Seismic Tools for Reservoir Management - Oilfield...

Seismic Tools for Reservoir Management

Reservoir engineers, geophysicists, geologists and managers agree that

the 3D seismic technique can shed light on reservoir structure. But

there’s more to seismic than faults and layers: with the right handling,

seismic data can predict rock and fluid properties across the whole field.

Here’s a look at some of the powerful probes in the seismic

toolbox—inversion, AVO, 3D visualization and time-lapse surveys—with

guidelines for use and some success stories.

4 Oilfield Review

Tajjul AriffinGreg SolomonSalehudin UjangPETRONAS CarigaliKuala Lumpur, Malaysia

Michel BéeSteve JenkinsCaltex Pacific IndonesiaRumbai, Indonesia

Chip CorbettHouston, Texas, USA

Geoffrey DornRobert WithersARCOPlano, Texas

Hüseyin ÖzdemirGatwick, England

Chris PearseAmoco NorwayStavanger, Norway

For help in preparation of this article, thanks to DavidCefola, Oryx Energy Company, Dallas, Texas, USA; BobKeys, Mobil Exploration & Producing Technical Center,Dallas, Texas; Andy Maas and Tiga Teilmann, GeoQuest,Duri, Sumatra, Indonesia; Robert North, GeoQuest,Anchorage, Alaska, USA; and Christopher Ross, PGSTensor, Houston, Texas.In this article RM (Reservoir Modeling) software is a markof Schlumberger.

Oil and gas companies large and small arerelying on 3D seismic data to better delin-eate fields and identify new reserves. Oper-ating companies have quantified and docu-mented the value a 3D survey can add to anexploration or development project, com-pared to a 2D survey.1 These testimonialsdescribe the key role seismic images play inrevealing reservoir locations and structuresand the importance of using the informationearly in the life of a field to derive maximumbenefit.

But some companies are asking more oftheir 3D seismic surveys, demanding knowl-edge beyond—in fact between—reflections,and getting it. A new science of reservoirgeophysics is emerging to provide this addi-tional information to reservoir engineers.2At the heart of the matter are reservoir geo-physicists, who rely on high-quality 3D sur-veys—available through advances in acqui-sition, processing and interpretationtechniques—for complete volume coverageof the reservoir. High-resolution boreholeseismic surveys help fuse the surface seismicwith log and core data to allow log proper-ties such as lithology, porosity and fluid typeto be mapped field-wide (for an update see“Borehole Seismic Data Sharpen the Reser-voir Image,” page 18). With this more com-plete understanding of the reservoir, produc-tion engineers can optimize developmentand recover additional reserves. This articlereviews case studies of four techniques thatshow promise—inversion, amplitude varia-tion with offset (AVO), 3D visualization andtime-lapse monitoring.

InversionInversion is one of the foundations uponwhich reservoir geophysicists are buildingtools to make seismic information more use-ful to engineers. Inversion is so namedbecause it acts as the inverse of forwardmodeling. Forward modeling takes an earthmodel of layers with densities and veloci-ties, combines this with a seismic pulse, andturns out a realistic seismic trace—usuallycalled a synthetic. Inversion takes a realseismic trace, removes the seismic pulse,and delivers an earth model of acousticimpedance (AI), or density times velocity, atthe trace location (next page). Seismic inver-

Winter 1995

nForward modelingand inversion. Forward modelingbegins with anearth model ofacoustic impedance(AI) and ends witha synthetic seismictrace. Inversionbegins with a realseismic trace andoutputs an AImodel.

800�

850�

900�

Tim

e, m

sec

Res

ervo

ir

Inputwavelet

Seismictrace

Forward modelingInversion

Estimatedwavelet

Earth modelof acousticimpedance

Earth modelof acousticimpedance

sion can be posed as a problem of obtainingan earth model for which the synthetic bestfits the observed data.3 The simplest earthmodels contain layers with densities andcompressional velocities, but more elabo-rate inversions yield models with shearvelocities as well. Ideally, inversions com-bine surface seismic, vertical seismic profile(VSP), sonic and density log data.

The main use of inversion for reservoirmanagement comes through log-propertymapping: the seismically derived AI valuesare tested for correlation with logs at thewell location—porosity, lithology, water sat-uration, or any attribute that can be found tocorrelate. Those log properties are then

1. Jeffers PB, Juranek TA and Poffenberger MR: “3-D versus2-D Drilling Results: Is There Still a Question,” pre-sented at the SEG 63rd Annual International Meetingand Exposition, Washington, DC, USA, September 26-30, 1993, paper IM1.5.Greenlee SM, Gaskins GM and Johnson MG: “3-D Seis-mic Benefits from Exploration Through Development:An Exxon Perspective,” The Leading Edge 13, no. 7 (July1994): 730-734.Knecht SW and Helgeson S: “Case Study: How a SmallCompany Adopted 3D Seismic Technology,” Oil & GasJournal 90, no. 42 (October 19, 1992): 54, 56-57.Nestvold EO: “The 3D Seismic Revolution: Cost Bene-fits and Their Implications,” presented at the SEG Sum-mer Research Workshop on 3-D Seismology: IntegratedComprehension of Large Data Volumes, RanchoMirage, California, USA, August 1-6, 1993.Williams P: “Aces in the Hole,” Oil & Gas Investor(October 1994): 94.

2.“The Emerging Science of Reservoir Geophysics,” PetroSystems World (September/October 1994): 18-20.

3. Pan GS, Young CY and Castagna JP: “Net Pay Delin-eation of Gas Sand Using Integrated Target-OrientedPrestack Elastic Waveform Inversion,” presented at theSEG 63rd Annual International Meeting and Exposition,Washington, DC, USA, September 26-30, 1993, paperIM1.4.

N

N o r w a y

S E A

Areaof detail

Denmark

Norw

ay

Denmark

UK

N O R T H

3°00'E 3°30'EAlbuskjell

Ekofisk

Tor

Gamma

TommelitenEdda

S. E. Tor

Eldfisk

Embla Valhall

MjØlner

WestHod

EastHod

Argyll

Norw

ay

Denmark

UK

West Hod East Hod

Norway

Denmark

2/11-3

2/11-A2

0 ft 3281

0 m 1000

6

4. Berg E, Brevik I and Buller AT: “Experiences Gainedusing a Seismic Inversion Method for Detailed Reser-voir Studies,” in Buller et al (eds): North Sea Oil andGas Reservoirs—II. London, England: Graham & Trot-man, Limited (1990): 129-138.Brac J, Déquirez PY, Hervé F, Jacques C, Lailly P,Richard V and van Nhieu DT: “Inversion With a prioriInformation: An Approach to Integrated StratigraphicInterpretation,” in Sheriff RA (ed): Reservoir Geo-physics. Tulsa, Oklahoma, USA: Society of ExplorationGeophysicists (1992): 251-258.Oldenburg DW, Levy S and Stinson KJ: “Inversion ofBand-Limited Reflection Seismograms: Theory andPractice,” Proceedings of the IEEE 74, no. 3 (March1986): 487-497.

5. SEG Workshop on Comparison of Seismic InversionMethods on a Single Real Data Set, Los Angeles, Cali-fornia, USA, October 28, 1994.

6. Schultz PS, Ronen S, Hattori M, Mantran P and Cor-bett C: “Seismic-Guided Estimation of Log Properties,”The Leading Edge 13, no. 7 (July 1994): 770-776.Ashcroft WA and Ridgway MS: “Early Discordant Dia-genesis in the Brent Group, Murchison Field, UKNorth Sea, Detected in High Values of Seismic-Derived Acoustic Impedance,” accepted for publica-tion in Petroleum Geoscience.

7. Pearse CHJ and Özdemir H: “The Hod Field: ChalkReservoir Delineation from 3D Seismic Data UsingAmplitude Mapping and Seismic Inversion,” presentedat the Norwegian Petroleum Society GeophysicalSeminar, Kristiansand, Norway, March 7-9, 1994.

nThe surface of the top chalk of the Hodfield in the Norwegian sector of the NorthSea. Most of the 66.9 MM barrels of oilequivalent attributed to the field liewithin the East Hod anticline. Additionalreserves have recently been provenbeyond the limit of structural closure,north of East Hod, where seismic ampli-tude, inversion and porosity mappingtechniques indicated the presence of ahigh-porosity reservoir zone.

extrapolated throughout the inverted 3Dseismic volume using the lateral variation ofseismically derived AI to guide the process.

Adequately processed seismic data are amust for inversion, but the optimum pro-cessing required to prepare data for inver-sion is the subject of much debate, as is theoptimal inversion calculation itself. Numer-ous processing chains have beendeveloped.4 A workshop was held recentlyto define the ultimate processing scheme,but to the surprise of the participants, noone method proved best.5 The trait that setsinversion apart from the standard processingchain for structural interpretation is the needfor preservation of true relative amplitudes.Changes in trace amplitude from one loca-tion to another may reveal porosity or otherformation property variations, but theseamplitude changes are subtle and may beobliterated by conventional processing.

Inversion can be performed before or afterthe seismic traces have been stacked—summed to create a single trace at a centrallocation—but care must be taken to ensurethat stacking does not alter amplitudes. Insome cases, such as regions where seismicreflection amplitudes vary with angle of inci-dence at the reflector, stacking does not pre-serve amplitudes, and inversion must be per-formed prestack. Only examples of poststackinversion results are presented in this article.

The simplest inversion scheme derives rel-ative acoustic impedance changes for oneseismic trace by computing a cumulativesum of the amplitudes in the trace. Thegradual trend of increasing AI with depth—invisible to seismic waves—is taken fromdensity and cumulative sonic travel times,and added to the relative AI results.6

Porosity Mapping in the Hod Field ChalksAmoco Norway in Stavanger has drawnupon seismic inversion followed by porositymapping as an aid to managing the devel-opment of the Hod field, the southern-most

Oilfield Review

Campbell SJD and Gravdal N: “The Prediction of HighPorosity Chalks in the East Hod Field,” Petroleum Geo-science 1 (1995): 57-69.Landrø M, Buland A and D’Angelo R: “Target-OrientedAVO Inversion of Data from Valhall and Hod Fields,”The Leading Edge 14, no. 8 (August 1995): 855-861.

8. The spillpoint is the point of maximum filling by hydro-carbon of a structural trap.

9. Acoustic impedance has the units of velocity timesdensity. Although the combination of English and met-ric units seems peculiar, ft/sec x g/cm3 is a commonunit.

nA crossplot show-ing correlationbetween porosityand acousticimpedance,derived from sonic,density and poros-ity logs from EastHod wells.

Por

osity

, %

60

50

40

30

20

10

010,000 20,000 30,000 40,000

Acoustic impedance, ft/sec x g/cm3

2/11-A2 East Hod

Two-

way

tim

e, m

sec

Well Trace

-0.1 0.1

Sonicµsec/ft

200 40 2 3g/cm3

Density AI

10,000 36,400ft/sec x g/cm3

SyntheticTrace

-0.1 0.1 -0.1 0.1

Zero-phaseWavelet

Topchalk

2550

2600

2700

2800

2650

2750

2850

2500

in the trend of chalk oil fields in the Norwe-gian sector of the North Sea (previouspage).7 The two separate oil-filled anticlinalstructures in the field—West and EastHod—were discovered in 1974 and 1977,respectively. However, reservoir uncertain-ties were not resolved by appraisal drilling,and marginal economics delayed produc-tion until 1990. Total estimated originalreserves for the field are 66.9 million barrelsof oil equivalent (BOE), of which 94% areattributed to East Hod. An unmanned pro-duction platform is tied to the Valhall facili-ties to the north.

The primary reservoir interval at East Hodcomprises allocthonous—reworked andredeposited—chalks of the Tor formation.The 2/11-A2 well encounters a prime chalkreservoir section, with 90 m [295 ft] of Torformation showing porosities of up to 50%(below, right). Although East Hod is associ-ated with a pronounced anticlinal closure,oil is trapped not only structurally, but alsostratigraphically. Moveable oil has beenobserved below the established spillpoint,with reservoir distribution controlled by acombination of depositional, structural anddiagenetic factors.8 The complex interplaybetween these factors results in a highlyvariable chalk reservoir.

The top chalk surface represents an ero-sional unconformity that exposes a variety ofchalk types from the Ekofisk, Tor and Hodformations to the overlying Paleocene shaleseal. Well data show that chalks contributingto the top chalk seismic event have porositiesranging from 20 to 50%, with impedancesranging from 30,000 ft/sec X g/cm3 to 10,000ft/sec X g/cm3 [9150 to 3050 m/sec X g/cm3](top).9 The high-quality reservoir rocksexhibit a decrease in acoustic impedancecompared to the relatively uniform acousticimpedance of the overlying shale, whilenonreservoir chalks show an increase.Therefore the acoustic properties of thechalk exert the primary influence on the

7Winter 1995

nThe 2/11-A2 East Hod well in a prime chalk reservoir section, with log and seismic datacompared to synthetics. Sonic slowness (track 2) and density data (track 3) are combinedto give acoustic impedance (track 4). This is combined with a seismic wavelet (track 6) toyield a synthetic trace (track 5), which matches the recorded surface seismic data at thewell (track 1). The acoustic impedance decrease at the top chalk interface produces ahigh-amplitude seismic peak, or swing to the right, in the polarity convention used here.

Acoustic impedance, ft/sec x g/cm310,000 36,00020,000 30,000

amplitude of seismic reflections, making itpossible to develop an effective method formapping the reservoir extent and qualityfrom inverted poststack seismic data.

Various 2D and 3D seismic inversion andporosity mapping techniques have beensuccessfully applied in the area. Because ofthe combination of the great range in chalkimpedance, and its predictable dependenceon porosity, the results of most inversiontechniques establish similar porosity trends,with the differences to be found in small

details and absolute porosity values.The first 3D porosity mapping at Hod field

was carried out using the Log-Property Map-ping module of the RM Reservoir Modelingsystem. Vertical well 2/11-3, with its excel-lent tie to the surface seismic data, was usedas the key well to calibrate the inversion(above). The other wells also provided inputto the low-frequency AI model and calibra-tion of AI to porosity.

This mapping supports the presence of azone of high porosity beyond the limit of the

East Hod structural closure (right). Subse-quent drilling in this area has confirmed theinversion predictions of commercial poros-ity, and a horizontal producing well is cur-rently draining the area which now repre-sents a proven extension of the Hod field.

An ever increasing functionality and qual-ity of applications are available for this typeof reservoir characterization. An example ofa significant refinement to the process usedin the Hod field area is a scheme calledspace-adaptive wavelet processing.10

Applied as a precursor to inversion, this pro-cess integrates information from many wellsto ensure that seismic data with a common,broadband, zero-phase wavelet are input tothe inversion.11 The resulting improvementin the resolution of the inversion and subse-quent interpretation have allowed porositymapping from seismic to become a standardpart of the chalk reservoir management pro-cess, and a primary means of identifying andquantifying the potential for extensions tothe field or separate accumulations nearby.

2/11-3 East Hod


Well Trace

-0.1 0.1

Sonic

µsec/ft200 40 2 3

g/cm3

Density AI

10,000 36,400ft/sec x g/cm3

SyntheticTrace

-0.1 0.1 -0.1 0.1

Zero-phaseWavelet

2550

2600

2700

2800

2650

2750

2850

Topchalk

Two-

way

tim

e, m

sec

2500

10,000 36,00020,000 30,000

nAverage porosity mapped from seismicdata at East Hod, generated by the Log-Property Mapping application in the RMReservoir Modeling system. The plottedporosity values correspond to the averagein a 32-msec time window of the seismicdata. This corresponds to an intervalabout 40 to 55 m [131 to 180 ft] thick atthe top of the interpreted chalk section,and includes the optimum Tor formationreservoir. The white line indicates the lim-its of structural closure.

nLog and seismic data compared to synthetics in the key well 2/11-3. Tracks are as inthe figure on previous page, bottom. At the shale-top chalk interface, acousticimpedance increases. The synthetic and data traces respond to this with a high-ampli-tude trough—a swing to the left. This exploration well on the western flank of the EastHod encountered water-wet nonreservoir chalks.

8 Oilfield Review

10. Poggiagliolmi E and Allred RD: “Detailed ReservoirDefinition by Integration of Well and 3-D SeismicData Using Space Adaptive Wavelet Processing,”The Leading Edge 13, no. 7 (July 1994): 749-754.

11. Broadband means the bandwidth, or range of frequencies present in the wavelet, is wide. Zero-phase means the shape of the wavelet is optimizedfor interpretation of inversion results: trace peaksindicate locations of AI changes—in contrast toother kinds of wavelets, in which trace zeroes canindicate AI changes.

12. Corbett C, Solomon GJ, Sonrexa K, Ujang S and Ariffin T: “Application of Seismic-Guided ReservoirProperty Mapping to the Dulang West Field, OffshorePeninsular Malaysia,” paper SPE 30568, presented atthe 70th SPE Annual Technical Conference and Exhi-bition, Dallas, Texas, USA, October 22-25, 1995.

13. Anderson B, Bryant I, Helbig K, Lüling M and Spies B:“Oilfield Anisotropy: Its Origins and Electrical Char-acteristics,” Oilfield Review 6, no. 4 (October 1994):48-56.Ayan C, Colley N, Cowan G, Ezekwe E, Goode B,Halford F, Joseph J, Mongini A, Obondoko G, Pop Jand Wannell M: “Measuring Permeability Anisotropy:The Latest Approach,” Oilfield Review 6, no. 4 (Octo-ber 1994): 24-35.

N

2/11-3

Line 800

Line 850

Porosity, p.u.18 38

Mapping Porosity in MalaysiaOnce thought to be useful primarily in car-bonate reservoirs because of a more recog-nizable porosity-acoustic impedance rela-tionship, inversion for porosity mapping hasalso proven powerful in sand reservoirs.PETRONAS Carigali, the upstream operat-ing arm of the Malaysian national oil com-pany, has used seismic inversion to opti-mize drilling locations in the Dulang Westfield in the Malay basin of the South ChinaSea (right).12

The Dulang field has an estimated 850million barrels original oil in place (OOIP).In the first stage of development, more than100 wells were drilled in the central area ofthe faulted anticlinal structure, producingfrom an oil and gas column of up to 150 m[492 ft] of stacked sandstones. The nextstage of development focuses on the DulangWest portion, in which plans call for 25wells from a 32-slot platform.

The four delineation wells indicate areservoir too complex to understand fromwell data alone. The main reservoirs arefine-grained, discontinuous sands interbed-ded with shales and coals. The sand bodiesare preferentially oriented, suggesting per-meability anisotropy on the scale of thefield.13 Porosity, permeability and their rela-tionship to each other show great variabil-ity—for example, permeability can varyfrom 50 to several hundred millidarcies for amedian porosity of 25%. In the central areadeveloped earlier, close well spacing per-mitted property mapping from logs. But inDulang West, engineers have relied oninversion of the 3D seismic data to extendinformation contained in the delineationwells to map porosity across the field.

