Fracture characterization through the use of azimuthally sectored attribute volumes Amanda Thompson* and Jamie Rich, Devon Energy Co. Mike Ammerman, Ammerman Geophysical

Summary

Azimuthal volumes have often been looked at in conjunction with velocity anisotropy to determine the strike of fractures or maximum stress direction. Due to variations in the directions of fractures, attributes calculated on azimuthal volumes may also provide insight into the strike and dip, location, and spatial variation of the fractures. In particular, attributes highlighting the differences in frequency, time, and amplitude can effectively isolate areas of fractures (Tod et al 2007). If these seismic experiments can detect naturally-occurring fractures in reservoirs then why not use the same principles to detect hydraulically-induced fractures? The techniques discussed in this paper will show how these azimuthally sectored volumes and their attributes can be used to map the hydraulically-induced fractures in the Barnett Shale in the Ft. Worth Basin. Areas of pre-existing hydraulic fractures are typically avoided because of the potential interference with other wells in the area. We demonstrate how prior geologic knowledge of areas of hydraulically-induced fractures will be used in conjunction with the azimuthal volumes in order to effectively map the fractures.

Acquisition and Processing

In April 2009, Devon Energy acquired a wide azimuth 51 km2 proprietary 3D seismic data over the study area. Overall, the P-wave seismic data are of high quality with frequencies approaching 100 Hz. Table 1 summarizes the acquisition parameters. Offsets equal to or greater than the target depth are acquired. During processing, we first computed the azimuthal velocity variation based on the far-offset azimuthal variation in traveltimes. We then prestack time-migrated the data, and generated four 450-wide azimuthally-sectored volumes centered about 00 (North), 450, 900, and 1350 for further analysis. Figure 1 shows the azimuth (spider diagram) of the midpoints inside representative CMP bins. Since all azimuths are present in the bins this justifies the azimuthal processing. Although Table 1 shows the bin size to be 110 ft by 110 ft (33 m by 33 m), we increased the bin size of the azimuthal sectored volumes to generate ‘super gathers’ at 220 ft by 220 ft (67 m by 67m) to increase fold. This processing has been implemented in other 3D surveys in the study area, so the fast and slow azimuth directions are known. The fast direction is usually the maximum horizontal stress direction, which for the basin is

approximately N45E. There was no need to use multiple azimuths with smaller ranges to first determine the fast and slow directions of the data.

Table 1: Acquisition parameters used to allow subsequent azimuthal processing

Number of live lines: 30

Number of stations per line: 120

Receiver line interval 660 ft (201 m)

Receiver group spacing 220 ft (67 m)

Shot line interval 880 ft (268 m)

Vibrator array interval 220 ft (67 m)

Patch size 26,180 ft by 25,520 ft (7,980 m by 7,778 m)

Nominal bin size 110 ft by 110 ft (33 m by 33 m)

Number of vibrator sweeps 8

Number of vibrators per array 3

Sweep range 10-110 Hz, 10 s duration, 3 db/octave

Number of geophones per group 6 in a 6 ft (2 m) circle around station

 

 

Figure 1. Spcorrespondishown in Fiand close to

Geologic an

Within this operated Baor directionaterms of spewells beforeApproximaton another s In addition tproduction lconventionaand one welCompletionLower Barnlanded and chowever, thcompletions

Azimuthal

Typically seoffset stackewere run byazimuthallyexploit geomdifferences Horizons weso differencreviewed.

The wells arfilled fractu

pider diagram ofng to the CMP bgure 4. Note the

o uniform.

nd Production D

survey there arearnett gas wells, al and the remaiecial processes, ie April 2009, andtely a dozen micsubset of wells b

to these specialtylogs have been aal (whole) cores,ll has both conve

ns for vertical wenett interval and completed in the

here are seven hos including one w

Attribute Anal

eismic attributesed data. In our s

y a third party vey sectored volummetric, phase, inwere calculated ere picked on ea

ces in time betwe

re considered to res, and if orthor

f the azimuthal cbins displayed ine azimuthal cove

Data

e several hundredtwo thirds of whnder are horizonimage logs wered two wells after

croseismic experbefore April 2009

y log experimenacquired, four we, eight wells haventional and sideells are within thhorizontal wellse Lower Barnett orizontal Upper Bwith microseism

yses

are run on full-astudy a full suiteendor on each of

mes. In total 70 anstantaneous and

on each sectoredach azimuthally-een the azimuths

be producing frrhombic symme

overage n the red box erage is wide,

d Devon-hich are vertical ntal wells. In e run on sixteen r April 2009. riments were run9.

nts, two ells have

ve sidewall coresewall cores.

he Upper and s are primarily

interval; Barnett

mic.

azimuth, full-e of attributes f the four attributes that d amplitude d volume. sectored volume

s could also be

rom open, gas-etry is present

n

s

e

(flat lfractudetecexpecKnowanisobehavthis e

Ellips

Gretcvelocsymmtime dapproto expindepvariatoverpinterpTheseeach attrib

θ(A

by ca

⎜⎜⎜⎜⎜

⎝

⎛

cos.

coscos

Equatattribbe sol

(C^=

In ordrotatewith tequat

)(θA

wherethe ge

layers and one dures), then we hyt open fracturesct the amount of wledge of fast antropy trends fromvior of the microxpectation.

