56 Oilfield Review

Localized Maps of the Subsurface

Optimizing recovery from existing reservesrequires a detailed picture of the reservoir sothat wells are placed where they will be themost productive. Imaging methods in usetoday—seismic, ultrasonic, resistivity andoptical—give either broad-brush images farfrom wells or detailed images valid only atthe borehole wall. Sonic imaging is anemerging technology that helps span the gapleft by these measurements, delineating inter-mediate-sized objects—on the order of a fewfeet—that are up to several tens of feet awayfrom the borehole. Sonic imaging resemblesa miniature seismic survey from inside theborehole.1 With its ability to view detailsbeyond the borehole, this system is beingused in horizontal wellbores to determine theproximity of reflecting bed boundaries, such

as reservoir caps, and map small structures,such as fractures, in the regions surroundingboreholes. Under favorable circumstances, itmay also image gas-oil contacts (GOC).

The concept of sonic imaging is notentirely new. As early as 1982, researchers atELF were developing imaging techniques fortheir full waveform Evaluation of Velocityand Attenuation (EVA) logging tool.2 Duringthe same time an experimental sonic imag-ing method was developed at Schlumberger-Doll Research, patented and tested in NorthSea and Alaskan reservoirs.3 An example ofearly experimental results using this evolvingtechnology, imaging a coal bed in a reser-voir, was first discussed in the OilfieldReview in 1990.4

Chung ChangDavid HoyleShinichi WatanabeFuchinobe, Japan

Richard CoatesMichael KaneRidgefield, Connecticut, USA

Kevin DoddsAberdeen, Scotland

Cengiz EsmersoySugar Land, Texas, USA

Jerome ForemanTexaco LimitedLondon, England

High-resolution sonic imaging is being used to accurately

locate bed boundaries and fractures up to 30 feet

away from a borehole. These measurements

complement seismic surveys and well

logs to help create a more complete

characterization of reservoirs.

Spring 1998 57

Between several tens of meters and 1 ft[30-cm] resolution, and the kilometer andfoot-sized range of investigation, a large gapis left in viewing reservoir structure. Verticalseismic profiles (VSP) and crosswell seismicsurveys provide a measurable improvementin resolution over surface seismic images.

But there remains a gap between these seis-mic techniques and well logs. Furthermore,the effectiveness of borehole seismictechniques has been limited in horizontalwells, and the impact of crosswell seismichas been restricted by the need for two wellswith appropriate geometries.5

Res

olut

ion,

m

100

10

1.0

0.1

00.1 1.0 10 100 1000

Range, m

Ultrasonic

Soniclogging

Sonicimaging

Crosswellseismic

VSP(3D)

3D seismic

2D seismic

Seismicimage

Spatially varyingreservoir

Borehole

Soniclog

■■Range versus resolution for acoustic methods. Many established acoustic methods areavailable to delineate reservoir characteristics. Surface seismic and vertical seismic pro-files (VSP) give good sectional and depth coverage but do not have enough resolution todescribe many reservoir features. Sonic and other well logs give excellent bed resolution,but their information is typically restricted to the immediate vicinity of the well. Sonicimaging is analogous to a 2D surface seismic survey with frequencies scaled up andacquisition geometry scaled down by two to three orders of magnitude. The net result isthe ability to see reservoir features as thin as a foot and at a distance of several tens offeet from the borehole.

For help in preparation of this article, thanks to Alain Brie, Masafumi Fukuhara and Hitoshi Mikada,Schlumberger Wireline & Testing, Fuchinobe, Japan; andAndrew Kurkjian, Anadrill, Sugar Land, Texas, USA.

DSI (Dipole Shear Sonic Imager) and FMI (FullboreFormation MicroImager) are marks of Schlumberger.

1. Esmersoy C, Chang C, Kane MR, Coates RT, Dodds Kand Foreman J: “Sonic Imaging: A Tool for High-Resolution Reservoir Description,” ExpandedAbstracts, 67th SEG International Meeting andExposition, Dallas, Texas, USA (November 2-7, 1997):278-281.

2. Arditty PC, Arens G and Staron P: “State of the Art onEVA Data Processing: An Improvement in SubsurfaceImaging,” Expanded Abstracts, 52nd SEGInternational Meeting and Exposition, Dallas, Texas,USA (October 17-21, 1982): 322-325.Fortin JP, Rehbinder N and Staron P: “ReflectionImaging Around a Well with the EVA Full-WaveformTool,” The Log Analyst 32, no. 3 (May-June 1991):271-278.