After the poststack seismic and log datawere tied at the right depths and inverted foracoustic impedance, log properties weretested for their correlation with the AI valuesat the respective well locations using theLog-Property Mapping module of the RMReservoir Modeling software (right). Onlyporosity was found to correlate significantlywith acoustic impedance, with a trend simi-lar to that of the chalks of the East Hod field.Extending the log porosity values away fromthe four wells using the seismic inversionresults as a guide produced a reservoirporosity map.

nCorrelation between average acoustic impedance (AI) and two log properties, poros-ity and gamma ray. Porosity shows an inverse correlation—AI decreases as porosityincreases, while gamma ray shows no significant correlation with AI.

Cambodia

Viet

nam

Thailand

Malaysia

Mala

ysia

Sumatra

Java

Singapore

Dulangfield

Borneo

SO U THC

HI N

AS

EA

GULF OFTHAILAND

Laos

Myanmar

N

Dulang field

Westernarea

Unitarea

Easternarea

6G-1.36G-1.4

6G-1.1B

6G-1.6

6G-1.2

0 miles 3.1

0 km 5

Por

osity

, p.u

.

20

15

10

5

0

18,000 20,000 22,000

6G 1.3

6G 1.6

6G 1.46G 1.2

6G1.1B

6G 1.36G 1.6

6G 1.46G 1.2

6G 1.1B

18,000 20,000 22,000

GR

, AP

I uni

ts

85

90

95

100

105

110


9Winter 1995

nThe Dulang West fieldoperated by PETRONASCarigali, the upstreamoperating arm of theMalaysian national oilcompany.

An integrated assessment of porosity andstructure allowed interpreters to proposedrilling locations (above). Areas of higherporosity in the south were deemed morepromising than lower-porosity areas in faultblocks to the north. The Well Prognosismodule of the RM system allowed severalpotential sites to be quickly investigated forreservoir quality and likely reserves.

The reservoir model built from the seismicdata included not only the traditional aspectof reservoir structure, but also the total vol-ume of porosity in each volume element ofthe seismic cube. This model was scaled upfor input to a fluid-flow simulator. Perme-ability was distributed throughout the modelby applying a porosity-permeability trans-form to the seismically guided porosity map.

The new model provided a better estimationof production over a simulated seven-yearperiod than that obtained by other methods.

In addition, areas of high acousticimpedance were interpreted to be shaly orto have poor reservoir development,enabling better placement of planned wells.Recent appraisal drilling southeast of well6G-1.3, testing oil potential downdip of gasinferred from an especially low AI anomaly,encountered 18 m [59 ft] of good quality,18% porosity gross sand. Although the sandwas wet, agreement with the model wasgood, with 18.8 m [62 ft] and 19% porositypredicted. Two development wells, D1 andD2, further demonstrate the predictivepower of the method (below).

AVOIn some environments, seismic reflectionamplitude variation with offset (AVO) can beused as a reservoir management tool to indi-cate hydrocarbon extent.14 The AVO tech-nique relies on the observation—backed upby physics—that pore fluid type imprints asignature on the amplitude of a seismicreflection. To see this signature, seismic datamust be viewed at different angles of reflec-tion. Depending on the type of pore fluid inthe juxtaposed rock layers, the amplitude ofthe reflection may increase, decrease, orremain constant as the reflection angle at theboundary increases (below). The incidentangle of the seismic wave can be expressedin terms of offset, or distance, between seis-mic source and receiver—a congruent quan-tity more easily measured than an angle atsome depth.

A common way to use AVO to character-ize reservoirs is to identify a hydrocarbonAVO signature—for example, the AVOresponse of a gas reservoir—and comb the3D seismic volume for other areas with sim-

10 Oilfield Review

nWest Dulang seismically guided porosity map and proposed drilling locations (greendots). Comparisons between predicted and actual drilling results are shown in the table(below).

nAmplitude variation with offset (AVO).Some interfaces show AVO signatures, orvariation of reflection amplitude withangle of incidence. In this case, theamplitude increases with offset.

aaaaaaaaaaaaaaaaaaaaAmplitude increases

Amplitude variation with offset (AVO)

CommonMidpoint

(CMP)

Shale

Gas sand

S4 S3 S2 S1 R1 R2 R3 R4

Offset 3

Offset 2

Offset 1

Offset 4

Offset 3

Offset 2

Offset 1

Offset 4

Appraisal6G1-7

Well

DevelopmentD1

DevelopmentD2

Thickness

18.8 m

15.4 m

13.4 m

Porosity

ActualPredicted ActualPredicted

18 m

16.6 m

6.3 m

19%

19.8%

17.5%

18%

20%

18%

Porosity, p.u.13.69 20.00

6G 1.3

6G 1.6

6G 1.4

6G 1-7

6G 1.1BD1

D2

ilar signatures. This can result in discoveriesof bypassed hydrocarbon as well as exten-sion or delineation of existing reservoirs.The practice assumes that lithology does nothave enough lateral variation to affect theseismic amplitudes, so that all AVO effectsare due to changes in pore fluid type. Theseismic data must be processed to preserverelative amplitudes, and also must be ana-lyzed before stacking.

Some lithologies show less obvious AVOsensitivity to pore fluid change than others.Carbonates and low-porosity sandstonestend to have less evident AVO signaturesthan high-porosity sandstones, and specialcare must be taken in applying the technol-ogy in these areas.15

In an example from the mature BK field inthe Gulf of Mexico, the successful incorpo-ration of AVO analysis helped Oryx EnergyCompany engineers identify extensions ofthe reservoir that might have gone undrilled.The quality of the AVO results convincedmanagement to free up money for drillingthat had been allocated elsewhere.

The BK field lies off the flank of a shallowsalt and shale diapir in 5 m [16 ft] of waternear the Louisiana Gulf Coast (above right).The reservoir, discovered in the late 1940s,has produced 300 billion cubic feet (Bcf) ofgas. The map of the 5000-m [16,400-ft]deep structure had been constructed primar-ily with well control, and the new 78-km2

[30-square mile] survey, designed to provideincremental structural and stratigraphicinformation, changed the structural map sig-nificantly (right).

AVO analysis was introduced to betterdelineate the gas reservoir and reduce riskin choosing drilling locations. The analysisrequired a seismic cube for two differentfamilies of offsets. Data processing followedthe same sequence as for the full 3D cube,except the data were separated into a near-

11Winter 1995

nStructural maps created before and after the 3D survey showing significant differ-ences. Well BK-15 produces gas. BK-16 is a proposed well location in an undrilled updipfault block. (Adapted from Ross CP, reference 14, courtesy of Blackwell Science.)

U N I T E D S T A T E S

LouisianaMississippi

Alabama

Texas Mis

siss

ippi

Riv

er

G U L F O F M E X I C O

3D surveyarea

BK-16Line 1235

Line 1215BK-15

Before 3D Survey After 3D Survey

14. Chiburis E, Franck C, Leaney S, McHugo S and Skid-more C: “Hydrocarbon Detection with AVO,” Oil-field Review 5, no. 1 (January 1993): 42-50.Ross CP: “Improved Mature Field Development with3D/AVO Technology,” First Break 13, no. 4 (April1995): 139-145.

15. Lu HZ and Lines L: “AVO and Devonian Reef Explo-ration: Difficulties and Possibilities,” The LeadingEdge 14, no. 8 (August 1995): 879-882.Ross CP and Kinman DL: “Nonbright-Spot AVO:Two Examples,” Geophysics 60 (September-October1995): 1398-1408.Hall DJ, Adamick JA, Skoyles D, DeWildt J andErickson J: “AVO as an Exploration Tool: Gulf ofMexico Case Studies and Examples,” The LeadingEdge 14, no. 8 (August 1995): 863-869.Peddy CP, Sengupta MK and Fasnacht T: “AVO Anal-ysis in High-Impedance Sandstone Reservoirs,” TheLeading Edge 14, no. 8 (August 1995): 871-877.

nThe BK field oper-ated by OryxEnergy Companyoff the LouisianaGulf Coast.

offset volume with offset ranges from zeroto 3800 m [12,468 ft] and a far-offset vol-ume with offsets from 3800 to 5800 m[19,024 ft] (left).

Forward modeling using logs from pro-ducing wells indicated the gas zones havean AVO signature of amplitude increasingwith offset. Interpretation consisted of find-ing other areas in which the near-offset vol-ume has low amplitudes and the far-offsetvolume has higher amplitudes.

The technique is demonstrated on a pair ofseismic lines extracted from the 3D volume.The AVO signature on Line 1215 at the gas-producing well BK-15 is the standard towhich Line 1235 is compared to determinethe likelihood of hitting gas at the proposedlocation BK-16. A color-coding system wasdevised to discriminate increasing AVOtrends from decreasing ones (below left ).Results of the analysis show the BK-16 loca-tion to be similar to, and perhaps even morepromising than, the producer BK-15 (below).

Initial production from the BK-16 well was15.4 MMcf/D and 210 barrels of condensateper day from 25 m [82 ft] of 20% porositysand. Sand quality is better than that foundin the BK-15 well, refuting speculation thatsand quality degrades to the northwest. Andfollowing the BK-16 well, two additionalsuccessful wells have been drilled within theregion of AVO gas signature.

12 Oilfield Review

Far offsets

Offset 3Offset 4

Offset 2Offset 1

21Near-offset

stack

43Far-offset

stack

Near offsets

CMP

CMP

Far-offset cube

Near-offset cube

BK-16Line 1235

Line 1215BK-15

nConstruction of near-offset and far-offset cubes. Offsets less than 3800 m [12,464 ft]are stacked to create a trace in the near-offset cube, and offsets from 3800 to 5800 m[19,024 ft] are stacked to form a far-offset volume.

nThe AVO signa-ture at the gas pro-ducer and at theproposed welllocation. Red indi-cates amplitudeincreasing with off-set (near offsetssmaller amplitudethan far offsets) atthe top of the reser-voir and yellowindicates the sameresponse but forthe bottom of thereservoir. Thedesired AVO signa-ture, as seen at theBK-15 location, is ared-over-yellowsequence at 3.7sec. The same sig-nature is present atBK-16, indicatingthe likelihood offinding gas there.(Adapted from RossCP, reference 14,courtesy of Black-well Science.)

nReservoir quality map for the BK fieldcreated from AVO analysis. Qualityincreases from gold to orange. (Adaptedfrom Ross CP, reference 14, courtesy of Black-well Science.)

3.2

3.4

3.6

3.8

4.0

3.2200 250 300 350

3.4

3.6

3.8

4.0

Tim

e, s

ecTi

me,

sec

West East

200180

180

250

BK-15

Line 1215

BK-16

300 350CDP number

CDP number

West East

Line 1235

Pickerillfield

UNITED KINGDOM

Edinburgh

Aberdeen

London

N O R T H S E A

N

nLocation of the Pickerill field, operatedby ARCO, in the southern gas basin of theNorth Sea.

nTop: seismic reflection strength colorcoded on the surface of the faulted reser-voir. Middle: visualization of drilling tra-jectories. Bottom: verifying the trajectoryof an alternative well path planned toavoid the Plattendolomit. Dolomite “float-ing” above the reservoir can be mappedto assess the risk of well paths intersect-ing this overpressure hazard.

16. Dorn G, Cole MJ and Tubman KM: “Visualization in3-D Seismic Interpretation,” The Leading Edge 14,no. 10 (October 1995): 1045-1049.

Plattendolomit

Plan View

Well Projection

Optimized Well Trajectory

Top Reservoir

Top Reservoir

Approximatereservoirboundary

1 km

0.6 miles

Seeing is BelievingSometimes just seeing the interpreted seis-mic data from a new point of view can shedlight on reservoir complexities and helpengineers plan and manage development.With the arrival of powerful graphics work-stations, visualization has become a key ele-ment in integrated reservoir characterizationstudies. Workstation visualization allowssimultaneous display of data from varioussources and enhances the communicationof ideas and problems among technical per-sonnel and management. Visualization itselfcan at times reveal something about thereservoir that was not previously suspectedor understood.

Examples from ARCO operations in thePickerill field in the southern North Sea gasbasin demonstrate the value of 3D visualiza-tion as a tool to help in well planning(above ).16 Early drilling revealed someobstacles to effective field development.First, reservoir porosity varies between lessthan 8% and greater than 20%, and the lat-eral variation is quite rapid—the reservoir is

Winter 1995

highly faulted, and even small-throw faultscan create barriers to flow of gas, due todiagenesis along the fault surfaces. Second,a discontinuous dolomite floats within theZechstein evaporites overlying theRotliegend reservoir. Exploration drillingmet with overpressure problems when thewells penetrated the dolomite, while over-pressure was not encountered when theborehole avoided the dolomite. The over-pressure represents a drilling hazard and apotential cost to be avoided.

Visualization techniques used as part of areservoir characterization study helpedtackle these problems. A highly detailedreservoir fault interpretation was developedby combining an attribute of the seismicdata—the reflection strength—with theinterpreted structure in a 3D display (left,top). By casting a light on the 3D surface,which was colored according to reflectionstrength, interpreters were able to pick faultswith a vertical component of displacementas small as 3 meters [10 ft].

Planning safe well trajectories in areaswith dolomite sheets, locally known as Plat-tendolomit, can be optimized with 3D visu-alization. By simultaneously viewing thesurface of the top of the Rotliegend reservoirand of the Plattendolomit, proposed welltrajectories can be assessed for safety. A pro-posed well that penetrates the Platten-dolomit can be redirected to avoid overpres-sure problems (left, middle and bottom).

As part of the reservoir characterizationstudy, interpreters derived a correlation

13

between seismic reflection strength andporosity determined from well logs, thenmapped the lateral variation in porosity onthe reservoir surface (below). These displaysintegrate the estimated porosity and struc-tural information and present it in an easilygrasped, intuitive manner.

The results have been used to help guidethe location of development wells in thePickerill field. In four wells drilled withinthe first year after completion of the study,the actual porosity encountered—between11% and 15%—was slightly higher thanthat predicted, but well within the error ofthe techniques used.

14

Porosity, p.u.6 19

Reservoir Top Structure

Northern half

Southern half

Time-Lapse SeismicIf conditions are favorable, seismic surveysacquired at different times in the history ofthe reservoir can show how fluid fronts havemoved.17 The knowledge of fluid distribu-tion can help engineers identify unsweptzones and plan infill or injection wells tooptimize recovery.

To detect fluid changes, the differences inreflection amplitude or travel time of theseismic waves must be discernible abovedata noise levels. The rock properties thatinfluence seismic reflection response—den-sity and velocity—must show significantvariation with fluid content, pressure or tem-

nColor-codedporosity on a 3Dimage of the reser-voir surface. Visu-alization helpedlocate wells byavoiding faultingand tapping highporosity. Actualporosity encoun-tered was slightlyhigher than pre-dicted.

perature. The seismic surveys must alsohave acquisition and processing as similaras possible to ensure that all observed differ-ences can be interpreted as changes relatedto production. Accuracy and repeatability oftime-lapse seismic surveys may be signifi-cantly improved by using permanent sen-sors, either on land or on the seabottom.

Time-lapse seismic, sometimes called 4D,has proved an important tool for reservoirmanagement for Caltex Pacific Indonesia inthe Duri field of central Sumatra (next page,top left).18 The Duri field, with 5.3 billion bblOOIP, was expected to produce only 8% ofthe OOIP under primary recovery. Optimizedsteamflooding could bring ultimate recoveryto 60%—an incremental recovery of morethan 2 billion barrels of oil. Knowing wherethe heat is being placed in the reservoir iscritical to optimize the recovery. With thesereserves at stake, Caltex turned to a seismicmonitoring pilot study to understand thecomplex flow patterns in shallow Duri reser-voir rocks, and to evaluate the method’s suit-ability for large-scale application.

Laboratory tests on core samples indi-cated the steamflooding would reduce seis-mic velocities by up to 40%—a 15%decrease due to increased temperature, andat the highest temperatures, an additional25% reduction because of a water-to-steamphase transition (below).

Oilfield Review

nLaboratory results showing effect of tem-perature and pressure on seismic velocitiesof core samples. Steamflooding can reducevelocities by up to 40%—a 15% decreasedue to increased temperature, and at thehighest temperatures, another 25% fromthe water-to-steam phase transition.

Com

pres

sion

al v

eloc

ity, m

/sec

Temperature, °F0 100 200 300 400

1200

1600

2000

2400Kedua

LowerPertama

UpperPertama

500 psi

600 psi

430 psi

17. Albright J, Cassell B, Dangerfield J, Deflandre J-P,Johnstad S and Withers R: “Seismic Surveillance forMonitoring Reservoir Changes,” Oilfield Review 6,no. 1 (January 1994): 4-14.

18. Bée MF, Jenkins SD, Lyle JH and Murhantoro E: “4-DSeismic: A Powerful New Technology for MonitoringSteam Movements in Duri Field—Central Sumatra,”presented at the 23rd Annual Convention of theIndonesian Petroleum Association, Jakarta, Indone-sia, October 4-6, 1994.

600

600

miles0

0 km

Cambodia

Viet

nam

Thailand

Malaysia

Mala

ysia

Sumatra

Java

Singapore

Borneo

SO U THC

HI N

AS

EA

GULF OFTHAILAND

Laos

Myanmar

N

Durifield

Duri field

Currentsteamflood

Area ofdetail

1000 m

3280 ft

ReceiverShotProducer

Observation well

InjectornCaltex Pacific Indonesia’s Durifield of central Sumatra. The 3Dseismic survey geometry calledfor high-density sampling overthe pilot steamflood area.

nSeismic data from the baseline survey and five monitor surveys. As steamflooding proceeds, decreased velocities in treated layerscause an apparent sag in reflectors. The top yellow line tracks the reflection at the top of the reservoir, and the bottom yellow linetracks the reflection at the oil-water contact.

Baseline 2-month lapse 5-month lapse 9-month lapse 13-month lapse 19-month lapse

The seismic survey required a specialdesign, and all source and receiver parame-ters were tested in the field prior to the base-line survey to allow resolution of the shal-low 500-ft [152-m] target depth andoptimum repeatability (top, right).

The baseline survey was recorded onemonth before steamflooding. Full acquisi-tion and processing for the small 0.06-km2

Winter 1995

[0.02-square mile] survey took about oneweek. The survey was repeated five times inthe next twenty months (above).

