

3-componentdownhole

receiver

FastS wave

Natural coordinateframe for

vertical rays

PS1

S2

σH

σH

SlowS wave

R.M. Alford (1986) Shear data in the presence of azimuthalanisotropy. 56th Ann. International Meeting Soc. Explor.Geophys. Expanded Abstracts, pp 476 - 479.

51Number 18, 1997

Iran

Qatar

Zagros crush zoneZagros fold belt

UAE

Yemen

Oman

Najd faultsystem

Dead Seatransform fault

Gulf of Aden rift

Makran fold belt

Red Sea rift

Orientation of induced hydraulic fractures

Orientation of borehole ellipticity

Fig. 3.15: STRESS ASSESSED: Regional stress patterns can provide a general indication of fractureorientation and borehole ellipticity across the region, but rock stresses vary on a local scale andmodels that rely on accurate stress determination will require the local information which sonicanisotropy measurements provide.

Regional stress determinations (figure3.15) give a general indication of thelikely fracture orientations, but local vari-ations and the effects of localized struc-tures, such as large faults, can alter thestress pattern completely, counteractingor adding to the regional stress.

Because the permeability of frac-tures is influenced by regional andlocal stresses both must be assessed toensure that wells do not collapse andthat water production is minimized.

Fracture orientation

Most natural fractures are vertical orsub-vertical. As a result, horizontal wellsencounter many more fractures than ver-tical wells. The orientation and charac-teristics of each fracture set can have aprofound influence on production effi-ciency and water entry into a well, sofracture strike is a crucial factor for any-one drilling horizontal wells (figure 3.16).

The DSI sonic anisotropy techniquealso helps to estimate fracture heightand azimuth more accurately afterhydraulic fracturing, providing a quanti-tative assessment of a ‘frac’ job. It alsoallows the geoscientist to predict frac-ture closing stress.

The DSI tool makes high-qualityStoneley wave measurements availablefor fracture evaluation. When a bore-hole Stoneley wave encounters anopen fracture that intersects the well-bore, some of its energy is reflected asa result of the large acoustic impedancecontrast caused by the fracture.Processing of the acquired Stoneleywaveforms measures the reflection and,therefore, fracture characteristics.

Open fractures

Closed fractures

Minimumhorizontalstress

Maximumhorizontal

stress

Fig. 3.16: FACING UP TO FRACTURES:Horizontal wells intersect many vertical andsub-vertical fractures. Water can move easilyalong open vertical fractures, leading to earlywater production and high water cut inhorizontal wells. Detailed anisotropymeasurements help to avoid this problem bycharacterizing these fractures more accurately.

Fig. 3.17: Traditionalwaterflooding methodswhich ignore the fracturepattern in this field (a) will leadto early water production andleave a large volume of bypassed oilin the reservoir. The poor sweepefficiency is the result of rapid watermigration along fracture planes. If thewaterflood is adjusted to takeaccount of the fracture pattern(b) a more efficient sweepand delayed waterproduction shouldresult.

52 Middle East Well Evaluation Review

SHEAR INSPIRATION FOR OLD CORE

Coresample

Fastest velocity

Slowest velocity

Stressreliefcracks

Maximum stress

Minimum stress

Exaggeratedstrain relaxationFig. 3.18: WHAT A RELIEF: When a core is

removed from a formation the stresses actingupon it are relieved. In many rocks this stressrelief causes time-dependent anelastic strainsand produces microfractures with strikesperpendicular to the maximum horizontalstress of the formation.

Oil

Oil

Water

Fractures

Injector

Injector

Injector

Injector

Producer

Hydraulic stimulation and natural frac-tures can greatly enhance oil and gasrecovery in tight (low matrix permeabil-ity) formations. However, they affectflow and can set up unusual drainagepatterns within a reservoir. Effectivereservoir management relies on a clearunderstanding of existing fracture net-works, particularly during in-fill drillingwhen the positioning of wells is critical.

Effective hydraulic fracture stimula-tion is strongly influenced by fracturedirection. The type of treatment selectedcan be modified by fracture azimuth,especially if there are geological struc-tures to be intersected or avoided by thestimulated fractures.