3. Hornby BE: “Imaging of Near-Borehole Structure withthe Array Sonic Tool,” Expanded Abstracts, 58th SEGAnnual International Meeting and Exposition,Anaheim, California, USA (October 30-November 3,1988): 124-128.Hornby and Wiggins R: “Borehole Logging Methods for Detection and Imaging of Formation StructuralFeatures,” U.S. Patent No. 4,817,059 (March 28, 1989).Hornby BE: “Imaging of Near-Borehole StructureUsing Full-Waveform Sonic Data,” Geophysics 54,no. 6 (June 1989): 747-757.Hornby BE, Murphy WF, Liu H-L and Hsu K:“Reservoir Sonics: A North Sea Case Study,”Geophysics 57, no. 1 (January 1992): 146-160.

4. Adoumieh R, Berneking D, Olsen B, Kumar R andPhillips J: “The MAXIS System: Imaging for ReservoirCharacterization,” Oilfield Review 2, no. 2 (April1990): 31-42.

5. Christie P, Dodds K, Ireson D, Johnson L, Rutherford J,Schaffner J and Smith N: “Borehole Seismic DataSharpen the Reservoir Image,” Oilfield Review 7,no. 4 (Winter 1995): 18-31.

In this article we discuss how the technol-ogy combines sonic and seismic methods toproduce detailed pictures of the reservoir.Using field examples, we illustrate threeimportant applications of this technique:locating neighboring beds near horizontalwellbores, mapping fractures close to bore-holes and imaging structures too small to beseen in typical surface seismic data.

Resolution versus RangeIn any acoustic measurement, there is atrade-off between resolution and range.Resolution is usually determined by the fre-quency bandwidth of the measurement,while range, or distance investigated, is con-trolled by signal attenuation. At higher fre-quencies more bandwidth is available andtherefore better resolution is obtained.Unfortunately, as the frequency increases sodoes the attenuation, which decreases theeffective range.

In the context of acoustic imaging, resolu-tion refers to the ability to identify that twoseparate events are distinct rather than a sin-gle composite object. Signal wavelength isone factor that controls resolution. Shortwavelengths, corresponding to higher fre-quencies, are capable of resolving smallerfeatures. The ratio of sensor aperture to targetdistance is also important. Other thingsbeing equal, objects that are closer yieldhigher resolution. A simple analogy illus-trates this concept. The aperture formed bythe eye—the distance between the recep-tors—is fixed, and objects such as printedcharacters on a page are resolved when heldclose to the eyes, but not typically whenviewed from across the room.

Acoustic source bandwidth affects bothpulse duration in the time domain and the spa-tial resolution that is achievable in imaging.Pulses from sources with wide bandwidths aremuch narrower in the time domain, and there-fore reflected signals are easier to separatefrom direct borehole modes.

Surface seismic waves have a range ofmany kilometers—one of the largest rangesof any oilfield measurement. Nevertheless,seismic surveys with their frequency band-width of 5 to 100 Hz can resolve objectsonly as small as tens of meters—greater thanthe thickness of many oil- and gas-bearingreservoirs (above right). At the other extreme,sonic logging provides information aboutformations at much higher resolution, typi-cally beds as thin as one foot, but the rangeis limited to a cylindrical volume surround-ing the borehole with a radius comparable tothat of the borehole diameter.

58 Oilfield Review

In terms of resolution and range, sonicimaging is filling the remaining critical gapbetween logs and these seismic methods. Toevaluate this technology, the BoreholeAcoustic Reflection Survey (BARS) tool,developed by Schlumberger, is available invarious parts of the world. The BARS tooluses the same processing techniques andprinciples as surface seismic surveys toimage reflectors around the wellbore (left).The resolution achieved is comparable tosonic logging because the logging tool is rel-atively close to target reflectors allowing useof frequencies in the sonic range—5 to 15kHz. The depth of investigation is limited pri-marily by intrinsic formation attenuation atthese frequencies and the maximum practi-cal tool length. Even with these limitations,this tool can reveal relatively distant featuresthat need not intersect the borehole. Sonicimaging can be used to locate overlying orunderlying reservoir boundaries after a hori-zontal section is drilled and thereby verifywellbore position within target formations.This technique has also successfully mappedfractures in the vicinity of wellbores.

Echoes from the FormationConventional sonic logging methods useacoustic waves refracted along the boreholeto measure formation properties. Theseinclude compressional headwaves, shearheadwaves and Stoneley waves—the maincomponents recorded by monopole soniclogging tools. However, a typical sonicsource also radiates pulses of energy deepinto formations surrounding the borehole inthe form of both P and S waves.6 In anunbounded, homogeneous formation, thesepulses propagate away from the boreholeand are not detected by the receiver array.

In actual formations, there are discontinu-ities, such as bed boundaries, that reflectthese pulses back to the borehole. Acousticreflections are caused by a sudden change inacoustic impedance, and can be detected bya sonic receiver array in the borehole.7

Analogous to a two-dimensional (2D) sur-face seismic survey, sonic imaging uses thesereflected signals to image discontinuities.