Zones affected by the introduction ofpressure, temperature and steam can be rec-ognized in the monitor surveys. Seismicvelocities in zones surrounding the injectorwere so much slower than before treatmentthat the layers appeared to thicken and sagin the seismic images—an illusion createdby the increased travel time in those layers.

15

nPlan view of front movement at the time of each monitor survey. Travel time to thetreated zone is compared between the baseline survey and each subsequent survey(first five blocks of figure). A decrease in velocity is seen as a pull down—the reflectorappears pulled down—while an increase in velocity pulls the reflector up. The asym-metrical spread indicates a high-permeability trend to the north and west. The oil satu-ration distribution that would result after 19 months of steamflooding was simulated,

12 msec 12 msec

Pull up Push down

10% 55%Oil saturation

After 2 months After 5 months After 9 months

After 13 months After 19 months After 19 months

Zones farther from the injector showed avelocity increase at early times, because ofthe passage of the pressure front precedingthe arrival of the fluid.

The high-pressure and trailing high-tem-perature fronts spread asymmetrically fromthe central injector, indicating a high-per-meability trend heading north and west(right). Temperature data from two observa-tion wells and core-calibrated permeabilitylog data corroborate the presence of thefront and the permeability anisotropy sug-gested by the seismic data.

The seismic data were further examinedfor evidence of vertical sweep efficiency.Thermal effects were tracked in three differ-ent layers by correlating the percentagevelocity change in each layer with a devia-tion from ambient temperature (next page).The top layer was interpreted to have thelowest sweep efficiency, the middle layer tohave the highest, and the bottom layer inbetween. Prior to steamflooding, an inde-pendent reservoir quality analysis on corefrom the formations showed the same hier-archy: the top layer, with the most clay con-tent, was estimated to be the worst zone, themiddle layer the best and the bottom layerin between.

Encouraged by the feasibility demonstratedin the pilot study, Caltex has begun to imple-ment the technology on a large scale in theDuri field. The next phase, with the base sur-vey already shot in April 1995 and the firstmonitor survey planned in early 1996, cov-ers 35 injector patterns instead of one.

16 Oilfield Review

and shows good agreement with the location of the steam front imaged at that time(last block of figure).

Winter 1995

TemperatureAmbient100° F

Steam375° F

Upper Pertama

Lower Pertama

Kedua (deep)

132° F 100° F

237° F 167° F

104° F 112° F

Growing FieldsSeismic surveys can provide a wealth ofinformation beyond the structural frameworkfor which they are best known. However, thisis not yet a routine operation, and has to becarried out with care, using seismic and welllog data together, and a technique appropri-ate to the problem. Porosity and lithology canbe mapped from inversion results, and fluidcomposition can be predicted using AVOanalysis. Visualization is key when structureis complex, and time-lapse monitoring isappropriate when reservoir rock propertiesare sensitive to fluid changes.

Surveys need to be acquired, processedand interpreted quickly to make significantcontributions to the study. Some of the newtechniques are expensive, and they must bejustified. The additional information createsno value unless it changes the way a field isdeveloped or managed. Typically, however,only a small fractional increase in hydrocar-bon productivity is required to justify a seis-mic project in reservoir applications.

As fields mature, operators place greateremphasis on improving the profitability ofexisting assets through increased productionand improved efficiency. With this trend,reservoir geophysics will become morewidely used to extend field life and maxi-mize recovery. The field of reservoir geo-physics is developing to address the integra-tion of data of many scales and of differentphysical properties. More advances will bemade by trying the techniques in untestedareas and pushing the limits. —LS

17

nThermal effects tracked in three differ-ent layers to assess sweep efficiency. Thetop layer, Upper Pertama, was interpretedto have the lowest sweep efficiency, andthe middle layer, Lower Pertama, to havethe highest. Sweep efficiency of the deep-est formation, the Kedua, is between theother two. Reservoir quality analysis oncore from the formations showed thesame hierarchy.

1

Borehole Seismic Data Sharpen the Reservoir Image

Johnny RutherfordAmerada HessHouston, Texas, USA

Jack SchaffnerSt. Louis, Missouri, USA

Phil ChristieCambridge, England

Kevin DoddsAberdeen, Scotland

Dick IresonGatwick, England

Lucian (Sonny) JohnstonCaracas, Venezuela

Nigel SmithStavanger, Norway

Seismic surveys in the borehole deliver a high-resolution quantitative measure of the seismic response of the

surrounding reservoir. Although these measurements may be used alone to image local features, they may

also be tied with well data—logs and cores—and then related to more extensive surface seismic data.

Advances in borehole geophysics are helping realize the full potential of existing data to create a sharper

image of the reservoir.

8

For help in preparation of this article, thanks to ShabbirAhmed and Antoine Track, Schlumberger Riboud Prod-uct Centre, Clamart, France; Philip Armstrong, Geco-Prakla, Gatwick, England; Bruce Cassell, Wireline &Testing, Dubai, UAE; Alex Cisneros, GeoQuest, Houston,Texas, USA; Ian Gollifer, GeoQuest, Aberdeen, Scotland;Jakob Haldorsen and Doug Miller, Schlumberger-DollResearch, Ridgefield, Connecticut, USA; Brian Hornbyand Colin Sayers, Schlumberger Cambridge Research,Cambridge, England; Mike Jones, Wireline & Testing,Calgary, Alberta, Canada; Scott Leaney, Wireline & Test-ing, Jakarta, Indonesia; William Underhill, Geco-Prakla,Hannover, Germany; and John Walsh, Wireline & Test-ing, Houston, Texas.

In this article, ASI (Array Seismic Imager), CSI (Com-binable Seismic Imager), DSI (Dipole Shear Sonic Imager)and Formation MicroScanner are marks of Schlumberger.

It’s a matter of resolution. Surface seismicsurveys deliver one of the few quantitativemeasurements of reservoir properties awayfrom wells, making the technique central tostructural mapping of the entire reservoirvolume. However, surface seismic wavescannot resolve features smaller than 30 to40 ft [9 to 12 m]. On the other hand, logsand cores resolve features on the scale of afew feet down to about 6 inches [15 cm].Reconciling these two measurement scalesto get the optimal picture of the reservoirvolume is a problem that has long chal-lenged the industry.

Borehole geophysics has a foot in boththe logging and surface seismic camps.From the vantage of the wellbore, seismicdata often have higher resolution than theirsurface seismic counterparts. Depths ofeach borehole receiver are also known, pro-viding a better tie to the formation proper-ties provided by petrophysical, core andother in-situ measurements and relatingthem to the 3D seismic volume.

The idea of locating a receiver downholeand a seismic source at surface is not new.For more than half a century, the check shot

has helped to correlate time-based surfaceseismic surveys with depth-based logs.Check shots check the seismic travel timefrom a surface shot to receivers at selecteddepth intervals. Subtraction of times, com-bined with the depth differences, yields ver-tical interval velocities and thus relates welldepths to surface seismic times.

In vertical seismic profiles (VSPs), thespacing between downhole geophone levelsis considerably closer than for check-shotsurveys. VSPs use high-quality full wave-forms that include reflection informationrather than just the time of first arrivals —orfirst breaks—to create an image of reflectionsnear the wellbore.1 Building on this tech-nique, 2D reflection images have beenobtained by offset and walkaway surveyswith sources and receivers in a variety ofconfigurations that address most reservoirproblems (see ”The VSP Family,” page 22).

Oilfield Review

1. Babour K, Joli F, Landgren K and Piazza J-L: “Elementsof Borehole Seismics Illustrated with Three Case Stud-ies,” The Technical Review 35, no. 2 (April 1987): 6-17.Belaud D, Christie P, de Montmollin V, Dodds K, JamesA, Kamata M and Schaffner J: “Detecting Seismic Wavesin the Borehole,” The Technical Review 36, no. 3(July 1988): 18-29.

Yet despite these and other developments,borehole geophysics has for many yearsfailed to gain the status in reservoir charac-terization that some industry specialiststhink it deserves. Now, thanks to improvedquality and increased confidence in thematch between borehole and surface seis-mic data, borehole geophysics seems to bemoving into an increasingly valued position.

Before examining how borehole seismicdata are being used to successfully integrateother data, this article will illustrate how thescope of VSP is broadening through thedevelopment of horizontal, 3D and through-tubing techniques.

Winter 1995

Broadening the Scope of VSP ApplicationsIn the deviated and horizontal wells of theNorth Sea, the most common type of bore-hole seismic survey is the vertical-incidenceVSP. These are often called walk-above sur-veys because, as the geophone is movedalong the deviated section of borehole, thesource is kept vertically above it, “walkingabove” the well. In VSP terms, a horizontalwell is an extreme version of a deviatedwell. Like other VSPs, deviated well surveysmay be used for locating the well in the 3Dsurface seismic volume and assessing thequality of surface seismic surveys. Also, thetechnique may be employed for measuringlateral velocity variations and for imagingfaults and structures below the wellbore.

The following example of a walk-aboveVSP was carried out in late 1994, in a NorthSea well with a 1.2-kilometer [0.75-mile]horizontal section. There were two mainobjectives. The first was to measure a sus-pected lateral velocity anomaly that mayhave been creating artifacts in the surface

19

nWalk-above VSP in a North Sea horizontal well to assesslateral velocity variations and local structure. The first-arriving seismic signals are tracked with a red line, whichalso delineates the borehole trajectory in two-way time.

nThe shortcomings of conventional processing for VSPs inhorizontal wells. In conventional methods of VSP process-ing, the lack of apparent velocity difference betweendowngoing and upgoing wavefields leaves little or noreflected upgoing energy after wavefield separation.Boosting the gain in the horizontal section is of little value.There is poor continuity of reflected events, and the fault-scattered energy further complicates the image.

Two-

way

tim

e10

0 m

sec

Level number2401

Two-

way

tim

e10

0 m

sec

Level number2401

seismic data (above). The second was toobtain a high-resolution seismic imagebelow the deviated portion of the well. Anadditional objective was to obtain a seismicimage in the horizontal part of the well.2

Data were collected in the vertical anddeviated portions of the cased well usingthe conventional wireline-conveyed ASIArray Seismic Imager tool. In the horizontalsection, a two-element CSI CombinableSeismic Imager geophone array was run ondrillpipe in combination with a cementbond log. By decoupling the sensor modulefrom the body of the CSI tool, the geo-phones are isolated from noise and distor-tions created by the drillpipe.

20

As with any survey, the desired seismicimage is produced using the reflected, orupgoing, wavefield. So the first processingtask was to separate downgoing waveformsfrom upgoing. For walk-above surveys inhorizontal wells, this is far from straightfor-ward, since unlike vertical and deviatedwells, there is no apparent time differenceacross the array between the downgoingand the reflected upgoing waves. It is there-fore impossible to use conventional tech-niques to distinguish between reflectionsand downgoing waves (above, right ). Toimprove the image a number of specialtechniques were used, including:• multichannel filtering to attenuate noise

and sharpen the desired signal3• downgoing wavefield subtraction

using a long filter length to estimate the downgoing wavefield

• median filtering techniques to estimateand subtract the energy scattered by faults

• enhancement of the desired upgoing signal

• equalization of the reflected wavefieldamplitudes from the horizontal and thebuild up sections.The final image showed three important

features: the two faults marked A and B,which appear where suspected in thereflected image, and the dip of the stratabelow the well (next page, left). FormationMicroScanner data acquired during open-hole logging were compared with the VSP,confirming the fault locations—seen aschevrons in the VSP—and the apparent dips.

In this case study, VSP processing wasperformed before Formation MicroScannerdata were ready to interpret, and the VSPhelped the interpretation by outlining the

Oilfield Review

2. Ediriweera K, Smith N and Prudden M: “BoreholeSeismic Surveys in Horizontal Wells—A Case Studyfrom the North Sea,” paper B018, presented at the57th EAEG Meeting and Technical Exhibition, Glasgow, Scotland, May 29-June 2, 1995.

3. Haldorsen J, Miller D and Walsh J: “MultichannelWiener Deconvolution of Vertical Seismic Profiles,”Geophysics 59 (October 1994): 1500-1511.

nHorizontal VSP data after improved processing. Thereprocessed data show three important features: twofaults marked A and B which appear as anticipated inthe reflected image, together with evidence of dipping.The apparent formation dips seem to be parallel to theborehole until very near total depth. This turned out to beentirely consistent with the Formation MicroScanner com-puted dips.

Two-

way

tim

e10

0 m

sec

Level number2401

BA

=-

Direct,downgoing wavelet

Reflected,upgoing wavelet

Geophone Hydrophone

nSeparating downgoing and upgoingwaves in horizontal VSPs using geophoneand hydrophone seismic signals. A wavearriving from above pushes the sensor inthe geophone down, recording a positivemotion (trace swing to the right). The samewave, having reflected below the geo-phone without changing polarity, pushesthe sensor upwards as it arrives frombelow (swing to the left). The resulting seismic trace consists of a positive down-going then a negative upgoing pulse. A hydrophone, by comparison, produces a seismic trace with two wavelets of thesame polarity. Therefore, the reflectedevent as seen by a geophone is of oppositepolarity to that seen by the hydrophone.Subtraction of one seismogram fromanother, after appropriate scaling, elimi-nates the downgoing wavelet, leaving the reflected signal.

major features. The two data sets were theninterpreted and refined together, providing amore complete description of near-wellgeology than was otherwise available. Theresults met the main objectives of the surveyand delivered an image below the horizon-tal section.

An alternative strategy for acquiring andprocessing horizontal VSP data exploits thedifferent responses of geophones andhydrophones to differentiate downgoingenergy from upgoing energy in horizontalwells. Geophones are clamped to the for-mation, and sense its motion. In contrast,hydrophones are suspended in the boreholefluid and are sensitive to fluid pressurechanges as a seismic wave passes in anydirection. When the two sensor types show

Winter 1995

the same signal polarity for a downgoingwave, they show different polarities for theupgoing wave. By taking the differencebetween signals received at the two types ofsensors—for a signal consisting of a directpulse followed by a reflected pulse—thedirect wave is canceled and the reflectionenhanced (above, right).

Complications arise from differences in thecoupling and impulse responses betweengeophones and hydrophones. However, thisapproach has recently been applied in thefield, enabling the extraction of reflectedwavefields in a horizontal well and the imag-ing of reflectors below the receivers.

21

22 Oilfield Review

The VSP Family

Reflected upgoing multipleReflected primary

Source

Reflector

Downgoing multiple

Direct wave

Geophone positionTime

Subsurface reflector

Zero-offset VSP

The members of the family of borehole seismic

measurements differ in the number and location

of sources and geophones used and how they

are deployed.1

The simplest of all services—in common prac-

tice by 1940—is the check shot, sometimes

called a velocity survey.2 Check shots measure

direct travel times from source to receiver, with

no reflections along the way. This provides a

measure of seismic velocity near the well and

relates seismic time to well depth. The check-

shot service deploys a stationary seismic

source, while a single downhole geophone is

moved to locations in the well indicated by the

well logs, measuring the travel times to specific

reflectors. The first arrivals—or first

breaks—recorded on the seismic traces are

picked to deliver the time-to-depth information.

This geophone and source configuration is

similar for the next member of the family, the

zero-offset VSP. The zero-offset VSP has its ori-

gins in the 1950s.3 The source is located directly

above the receivers (above, right). However, to

obtain an image of subsurface reflectors, a

higher density of receiver positions is used than

in a check-shot survey and trace recordings

extend beyond the first breaks to include later-

time reflections.

Next in the family comes the offset VSP, in

which a single surface source is positioned at a

substantial distance—termed offset—from the

well (next page, top). This shifts the reflection

points away from the well and extends the

subsurface coverage, helping to detect faults,

for example.

The check shot and the two VSP techniques

described above are multireceiver, single-source

techniques. The walkaway VSP departs from

this. In its simplest form, a receiver array of five

to seven geophones collects data from multiple

surface source locations along a line that

extends from the well. Each line typically has

hundreds of source positions. Reflections from

each horizon below the geophone offer an

umbrella-shaped coverage of the formation

alongside and beneath the well. These data may

then be processed to create an image that usually

has higher resolution than that from surface seis-

mic surveys. The acquisition of 3D VSP involves

multiline walkaway profiles.

Somewhere between single-source and walka-

way VSP is the VSP in deviated and horizontal

wells—often called a walk-above VSP. In this

technique a source may be positioned directly

over the receiver to map a deeper reflector and to

map a deviated well onto a seismic section.

Three special members of the VSP family are

salt-proximity surveys, shear-wave VSPs and

drill-noise VSPs. Salt-proximity surveys, which

originated in the 1930s, are recorded in wells

adjacent to salt domes with the source placed

immediately above the salt dome.4 Travel-time

information and the polarization of first arrivals

are measured by the downhole geophone at vari-

ous locations in the well.

Knowing the location of the receivers and the

source, the velocity of the salt, the velocity of the

sediment layers and the distance to the top of the

salt dome, a travel-time inversion may be per-

formed to determine the locations of points where

rays exit the salt dome. This allows a profile of

the salt dome to be constructed, which may be

used to determine the lateral distance from the

well to the salt, and also to identify possible over-

hangs and potential traps along the salt flank.

As the name suggests, shear-wave VSPs are

VSPs recorded with shear-wave sources, usually

shear-wave vibrators (next page, middle). These

may be used in a manner similar to P-wave VSPs

to create a high-resolution image of reflectors.

However, another application is to measure a

phenomenon known as shear-wave splitting. This

is associated with anisotropy due to stress or ver-

tically aligned fracture systems. Shear waves

travel faster when their particle motion is polar-

ized in the plane of fractures than when it is per-

pendicular to the fracture plane. Shear-wave VSPs

have been used to determine the expected orien-

tation of induced fractures, the orientation of nat-

ural fractures and stress directions, and a qualita-

tive indication of fracture density.5

The drill bit seismic technique, sometimes

called drill-noise VSP or seismic-while-drilling,

reverses the geometry of the source and receiver

(next page, top). The drill bit itself is the seismic

source, and receivers are placed on the surface.6

Clever processing can image the reservoir or a

drilling hazard, such as overpressure, as the well

is being drilled.

23Winter 1995

1. For a review: Oristaglio M: “A Guide to Current Uses of Vertical Seismic Profiles,” Geophysics 50 (December 1985):2473-2479.

2. Heiland CA: Geophysical Exploration. New York, NewYork, USA: Prentice-Hall, 1940.

3. Jolly RN: “Deep-Hole Geophone Study in GarvinCounty, Oklahoma,” Geophysics 18 (July 1953): 662-670.

4. McCollum B and LaRue WW: “Utilization of ExistingWells in Seismograph Work,” AAPG Bulletin 15(December 1931): 1409-1417.Gardner LW: “Seismograph Determination of Salt DomeBoundary Using Well Detector Deep on Salt DomeFlank,” Geophysics 14 (January 1949): 29-38.