Waterfloods and other enhanced oilrecovery (EOR) projects require modelsof fracture direction. Waterfloodsdesigned without reference to the reser-voir's fracture pattern may lead to pre-mature water breakthrough and reduceultimate recovery rates. Waterflooddesigns which incorporate fracture datacan maximize sweep efficiency.

In a typical waterflood pattern theproducing well is surrounded by a ringof injectors. If the fracture direction inthe reservoir coincides with the routefrom injector to producer, then injectedfluids will travel along the fractures, by-pass the non-fractured rock, break intothe producer very early and sweep onlya small portion of the reservoir (figure3.17a). However, if the injectors lie alonga line parallel to the fracture directionand the producers along another parallelline (figure 3.17b) the fractures can beused as a line injector, greatly increasingsweep efficiency across the reservoir.

Most hydraulic fractures and manynatural fractures are near-vertical andtheir propagation direction is parallel tothe direction of maximum horizontalstress. Horizontal fractures are foundonly in shallow reservoirs where theweight of overlying rocks (the overbur-den) is relatively small. Fracture direc-tion can therefore be predicted bymeasuring the direction of maximumhorizontal stress before stimulation frac-tures are generated.

Oil

Water

Fractures

Injectors

Producers

(a)

(b)

53Number 18, 1997

Core samples can also provide impor-tant information about reservoir stresspatterns. When a core sample isremoved from a formation the stresseswhich acted upon it are relieved and therock expands (figure 3.18). In the hori-zontal plane, the greatest expansionoccurs in the direction of the maximumhorizontal in-situ stress.

The stresses which affected the corewere anisotropic and the rock expandsless in some directions and more in oth-ers. This differential expansion causescracks in the core. The cracks are notdistributed uniformly through the rock;the largest proportion have strikes per-pendicular to the maximum horizontalstress of the formation.

One technique of core analysis (thevertical shear velocity anisotropymethod) involves cutting two parallelsurfaces on an oriented core sample andpropagating a shear acoustic wave in adirection parallel to the vertical directionof the formation. The transmitting andreceiving planar shear transducers arearranged so that their directions of polar-ization are parallel. Rotating the coresample relative to the transducer polar-ization shows the variation in shearvelocity through the core as the direc-tion of shear wave polarization varieswith azimuth. When the direction ofpolarization of the transducers is parallelto the microfracture direction shearvelocity is at a maximum; when perpen-dicular, it is a minimum. When the trans-ducers are in any other orientation withrespect to the fractures two orthogonallypolarized shear waves are propagated inthe core and shear wave splitting isobserved at the receiving transducer.

This technique is sensitive and ac-curate, but the shear anisotropy is usu-ally small. Few rocks show vertical shearvelocity anisotropies greater than 10%and most are in the range 1- 5%.

Heading for extinction

To overcome the problem of measuringsmall anisotropies a method of cross-polarizing the transmitting and receivingshear transducers has been developed.When the core sample is rotatedbetween these cross-polarized trans-ducers ‘acoustic extinction’ patterns(similar to the optical extinction patternsin light microscopes which use cross-polarized lenses) are produced. Energyfrom the transmitting transducer has itspolarization split and rotated as it entersthe sample so that the energy is polar-ized parallel and perpendicular to themicrofracture direction. As the corerotates, the microfracture direction willbe aligned with the polarization direction

D.P. Yale and E.S. Sprunt (1989) Prediction of fracturedirection using shear acoustic anisotropy. The Log Analyst(30) pp. 65-70.

Fig. 3.20: Comparisonof fracture directions derived

from acoustic anisotropymeasurements on oriented core samples

and results from other measurementmethods.

of one of the transducers (figure 3.19). Atthat orientation, the energy is a singlewave as it enters the rock. This wave hasno component of motion in the directionof the receiving transducer and noenergy will be received - resulting inacoustic extinction. If the core is rotatedthrough 360° there will be four suchextinctions, as in optical microscopy.Two of these four directions will be para-llel to the microfractures which cause theanisotropy and the others will be perpen-dicular.

The technique has been tested oncores from a number of wells (figure3.20) and the fracture directions deter-mined from acoustic anisotropy havebeen within 15° of other fracture direc-tion measurements including anelasticstrain relaxation, tiltmeter surveys, corefracture descriptions, overcoring of mini-fracs and horizontal velocity anisotropy.Shear acoustic anisotropy is a simple lab-oratory technique - easier to performand more cost-effective than field tech-niques such as anelastic strain relax-ation, where results will vary over time.Acoustic anisotropy is not time-depen-dent: old core samples still display mea-surable shear acoustic anisotropies even20 years after collection.