The arrival time of each signal depends onthe velocity of acoustic waves in the forma-tion, the source-receiver offset and the loca-tion of target reflectors (next page, top). Byvarying the length of spacer in the tool,which determines the receiver array offsetspacing, a useful time window to detectthese reflections—echoes—can be found.

Reflected signalBed boundary

Borehole signals

Eight-element receiver arrayVariable length

spacer Isolation joint Monopole source

6 in.

■■Sonic imaging principle. Acoustic impedance contrasts, such as bed boundaries and otherdiscontinuities, reflect sonic waves back to the borehole where they are detected by thereceiver array. Signal processing removes the strong borehole signals from the reflections.

30

32

28

26

24

22

20

Sou

rce-

rece

iver

spa

cing

, ft

0 1 2 3 4 5 7 8 9 106

Time, ms

P-P P-S and S-P

S-S

StoneleyShearheadwave

Compressionalheadwave

Receiver Waveforms

■■Prejob modeling results. A typical prejob model shows the expected waveforms at eachreceiver for a single-source pulse, called a shot gather. The delays of the direct and reflectedpulses, seen in each receiver, increase with source-receiver spacing and depend on wavevelocity. The borehole mode signals (blue wavelets)—compressional and shear headwavesand Stoneley waves (clipped in this display)—were computed from model formation acous-tic properties. Reflected signals (red wavelets) were computed for a parallel interface 20 ftfrom the borehole. These include P-to-P reflections (P wave radiated from the borehole andreflected as a P wave from the interface), P-to-S and S-to-P wave mode conversions and S-to-Sreflections. P-to-S and S-to-P signals arrive at the same time. If a P-to-P reflection is to be usedfor imaging, then the modeling suggests that a source-receiver spacing of 20 to 30 ft wouldbe optimum, since the P-to-P reflection has high amplitude and arrives equidistant from thecompressional and shear headwaves.

Spring 1998 59

Knowing When to ListenMaking use of the reflected signals to imagediscontinuities requires enhancing low-amplitude parts of the sonic waveformrecord that were previously considerednoise, while suppressing the high-amplitudesignals that are used for conventional sonicwell logging. Knowing when to listen for thereflected energy is a vital part of successfulimaging. This is achieved by prejob model-ing. The prejob, or planning, model answersthe questions: will there be any detectablereflections from nearby formation disconti-nuities and what are the optimal acquisitionparameters needed to see the reflections?Modeling is used to optimize the source-receiver spacing so that reflected pulses, orwavelets, from remote targets arrive in a timewindow that doesn’t contain borehole sig-nals (previous page, bottom).

Two parameters are essential. First, thesource-receiver spacing determines the rela-tive arrival times of the borehole modes andthe reflections. Consequently, this parameterdetermines the range of distances in whichreflectors, if present, may be detected.Second, the signal frequency bandwidth

controls whether the radiated pulse is com-pact or extended and thus provides good orpoor imaging. Model inputs also include for-mation velocities, density and attenuationfrom both sides of a target interface, bore-hole radius, mud acoustic properties, andmost importantly, the range of distances toexpected target interfaces based on seismicsections and drilling plans.

Approaching the TopAn important application of sonic imaging isevaluating the position of horizontal wellsrelative to reservoir bed boundaries that liewithin the proximity of the wellbore. TheTexaco North Sea Captain field has twomain reservoirs located in high-porosity (to34 p.u.), superpermeable (7-darcy) sandsseparated by a thin shale zone (below).

0

1

2

3

4

5

6

7

8

9

10

Trav

eltim

e, m

s

0 5 10 15 20 25 30 35 40 45 50Source-receiver offset, ft

Stoneleymode

Shearheadwave

Compressionalheadwave

Nearreflector

Deepreflector

■■Moveout curvesfor borehole modesand reflected sig-nals. The boreholemodes (graycurves) typicallyhave amplitudesat least an order ofmagnitude largerthan the reflections(red curves). Thereflected curveswere computed for reflectorsspaced 5 ft and 20 ft from the bore-hole. With anappropriatesource-receiverspacing, thereflections can beseen within thetime intervalbetween the com-pressional andshear headwaves.

Wellbore

Mid-Captain shale

Lower Captain sand

Lower Aptian shale

Lower Barremian shale

Lower permeabilityunit

Oil-water contact

Casing shoe

Captain Field Cross Section

Horizontal distance, 400 ft

Vert

ical

dis

tanc

e, 4

00 ft

N O R T H S E A

Captain field

Upper Captain sand

NORWAY

UK

■■Captain field location and cross section. The region that was imaged is shown bythe red ellipse.

6. Compressional waves are traditionally called P waves,for primary waves. Shear waves are called S waves,for secondary waves.