5. Sun Z and Jones M: “Shear Anisotropy Analysis fromDirect and Converted Wave VSP Data Using VariousAlgorithms,” paper BG4.5, presented at the 64th SEGAnnual International Meeting and Exposition, Los Angeles, California, USA, October 23-28, 1994.Winterstein DF and Meadows MA: “Shear-Wave Polar-izations and Subsurface Stress Directions at Lost HillsField,” Geophysics 56 (September 1991): 1331-1348.

6. Meehan R, Miller D, Haldorsen J, Kamata M and Under-hill B: “Rekindling Interest in Seismic While Drilling,”Oilfield Review 5, no. 1 (January 1993): 4-13.

■■Prediction to depthof a drilling hazard,continuously updatedwith drill bit seismicwellsite processing.Acoustic impedance intwo-way time from aVSP acquired at a shal-lower drilling depth(top) shows a drillinghazard in the form of a decrease in acousticimpedance at 2.19 sec(blue line). This is con-verted to depth usingtime-depth pairsderived from drill bitseismic processing(bottom). The earlyprediction of the depthto the drilling hazard at2673 m (gold line) isupdated to a new,deeper prediction, 2684 m (red line).

Shear-Wave VSP

Fractured layer

Two shear-wavepolarizations

2200

Dep

th, m

2100

2000

2300

2400

2500

2600

2700

2800

6000

7000

8000

Aco

ustic

impe

danc

e,m

/sec

x g

/cm

3

Two-way time, sec

1.8 2.0 2.2 2.4

Old hazard depth = 2673 m

New hazard depth = 2684 m

Offset VSP Walkaway VSP Walk-above VSP Salt-Proximity VSP

Salt

Drill-Noise VSP

Receivers

Drillbit

Ordinary VSPs can give an indication of such

features below the current total depth of the well,

by flagging zones of anomalous acoustic

impedance. But VSP data can be converted to

depth only as far down as the lowest borehole

receiver level. After that, results are in time, and

so not very useful to drillers.

The drill bit seismic technique can complement

the time-based acoustic impedance profile

derived from VSPs by providing a means to mea-

sure the time-depth information below the bottom

VSP receiver level as the drilling progresses.

Thus, the depth index of the acoustic impedance

prediction can be continuously updated at the

wellsite, showing clearly when a suspected

drilling hazard is about to be hit (left).

North

South

To V20(E)

End of Survey

fromV17(W)

A 591762 N591576 N

591949 N

24 Oilfield Review

nMarine 3D VSP acquisition.Sail lines are planned tominimize turning, and satel-lite navigating systems trackthe shot points.

4. Christie PAF and Dangerfield JA: “Borehole SeismicProfiles in the Ekofisk Field,” Geophysics 52, no. 10(1987): 1328-1345.

5. Johnstad SE and Ahmed H: “Reservoir MonitoringUsing the Vertical Seismic Profiling (VSP) Technique, ACase Study in the Oseberg Field,” presented at the 54thEAEG Annual Meeting and Technical Exhibition, Paris,France, June 1-5, 1992.

6. Belaud et al, reference 1.

7. Van Der Pal RC, Bacon M and Pronk DW: “EnhancingSeismic Resolution in the Brent Field by a 3D VSP,”paper B017, presented at the 57th EAEG Meeting andTechnical Exhibition, Glasgow, Scotland, May 29-June 2, 1995.

8. GPS is Global Positioning System.A description of the GPS technique may be found at:http://www.utexas.edu/depts/grg/gcraft/notes/gps/gps.html

9. Johnston LK: “Borehole Seismic Applications Using a Slim Well Seismic Receiver in a Production Environ-ment,” presented at the 1st Latin American Geophysi-cal Conference and Exposition of the LAGU, Rio deJaneiro, Brazil, August 20-24, 1995.

3D VSPsVSP imaging surveys, such as walkaways,have been used for a number of years toimage structural complexity away from theborehole.4 These walkaway profiles areessentially two-dimensional, confined to thevertical plane containing the surface sourceand the borehole.

Because of the proximity of the receiversto the target, like all VSPs, these 2D imagesusually have the advantage of being ofhigher resolution than their surface seismiccounterparts. But, by definition, 2D walka-ways don’t describe the full volume of thereservoir. Fortunately, the acquisition princi-ple may be extended to cover three dimen-sions by repeated profiling in parallellines—in effect, by collecting a series of 2Dwalkaway surveys similar to marine 3D seis-mic data acquisition (below).

The progression from 2D to 3D in VSPsurveys is similar to the progression in thesurface seismic technique, and offers equiv-alent benefits. Thus, 3D VSPs allow high-resolution imaging to augment surface 3Dsurveys and make it possible to obtain

images beneath surface obstacles, such asplatforms, and near-surface obstructions,such as shallow gas zones. In addition,because the acquisition conditions and pro-cessing steps of VSP surveys are accuratelyreproducible, 3D VSP opens up the possibil-ity of time-lapse, or 4D, seismic surveying.5

However, progressing from 2D to 3D sub-stantially increases the need for planningand logistics control. Similarly, the process-ing requirements are almost an order ofmagnitude greater.

The first 3D VSP survey was run in 1987in the Adriatic Sea Brenda field, operated byAGIP.6 Since then, there have been two 3DVSP surveys in the Norwegian Ekofisk fieldfor Phillips Norway—where a large gasplume over the center of the structure pre-vents imaging using conventional 3D surfaceseismic techniques. Other Norwegian sur-veys probe the Eldfisk and Oseberg fields.

In the UK North Sea, a 41-line, 3D walk-away VSP survey has been carried out inShell Expro’s Brent field. In this case, theaim was to acquire a survey with improvedresolution compared with the 3D surface

seismic survey. The image was then to beused to produce an accurate structural mapto aid the planning of horizontal develop-ment wells in the Brent slump—a crestalzone of complex faulting and collapsewhich contains a significant portion of thefield’s remaining oil reserves.7

The survey was executed from a well witha trajectory that allowed positioning the geo-phones to give three-dimensional illumina-tion of the slump zone. The receivers con-sisted of five shuttles with fixed triaxialsensors, clamped 2000 ft [606 m] above thetarget during the entire survey. Once in thewell but prior to shooting, the couplingbetween each of the shuttles and the forma-tion was evaluated using internal shakers toensure distortion-free data.

The seismic source consisted of a clusterof three 150-in.3 [2460-cm3] sleeve guns. Tosupply sufficient gas for 41 lines of 200shots per line, four 5100 cubic meter nitro-gen-filled tube skids were used. Simultane-ously with the downhole data acquisition,each shot location on the surface wasrecorded using two differential GPS naviga-tion systems.8

To make the survey cost-effective, it wasvital to minimize time spent acquiring data—every extra minute per sail line meant anadditional 41 minutes of rig time. For exam-ple, to reduce the time the vessel took tomaneuver between lines, a strategy wasdevised to wrap each line efficiently into thenext (far left ). In the end, the data wereacquired within the planned survey time oftwo and a half days, including a conven-tional VSP.

The 3D processing involves an extensionof methods already developed for 2D walk-aways—data preparation and navigationcheck, triaxial projection, wavefield separa-tion, deconvolution and migration (see “3DVSP Processing,” page 26).

In this case, the processing consisted ofseparate preparation and processing of all 41lines up to the deconvolution stage. Then all41 reflected energy profiles were accessed bythe 3D VSP migration algorithms to place thereflections correctly in space.

The successful processing of these surveysrequired an experienced geophysicist withstrong interpretative skills to make the cor-rect decisions at each stage of the process-ing—for example, to ensure that all possiblequestions related to the influence of dataquality had been resolved. These skillsensured that the image was interpreted interms of reservoir structure without process-ing artifacts.

The migration process requires the compu-tation of raypaths from each source and every

25Winter 1995

nThrough-tubing, through-casing VSP.The offset VSP survey was designed toconfirm the location of a low-angle fault(yellow) that could not be seen on the sur-face seismic images. It was easilylocated from the offset VSP image.

250

mse

cTi

me

Distance, ft 01650

receiver to every reflection point in thesubsurface. The rays are traced through avelocity model of the subsurface that canvary in complexity between flat layers (a 1Dlayercake) to complex structures in 2D or 3D.

For simple structures, a layercake velocitymodel, which reduces computation time, issufficient. However, using this model inmore complex subsurface may lead to erro-neous positioning of reflections and theincorrect focusing of real events. Morecomplex velocity models increase the num-ber of ray-trace computations required, butare better able to position reflected eventsand focus the wave energy.

The Brent structure varies in the dipdirection but changes very little alongstrike. Consequently, the velocity model ismore complex than a plain 2D model butnot as complex as a full 3D model; thestructure varies in one horizontal directionand is extruded into the other horizontaldimension to form a so-called “2.5D”model. In this, the volume may be thoughtof as filled with an infinite number of 2Dsections. This allowed computational effi-ciency due to symmetry and ensured aclose match with the actual Brent structure.

Shell concluded that the Brent 3D VSPimproved vertical resolution and signifi-cantly improved horizontal resolution—resolving features on the order of 100 to 150ft [30 to 45 m] as opposed to the original3D surface seismic resolution of 200 to 300ft [60 to 90 m]. The interpretation of theslump features has confirmed conclusionsreached independently, demonstrating thetechnique’s potential and reducing the riskof a proposed new 3D surface survey.

Through-Tubing VSPsThe third application broadening the scopeof borehole geophysics is the VSP throughtubing. Thanks to hardware developments,cost-effective VSPs can be run in maturefields that promise significant economicbenefits.

Traditionally, borehole seismic surveys areacquired in exploration wells when they aredrilled. However, in older fields, boreholeseismic information is often needed to aidthe reservoir engineer in areas where nonew wells are planned, or to plan a newwell. Now a slim seismic receiver may bedeployed by a simple masted logging truckto acquire borehole seismic data throughproduction tubing and inside casing duringworkover or while the well is still on pro-duction. This reduces acquisition costs and

makes surveys in multiple wells possibleduring the same mobilization.

In this way, a full range of borehole sur-veys may be carried out and the data maybe used to tie log and production informa-tion to new 3D surface seismic surveysbeing run in older producing fields.

The slim seismic tool has a 111/16-in. out-side diameter and may carry one single-axisgeophone group or three orthogonallymounted accelerometers. The mechanicallyactuated anchor has a maximum opening of7 in. [17 cm]. The tool is adapted for opera-tion with a monocable wireline andthrough-wellhead pressure fittings. Thisallows for operations in producing wellswith surface pressure. As with any system, arange of seismic information may beobtained in vertical or deviated wells, fromcheck shots to walkaway VSP images.

For example, an offset VSP survey wasacquired through tubing and through casingin an abandoned well in an inland shallow-water field in south Louisiana, USA, using amarine vibroseis unit as a source to acquirehigh-resolution data.

9 The offset VSP surveywas designed to confirm the location of alow-angle fault—indicated by logs—whichcould not be seen on the surface seismicimages. The fault’s orientation was neededto reduce the risk of an infill developmentwell and was easily spotted using the offsetVSP image (above).

Using Borehole Geophysics to Integrate DataAt the heart of developments to improvedata integration is the recognition of thecomplementary nature of some measure-ments. Perhaps the best example of this isthe relationship between sonic logs andseismic data. In these two measurements,the physical interaction with the reservoir isthe same, but at a different scale of resolu-tion. The sonic tool measures formationcompressional slowness, which is depen-dent on many factors, including the forma-tion porosity and lithology. Compressionalslowness combined with density providesthe one-dimensional acoustic impedance ofthe formation, the same property that under-lies seismic reflections.

But seismic waves are sensitive only torelative changes in acoustic impedance,unlike sonic slowness measurements, whichsample absolute values. Therefore, acousticimpedances from logs provide sufficientinformation to model most, but not all fea-tures of the seismic response. The totaltravel time measured by sonic logs is arequired contribution to the bulk responseof the low-frequency surface seismic sur-veys. Then, synthetic seismograms may beconstructed and the response of the forma-tion simulated by altering parameters suchas porosity, fluid type and lithology. The syn-thetics can be used to interpret real data.

Although the scope of VSPs is expanding,the wealth of information relating to lithol-ogy, fluid contacts and the seismicresponses that they produce is not alwaysused to its fullest extent. This is particularlytrue when it comes to evaluating andimproving the information content of sur-face seismic data. Now, existing technolo-gies are being used in new ways to provideadditional direct quantitative measurementsof the seismic response of the reservoir adja-cent to wells.

The next two examples clearly indicatehow the integration of all available data mayimprove understanding of the reservoir. Thefirst example looks at how structural andstratigraphic interpretations may beimproved. The second shows how reflectionamplitude variation with offset (AVO) fromVSPs may be used to calibrate surface seismic AVO.

26 Oilfield Review

■■Synthetic offset VSP generatedfrom a model including the abrupttermination of Hackberry sand.This, and another model based ongradual termination of the Hack-berry (not shown) indicate that anamplitude decrease in the Hack-berry can be detected, as distancefrom the well increases.

Well #10 Well #8

1.0

1.5

2.0

Above 0.1060.079 to 0.1060.053 to 0.0790.026 to 0.053

-0.001 to 0.026-0.027 to -0.001-0.054 to -0.027-0.081 to -0.054 -0.107 to -0.081Below -0.107Undefined area

Tim

e, s

ec

2000 0

Hackberrysand

Distance, ft

Amplitude

Morgan’s BluffIn the Morgan’s Bluff field of Orange County,Texas, USA, the operator IP Petroleumneeded to map the shale edge of its Hack-berry reservoir to design a secondary recovery program.

Substantial existing 2D surface seismicdata did not adequately image the reservoir.Therefore, vertical incidence and offset VSPswere shot within a production well. Theseresults were combined with logs and geo-logic information to map the edge of theshale. Further, the surface seismic lines werereinterpreted, resulting in an extensiveremapping of the Hackberry sand.10

The aim was to drill a sidetrack from theshut-in producing Well 8 toward the adja-cent Well 10, depending on the exact reser-voir boundary, to be determined using theVSP—the Hackberry sand was originallymapped on the strike line that runs throughboth of these wells.

First, the feasibility of this plan was testedand detailed survey models were con-structed using structure maps, log data fromthe two wells and velocity data from a thirdwell. Borehole seismic data shot in 1986 inthe central part of the field were used toconstruct the general velocity model. InWell 8, sonic logs were available to about8000 ft [2440 m], and only nuclear andresistivity logs from there to total depth. Apseudosonic log was constructed from theselogs and compared to the velocities from

3D VSP Processing

Data Preprocessing—This includes loading and

verification of the navigation or geographical sur-

vey data for each shot location and the display of

the data for each receiver level and for each of

three components for quality control purposes.

Triaxial Projection—To acquire data rapidly, a

tool with multiple triaxial geophone “shuttles” is

deployed in the well. However, each shuttle can

have a different, unknown orientation in the

borehole, and this orientation may change if the

tool is moved and clamped at a new level. Thus,

waves arriving from one direction, say a reflec-

tor, may have a different appearance on each

receiver. Triaxial projection converts the data

from the recording geometry to the geometry of

the arriving waves, to make the data suitable

for processing.

Wavefield Separation—The energy arriving at

the geophone consists of energy arriving from

above overprinting the energy reflected from the

target formations below. Wavefield separation

discriminates between this “downgoing” and

“upgoing” energy in the received wavefield. This

is achieved by velocity filters, which enhance the

coherency of events aligned with a given appar-

ent velocity relative to the receiver geometry.

The same technique may be used to separate

compressional from shear waves, since they

travel at different velocities. A popular method

for wavefield separation is called parametric

wavefield decomposition.1

Deconvolution—In addition to reflections, the

earth creates unwanted additional seismic events

called multiples, and it attenuates higher fre-

quencies more than lower frequencies, producing

signals with both desirable and undesirable infor-

mation. The process of deconvolution, of which

there are a number of varieties, attempts to undo

the excess work of the earth so that only signals

related to reservoir features remain. Through

deconvolution, the recorded downgoing wavefield

is converted into an idealized downgoing wave-

field. The filter that accomplishes this is then

applied to the upgoing wavefield to produce sig-

nals that would have been recorded if the experi-

ment had been perfect. Special deconvolution

methods have been developed for walkaway data.2

3D VSP migration—The migration technique used

for 3D VSPs allows the repositioning of events and

focusing of energy to their appropriate positions in

space. Possible raypaths and travel times are com-

puted by tracing rays through a velocity model of

the subsurface.3 Then all source-receiver pairs in

the seismic data volume are checked for energy

that could satisfy such path and time constraints.

In contrast to the migration of 3D surface seismic

data, the VSP migration algorithm includes spe-

Tim

e, s

ec

1600 ft

Well 10 Well 8

1600 ft

Well 10 Well 8

0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

0

0.5

the VSP survey. A synthetic offset VSP wasthen generated using the same wavefieldseparation, deconvolution and migrationprocessing to be used with the real data.

Two scenarios were forward modeled: agradual shaling out and an abrupt, orfaulted, sand termination (previous page).From this it was agreed that in either casethe shale boundary should be interpretableto within 100 ft using the offset VSP sec-tions, and the go-ahead for the survey wasgiven. Additionally, a second offset VSP tothe west of Well 8 was designed to confirmthe interpretation. A VSP was also to be car-ried out in Well 8 to build an updatedvelocity model for migration.

The three downhole surveys were acquiredwith sources located 4000 ft [1212 m] to thewest-southwest, 4300 ft [1300 m] to thesouthwest and 400 ft [121 m] to the east-southeast. An eight-level downhole receiversystem was deployed to record 110 levels at50-ft [15-m] spacing from 8500 to 3000 ft[2575 m to 909 m]. Across each interval, thetop and bottom shuttles were overlapped tocheck for any source amplitude, signature orphase changes during the survey.

Following a standard processing sequenceusing a flat-layer velocity model and somesmall velocity changes to match the modelto the observed transit times, each of the off-set VSPs was migrated. Logs from Well 8were correlated with the offset VSPs.

Formerly, the dominant reflection at 2.2 to2.25 sec on the surface seismic line had

■■ Original (top) versus revised interpreta-tion. The reinterpretation (bottom) basedon results from the borehole seismic sur-veys suggested that during the Hackberrydeposition, a small slump fault occurredto the southwest of the present field. Thiswas scoured out and filled with shale,serving as a trap.

Winter 1995

1. Leaney, main text reference 19.

2. Haldorsen et al, main text reference 3.

3. For an explanation of this migration technique:

Van Der Poel NJ and Cassell BR: “Borehole Seismic Surveys for Fault Delineation in the Dutch North Sea,” Geophysics 54 (September 1989): 1091-1100.