Sample rotation

Acoustic extinction cross- polarized transducers

Polarization direction oftransmitting acoustic transducer

Microfracture strike

Polarization directionof receiving transducer

Fig. 3.19: The acoustic extinction techniquepropagates plane-polarized shear waves alongthe vertical axis of the core and receives themat a shear transducer which is cross-polarizedrelative to the transmitter. These polarizationsremain fixed as the sample is rotated.

Acoustic anisotropyStrain relaxationCore fracturesOther methods

(A)

(B)

(C)

Field A

Field B

Field C

37°

101°

269°

44°

91°

283°

39°

95°

280°

38°

105°

272°

Acousticanisotropy

Strainrelaxation

Corefractures

Othermethods

54 Middle East Well Evaluation Review

Microfracextensometer

Boreholebreakout

Core recoveredmicrofrac

Anelasticstrain recovery

N N NN

Fig. 3.21: Several methods were used to determine the various rock mechanical properties requiredto optimize the propped hydraulic fracture treatments in the ALS-1 well. The methods gave resultswith varying degrees of accuracy and reproducibility.

Faults found and followed

Borehole images reveal information aboutfaults that dipmeters cannot provide. Indipmeter analysis, the strike of the fault isdefined by the drag along normal orreverse faults or rollover structure withgrowth faults. However, if there is a com-ponent of strike slip even the strike analy-sis of the fault can be in error by asignificant angle. Imagery can reveal,without ambiguity, the type of fault pre-sent, its exact dip angle and direction,allowing an estimate of the location wherethat fault will intersect a reservoir zone.

Many faults, especially late-stage faultsin brittle rocks, have little or no deforma-tion or drag zone, so they may not bedetected by dipmeters. In areas domi-nated by wrench tectonics, vertical dis-placement of beds on opposite sides ofthe fault may obscure it. Using imagerythese faults can be found and investigatedfor aperture and mineralization.

In a reservoir it is very important tounderstand faults around the wellbore.Large open faults can produce water andshould be avoided. Smaller faults andtheir associated fractures, however, canboost oil output without early water pro-duction. Integration of 3D seismic data andimagery is helping to locate wells close tosmaller faults which cannot be detectedusing conventional 2D seismic methods.

However, fault characterization is mosteasily and reliably achieved by examin-ing the fault plane. This can be critical forthe detection and geometrical analysis ofgrowth faults and normal faults which arecommon in deltaic reservoirs. A review ofmany dipmeter surveys has shown thatthe actual fault plane of growth faults israrely correlated. This is because the bed-ding dip direction is opposite to the fault

direction (a result of fault rotation andbedding dip into the fault plane). For asso-ciated normal faulting it was found that ifthe correlation lengths used during dip-meter processing were too long, the dip ofthe fault and the deformation immediatelyaround it would be averaged with smallerdips in beds further from the fault.

The borehole image, however, allowsthe geoscientist to examine each faultzone individually in sufficient detail fordip and orientation of the fault to bedefined accurately, even if the profile isproduced using dipmeter software.

Stresses old and new

Mærsk Oil Qatar AS has carried outappraisals in the Cretaceous oil-bearingreservoirs of the Thamama and Wasiagroups, offshore Qatar.

These sedimentary groups are domi-nated by a series of upward shallowingcycles of shelf carbonates, notably theKharaib, Shuaiba, Khatiyah andMauddud formations. The reservoirs inthis study occur at relatively shallowdepths (2700-3600 ft).

One of the most important aspects ofthe programme involved determining theproductivity of multiple, hydraulically-frac-tured horizontal wells. A high-resolution3D seismic survey was designed to gener-ate a migrated stratigraphic and structuralimage of the oil-bearing Cretaceous reser-voirs at Al Shaheen. Cross-sectionsextracted from the interpreted 3D seismicdataset were used to place the horizontalborehole trajectories within the reservoirsection accurately and to predict faultpositions along the planned trajectories.