7. The acoustic impedance is equal to the product ofbulk density and the acoustic velocity. The reflectioncoefficient at a discontinuity is equal to the differencebetween the impedances on each side divided by thesum of their impedances.

60 Oilfield Review

These sands vary in thickness across theblock, but generally are 30- to 200-ft [9- to61-m] thick in the development area. Thetotal estimated oil in place is 1500 millionbbl [238 million m3] and the Upper Captainsand contains a gas cap with an estimated29 Bcf gas in place.

The viscous nature of this oil and the off-shore location of the field required extended-reach horizontal drilling with reservoircompletion lengths of up to 6500 ft [1980 m].Artificial-lift technology is required to helpachieve economic field development.

The first development wells were drilled in1995. These were designed to stay within thetop 20 ft [6.1 m] of the reservoir—just belowthe shale—to optimize total production anddelay water breakthrough from the underly-ing aquifer. When the borehole trajectorydeviated into the shale, either because of thepresence of a subseismic fault or because ofinaccuracy in determining the shale bound-ary from surface seismic data, the drill bitwas pulled back into the sand and an open-hole sidetrack was drilled.

The main purpose of sonic imaging in thesehorizontal wells was to verify wellbore loca-tion within the top of the reservoir. Theimage also permits optimization of the wellcompletion, and aids interpretation of mea-surements-while-drilling (MWD) and pro-duction logging results. In addition, thenear-borehole formation images help engi-

neers regulate production rates to managewater and gas breakthrough. In extremecases, the images show where there wouldbe a need to redrill sections of the well.Finally, the images help enhance structuralinterpretation and reservoir description, andalso assist in planning future wells.

Data processing for this well achieves twoobjectives. First, slowness estimation—drawnfrom conventional sonic processing—is usedto identify and remove strong boreholemodes, especially compressional headwaves,to enhance reflections.8 Next, images of thereflected events are made using seismic pro-cessing techniques—such as normal move-out (NMO) and migration (see “Creating theImages,” next page).9 The resulting image,using wellbore trajectory information, showsa 400-ft [122-m] section along the nearlyhorizontal wellbore (left). The reflectedimages from the mid-Captain shale bed,highlighted by the yellow box, can be seenout to about 25 ft [8 m]. Reflections can alsobe seen to within about 3 feet [0.9 m] alongthe borehole trajectory. At shorter ranges thealgorithms used to attenuate the direct com-pressional headwave reduce sensitivity toreflected waves. The wellbore, shown aswhite parallel lines, intercepts the shalereflector at X1025 to X1050 ft. The gammaray and density readings do not vary whilethe wellbore stays within the reservoir sand,but dramatically increase as it movesupwards into the shale bed, verifying well-bore passage into the shale.

An important problem is determiningwhether the reflector is above or below thewellbore. Two methods are commonly usedto resolve this ambiguity. In many cases, litho-logical barriers are essentially planar, whilewellbore trajectory is rarely completelystraight and parallel to the bedding. Therefore,a simple assignment can be made by com-paring changes in reflector arrival time withthe upward or downward pitch of the well. Ifthe reflection arrives earlier in time as the well trajectory rises, then the reflector is abovethe well. If it comes in later, the reflector isbelow. A planar reflector placed on the cor-rect side of the well will appear planar, but ifplaced on the incorrect side, the reflectionswill undulate in a way that is closely corre-lated with well trajectory. In other cases wherea consistent and significant relative dip existsbetween the borehole trajectory and thereflector, the apparent dip of a correctlyplaced reflector—above or below the well-bore—is more likely to agree with theregional formation dip projected onto the ver-tical plane occupied by the borehole.

True

ver

tical

dep

th, f

tX896

X904

X912

X920

X928

100

80

60

40

20

2.3

2.2

2.1

2.0

Gam

ma

Ray

, AP

ID

ensi

ty, g

/cm

3

Downhole distance, ftX1100X1050X1000XX950XX900XX850XX800XX750XX700

Migrated Image

Gamma Ray

Density

Wellbore

Shale bed response

■■Migrated image of the reservoir cap in Captain well (top). The principal reflector (yellowbox) is interpreted as the lower sand-shale interface, which is dipping downwardstoward the deviated wellbore. At short distances from the borehole, both reflectorsbecome attenuated by the processing to remove compressional headwaves. Far from thewellbore, at distances approaching 25 ft, the remnants of the shear wave, which werenot attenuated, are visible. They appear as “ghost” reflectors that are parallel with theborehole. The gamma ray and density logs (bottom) confirm that the shale bed is crossedby the wellbore at X1025 to X1050 ft. The gamma ray log has a greater depth of investi-gation than a density measurement, and therefore starts to respond to the shale bedbefore the density log does.