Miller D, Oristaglio M and Beylkin G: “A New Slant onSeismic Imaging: Migration and Integral Geometry,” Geophysics 52, (July 1987): 943-964.

Tim

e, s

ec

1.0

1.5

2.0

2.5

3.0

3.5

been thought to be the Hackberry sand.Now it has been interpreted as a calcareousstreak that runs throughout the field. It isseen on the logs at 8310 to 8360 ft. Thereflection package from the Hackberry sandactually begins at 2.31 sec and ends at 2.36sec. While this was a subtle reflection in thesurface seismic data, it clearly shows up onthe offset VSPs and was confirmed by thetime-based logs from Well 8. The sandsshale out abruptly around 500 ft [151 m] tothe southwest.

This interpretation was then combinedwith existing offset and seismic data to yielda new interpretation of the shale boundary.The termination of the Hackberry sand,along with images of reflections from olderstrata some 150 msec below, generated anupdated seismic interpretation (left). Basedon this survey—at a cost of about $50,000—the IP Petroleum engineers decided not tosidetrack Well 8, saving about $500,000.

The remapping of the Hackberry sandusing the seismic tie developed above sug-gested that a secondary high might exist inthe field in a zone that had previously beenmapped as a water leg. When a well target-ing this zone was drilled, 40 ft [12 m] of oilwas discovered on top of the water, addingnew reserves to the field.

AVO in VSPsWhen a wavefront hits a boundary at verti-cal incidence, the amount of compressionalenergy reflected and transmitted is depen-dent only on the contrast of acousticimpedance—density times compressionalvelocity—of the rocks at that boundary. Butwhen the incident angle is not 0°, theamount of compressional energy reflectedor transmitted depends on the angle of inci-dence, or source offset, and contrasts indensities and shear and compressionalvelocities.11 In such cases, the reflectionAVO can be measured and analyzed toyield information about lithology and porefluid through their effects on density andcompressional and shear velocities.12

Carrying out a walkaway VSP with thereceivers straddling such a boundary allowsdirect measurement of the variation in

cific provision for the different source and down-

hole receiver geometry.

The migration applied to 3D VSPs is a full 3D

migration, meaning it considers the positions of

reflectors in the imaged volume, rather than in a

series of 2D slices. The migration algorithm is

based on the seismic wave equation, and applied

prestack, to preserve true reflection amplitudes.

The velocity model used for migration may be

completely 3D with structural complexity and

velocity heterogeneities.

27

10. Schaffner J, Reisinger M and Rutherford JW: “OffsetVertical Seismic Profiles Define Shale Boundaries inMorgan’s Bluff Field,” The Leading Edge 13, no. 1(November 1994): 1095-1100.

11. This new dependance on shear velocity arisesbecause some compressional energy is converted toshear energy at the boundary.

12. For an overview of the AVO technique: Chiburis E, Franck C, Leaney S, McHugo S and Skidmore C: “Hydrocarbon Detection with AVO,”Oilfield Review 5, no. 1 (January 1993): 42-50.Leaney WS: “Anisotropy and AVO from Walk-aways,” paper BG4.4, presented at the 64th SEGAnnual International Meeting and Exposition, LosAngeles, California, USA, October 23-28, 1994.

nData collection strategy. To evaluate the AVO responses contained within surfaceseismic data, the shooting direction of the walkaway was designed to be the same asthe sail line direction of the 3D surface seismic acquisition. The position of the five-level,three-component downhole geophone array was designed both to provide reflectioninformation from the reservoir and to provide a measure of the TI anisotropy in theshales overlying the reservoir sands, at the same location.

nHorizontal andvertical slow-nesses observedin a walkawayVSP. Anisotropicvelocities, indi-cated by the non-circular curve fitto the data points,can be measuredin situ at the seis-mic scale bywalkaways.

Sail line ofexisting 3D

Walkaway source pointsVertical incidence source points

Calibration Walkaway in a Producing FieldVe

rtic

al s

low

ness

, sec

/km

Horizontal slowness, sec/km

0.4

0.3

0.2

0.1

0-0.3 -0.2 -0.1 0 0.1 0.2 0.3

amplitude with offset that arises from lithol-ogy and fluid properties above and belowthe reflector.13 The results can be analyzedfor fluid and lithology identification in awide zone around the well. Formation prop-erties inferred from VSPs can be integratedwith those interpreted from well logs andmeasured directly from cores. In this waythe VSP can also provide independent cali-bration of the same amplitude variation seenacross a surface seismic reflection pointgather—a gather is the collection of tracesthat reflect at the same point, but at differentangles, or offsets.

Calibrating the surface seismic AVO datawith the VSP AVO response brings addedvalue by:• establishing viability of using AVO to

map a reservoir• reducing the risk involved with the added

cost of AVO studies• improving the reliability of AVO

interpretations• quantitatively assessing the effects of

processing on the AVO response.To establish whether AVO is applicable

as an interpretation tool for a particularreservoir, the expected AVO response isusually modeled. This requires knowledgeof the model parameters, including shearvelocity. Dipole shear sonic logging toolsare used to measure shear velocities evenwhere this velocity is slower than the bore-hole fluid velocity.

However, use of density and velocity logdata to model anticipated AVO anomalieshas not always succeeded in fully explain-ing the AVO response observed on surfaceseismic gathers. The reasons for this aremany and include reflectivity mismatchesbetween surface seismic and log data, wavepropagation effects through fine layers, tun-ing effects (constructive and destructiveinterference at seismic wavelengths), geo-metric effects, processing-related issues andintrinsic anisotropy.14

Borehole seismic data can quantify theseeffects. VSPs provide an independent mea-sure of the seismic AVO response and theability to include necessary effects in theforward modeling to satisfactorily explainthe origins of the surface seismic AVOresponse. Anisotropy is one such effect—one that can both mimic and mask AVOresponses, giving false hope for or conceal-ing the presence of hydrocarbons.

Information about anisotropic velocitiesfor forward modeling often comes from

28

measurements made on cores. But beingscale-dependent, anisotropy may be differ-ent at the seismic wavelength scale. There-fore, it is better to measure the elasticanisotropy at the seismic scale.

In 1994, at Schlumberger CambridgeResearch in Cambridge, England, DougMiller proposed a method to do this usingthe arrival times from a walkaway survey toprovide a measure of compressional veloc-ity anisotropy in a shale, and from this tocharacterize the elastic properties of thatshale, governing compressional and verti-cally polarized shear waves.15

Shale consists of finely-layered clayplatelets and exhibits an anisotropy called

transverse isotropy (TI). The acoustic proper-ties vary depending on whether wavespropagate with particle motions parallel orperpendicular to the platelet layers—oftenthought of as horizontally or vertically,because the clays usually lie flat.16

Miller proposed that the vertical slow-ness—the inverse of velocity—of a shalemay be measured across an array of geo-phones for each shot point offset along awalkaway profile. And the horizontal slow-ness can be measured at a single receiverlocation for adjacent shots in the same pro-file, providing the subsurface layers are

Oilfield Review

nComparing walkaway, walk-above and synthetic data. Thedeconvolved upgoing wavefield of the walkaway is displayedwith two other types of traces spliced in at the zero-offset location.The walk-above trace (left of center) is from the same receiver levelas the central receiver in the walkaway VSP. The walk-abovetrace has been repeated four times for visibility.

A synthetic seismogram (right of center) is derived from densityand calibrated sonic logs corrected to the vertical. For this display,the zero-offset walkaway trace has been shifted to two-way timeand the source wavelet in the synthetic has been chosen to matchthe walkaway source wavelet.

The walk-above VSP data have been processed through a tradi-tional processing sequence. The near-perfect match betweenthese data and the zero-offset walkaway trace gives confidence inthe processed walkaway results, particularly at relatively shortoffsets. The match with the synthetic is complicated by the factthat the well is in a plane that is almost perpendicular to thewalkaway line. Nevertheless, for the top and base reservoir sandinterfaces (black peak at 2.105 sec, white trough at 2.190 sec), the agreement is excellent.

Upgoing Walkaway with Walk-above VSP & Synthetic

Offset, m

Two-

way

trav

el ti

me,

sec

Left Walkaway Right Walkaway

Walk-above VSP Synthetic

1804 1804

2.4

2.3

2.2

2.1

2.0

13. Coulomb CA, Stewart RR and Jones MJ: “Elastic WaveAVO Using Borehole Seismic Data,” paper SL3.2,presented at the 62nd SEG Annual InternationalMeeting and Exposition, New Orleans, Louisiana,USA, October 25-29, 1992.

14. See Chiburis et al, reference 12.15. Miller D, Leaney WS and Borland WH: “An In Situ

Estimation of Anisotropic Elastic Moduli for a Subma-rine Shale,” Journal of Geophysical Research 99, no.B11 (November 10, 1994): 21,659-21,665.

16. For an overview of elastic anisotropy:Armstrong P, Ireson D, Chmela B, Dodds K, EsmersoyC, Miller D, Hornby B, Sayers C, Schoenberg M andLynn H: “The Promise of Elastic Anisotropy,” OilfieldReview 6, no. 4 (October 1994): 36-47.

17. This combination allowed sonic measurements to becompared with core data, and anisotropy parame-ters from VSPs to be compared with core anisotropy.In both cases good agreement was found.Armstrong PN, Chmela W and Leaney WS: “AVOCalibration Using Borehole Data,” First Break 13,no. 8 (August 1995): 319-328.

18. Tuff is lithified volcanic ash.19. Esmersoy C: “Inversion of P and SV Waves from

Multicomponent Offset Vertical Seismic Profiles,”Geophysics 55 (1990): 39-50.Leaney WS: “Parametric Wavefield Decompositionand Applications,” paper SE2.40, presented at the60th SEG Annual International Meeting and Exposi-tion, San Francisco, California, USA, September 23-27, 1990.

essentially flat. A crossplot of these mea-surements for each shot position defines thecompressional anisotropic response of theshale. A curve fitted to these data pointsprovides a solution to the equations thatdeliver shear anisotropy through a completedescription of the elastic properties of theshale (previous page, top).

The constraint of flat layers has now beengeneralized to dipping layers by work atSchlumberger Cambridge Research.

These research efforts have been put topractical use in the BP-operated Forties fieldin the UK sector of the North Sea. The ulti-mate aim is to enable AVO attributes to bemapped with confidence from 3D surfaceseismic data. To achieve this, a detailed eval-uation of shear velocity anisotropy in the for-mations overlying the Forties sand has beenundertaken to build a velocity model. Thedata used included acoustic measurementsfrom preserved shale and sand cores, a fullsuite of logs—including standard density andDSI Dipole Shear Sonic Imager logs—inaddition to walkaway, rig source and verti-cal-incidence VSP data.17

To evaluate the AVO responses exhibitedby the surface seismic data at the target, theshooting direction of the walkaway wasdesigned to be in the same as that of the saillines of the 3D surface seismic acquisition(previous page, middle).

To sample a wide range of incidenceangles at a common reflection point for AVOevaluation of the reservoir, the geophonearray must be close to the reservoir. Typi-cally, this will be in the caprock just abovethe reservoir, facilitating the measurement ofanisotropy in the caprock at the same time.

In the Forties case, an ASI tool consistingof five magnetically clamped, three-compo-nent receivers, spaced at 15-m [49-ft] mea-sured depth intervals, was positioned so thatthe upper three receivers were in a tuff layerabove the shale that contains the two deep-est receivers.18 The interface at the base ofthe shale represents the top of the reservoirsand, which contains an oil column of only7 m [23 ft] in this well. The source used forthe walkaway acquisition was a small clus-tered sleeve gun array.

The walkaway processing for AVO analy-sis involved three steps: the standard stepsfor data orientation through triaxial projec-tion, parametric wavefield decompositionfor wavefield separation and wavelet stabi-lization through deconvolution.19

With the complete data set, several com-parisons could be made. The walk-abovesurvey assessed the vertical-incidence seis-

Winter 1995

mic response near the well, for comparisonwith the vertical-incidence synthetic seis-mogram generated from density and cali-brated compressional-velocity logs (above).The walk-above data and synthetic tracesboth match the nearest offsets of the walka-way data. The walkaway data show varia-tion of reflection amplitudes as a function ofoffset, and can be compared with the sur-

face seismic AVO, when the data becomeavailable.

The next phase of the project was tomodel the walkaway VSP using density andcompressional (P)- and shear (S)-wave soniclogs, but also allowing nonnormal inci-dence angles. The predicted AVO responsewas computed using an algorithm devel-oped at Schlumberger Cambridge Research

29

nIsotropic and TImodels. In the firstmodel, the shaleoverlying the reser-voir sand is assumedto be isotropic (left).For the second model,key anisotropy factorscalculated from thewalkaway travel-timesurface were used tomake the shale trans-versely isotropic (TI)(right). Differences inAVO behavior are vis-ible in the reflectionat 1.07 sec, zero-offsettime. The TI modelshows an earlierincrease in amplitudethan the isotropicmodel.

nAVO calibration.Measured walkawayAVO response at thecaprock/oil-sandinterface is shown asa red line (top), andthe response at thebase of the sand as apurple line (bottom).The equivalent mod-eled response utiliz-ing an isotropiccaprock shale isshown in orange,and provides a poorfit to the measuredresponse at longeroffsets. Includingtransverse isotropy(TI) in the caprockshale (green) gives abetter match withthe observed data.However, the shalebelow the base of thesand can be ade-quately modeled asisotropic. The sand ismodeled as isotropicin all cases.

Walkaway VSP Models

Tim

e, s

ec

20 400 800 1200 1600 20 400 800 1200 16001.0

1.1

1.2

1.3

1.4

1.5

Isotropic shale TI shale

Offset, m Offset, m

4 11 18 26 33 41 48 56100 300 500 700 900 1100 1300 1500

Base sand interface

AVO Calibration

Top sandinterface

10

8

6

4

2

0

-2

-4

-6

-8

-10

Am

plitu

de

that accounts for fine layers, tuning effectsand shale anisotropy.

Initially, two models were generated, oneassuming the shale overlying the reservoirsand was isotropic and another in which TIanisotropy was introduced (right). Differ-ences in amplitude response between thetwo models were immediately observed,particularly at far offsets for the interfacebetween the shale and the reservoir sand at1.07 sec normal incidence time.

The predicted response assuming ananisotropic shale was validated by theamplitude measured in the calibration walk-away (below, right).20 This implies that theeffect of the anisotropic velocity in the shalemust be taken into account before attribut-ing the AVO response in the surface seismicdata to effects of fluids in the reservoir.

It is clear from this study that the combi-nation of AVO measurement from VSP andlog-based, anisotropic forward modelingprovides a powerful methodology for cali-brating AVO responses observed on surfaceseismic data near wells in low dip struc-tures. Where AVO analysis is used as thebasis for hydrocarbon indication in fieldswith existing wells, the method helps iden-tify the origin of observed AVO effects,determining whether large-scale AVO analy-sis and reprocessing effort are worthwhile interms of achieving the desired objectives.The greater understanding of observed AVOeffects should minimize the risk of missinggenuine hydrocarbon-related AVO anoma-lies or of misinterpreting anomalies causedby other factors, such as anisotropy.

The same combination of data can beused to evaluate and calibrate the seismicresponse to other properties such as porosityvia amplitude inversion (see “Seismic Toolsfor Reservoir Management,” page 4).

An intriguing prospect is the integration ofthe information described above to improvethe quality of surface seismic data. Geco-Prakla uses borehole seismic data in theSurvey Evaluation and Design process tooptimize acquisition and processing param-eters for new 3D surveys and to providequantitative quality control during theacquisition and processing phases.21

30Oilfield Review

20. See Leaney, reference 12.21. Ashton CP, Bacon B, Mann A, Moldoveanu N,

Déplanté C, Ireson D, Sinclair T and Redekop G:“3D Seismic Survey Design,” Oilfield Review 6, no. 2 (April 1994): 19-32.

Top sand (measured from walkaway VSP)

Model top sand overlain by isotropic shale

Model top sand overlain by transverselyisotropic (TI) shale

Base sand (measured from walkaway VSP)

Model base sand (isotropic)

Angle, degree/offset, m

nAssessing the quality of surface seismic surveys using boreholegeophysical data in addition to log-derived synthetics. Data fromDSI measurements have been used to produce a synthetic zero-off-set seismogram (track 2) and a stacked synthetic offset seismogram(track 3). These traces do not compare well with the migrated 3Dsurface seismic data from the vicinity of the well (track 5). The zero-offset traces from the walkaway survey (track 1) show a better, butless than desirable correlation with the surface seismic data thando the stacked walkaway data (track 4). This example reinforcesthe fact that stacked surface seismic data may not provide accu-rate estimates of properties that can be derived from logs unlessborehole seismic information is incorporated.

Tim

e, m

sec

Migrated 3Dsurface seismic

section

Walkawayzero-offset

trace

Syntheticzero-offset

trace

Stackedsynthetic

Stackedwalkaway

2200

2100

2300

2400

2500

In order to use borehole data to improveand quantify the information content of sur-face seismic data, it is necessary to under-stand the relationship between the differentmeasurements. The potential role of bore-hole seismic data in this respect may beillustrated using the Forties data setdescribed above (right).

In this example, the same Vp and Vs logsobtained from DSI measurements have beenused to produce a synthetic seismogram tobe compared with the measured zero-offsetresponse from the walkaway, then with tracesfrom the 3D survey. The synthetic must be“corrected” for geometric effects due to thelateral offset from source to receiver and thenstacked using the same number of traces andoffset range as used in the surface seismicprocessing. Similarly, the walkaway VSP dataset is corrected and stacked.

The stacked walkaway matches the sur-face seismic data better than the log-basedsynthetic. The first thing this indicates isthat the walkaway more closely resemblesthe physical experiment of the seismic sur-vey. But second, and more important, thesurface seismic traces, although processedto deliver an estimate of the vertical-inci-dence response, do not match a normalincidence model.

The borehole seismic measurement con-nects the log-generated synthetic and thesurface seismic data. The addition of such ameasurement provides a mechanism foridentifying and reducing the uncertaintyassociated with the log and surface seismicmeasurements through evaluation and cali-bration using the borehole seismic data as acontrol. Upscaling the log information tothe seismic scale—not always a trivialstep—is facilitated through the comparisonof the synthetic seismogram with the bore-hole seismic. Last but not least, the indepen-dent measure of the seismic response in theborehole can be used to quantitatively eval-uate the processing of the surface seismicdata at the well location to assure an opti-mum result.

Winter 1995

Looking to the FutureWe have seen a distinct evolution in thecomplexity of information provided by bore-hole geophysics. First, direct travel-timecheck shots simply related time to depth.Then, VSP reflection information improvedstructural imaging. Now, AVO and anisotropyare probing the fine-scale properties of theformation within that structure.