A large amount of rock mechanicaldata was gathered from the verticalappraisal well, ALS-1, to optimize thedesign of hydraulic fractures. The com-

bined surface seismic and rock mechani-cal datasets form a unique resource forunderstanding palaeo and present daystress fields and fracture patterns.

The full-fold coverage of the 3D seismicsurvey amounts to 50 km2 centred on wellALS-1. The 2D survey - 2530 km in total -was designed to incorporate existing wellinformation and to image and identifyJurassic and Cretaceous exploration tar-gets within the area. The 3D seismic datawas ana-lysed for fault and fracture trendsat the Cretaceous reservoir levels and atshallower Tertiary levels. Horizontal timeslices and seismic attribute maps wereused to develop a detailed interpretationaround the ALS-1 well and along theplanned ALS-2 and ALS-3 trajectories.

Mechanical responses

Having gathered the seismic data, thenext step involved assessing rockmechanics in the reservoir. Many differ-ent methods were used, including:• Microfrac pressure analysis• Borehole extensometer used duringmicrofracture testing• Anelastic strain recovery of orientedcore sample• Borehole ellipticity from FMS caliperrecords• Over-coring and imaging of inducedmicrofractures.

The results obtained (figure 3.21) indi-cate that there is close agreementbetween some methods, notably the bore-hole breakout and core recoveredmicrofrac, while others (such as anelasticstrain recovery) proved to be of littlevalue for this type of study.

The observed stress orientation fromMærsk's detailed testing programme wasfound to coincide with the regional‘Zagros Stress’ direction (see figure 3.15) -

55Number 18, 1997

N

S

W E

Orientation of natural fracturesinterpreted from FMS

NE

SW

N

N

S

N

Fractures causedby folding

Fractures caused by regional stress

σ max

σ min

NE-SW

W-E E-W

Fig. 3.23: The regional stress pattern produces fractures which are oriented NE-SW perpendicular tothe minimum stress direction. In the anticline, however, stresses associated with folding have ‘over-printed’ the regional pattern with fractures aligned N-S. Waterflooding or horizontal drilling in thisanticline would encounter fracture orientations and potential problems which could not be predictedfrom the regional stress pattern.

Fig. 3.22: The orientation of naturalfractures derived from FormationMicroScanner (FMS) data. Theorientations appear to match thestress field orientation, with amarked dominance of ENE-WSWoriented fractures. On a largerscale, however, the results of 2Dand 3D seismic mapping show apredominance of WNW-ESEtrending features with asubordinate N-S trend also present.

the dominant stress component in thispart of the Arabian Plate since the earlyTertiary.

The modern stress field measured inALS-1 predicts two primary shear compo-nents oriented N-S and ENE-WSW. Thefractures observed in a wellbore using theFormation MicroScanner* (FMS) methodagree with this interpretation and suggestthat the ENE-WSW fractures are dominant(figure 3.22). On a larger scale, the resultsof 2D and 3D seismic mapping show a pre-dominance of WNW-ESE trending featureswith a subordinate N-S trend. This sug-gests that while the regional scale faultingwas triggered by the Zagros Stress, at leastone of the shear components has initiatedmovement along pre-existing fault direc-tions in older rocks. A study of ZakumField in Abu Dhabi has shown that theE-W oriented Oman Stress influenced frac-ture directions in Thamama Group reser-voirs deposited after the Oman Stressbecame inactive.

Fault and fracture trends

Regional fault and fracture trends can beuseful for predicting fracture orientationin some fields, but the regional values areonly certain to agree with induced frac-ture orientations if there is no local per-turbation in the stress field. For example,localized compressional folding may gen-erate an induced fracture set which iscompletely different to the regional aver-age (figure 3.23). In addition, natural frac-ture sets can be influenced by olderstress regimes and so be completely dif-ferent to existing regional stress patterns.

Fold-related fault and fracture orient-ations that do not coincide with regionalpatterns must be identified when reser-voirs in structural traps are the target.

Having proved its value in field tests,dipole shear anisotropy logging is nowbeing made available to the industry. Theactual logging time for sonic anisotropymeasurements is only slightly greater thanfor a standard DSI logging run.

Future development of this techniquewill be controlled by the rate at whichinterpretation skills evolve. This, in turnwill require data on the operational limitsof the method in fast formations, carefulinvestigation of the influence of boreholeshape, comparisons with other methodsand refinements in data processing.