8. Ellis D: Well Logging for Earth Scientists. New York,New York, USA: Elsevier, 1987.Tittman J: Geophysical Well Logging. Orlando,Florida, USA: Academic Press, Inc., 1986.

9. Sheriff RE and Geldart LP: Exploration Seismology.Cambridge, England: Cambridge University Press, 1995.Farmer P, Gray S, Whitmore D, Hodgkiss G,Pieprzak A, Ratcliff D and Whitcombe D: “StructuralImaging: Toward a Sharper Subsurface View,”Oilfield Review 5, no. 1 (January 1993): 28-41.

(continued on page 64)

Spring 1998 61

A look at the data clarifies the nature of the imaging

problem. First, a shot gather presents raw traces

from each receiver in the BARS prototype tool

(below left). This display shows traces from a depth-

derived synthetic receiver array, as if they were

received from one firing of a single transmitter.

Here, the dominant direct compressional and shear

headwave progressions, called moveout, can be

seen to increase with increasing receiver spacing.

Next, a constant-offset gather displays raw

traces from one receiver at each depth in the well

(below right). This raw data display gets its name

from the constant distance, or offset, maintained

between the source and receiver. It is the first

unprocessed overall image of the formation.

Borehole modes are clearly seen—direct com-

pressional waves, from 2.9 to 3.5 ms, and direct

shear waves, from 4.9 to 6 ms. Between 3720 and

3760 ft [1134 to 1146 m], pulse arrivals from a

distant reflector are becoming discernible in the

time interval bounded by the direct compressional

and shear events. This interval will be the image

processing window for the reflections.

The most troublesome source of interference in

the imaging time window comes from extended

compressional headwave transients, which can be

3700

3720

3740

3760

3780

3800

3820

3840

3860

38802 3 4 5 6

Time, ms

Time, ms1.5 2 2.5 3 3.5 4 4.5 5 5.5 6

52

50

48

46

44

42

Source Depth 3733 ft

52-ft Offset Receiver

Reflectedevents

Directcompressionalheadwaves

Direct shearheadwaves

Directcompressional

headwaves

Direct shearheadwaves

Rec

eive

r sp

acin

g, ft

Depth, ft

■■Raw sonic logging data. This display, called a shot gather, displays the raw waveforms from each receivergenerated by a single-source pulse. The source is located at depth of 3733 ft and the receivers are spacedfrom 41 ft [12.5 m] to 52 ft [15.8 m] from the source. First arrivals (compressional headwaves) are followedby reflected events (red box). The dominant direct compressional headwave can be seen starting at about 2.3 ms on the nearest receiver at 41 ft. Large, direct shear headwaves can be seen starting at 3.9 ms.

■■Constant-offset gather. The display presents theraw waveforms from one receiver as it moves alongthe entire logging interval. Images from a distantreflector (yellow box) are perceptible between thedirect borehole compressional and shear headwaves.

Creating the Images

62 Oilfield Review

Raw Data After Preprocessing

28

27

26

25

Sou

rce-

rece

iver

spa

cing

, ft

Time, ms2 3 4 5 6 7 8

Time, ms2 3 4 5 6 7 8

P-P reflectionP-P reflection

Suppressedcompressional headwave

■■Shot gathers with reflected signals. The eight-receiverraw signal shot gather (left) demonstrates compressionalheadwaves interfering with the reflected wavelets. Thehigh-amplitude, high-frequency compressional headwavearrives at the closest receiver (lowest trace at 25 ft) atapproximately 3.5 ms. The lower frequency signal arriving at 6.5 ms is a combination of shear headwaveand the Stoneley wave. The arrivals marked by the yellowshading are P-to-P reflections. After preprocessing tosuppress the compressional headwaves (right), reflectedsignals are easier to see in the shot gather starting at3.9 ms in the closest receiver.

10

Ran

ge, f

tR

ange

, ft

3841 3840 3839 3838

Migrated CMP Gathers Before Stacking

Midpoint depth, ft3900 3880 3860 3840 3820 3800 3780 3760 3740 3720

20

30

10

20

30

Migrated CMPGathers After Stacking

■■Migrated common midpoint (CMP) gather before andafter stack. Traces from the migrated image generatedfrom each receiver offset are plotted before stacking—summing together—for a small depth interval (top).Events appearing with equal amplitudes and at the same range indicate that the migration has been performed with the correct velocity. When the traces are stacked (bottom) they add constructively to producea focused reflection such as the one at the 3840-ft depth and a 27-ft range.

Spring 1998 63

seen in the next shot gather (previous page, top).Preprocessing of the raw data removes direct

compressional arrivals using a three-step filtering

process. First, the amplitude of each leading

wavelet is normalized and time shifted so that all

are in phase. This corrects amplitude and phase

changes caused by borehole coupling effects.