But it doesn’t stop there. Borehole seismicsurveys bridge the gap between the log andseismic measurements, merging data atscales as fine as borehole resistivity imageswith reservoir-scale pictures of seismic reflec-tions. Moreover, VSPs span the life of thereservoir, from VSPs while drilling to surveysthrough production tubing. It is clear thatwhile wellbore geophysics is already deliver-ing valuable information, the techniquepromises a great deal more in the future.

—CF, LS

31

32

Permanent Monitoring—Looking at Lifetime Reservoir Dynamics

Alan BakerJohn GaskellAberdeen, Scotland

John JefferyElf Enterprise Caledonia Ltd.Aberdeen, Scotland

Alan ThomasTony VenerusoClamart, France

Trond UnnelandStatoilBergen, Norway

For help in preparation of this article, thanks to ElwinZaimul Arifin and Barry Nicholson, Wireline & Testing,Jakarta, Indonesia; Mary Ellen Banks, Schlumberger-Doll Research, Ridgefield, Connecticut, USA; ThomasBundy, Conoco Indonesia, Jakarta, Indonesia; LilianneChérière, Alp Tengirsek and Imran Kizilbash, Wireline& Testing, Montrouge, France; Gilbert Conort, BernardGlotin and Peter Soroka, Schlumberger-Riboud ProductCentre, Clamart, France; Eric Decoster, Wireline &Testing, Rio de Janeiro, Brazil; Marcelo Prillo, Wireline& Testing, Anaco, Venezuela; and John Pucknell, BPExploration, Aberdeen, Scotland.

In this article, UNIGAGE (downhole recorder) is a markof Schlumberger and Windows is a mark of MicrosoftCorporation.

Permanent monitoring systems measure and record well performance

and reservoir behavior from sensors placed downhole during completion.

These measurements give engineers information essential to dynami-

cally manage hydrocarbon assets, allowing them to optimize production

techniques, diagnose problems, refine field development and adjust

reservoir models.

Oilfield Review

nPermanent moni-toring system. Datarecorded by per-manent gaugesmay be transmit-ted via satellite to oil companyoffices for use inreservoir modeling.

1. Lilley IJ, Douglas AA, Muir KR and Robinson E: “Reser-voir Monitoring and Wireline Logging in Subsea Wells,”paper SPE 18357, presented at the SPE EuropeanPetroleum Conference, London, England, October 16-19, 1988.Shepherd CE, Neve P and Wilson DC: “Use and Appli-cation of Permanent Downhole Pressure Gauges in theBalmoral Field and Satellite Structures,” SPE ProductionEngineering 6, no. 3 (August 1991): 271-276.Carter PJ and Morel EH: “Reservoir Monitoring in theDevelopment of Marginal Fields: Ivanhoe, Rob Roy andHamish,” paper SPE 20978, presented at Europec 90,The Hague, The Netherlands, October 22-24, 1990.

Reservoir development and managementtraditionally rely on early data gathered dur-ing short periods of logging and testingbefore wells are placed on production.Additional data may be acquired severalmonths later, either as a planned exercise orwhen unforeseen problems arise. Such dataacquisition requires well intervention andnearly always means loss of production,increased risk, inconvenience and logisticalproblems, and may also involve the addi-tional expense and time of bringing a rigonto location.

Permanent monitoring systems allow adifferent approach (above ). Sensors areplaced downhole with the completion stringclose to the heart of the reservoir. Moderncommunications provide direct access to

Winter 1995

sensor measurements from anywhere in theworld. Reservoir and well behavior maynow be monitored easily in real time, 24hours a day, day after day, throughout thelifetime of the reservoir. Engineers canwatch performance daily, examineresponses to changes in production or sec-ondary recovery processes and also have arecord of events to help diagnose problemsand monitor remedial actions, rather likemonitors in a power plant’s control room.

Most systems in operation record bottom-hole pressure and temperature, but othermeasurements, such as downhole flow rate,are being introduced and may becomecommon in the future. However, pressureand temperature provide dozens of benefi-cial applications.1 This article reviews thedevelopment of permanent monitoring,looks at applications with several examplesand describes the hardware.

33

1960 Technology

Current Technology

Insulatedoutlet forwire

Pressuregauge

Combinationwire clamp &wire protectoron eachtubing collar

Wire Cableprotector

Gaugemandrel

Pressure/temperaturegauge

Cable

nPermanent moni-toring system—circa 1960. Thepressure gaugewas hung on thebottom of the tub-ing and communi-cated with the sur-face via a cablestrapped to theoutside. Connec-tion to the gaugedownhole andrecording equip-ment at surfacewas through insu-lated ports. Thissetup causedrestrictions in flowand also preventedaccess to the wellbelow the gauge.With current per-manent monitoringsystems the gaugeis housed outsidethe tubing in amandrel allowingfullbore flow andunrestrictedaccess.

Early DaysPermanent monitoring has its roots in theearly 1960s on land wells in the USA.2 Pres-sure gauges were needed to monitor theperformance of secondary recovery pro-jects, such as waterfloods or artificial liftschemes, where they were required down-hole for several weeks. In many cases, theonly option available was to run a standardpressure gauge on the end of the comple-tion string (above). The cable for power anddata transmission was passed through aninsulated connector in the Christmas tree,strapped to the outside of the tubing andthen ported back inside the tubing justabove the gauge leaving the bore free of anyobstructions. Even though the hardware wassimple by today’s standards, these earlyexamples proved invaluable to oil compa-nies and showed the diverse use of and ben-efits from the pressure data gathered.

One example from 1962 is typical of theperiod. Henderson 6 was the second wellcompleted by the Coronado Company in theBell Sand of the Old Woman Anticline,Wyoming, USA. A permanent pressure

34 Oilfield Review

gauge was placed below a conventionalpump in a 2400-ft [720-m] well for interfer-ence testing and to determine the productiv-ity index.3 Initial bottomhole pressure (BHP)was 680 psi.

The well produced 340 barrels of oil perday (BOPD) [54 m3/d] with a 60-psi draw-down, but quickly suffered from increasingwater cut. Bottomhole pressure returned to680 psi indicating complete water break-through—possibly by water coning. Bymodifying production and monitoringdownhole pressure changes it quicklybecame apparent that the coning problemwould not repair itself and that the wellwould have to undergo workover. Afterwardthe well was put back on production and,this time, the pressure gauge measurementswere used to control drawdown to just 40psi to prevent recurrence of water coning.

Other examples from the 1960s showhow pressure gauges were used to monitorprogress of secondary recovery fronts acrossfields, to check the operation of subsurfacepumps, to provide reservoir data and to cal-culate individual well drainage during thelife of the reservoir.

The Technical Challenge: How Permanent is Permanent?Although permanent monitoring systemshave been around for a number of years, thetechnology has evolved fairly slowly. Relia-bility was a major issue with early installa-tions (next page, middle).4 The first perma-nent pressure gauge run by Schlumbergerwas for Elf in Gabon (Africa) in 1972 fol-lowed one year later by the first North Seainstallation on Shell’s Auk platform. Theseearly systems were essentially adaptations ofelectric wireline technology. A standardstrain pressure gauge was clamped to thetubing and ported to monitor tubing pres-sure. A stranded single-conductor loggingcable was strapped to the outside of the tub-ing exiting at the wellhead. Data wererecorded on a standard acquisition unit.

Many early failures were caused by dam-age during installation or by cable problemsat a later date—either by loss of electricalcontinuity or breakdown of insulation caus-ing a short circuit (next page, top). Statoilreport that many cable failures occurred atsplices and now request splice-free cables.Detailed analysis, such as that performed byPetrobras on systems run in Brazil and theNorth Sea, shows how reliability hasimproved.5 More recently a detailed researchand development project by Schlumberger,40% funded by the European CommunityTHERMIE project, has resulted in develop-ment of a new generation permanent gaugeand its associated components for evengreater reliability.6

Present systems are engineered specifi-cally for the permanent monitoring marketand have a life expectancy of several years(see “Hardware,” page 37).7 Gauges havedigital electronics designed for extendedexposure to high temperature and undergoextensive design qualification life tests andstrict quality checks during manufacturebefore being hermetically sealed.8 They arenot designed for maintenance.

Cables for permanent installations areencased in stainless steel or nickel alloypressure-tight tubing that is polymer-encap-sulated for added protection. All connec-tions are verified by pressure testing duringinstallation.

Connections through tubing hanger andwellhead vary depending on the type ofcompletion—subsea, platform or land—butcomponents are standard, tried and testeddesigns made in conjunction with the tub-ing hanger and wellhead manufacturers.

Data transmission and recording are tai-lored to oil company needs, and wherever

35Winter 1995

2. Nestlerode WA: “The Use of Pressure Data From Per-manently Installed Bottom Hole Pressure Gauges,”paper SPE 590, presented at the SPE Rocky MountainJoint Regional Meeting, Denver, Colorado, USA, May27-28, 1963.

3. An interference test measures the pressure response inone well to changes in production or injection in asecond well. The objective of the test is to assess com-munication between the wells.Productivity index is a measure of producibility of thereservoir and is equal to flow rate divided by pressuredrawdown.

4. Bezerra MFC, Da Silva SF and Theuveny BC: “Perma-nent Downhole Gauges: A Key to Optimize DeepseaProduction,” paper OTC 6991, presented at the 24thAnnual Offshore Technology Conference, Houston,Texas, USA, May 4-7, 1992.

5. Bezerra, Da Silva and Theuveny, reference 4.6. European Community THERMIE project on Improving

the Reliability of Permanent Down Hole PressureGauges.

7. Catherall R, Spence JR and McKee P: “PermanentDownhole Instrumentation: Recent Developments inEngineering for Reliability,” Transactions of the 3rdLatin American Petroleum Congress, Rio de Janiero,Brazil, October 18-22, 1992, paper TT-219.The life expectancy of a permanent monitoring systemdepends on many factors, the most important beingbottomhole temperature. High temperature increaseselectronic aging and failure rates. However, there arefunctioning gauges that have been in wells for morethan 10 years and these are obviously not the latesttechnology.

8. Levera R, Pohl D and Veneruso A: “PermanentGauges Enter the Digital Era,” Transactions of the 3rdLatin American Petroleum Congress, Rio de Janiero,Brazil, October 18-22, 1992, paper TT-154.

nPermanent downhole cable evolution.Standard monoconductor logging cableswere used with the first installations (1).Later, cable bumpers and polymer encap-sulation improved protection (2). In theNorth Sea there was a shift to tubing-encased cables in the mid-80s. The firsttype used Teflon insulation whichallowed the cable to slip inside the tub-ing, breaking connections (3). Teflon wasreplaced by a friction material to allevi-ate this problem (4). This type of cable isnow standard in the North Sea. Petrobrasuses a combination of tubing-encapsu-lated cables and bumpers (5).

nPermanent moni-toring system relia-bility in the NorthSea. Data onSchlumberger per-manent monitoringsystems installedin the North Sea inthe last nine yearsshows improvingreliability. Mostworking systemsare less than threeyears old (top), buta significant num-ber have beenworking longer.Most failuresoccurred duringthe first year andwere likely causedby cable or con-nection damageduring installation(middle). Ideally,systems shouldremain workingduring the life ofthe well so thataverage system lifedivided by aver-age well life equals100%. System lifereached only 54%of well life forinstallations in1988 (bottom). Thisfigure has steadilyimproved and, notsurprisingly, is100% for systemsinstalled this year.

Outer armor

Inner armor

Cable bumpers

Insulation

Central copper core

Polymerencapsulation

Outer armor

Inner armor

Insulation

Central copper conductor

Cable bumpers

Similar tocable 4


Stainlesssteel tubing

Inner insulation

Outer insulation




Stainlesssteel tubing

Insulation


Friction material

1 2 3 4 5

19

10

6

3

1

Age of Working Permanent Monitoring (PM) Systems

Num

ber

Years

Ave

rage

PM

life

/ave

rage

wel

l life

, %

1987 1988 1989 1990 1991 1992 1993 1994 1995

Year

Working when removed

1 2 3 4 5 6 7 8

Age of Failed PM Systems

Years1 2 3 4 5 6 7 8

Num

ber

41 21 20 5 6 3 8 3

14

1

84

1

79%

54%

70%74%

80%88% 86%

91%100%

5

12

1918

87 36

3334

ClaymoreComplex

Elf Enterprise

ScapaElf Enterprise

HighlanderTexaco

PetronellaTexaco

ChanterElf Enterprise

SaltireElf EnterprisePiper B

Elf Enterprise

AberdeenPeterhead

St. Fergus

Flotta terminal

MCP01

TartanTexaco

SCOTLAND

Orkney Islands

Gas pipelineOil pipeline

Ivanhoe/Rob RoyAmerada Hess

n Location of Scapa, Saltire and Chanter fields. (Courtesy of Elf Enterprise Caledonia Ltd.)

36 Oilfield Review

9. Ellison RL, Simlote VN and Ko RS: “ContinuedDevelopment, Reservoir Management and ReservoirModelling Have Increased Proven Developed OilReserves by Over 35% in the Marginal Scapa Field,”paper SPE 25063, presented at the SPE EuropeanPetroleum Conference, Cannes, France, November16-18, 1992.

10. A wet connect is an electrical connector—a plugand socket—that is pressure-tight and waterproof. Itallows a connection to be made in any fluid underhigh-pressure and over a range of temperatures. Awiper system cleans the male pin as it is insertedinto the female socket ensuring good electrical con-tact. O-rings prevent fluid entry as the connection ismade, ensuring insulation.

possible industry standards are used so thatsignals may be integrated with other sys-tems. For example, many subsea comple-tions have memory modules called data-loggers that record, for instance, wellheadpressure or the status of control valves. Per-manent gauge data may be fed to interfacecards located in the data-logger so that datatransfer may be executed in one step.

Of equal importance are planning andproject management for each installation.Although most permanent monitoring hard-ware may be considered off-the-shelf, sev-eral parts may have to be customized forspecial types of wellhead. Longer lead timesmay be needed if the project requires cus-tom-built equipment. For example, in sub-sea well completions, the permanent gaugesare connected to subsea-mounted electronicpods with acoustic data links to surface.

Specialist teams of engineers and techni-cians install permanent monitoring systemsand work closely with rig crews who arefully aware of the importance of installing aworking system. Pressure gauges are usuallyconnected to the cable at the workshopwhere pressure or welded seals can easilybe made and pressure tested. At the well-site, the gauge is mounted onto a mandrel,which is connected to the tubing. The cableis supplied on a reel and is run in the holewith the tubing. Great care must be taken toavoid damaging the cable at this stage.Cable protectors placed on every tubingjoint help prevent damage as the system isrun in the well. Checks on both pressureintegrity and gauge operation during theentire procedure ensure a working system.

For certain subsea installations, hookup tosurface acquisition equipment may involvedivers and diver-matable connections orremotely operated vehicles (ROVs) and con-nections to acoustic data-loggers ortelecommunication equipment.

Once they are connected and running,permanent monitoring systems begin pay-back in many different ways as the follow-ing case studies show.

A Decade’s Experience in the North SeaElf Enterprise Caledonia Limited (EEC) hasused permanent monitoring systems in itswell completions since 1983. The first appli-cation was on Scapa, a small satellite field 5km [3 miles] from the Claymore platform inthe UK sector of the North Sea (above). TheScapa well C-47 was drilled at a highangle—67° to 68° deviation—from theClaymore platform.

A one-year extended well test (EWT) con-ducted by EEC helped determine the long-term deliverability of Scapa. Because of thedifficulties in running wireline operations insuch a high-deviation well and with theClaymore platform rig working on otherwells, it was not possible to enter C-47 dur-ing this period. The only way to obtaindownhole pressure data to evaluate theEWT was by using permanent monitoringsystems. The outcome of the EWT paved theway for the development of the field with amultiwell slot subsea template.

The second application was in the samefield, but this time in a newly discovered,lower sand body in subsea well S-20. Pres-sure data from the permanent monitoringgauges installed in this well contributed tothe estimated reserves being increased fromapproximately 40 million barrels to 60 to 70million barrels. More recently the estimatehas been increased again to 100 millionbarrels.9 Four more wells were drilled andcompleted with permanent gauges.

Permanent pressure data have been usedto model the interaction between the threeoil accumulations of the Scapa field—directly through extensive interference test-ing and indirectly through use of the data inmaterial balance and simulation studies.This has resulted in a more thorough under-standing of field behavior, leading to opti-

mized recovery of reserves and continueddevelopment drilling.

Scapa has also seen the use of some novelhardware applications. A conventional well-head uses a dual-bore Christmas tree, whichhas to be oriented to allow the use of thetraditional wet connect system (see “Hard-ware,” next page).10 However, EEC’s con-centric completion system does not need tobe oriented. To realize the full benefits ofthis tree, EEC has successfully used aninductive coupling system at the interfacebetween the tubing hanger and the Christ-mas tree. Also, because the umbilical con-necting Scapa electrically to the Claymoreplatform has reached its capacity, data fromall the wells are stored in subsea data-log-gers. These are periodically interrogatedacoustically during one of the many trips bysupply boats to the area. Recently, the samemethod has been applied to collect datafrom flowmeters mounted on a subsea waterinjection line.

Saltire field was the next application, com-ing on stream in 1993 on the back of thePiper redevelopment. Wells were drilledfrom a minimum facility platform that would

(continued on page 41)

■■Permanent quartz gauge. The permanent quartzgauge measures downhole temperature and pres-sure. High measurement stability and long life areachieved by using hermetically sealed quartz crystalresonators, digital electronics and proprietarymechanical seals.

Pressure connection

or

Digital pressure,temperature and self-test

1⁄4-in. encased cable

Cable driver andfault-tolerant regulator

Metal-to-metal sealedcable head

Hermetically sealedwelded housing

Autoclave AxialConnection

Gland RadialConnection

Quartz crystal resonatorsmeasure pressureand temperature

Protection bellows

…11

010

P/T

37Winter 1995

A permanent gauge installation is an engineered

product tailored to the well completion. The sys-

tem is built of standard components carefully

chosen to fit oil company requirements. The com-

plete system ranges from pressure gauge to

cable, from wellhead to data transmission.

Hardware

Pressure Gauges

Pressure gauges are built to more exacting life-

time specifications than wireline- or drillstem

test-conveyed systems. A temporary completion

may contain gauges that stay downhole for sev-

eral months during a long-duration well test.

However, permanent gauges have to stay in wells

for several years. Reliability is a key feature and

this is inversely proportional to temperature,

time and wellbore chemistry. Gauge electronics

are designed with this in mind.