se Fitting to Az

chka and Tsvankcity in a medium metry can be fit bdelays in azimut

oach to azimuthaploit the azimuth

pendent of the mtion with azimutprint of the time-preted on each ofe horizons are thcorresponding aute values, A(θ),

θ 2cos) ba +=

asting equation 1

nn sincos.....sincossincoss

2

222

112

θθ

θθθθ

tion 2 has the foutes, and C are tlved using least-

( ) ATTT T1T −.

der to find the me the ellipse by sthe local coordintion of the ellipse

(cos) 21 βθλ −=

e λ1 and λ2 are theneral coefficien

ominant set of vypothesize that wwith seismic (Ly

f fractures detectnd slow azimuth m previous seismoseismic data wi

zimuthal Volum

kin (1998) showewith horizontal

by an ellipse. Fithally-sectored dal velocity analyhal variations in ore common timth (AVAz). In o-delay anisotropyf the azimuthally

hen used to extraazimuthally-secto, are then fit to a

θθ sincosb +

for each azimut

⎜⎜⎜

⎝

⎛

⎟⎟⎟⎟⎟

⎠

⎞

nn sinn...

sinnsinn

2

22

2

12

1

θθ

θθθθ

rm A=TC, wherthe unknown coe-squares:

major and minor aome angle β suc

nate system. For e can be written

(sin) 22 θλβ −+

he eigenvalues onts

vertically-alignedwe should be ablynn 2004). We ed to vary by azdirections, veloc

mic experimentsll aid in confirm

mes

ed how the NMOly transverse itting such an elldata is the standasis. Our goal hewaveform,

me shifts or amplorder to avoid they, the same horizy sectored volumact a given attribuored volume. Than ellipse using

θ2sinc , (1)

th, θj, in matrix f

⎟⎟⎟⎟⎟

⎠

⎞

⎜⎜⎜⎜⎜

⎝

⎛

=⎟⎟⎟

⎠

⎞

nA

AA

cba

...2

1

. (2)

re A are the meaefficients, which

(3)

axes of the ellipsch that these liner our problem, th

as

)β− , (4)

of the matrix form

d open le to also imuth. city and

ming

O

lipse to ard ere is

litude e zon is

mes. ute for

hese

form

asured h can

se, we e up he

med by

 

 

Figure 5. Exthe Ordovic

Incorporati

Once the seierrors in theevaluated accompared toand Januarysomewhat sgenerate a mmore challelength and twas possibleconsistent w

Next, the mcore informFor wells wcompleted wApril 2009 pthree valuesproduction hfractures. Uwe have beethat have a g

Since there traditional tiseveral timelogged befowells produ

xample of Reliabcian Unconformi

ion of Producti

ismic data has be fit of the ellipseccording to theiro the monthly pry 2010. The data impler to analyz

map. The EUR frnging and neede

the number of stae to scale the ho

with the vertical w

icroseismic dataation for the app

with microseismicwas compared toproduction and Js should be an inhas declined by

Using the ellipse en able to identifgood correlation

is only one seismime-lapse experie-lapse componeore fracturing. Seced well and wh

bility calculated ity.

on Data

een fit to an ellipe are calculated,r EUR. These varoduction rate fo

from vertical wze and have beenrom the horizonted to be normalizages. After normrizontal well EUwell EUR.

a, formation imagplicable wells wec, the date the wo its first month’January 2010 pro

ndication of how potential closurefitting to the azi

fy the subset of sn with the produc

mic experiment, iment. Neverthe

ents. First, all theecond, knowledghich produced po

on the top of

pse and the the wells were

alues have been r April 2009

wells were n gridded to tal wells were zed by lateral

malization, it UR to be

ging logs and ere analyzed.

well was s production, oduction. Thesethe well’s

e of the induced imuthal volumesseismic productsction data.

this is not a less, there are

e wells were ge of which oorly was used.

e

s s

Thirdseismwere 2009,anticiinsencurvathougunstimthe enthe inbeforimageused tTreatformaused trelativindicafractu

Conc

Azimthrouthe revariatfractuisolatcorrelthe sevariogalternkrigg

Ackn

GrateProf. insighdiscu

ddm,dieapgmfncndettnadt dvat o r s o f t h e a m ouufcmtebtnuntulaeglndneeUhdnemdnyetpycfwctedhtnrpdeduufadtnbenrnouraelfdyeUm.neemcnwytyecwotwerhrfor1the wells werpnewuoaednnesreon d a t a h a s h e l pertaefytum.ecweowrrenwoenseetatu

 

 

References

Barnes, Arthur E., 1993, Instantaneous spectral bandwidth and dominant frequency with applications to seismic reflection data, Geophysics, 58, 419-428.

Grechka, Vladimir and Ilya Tsvankin, 1998, 3-D description of normal moveout in anisotropic inhomogeneous media. Geophysics, 639, 1079-1092.

Lynn, Heloise, 2004, The winds of change Anisotropic rocks-their preferred direction of fluid flow and their associated seismic signatures-Part 1, The Leading Edge, 1156-1162.

Tod, Simon, Taylor, Brian, Johnston, Rodney, and Allen, Tony, 2007, Fracture prediction from wide-azimuth land seismic data in SE Algeria: The Leading Edge, 1154-1160.