Next, direct arrivals are removed using a moving

time-average filter leaving residual reflections and

direct shear waves. Finally, the scaling and offset

shifting in the first step are reversed to restore the

original amplitude and time of the remaining

reflection data. P-to-P reflections and P-to-S or S-

to-P wave conversion events, if present, can be

seen in the processing window between the

removed direct P-wave and the S-wave arrivals.

The improved image quality can be seen in the

constant offset gather preprocessed by this tech-

nique (above).

After Preprocessing

2 3 4 5 6

Time, ms

3700

3720

3740

3760

3780

3800

3820

3840

3860

3880

Dep

th, f

t

■■Constant-offset gather after preprocessing. Imagesfrom reflectors (yellow boxes) are enhanced by pre-processing that removes direct compressionalarrivals. These reflections are barely visible in theraw constant-offset gather (page 61).

S

P

P

S

■■Wave conversions to enhance visibility of dippingfractures. When waves encounter fractures in the for-mation, energy propagating as P waves (red rays) canbe converted to S waves (blue rays) and vice versa.The fact that P- and S-waves travel at different veloci-ties means that wave refraction angles occurring atconversion points are asymmetrical. P-to-S conver-sions can be detected only on the updip side of thefracture, and S-to-P conversions can be detected onlyon the downdip side. In both cases, the convertedenergy is detected when the source and receivers lieon opposite sides of the fracture. The resulting signalsare converted to a migrated image using methodssimilar to those for P-to-P reflection imaging.

The next step in forming the image is called

migration. This process redistributes the recorded

events to their true geometrical position satisfying

the physics of acoustic elastic-wave propagation in

the formation. This step uses a velocity model

derived from the appropriate compressional and

shear slowness logs to compute the traveltime of

each signal from the source to the reflector and

back to the receiver. One constant-offset gather

image is produced for each receiver, and these are

summed, or stacked, to produce the final image.

The final processing step is concerned with

quality control—assuring that the images are real

and correctly positioned in space. A common mid-

point gather (CMP) is migrated assuming a reflec-

tor dip and velocity. Traces from the migrated CMP

image generated from each receiver are plotted

next to one another for comparison (previouspage, bottom). Each of these images shows reflec-

tions located in space. Each event should be visi-

ble on adjacent traces and positioned at the same

depth. The reflection image will appear flat, and

all traces in the CMP gather can be stacked to pro-

duce the final image. Dipping events in the

prestack gather are indicative of either an incor-

rect velocity or an incorrect reflector dip used in

the migration. This results in degraded focusing of

the final image and inaccurate location of the

event in depth. Incoherent, or poorly focused,

events also suggest that unsuppressed noise may

be corrupting the image.

Somewhat different fracture images are pro-

duced from the P-to-P reflections and P-to-S or

S-to-P mode conversions because these events

illuminate different parts of the fractures (aboveright). P-to-P events are reflections and are

observed only when the source and receivers are

on the same side of the reflector. P-to-S and S-to-P

wave conversion events occur when the sonic

waves pass across a fracture, and therefore are

observed when the transmitter and receiver are on

opposite sides of the discontinuity.

64 Oilfield Review

Verifying Storage IntegrityAn early test of the value and effectiveness ofsonic imaging was done for Nirex in one ofits Sellafield, Cumbria, England test wells.The formations—hard volcanic rocks—arebeing considered for long-term intermediateand low-level radioactive waste storage. Theobjective was to image fractures within theproposed burial zone.10

This example shows how P-to-S and S-to-Pwave conversions can be used to help imagefractures beyond the borehole. Previouscrosswell surveys, the FMI FullboreFormation MicroImager tool and theBorehole Televiewer tool (BHTV) indicatedthe presence of many near-vertical fracturesin the freshwater-filled 61⁄4-in. openhole sec-tion from 2950 to 3940 ft [900 to 1200 m].Of special interest for sonic imaging wasdetermining the connectivity between adja-cent fractures and whether fractures observedat the wellbore extended away or not.

A prototype of the Borehole AcousticReflection Survey (BARS) sonic imaging toolwas used to log the interval suspected tocontain fractures. Waveforms were takenfrom each of eight receivers spaced at 6-in.[15-cm] intervals. The prototype, using threetransmitters, could make a depth-derivedlong-array measurement equivalent to24 receivers. By inserting different spacersand isolation joints between the transmitterand receiver array, tool source-receiver offsetconfigurations could be varied from 14.5 to52 ft [4.4 to 16 m]. For each array spacing, aseries of stationary measurements was made

Range from borehole, ft

S-to-P Image

-20 -10 0 10 20-20 -10 0 10 20

Range from borehole, ft

P-to-S Image

Dep

th, f

t

3720

3760

3800

3840

3880

3920110 90 70 50

Slowness, µs/ft

Slowness Logs

Compressionalslowness

Sheerslowness

■■Migrated P-to-S andS-to-P images. Theupdip fractures aremore evident in the P-to-S conversionimage, and thedowndip fractures areclearer in the S-to-P conversionimage. On both the P-to-S and the S-to-Pmigrated images,details closer to theborehole are easier to image than withthe P-to-P reflections.The compressionaland shear wave slow-ness logs (left) wereused to redistributethe recorded reflectedevents to their truegeometrical positions.