Schlumberger permanent gauges use a modi-

fied version of UNIGAGE electronics. These elec-

tronics are used in Schlumberger memory gauge

recorders and are designed for rugged, long-

duration operations. Modifications include opting

for totally soldered and hermetically sealed,

solid-state electronic components that may be

bigger or more expensive, but are temperature-

stable for long periods. Any drift in the electron-

ics is automatically corrected. Once installed, the

gauges are not going to be used on other wells,

so there is no need to consider maintenance. To

this end, all internal connectors and sockets are

eliminated and, after 100% burn-in and calibra-

tion testing, the gauge housings are welded shut

during manufacture. Connections to the outside

are provided by a feed-through connector (right).Sensors used in permanent gauges have

slightly different specifications than pressure

gauges used in well testing. The emphasis is on

long-term gauge stability rather than fast

dynamic response. Quartz crystals are most often

used although other types of sensor, such as sap-

phire sensors, may also be used.

38 Oilfield Review

Section A-A

Cable head

Permanent gauge

Exploded view ofmetal-to-metal seal

A A

Tubinghanger

Threaded wellhead outlet

Cable penetratorwith sealingglands ateach end

Optional flanged wellhead outlet

■■Gauge mandrel. The gauge mandrel provides aprotective recess for the permanent gauge.

■■Surface wellhead con-nector. Signals from thepermanent downholegauge pass from thecable, clamped to theoutside of the tubing, toan external terminal onthe wellhead. Holes aremade in the tubinghanger and wellhead todo this. Threaded con-nections are made oneach, so that compres-sion fittings can sealeach side of the holes to maintain pressureintegrity of the wellhead.

Gauge Mandrel

Gauges are housed in the protective recess of a

gauge mandrel (above). This provides complete

gauge protection against mechanical damage

along the entire gauge length. Gauge protection

is especially important in deviated wells, where

the gauge has to pass through liner hangers, or

during completions from floating vessels.

Bottomhole Connectors

There are two connections to the permanent

gauge: electrical connection to the cable for

power and data transfer, and hydraulic to connect

the sensor to tubing pressure. Electrical connec-

tion is usually made at the workshop. The con-

ductor is soldered to the feed-through connector.

The pressure connection is made at the wellsite

with metal-to-metal seals.

Metal-to-metal seals are also made between

the gauge and its gauge carrier or gauge man-

drel. At the wellhead end of the cable, metal-to-

metal seals are again made to ensure that con-

nections are pressure tight. Each connection is

pressure tested and verified during installation

at the wellsite.

Cable

Cables form a major part of the budget for a per-

manent monitoring system—up to 30% of the

cost. Permanent downhole cables have to with-

stand pressure, temperature and exposure to

highly corrosive wellbore fluids during the life of

the permanent installation. They also have to be

mechanically rugged so that they are not dam-

aged during installation. Cables consist of copper

conductors surrounded by Teflon insulation mate-

rial, antislip filler, standard 1/4-in. stainless-

steel or nickel alloy tube and thermoplastic

encapsulation material (page 35, top).1 The filler

material supports the cable inside the tube pre-

venting the entire weight of the cable from being

supported by the top connector. It also allows

39Winter 1995

1. Much has been published about the potential of fiber-optic cables. For an operator’s perspective of where thistechnology currently stands:Botto G, Maggioni B and Schenato A: “Electronic, Fiber-Optic Technology: Future Options for Permanent Reser-voir Monitoring,” paper SPE 28484, presented at the 69thSPE Annual Technical Conference and Exhibition, NewOrleans, Louisiana, USA, September 25-28, 1994.

Tubing joint

Standard designProtectorfor twin

encapsulatedcable

Cross-Coupling Cable Protectors

Diver-matable subseaelectric connection

Flanged wellhead outlet

Female wet-connect

Male wet-connect

Tubing hanger

1⁄4-in. encased cable

Wellheadvalve block

■■Permanent cable protector.

■■Subsea wellhead connectors. Signals have to passthrough the tubing hanger and Christmas tree toemerge at a suitable connection—diver or ROV mat-able. A male electrical wet-connect makes the con-nection through the tubing hanger to the permanentdownhole cable. The wellhead valve block is pre-pared with a flanged outlet and female wet-connect.Contact is made when the oriented wellhead is low-ered onto the tubing hanger.

some movement inside the stainless-steel tube so

that the cable is not exposed to thermal stresses.

The metal tube has up to 20,000-psi collapse

pressure and prevents wellbore fluid contamina-

tion which could short circuit the insulation.

Encapsulation helps prevent cable damage such as

nicks and crimping during installation. Even so,

the cable requires careful handling.

Cables usually have single conductors, but can

be manufactured with more. Encapsulation materi-

als and sizes can also be tailored to oil company

requirements.

Cable Protectors

Cables are clamped to the tubing string using

cable protectors. These are clamped across tubing

joints—the place where the cable flexes slightly

over the collar (above right).

Tophole Connectors

Connections are made by pressure-tight, compres-

sion-fit, metal-to-metal seals between the down-

hole cable and the tubing hanger and downhole

cable and gauge. Other elements of the comple-

tion may also require connections, for example, if

the cable has to pass through a packer.

Wellhead Connectors

There are many types of wellheads and the cable

from a downhole permanent gauge must pass

through to an exterior terminal. Connections are

first made to the tubing hanger. Connections to the

other side of the hanger depend on the type of

wellhead. If the wellhead is at surface—for exam-

ple, a wellhead on a land well or a wellhead

exposed above the sea on a platform—then a con-

nection has to be made through the wellhead to a

terminal block (previous page, bottom). The signal

is then routed to the surface acquisition system.

For subsea wellheads, the connection is more

complicated (right). An electric wet-connect (EWC)

system is commonly used enabling a direct link

across the wellhead. The EWC consists of a male

pin situated in the tubing hanger. The female

socket sits below the valve block and is oriented to

align with the male pin. On the outside of the well-

head valve block is a flanged outlet to either a

diver-matable subsea electrical connection or a

remote-operated vehicle connection. The signal

is then routed to an acoustic transducer, an inte-

grated control pod or a subsea umbilical.

Acquisition Systems

There are a number of different methods for col-

lecting data. Often on subsea completions, it is

possible to hook into existing data-gathering sys-

tems. These have been set up to monitor subsea

wellheads providing such data as surface flow

rates, temperature and pressure as well as valve

positions and status. Permanent gauge interface

cards are now available for most data gatherers,

which are normally connected to platforms by

seabed umbilical cables.

A system that does not use an umbilical cable

is a hydro-acoustic system (next page). In this

approach, the permanent gauge signal is col-

lected at a data acquisition unit (DAU) that logs

and performs a quality check of each measure-

ment. The DAU can be periodically interrogated

40 Oilfield Review

■■Acoustic data link. In subsea completions signals may be stored in abattery-powered data acquisition unit (DAU). The DAU communicatesacoustically with the surface using a remote transducer. The trans-ducer may be hung over the side of a rig or supply boat or even hungbeneath a helicopter.

Surfacetransducer

Remotetransducers

Subsea wellhead

Data acquisition unit (DAU)

Battery pack

PC-controlledsurface interrogation system

using an acoustic transducer that may be hung

over the side of a boat, rig, platform or even from

a helicopter. The subsea equipment is powered by

a battery pack that can be replaced by divers or a

ROV without losing the DAU memory.

For platforms, several permanent gauges may

be connected to an autonomous surface unit that

is rack-mounted in the cabin or packaged in an

explosion-proof box near the wellhead. This

acquires and records the raw measurements and

communicates with the oil company’s computers

via standard modem data links or local area net-

works. Communication may be via satellite to the

oil company office anywhere in the world.

Software

Permanent gauge monitoring software enables a

user to control and monitor permanent gauges

from anywhere in the world. This Windows-based

PC software makes full use of standard communi-

cations networks and straightforward point and

click menus and icons. With this software, a user

can view the real-time downhole gauge measure-

ments directly or display recorded data files. In

addition, the data can be shared via networks with

other users for further analysis and interpretation.

Power Supply

Gauge power is provided from surface directly

from subsea umbilicals, platform supplies or from

subsea battery packs. On land in sunny areas,

batteries may be recharged using solar panels.

nInterference test. Pressure pulses recorded in Saltire A01 (top) are seen as smallchanges in pressure recorded by the permanent gauge in Saltire A04 (bottom). (Courtesyof Elf Enterprise Caledonia Ltd.)

Saltire A01

Saltire A04

4900

4800

4700

4600

4500

4400

4300

4200

4100

4000

13 14 15 16 17 18 19 20 21 22 23 24 25June

Pre

ssur

e, p

si

Pressure pulses

Effect ofshort pulses

5060

5040

5020

5000

4980

4960

Effect oflong pulse

Pre

ssur

e, p

si

not allow concurrent well intervention dur-ing a drilling program lasting several years.When drilling stopped, the platform wouldbecome unmanned. Any well reentry fordata gathering would then be extremelycostly. So EEC decided to incorporate perma-nent monitoring systems in the completionsfrom the first stages of field development.

The reservoir proved to be complex andpermanent pressure data served to optimizeproduction (below). For example, the bub-blepoint of the crude oil in one of the reser-voir members is 3700 psi and the initial for-mation pressure, 4600 psi. So drawdownhad to be less than 900 psi to sustain gas-free production. High skin factor in the firstwell meant that as large a drawdown as pos-sible would be needed for adequate produc-tion—introducing a further complication.However, the pressure could be carefullymonitored and production optimized tomaintain reservoir pressure at around 40 psiabove bubblepoint.

Additional benefits of continuous pressurerecording have included cross-field interfer-ence testing that has shown that althoughthe reservoir is mapped as being compart-mentalized, there is generally pressure com-munication between compartments. Forexample, pressure changes of less than 5 psiare detected in a well approximately 600 m[2000 ft] away from one being pulsed(right ). These data helped optimize welllocations and water injection strategy tomaintain reservoir pressure, and have alsoprovided a useful history-matching parame-ter for the reservoir simulator model.

41Winter 1995

n Permanent pres-sure data used tooptimize production.Production from oneof the Saltire reser-voir members wasadjusted severaltimes until anacceptable bottom-hole flowing pres-sure was achievedin Well A07. Abruptchanges in pressuremay be seen eachtime the productionwas adjusted. (Cour-tesy of Elf EnterpriseCaledonia Ltd.)

2960

2940

2920

2900

2880

2860

2840

2820

2800

2780

18 23 28 02 07 12 17 22 27 31 05 10 15 20 25 30 05

July August September

Pre

ssur

e, p

si

Well A07

Adjustments in production

A more unusual use of permanent pres-sure data confirms the successful isolationof an underlying higher pressure intervalthat would otherwise have been difficult todemonstrate (below).

The final EEC field application is on a sin-gle subsea well development—the Chanterfield. This has separate reservoirs of oil andcondensate. Initially the well produced oiland was later converted to produce conden-sate. Continuous pressure monitoring keptreservoir pressure from dropping below oilbubblepoint during that particular phase ofproduction and also helped optimize the tim-ing of conversion to a condensate producer.

The pressure data also provided input tocalculate the accumulation in contact withthe well and evaluate the effectiveness ofthe aquifer as a reservoir drive mechanism.This information will help establish the

42

Tim

4500

4400

4300

4200

4100

4000

3900

3800

37000 12 2

Piper p

Pre

ssur

e, p

si

Galley zone shut-in

Well shut-in;pressure increasesduring crossflow

11. Schmidt H, Stright DH and Forcade KC: “MultiwellData Acquisition for Permanent Bottomhole PressureGauge Installations,” paper SPE 16511, presented atthe Petroleum Industry Applications of Microcom-puters, Montgomery, Texas, USA, June 23-26,1987.Unneland T and Haugland T: “Permanent DownholeGauges Used in Reservoir Management of ComplexNorth Sea Oil Fields,” SPE Production & Facilities 9,no. 3 (August 1994): 195-203; also paper SPE26781, presented at the SPE Offshore Europe Con-ference, Aberdeen, Scotland, September 7-10, 1993.

12. Statoil has looked into the effects of producing

nIsolation of a high-pressure interval. Pressurtypical buildup when the Galley sandstone issandstone. Pressure increases rapidly when tA bridge plug is set to isolate Piper and this isthe original shut-in pressure buildup trend of Caledonia Ltd.)

requirement for a possible additional well inthe field. In subsea wells such as Chanter,the cost of the permanent monitoring systemis immediately recouped if only one wellreentry operation is avoided.

Apart from the major applicationsdescribed above, EEC has also found perma-nent monitoring data useful in other circum-stances. Knowing bottomhole pressureallows calculating the correct weight of killfluid. This minimizes formation damagewhile ensuring an effective kill. Knowledgeof bottomhole pressure also allows optimalcontrol of underbalanced perforating.

Based on their experience in the NorthSea, EEC considers permanent monitoringsystems beneficial whenever developmentinvolves satellite fields, subsea completions,difficult access well completions or limitedaccess platforms.

e, hr4 36 48

erforated

Confirmation ofisolation

Galley backon production

Piper plugged

below bubblepoint in the near wellbore region.13. Unneland T and Waage RI: “Experience and Evalua-

tion of Production Through High-Rate Gravel-PackedOil Wells, Gullfaks Field, North Sea,” SPE Produc-tion & Facilities 8, no. 2 (May 1993): 108-116; alsopaper SPE 22795, presented at the SPE Annual Tech-nical Conference and Exhibition, Dallas, Texas,USA, October 6-9, 1991.

14. Bale A, Owren K and Smith MB: “Propped Fractur-

e data recorded in Saltire A04 shows a shut in prior to perforating the Piperhe high-pressure Piper zone is perforated. confirmed as successful by the return tothe Galley zone. (Courtesy of Elf Enterprise

Norwegian ConnectionTwo fields in the Norwegian sector of theNorth Sea highlight several more applica-tions of pressure data recorded by perma-nent monitoring systems.11 Gullfaks andVeslefrikk fields operated by Statoil are com-plex and require careful reservoir manage-ment. Gullfaks is in the central part of theEast Shetland basin, 175 km [109 miles]northwest of Bergen, Norway. Veslefrikk isabout 30 km [19 miles] south of Gullfaks(next page, top).

Gullfaks is heavily faulted with a numberof sealing or partially sealing faults. Oneimportant reservoir monitoring objective isto measure the degree of communicationbetween the fault blocks. Veslefrikk startedproduction with commingled wells. Heregauges are used in dedicated wells to moni-tor the two reservoirs independently. Dataare used in both fields to ensure single-phase oil flow in each fault block, to moni-tor and optimize well performance withtime, for transient well test analysis and formatching numerical models.

At present, Statoil has more than 50 per-manent gauge installations. Each is con-nected to a communications system thatallows gauge control from PCs located any-where in the world. For example, a well testcan be monitored remotely and the datasampling rate adjusted during the test.

Gullfaks—Gullfaks field development isbased on single-phase oil flow without free-gas in the reservoir. In wells with permanentmonitoring systems, bottomhole flowingpressure (BHFP) is maintained slightly abovesaturation pressure by adjusting the flow rate(next page, bottom).12 This results in a poten-tial increase in the individual well productionrate of 100 m3/d to 500 m3/d [630 B/D to3150 B/D]. In wells without permanent mon-itoring, calibrated curves based on empiricalmultiphase equations and permanent pres-sure data from nearby wells are used.

Oilfield Review

ing as a Tool for Sand Control and Reservoir Man-agement,” paper SPE 24992, presented at the Euro-pean Petroleum Conference, Cannes, France,November 16-18, 1992.Bale A, Smith MB and Settari A: “Post-Frac Produc-tivity Calculation for Complex Reservoir/FractureGeometry,” paper SPE 28919, presented at the SPEEuropean Petroleum Conference, London, England,October 25-27, 1994.

Winter 1995

1000 1500 2000 2500 3000 3500 4000 4500Time, hr

Flow

rat

e, m

3 /d

2

500

3000

Pre

ssur

e, k

Pa

27,

000

28,0

00

Bergen

PC Modemtelephone

Gullfaks

Veslefrikk

Permanentgauge

Murchison

Huldra

Statfjord

Gullfaks South

Veslefrikk

Snorre

Alwun

Brent

Gullfaks

Norway

Denmark

Germany

nAdjusting bottom-hole flowing pres-sure (BHFP) tomaximize oil pro-duction. As BHFP isadjusted to slightlyabove saturationpressure on Gull-faks C-3 (top), dailyoil production rateincreases (bottom).(Adapted fromUnneland and Haug-land, reference 11.)

nLocation and permanent monitoring system setup of Gullfaks and Veslefrikk fields.(Adapted from Unneland and Haugland, reference 11.)

Data from permanent gauges are used tohistory match numerical models for eachproduction area, to identify the degree ofcommunication between wells and to con-trol the flow into and out of each block tomaintain material balance. For example,geological interpretation indicated a faultbetween two wells—a producer and aninjector. The producer was shut in duringstart of injection. Permanent gauge datafrom the producer showed an increase inpressure during this start-up period indicat-ing excellent communication across thefault. Combining openhole pressure dataand permanent pressure data has revealedsuch interwell communication in a numberof wells.

About 40% of Gullfaks producers aregravel packed and contribute more than50% of production.13 In the majority ofthese wells, permanent gauges continuouslymonitor downhole flowing pressure andtemperature. These data provide input tomonitor gravel-pack performance and maybe used to analyze and identify problemscaused by a variety of phenomena, includ-ing migration of fines and scales.

As an alternative to gravel packing forsand control, Statoil has used indirect verti-cal fracturing to complete several wells.14

This method allows production from uncon-solidated sands through less productive,fractured, consolidated intervals. Availabilityof real-time downhole pressure data allowsfracturing operations to be optimized and,for operational reasons, these data can beobtained only from permanently installedmonitoring systems.

43

Comparison of Raw Data

Wireline gauge data

Effects of phase segregation atdepth of permanent gauge

24,0

0025

,000

26,0

0027

,000

Per

man

ent p

ress

ure,

kP

a

Permanentgauge data

Wire

line

pres

sure

, kP

a27

,000

28,0

0029

,000

30,0

00

pre

ssur

e, k

Pa

5,50

026

,000

26,5

00W

ellb

ore

pres

sure

, kP

a27

,500

28,0

0028

,500

29,0

00

By visual fitp = 26086.051 kPam = 271.685 kPakh = 17.1477 µm2.ms = - 0.87 storage units

By least squaresp = 29116.602 kPam = 271.685 kPakh = 17.1477 µm2.ms = - 2.04 storage units

Interpretation of Wireline Pressure Data

Interpretation of Permanent Pressure Data

-5 -4 -3 -2 -1 0Multiple-rate Horner time

0 4 8 12 16 20 0 4 8 12 16Time, hr

nComparisonbetween permanentand wireline gaugepressure data. OnGullfaks A-10buildup pressuredata were recordedby a wireline pres-sure gauge set closeto the perforationsand also by the per-manent gauge (top).The permanentpressure data showthe effects of phasesegregation, butdespite this bothdata sets may beused for analysis(middle and bottom).The difference incalculated skin, s,in the analysis ofthe two data sets isattributable to fric-tion pressure lossbetween the perfo-rated interval andthe permanentgauge. (Adapted fromUnneland and Haug-land, reference 11.)