Dep

th, f

t

3720

3760

3800

3840

3920110 90 70 50

Slowness, µs/ft-20 -10 0 10 20

Range from borehole, ft

3880

-30 30

P-to-P ImageSlowness Log

Compressionalslowness

■■Migrated P-to-P image (right). The bright reflector from 3805 to 3840 ft is from a fractureabout 27 ft (yellow box) into the formation from the borehole. Markers in the boreholeindicate the location and dip of fractures determined from BHTV images of the boreholesurface. The BHTV-observed fractures confirm projections of the reflections seen withsonic imaging. The compressional wave slowness log (left) was used to redistribute therecorded P-to-P reflected events to their true geometrical positions.

Spring 1998 65

while varying the transmitter frequency todetermine the best operating frequency formaximizing compressional wave excitationsrelative to shear and Stoneley waves.Continuous logging was then performedover the entire openhole section using afixed source-receiver array offset range from41 to 52 ft [12.5 to 16 m].

The P-to-P image shows fractures with aconstant dip over a 220-ft [67-m] section(previous page, top). The dominant reflectionappears on the right side of the imageapproximately 27 ft [8.2 m] from the bore-hole at a depth of 3840 ft [1170 m]. Near theborehole the P-to-P image begins todegrade—meaning that observed events aresmall or absent. As in the Captain fieldexample, arrival times of near-boreholereflections interfere with direct borehole sig-nals. The processing algorithms used to sup-press the direct borehole signals attenuatethese reflections.

However, in this well converted P-to-S andS-to-P signals provide complementary high-resolution images of the fractures close to theborehole (previous page, bottom). Wave-con-version events from fractures can be detectedwhen the source and receiver are located onopposite sides of the fracture. Although geom-etry limits the distance from the borehole thatwave conversions can occur on fracture sur-faces, it also tends to enhance the ability tosee details of fractures closer to the borehole.Another property of these conversions isasymmetry. P-to-S conversions can bedetected only on the updip face of the frac-ture, and S-to-P conversions can be detectedonly on the opposite downdip side. This resultis simply due to Snell’s law—requiring theincident and the converted waves to have thesame apparent velocities along the fracturesurface—and the fact that P-wave velocitiesare larger than S-wave velocities.

Combining P-to-S and S-to-P images pro-vides a picture of both the updip anddowndip parts of fractures near the borehole(left). Again, the resulting images close to theborehole agree well with the BHTV-observed fractures. The sonic images confirmthe location of fractures in the proposedburial zone. More importantly, the imagesdemonstrate whether the detected fracturescontinue for significant distances on eitherside of the well or not, and therefore whetherthey are minor localized or shallow, drilling-induced fractures or not.

Small Structures Influence ProductionA North Sea offshore well illustrates thevalue of sonic imaging in locating small fea-tures that may not be visible on a surfaceseismic survey. A DSI Dipole Shear SonicImager tool was used for sonic imaging in ahighly deviated Troll field wellbore operatedby Norsk Hydro. In this field, thin calcitestringers are known to be vertical permeabil-ity barriers, and characterization of their dis-tribution and horizontal extent is importantfor optimum reservoir development. Themigrated image shows at least two smallreflectors about 30 ft away from the wellbore(below). The left reflector dips about 8°, andits projection intersects the wellbore a fewfeet above the high-velocity layer shown in acompressional slowness log. A strongerreflector, seen on the right side of themigrated image, also dips about 6°.

The reflections are believed to be due tocalcite stringers, confirmed by the increasedcompressional velocity log readings. Thereare weak events between these two reflectorsindicating that they may be connected.However, the image quality is not goodenough to be conclusive. As in other exam-ples, interference from borehole wave atten-uation tends to obscure any reflections thatmay be located in the region near the bore-hole trajectory. The conclusion is that the twostrongest reflectors are probably separatestructures, each with short horizontal extent.

0 10 20-10-20

Range from borehole, ft

Combined ImagesD

epth

, ft

3720

3760

3800

3840

3880

3920

■■Combined migrated P-to-S and S-to-Pimages. An integrated image is obtainedby combining the P-to-S and S-to-Pimages. Images from converted wavesilluminate parts of the fractures closer tothe borehole complementing the P-to-Preflections that look deeper.

X100

X150

X200100 150

Dep

th, f

t

900Lateral section, ft1000 1100950 1050 1150

Gas-oil contact

Compressional Slowness Log Well Trajectory

Migrated image

Wellbore

High-velocitylayer

X60

Slowness, µs/ft

A

B

■■Migrated image of beds from a Norsk Hydro well. The migrated sonic image (inset, right)shows two distinct, small reflecting beds located approximately 10 ft (A) and 30 ft (B)above the highly deviated wellbore. The known GOC (solid blue line) at X142 ft is not seenin the migrated image. The compressional slowness log (left) shows a thin zone—stringer—at X100 ft with high velocity, but also shows no significant evidence of the GOC at X142 ft.

10. Ellis D, Engelman B, Fruchter J, Shipp B, Jensen R,Lewis R, Scott H and Trent S: “EnvironmentalApplications of Oilfield Technology,” Oilfield Review8, no. 3 (Autumn 1996): 44-57.

66 Oilfield Review

Imaging Fluid ContactsAcoustic imaging depends on the ability toobserve reflections, and reflections arecaused by an abrupt and significant changein acoustic impedance. Since oil and waterhave small differences in acousticimpedance, oil-water contacts (OWC) areexpected to be difficult to image. However,gas and oil have a large difference in acous-tic impedance, and therefore gas-oil contacts(GOC) might be expected to producereflections. Bright spots—seen in surfaceseismic surveys—are caused by high-ampli-tude reflections and are commonly used forhydrocarbon identification.

One of the original objectives in the NorskHydro well was to image a GOC known tolie about 36 ft [11 m] above the near-hori-zontal section. In this well, a GOC in cleansandstone with over 30 p.u. porosity wasexpected to be a good reflector for compres-sional waves.

Openhole logs confirmed that the wellintersects the GOC at X142 ft in the deviatedsection. This provided an opportunity topotentially track the contact as the wellboremoves away from the GOC. Unfortunately,neither the migrated image nor the compres-sional log show evidence of the GOC. Thesonic imaging is working, and it shows twosmall reflectors about 30 ft [9.1 m] awayfrom the wellbore. Absence of any signifi-cant GOC on the compressional log may beexplained by mud-filtrate invasion, butinability to image the GOC was unexpected.Extended images over the entire logged hor-izontal section produced no further evidenceof the GOC. Why wasn’t the GOC seen inthe sonic image?

A GOC, especially after a long period of production, is likely to be a zone of slowly changing saturation—from oil at thebottom to gas at the top. As a result, this is also a zone of slowly changing acousticimpedance. The transition zone can be quitelong—up to several meters or more. If theacoustic wavelength is much larger than thetransition zone, then the interface stillappears abrupt, reflects the signal and makesimaging possible. This is certainly what hap-pens in low-frequency—long-wavelength—surface seismic data.

However, if the wavelength is small—sonicwavelengths are typically 1 to 2 ft [0.3 to0.6 m]—compared to the transition zone,then the impedance changes slowly relativeto the wave and a reflection does not occur.Instead, sonic waves pass through the inter-face with only a slight bending (refracting).Only in cases where the acoustic impedancechanges suddenly with respect to the wave-length can strong reflections be expected tooccur. In this well, openhole neutron anddensity logs show that the gas-oil transitionzone occurred over a 10-ft zone—insuffi-ciently abrupt for sonic imaging.

The Next StepImages obtained from sonic imaging surveyscomplement seismic and log information inreservoir descriptions. Field tests verify thatthe effective depth of penetration for thismethod is approximately 30 ft depending onformation attenuation, relative dip and thereflection coefficient of the interface. Theexamples discussed here demonstrate that it

is possible to image distant formation fea-tures, such as reservoir caps, with high reso-lution in deviated and horizontal wells.

To date, this sonic imaging method hasbeen applied only in openhole. Effectivenessof the method in cased wells will dependheavily on cementing quality. As seen in thelast example, direct observation of fluid con-tacts and ability to track their movements ina time-lapse mode is problematic with sonicfrequencies. These limitations, together withthe goal of increasing effective range of themethod, are subjects of continuing research.

Sonic imaging is an evolving technologywith a growing market for its application(above). Subsurface images are providingnew insight into many aspects of reservoirmanagement. As always, with new technolo-gies, several technical challenges remain.But with continuing development, the out-look is promising. —RCH

Saltflank

Fracture

Subseismic-scale structure

Structure

Reef

D

C

B

A

E

■■Applications for sonic imaging. The proven applications include: (A) imaging reservoirstructure for high-resolution verification of the wellbore trajectory relative to geologicalmarkers, (B) imaging subseismic scale reservoir structure for understanding overall poros-ity and permeability characteristics and (C) detecting fractures away from the wellboreto distinguish them from localized drilling-induced cracks. Other potential applications—such as (D) imaging salt flanks and (E) reefs, generally requiring greater range—may bepossible with improved technology.