Even though permanent sensors weredeployed several tens of meters above perfo-rations, transient analysis of their pressuredata gave satisfactory results when com-pared to data from wireline pressure gaugeslocated much closer to producing zones. Inthis example, the wellbore storage effect didnot dominate transient analysis, whichallowed a comparison of results. Differencesin calculated values of skin were attributedto frictional losses along the tubing to thepermanent gauge. This allowed correctionsto be made to other permanent gauge datasets in the area to estimate true formationskin (right).

Gullfaks produces from a mixture of weakformations and exhibits large variations indepletion in the various fault blocks andreservoirs. In many cases, only a small mar-gin exists between formation fracture pres-sure and pore pressure. Safe drillingdepends on obtaining an estimate of porepressure for each zone before penetratingthe reservoir. Permanent pressure data areused to calculate pore pressure and hencedetermine the optimum mud weight for wellcontrol without fracturing the formation.

Veslefrikk—The 12,000-m3/d [75,000-B/D] Veslefrikk field, located 145 km [90miles] northwest of Bergen, was considereda marginal field. To reduce total investment,commingled production and injection wasplanned from the Brent and Intra DunlinSand (IDS) reservoirs. Control is obtained byselective perforation in producers anddownhole chokes in injectors. A carefullyplanned data acquisition program duringthe initial production phase provided infor-mation about reservoir properties, produc-tion potential and well behavior. In addi-tion, two of the largest uncertainties werepartially resolved: the degree of communi-cation across the main arcuate fault and thevertical transmissibility between the Lowerand Middle Brent through the low-qualityRannoch sand.

This information led to improved reservoirdescription and allowed adjustments to be

44 Oilfield Review

Wel

lbor

e24

,500

25,0

002

-5 -4 -3 -2 -1 0Multiple-rate Horner time

Reservoir Management ApplicationsApplication

Interference testing

Reservoir pressure control

Transient well testing

History matching

Well performance

Hydraulic fracturing

Bottomhole pressure data

Description

Establishes the degree of commu-nication across the field, betweenwells, between fault blocks and also the vertical transmissibilitybetween reservoirs.

Maintains bottomhole flowing pressure above a threshold bymonitoring permanent gauge datawhile adjusting water or gas injec-tion, or while varying production.

Performs a pressure buildup testautomatically whenever a well containing a permanent gauge is either deliberately or inadver-tently shut-in. Similarly a pressuredrawdown test is performed whenthe well is opened up.

Provides continuous recording ofpressure data during the lifetime of the well.

Provides continuous recording ofpressure data.

Monitors downhole pressure duringhydraulic fracturing.

Provides continuous knowledge of bottomhole pressure.

Benefits

• Wireline intervention eliminated.• Limited planning involved.• Observation of effects caused

by any change in production orinjection in wells where perma-nent gauges are installed.

• Individual well production maximized.

• Injection rates optimized.• Sand production eliminated by

controlling drawdown.• Completion costs optimized.

• Transient analysis of problemwells with minimum intervention.

• Real-time, early reservoir datawith no cable in the tubingduring an extended well test orduring early production.

• Remote control and analysisof data.

• Limited production loss byeliminating wireline operations.

• Verification or adjustment ofreservoir models.

• Improved reservoir description.• Improved estimation of reserves.

• Completion performance established.

• Gravel-pack performanceestablished.

• Monitoring of migration of fines or scale buildup.

• Wellbore hydraulic curves calibrated for optimizing gas lift.

• Fracture length optimized.• Real-time surface readout of

downhole pressure data during a frac.

• Pore pressure calculated for safety while drilling developmentwells.

• Computation of accurate kill fluidweight.

• Calculation of accurate under-balance or over-balance beforeperforating.

Field or Well Condition ApplicationsApplication

Restricted access

Highly deviated wells

Pumping wells

Description

Installs permanent monitoring sys-tems whenever access is restrictedby subsea completion wells, smallplatforms or single wells, remoteland wells or when other activitieson a platform, such as continuousdrilling, prevent well access.

Installs permanent monitoring sys-tem with completion.

Runs gauges to monitor pumpinlet pressure and pump outletpressure.

Benefits

• Elimination of costs associatedwith wireline intervention.

• Operational hazards reduced.• No personnel required.• No rig required.• Only method to record data in

many cases.

• Elimination of costs of coiled tubing or snubbing equipment to convey wireline pressure gauges.

• Pump efficiency established.• Pump rate optimized.• Pump maintenance planned.• Only practical method to record

downhole pressure data.

made to the reservoir development planbefore drilling the first injection well. A full-field numerical model was constructed,enabling careful planning and effectivereservoir management. Permanent gaugedata play a vital role for continuous historymatching and refinement of this model.

History matching has shown that lateralcommunication across the IDS reservoir isconsiderably more complex than describedby the geological model. This complexityhas been attributed to lateral changes inlithology.

Permanent gauges were run in dedicatedIDS wells. A falloff analysis showed that oneproducer was isolated on three sides fromwells farther south and that communicationto an injector across the main arcuate faulton the fourth side was unexpectedly good.This information allowed the injection rate tobe decreased and the production rateincreased until the pressure balancedbetween the two. This resulted in an increasein production of 200 m3/d [1260 B/D].

Gauge Drift—To monitor permanentgauge performance over time, Statoil hascompared permanent gauge pressure datawith data from wireline pressure gauges(next page). A radioactive marker installedin the permanent gauge mandrel allowsaccurate depth control of the wirelinegauges run during periodic production log-ging operations. Several comparisons havebeen made in the same wells over severalyears and no significant drift has beenobserved on any of the gauges.

Permanent monitoring systems are consid-ered a good investment by Statoil and instal-lations have recently been made in severalsatellite fields around Statfjord and Sleipner,with plans to install gauges on the Heidronplatform.

45

nApplications andbenefits of perma-nent monitoringsystems.

46 Oilfield Review

15. Kilvington LJ and Gallivan JD, “Beatrice Field: Elec-trical Submersible Pump and Reservoir Performance1981-83,” Journal of Petroleum Technology 36(November 1984): 1934-1940; also paper SPE11881, presented at the Offshore Europe 83 Confer-ence, Aberdeen, Scotland, September 6-9, 1983.Gallivan JD, Kilvington LJ and Shere AJ: “ExperienceWith Permanent Bottomhole Pressure/TemperatureGauges in a North Sea Oil Field,” paper SPE 13988,presented at the Offshore Europe Conference 85,Aberdeen, Scotland, September 10-13, 1985.

16. Brodie AD, Allan JC and Hill G: “Operating Experi-ence With ESPs and Permanent Downhole Flowme-ters in Wytch Farm Extended-Reach Wells,” Journalof Petroleum Technology 47, no. 10 (October 1995):902-906.

150

100

50

0

-50

-100

-150

Pre

ssur

e di

ffere

nce,

kP

a

21,500

Permanent gauge pressure, kPa23,500 25,500 27,500 29,500

System 1System 2System 3

nPermanent gauge drift. Whenever wireline pressure gauges havebeen used in wells with permanent gauges installed, gauge drift maybe measured. Radioactive markers help avoid depth mismatch errorsenabling a good comparison. Results show that differences in mostcases are acceptable especially when gauge accuracy is taken intoaccount. Data presented are from gauges installed by three compa-nies. The discrepancy for System 3 is believed to be caused by wire-line gauge error rather than permanent gauge drift. (Adapted fromUnneland and Haugland, reference 11.)

Multisensor ApplicationsAll the applications described above requireonly one permanent downhole pressuregauge—even though in some cases a secondgauge has been used for redundancy. Thereare many other applications for permanentmonitoring systems and some require morethan one sensor (previous page).

Many oil companies have wells equippedwith electrical submersible pumps that areimpractical to log by wireline methods.15 Asingle permanent pressure gauge may provide useful information about well orreservoir performance, recording formationpressure when pumps are switched off.However, monitoring pressure during pump-ing—at the pump inlet and outlet—providesadditional information about pump effi-ciency. Pump efficiency has an impact not

only on production, but also on pump lifeand workover schedules. Tracking pump per-formance and adjusting pump speed tomatch reservoir conditions increase effi-ciency and pump life.

A relatively new technique uses perma-nent pressure gauges and a venturi to moni-tor downhole flow rates.16 A venturi isessentially a restriction placed in a flow-line—in this case the tubing. The venturicauses a small change in pressure—typi-cally less than 10 psi—which is related tothe fluid velocity. Often three pressuregauges are used, two for the venturi—tomeasure differential pressure—and the thirdone for standard pressure measurementssome distance away from the other two, sothat fluid density may also be calculated.

A Permanent Seat in Completion Plans?Driven by the trend toward unmanned plat-forms, subsea completions, limited accesswells—either because of their remote loca-tion on land, top-side activity offshore orwell deviation in general—permanent moni-toring systems are becoming an establishedpart of well completions. Now that reliabilityissues have been resolved by sound projectmanagement and the introduction of newtechnology, permanent monitoring systemsare a proven cost-effective and safer alterna-tive to intrusive data acquisition methods.

Applications for pressure data gathered bypermanent monitoring systems are numer-ous, ranging from reservoir evaluation dur-ing extended well tests or early in the pro-duction cycle, to lifetime reservoir and wellmanagement. Modern communication sys-tems enable remote control of the sensorsand make the data accessible from officesanywhere in the world, increasing the valueof permanent monitoring systems. —AM

Winter 1995

David AllenMontrouge, France

Bob DennisMuscat, Oman

John EdwardsJakarta, Indonesia

Stan FranklinJack LivingstonSouth Pacific ChevronBrisbane, Queensland, Australia

Andrew KirkwoodJim WhiteAberdeen, Scotland

Lee LehtonenMobil Exploration and ProducingNew Orleans, Louisiana, USA

Bruce LyonJeff PrillimanNew Orleans, Louisiana

Gerard SimmsAmoco Production CompanyNew Orleans, Louisiana

Modeling Logs for Horizontal Well Planning and Evaluation

For help in preparation of this article, thanks to BülentBaygün, Charles Flaum, Martin Lüling and RaghuRamamoorthy, Schlumberger-Doll Research, Ridgefield,Connecticut, USA; Kees Castelijns, Schlumberger Hous-ton Product Center, Sugar Land, Texas, USA; Huck HuiChang, Anadrill, Kuala Lumpur, Malaysia; Patricia Hall,Keerthi McIntosh and Jessie Lopez, Amoco ProductionCompany, New Orleans, Louisiana, USA; Scott Jacobsen,Schlumberger Wireline & Testing, Muscat, Oman; ChinLeong Lim, Anadrill, Jakarta, Indonesia; Ed Stockhausen,Chevron, New Orleans, Louisiana; and Wendy Tan,Anadrill, Singapore.

Before drilling a horizontal well, the important question for the operator

is “How can I land the well?” Once the well is drilled, the important

questions are “Where is the well?” and “How good is the reservoir?”

Answers to these questions come from an integrated forward modeling

system that simulates log responses along planned or drilled well tra-

jectories to guide drillers and help interpreters evaluate the formations.

The 1990s may become known in the oilfield as the decade of the horizontal well.During the past six years, the number ofhorizontal wells drilled annually worldwidehas jumped 1200%, from 250 to 3000. Thereasons for this dramatic growth are many.Horizontal wells can increase productionrates and ultimate recovery, and can reducethe number of platforms or wells required todevelop a reservoir. They can also helpavoid water or gas breakthrough, bypassenvironmentally sensitive areas and reducestimulation costs.1

As exploration and development budgetstighten, companies are becoming more effi-cient by drilling fewer, well-placed holes.Reentry and multilateral wells are growingin number, along with short-radius wells.There are greater expectations and smallermargins for error in drilling today’s horizon-tal wells.

In this article ADN (Azimuthal Density Neutron tool), AIT(Array Induction Imager Tool), ARC5 (Array ResistivityCompensated tool), ARI (Azimuthal Resistivity Imager),CDN (Compensated Density Neutron tool), CDR (Com-pensated Dual Resistivity tool), DIL (Dual Induction Resis-tivity Log), DLL (Dual Laterolog Resistivity), DSI (DipoleShear Sonic Imager), FMI (Fullbore Formation MicroIm-ager), GeoSteering, IDEAL (Integrated Drilling Evaluationand Logging), INFORM (Integrated Forward Modeling),IPL (Integrated Porosity Lithology), ISONIC (IDEAL sonic-while-drilling tool), Litho-Density and RAB (Resistivity-at-the-Bit) are marks of Schlumberger.

Drilling horizontal wells presentsformidable challenges. Planning trajectories,choosing fluids, steering, formation evalua-tion and completion—each stage is a hugetask. Several stages—planning, steering andformation evaluation—benefit from combin-ing the efforts of geologists, log analysts anddirectional drillers.

A powerful partner in all these stages isforward modeling, or log simulation. Otherindustries are using simulation to help trainpilots, model aircraft and automobile relia-bility and response, design buildings, testweapons, record music, predict weather—the list is endless. In the oil field, modelinghelps make efficient use of logs in horizon-tal wells in two ways—first by predictinglogging-while-drilling (LWD) tool responseto guide directional drilling, and second by

47

1. Teel ME: “Longer Reach is Key to Future Develop-ments,” World Oil 215, no. 4 (April 1994): 27.Deskins WG, McDonald WJ and Reid TB: “SurveyShows Successes, Failures of Horizontal Wells,” Oil &Gas Journal 93, no. 25 (June 19, 1995): 39, 42-45.

constraining formation evaluation when theconventional assumptions of a vertical wellno longer hold.

Directional drilling practice and technol-ogy have evolved to the point where, given agood plan, the target can be hit with highaccuracy. The drill bit can be placed within atarget the volume of an engineer’s office at adepth and lateral offset of a few miles. Trajec-tories are becoming more complex as direc-tional drillers push the technology to its limitsin “designer” wells (below).2 To improve theodds of these wells hitting the target, they arecarefully planned in two steps: definition ofthe target from maps and logs, then design ofa wellbore trajectory to hit it.

No plan, unfortunately, is foolproof.Uncertainties in the position of the target,combined with unpredictability of structuraland stratigraphic variations, even in devel-oped fields, can cause directional drillers tolose their way. The chance of going astraydeclines significantly, however, with the useof real-time formation evaluation logs andcomparison of the logs with modeled casesto gauge the position of the tool within thesequence of beds. The INFORM IntegratedForward Modeling program provides aninterface for building a formation model andsimulating log response, allowing drillers toanticipate what’s ahead. We look first atmodeling for horizontal well planning, thenexplore how the INFORM system facilitatespostdrilling visualization and formationevaluation of LWD and wireline logs in hor-izontal wells.

48

nA “designer” well with tu(Adapted from Teel, reference

Model First, Then DrillOften the objective of drilling a horizontalwell is to penetrate the reservoir but stayclose to a caprock shale or gas-oil con-tact—to drill parallel to a boundary or acontrast in material properties—for thou-sands of feet. Such a viewing angle isunusual for electromagnetic tools, the toolsmost commonly used for steering.3 Othermeasurements, such as gamma ray and den-sity, are also affected by the horizontalgeometry, giving an asymmetric response asthey lie against the floor of the borehole.

Because most resistivity tools probe sev-eral feet into the formation, they are affectedby resistivity inhomogeneities in the vicinityof the well and even ahead of the drill bit.This early warning feature is beneficial todirectional drillers, who harness it to steerwells into target layers or away from prob-lem zones before they are encountered bythe bit. This “proximity effect” can beaccurately modeled during predrilling plan-ning to provide a road map for drilling.

In a planning example from the NorthSea, Jim White of Schlumberger Wireline &Testing in Aberdeen, Scotland, used logmodeling to demonstrate the feasibility oflanding the well in a thin sand and avoidinghigh-resistivity, calcite-cemented, tightstreaks (next page ).4 Forward modelingcomputed the response of the CDR Com-pensated Dual Resistivity tool with its twodepths of investigation—shallow from thephase shift measurement and deep from theattenuation log.5 When the wellbore cameto within 3 ft [0.9 m] of the calcite zone, the

rns in the horizontal plane. 2.)

modeled attenuation and phase shift curvescrossed, because the deeper-reading attenu-ation measurement senses the high-resistiv-ity calcite.

The CDR logs acquired when the wellwas drilled corroborated the modeled pre-dictions. Based on the simulations, the sig-nature of the lower boundary—the deepreading crossing over the shallow—was rec-ognized while drilling, and the well wassteered away. Had the well entered thecemented zone, drillers estimated theywould have spent several days trying to getback on target.

Geologists from Chevron Niugini areusing INFORM forward modeling to planand geosteer horizontal wells in the Iagifu-Hedinia field, within the Southern High-lands Province of Papua New Guinea.Located in the Papuan Fold and Thrust belt,this field is part of a double anticline com-plex in the Hedinia thrust sheet. The majoroil reservoir is the Lower Cretaceous Torosandstone. Within the Toro, the hydrocar-bon accumulation consists of an oil band upto 218 meters [715 ft] thick overlain by agas cap. Gas cap expansion and gravitydrainage are the major drive mechanismsfor the field, with support from the Toroaquifer making a minor contribution.

Development well planning and drillingare complicated by the complex fold geom-etry. Unfortunately, the rugged karst topog-raphy created in the Darai Limestone at sur-face prohibits the acquisition of usableseismic data.6 For predicting the subsurfacereservoir geometry, geologists rely on sur-face geological mapping, side-scan radarimagery, dipmeter data and correlation logsfrom adjacent wells.

In order to maximize productivity and ulti-mate recovery from the horizontal wells,wells are programmed to be horizontal in theToro oil reservoir at a level of 15 m [50 ft]above the oil-water contact. This enables thewells to produce oil at lower solution gas/oilratios (GORs) and should delay breakthroughfrom the advancing gas front.7

During drilling to the Toro objective, thelanding phase is critical to the success of thehorizontal well program. With an unstableAlene shale section overlying the Toro, it isimportant to minimize the amount of hori-zontal section drilled before encountering thetop Toro. Conversely, encountering the Toroduring the build section of the well course,before reaching horizontal, can result in lossof productive interval since this hole section

Oilfield Review

2. A designer well is defined as one with turns greaterthan 30° in the horizontal plane, combination rightand left turns, and turns not restricted by inclination.From Teel ME: “Extended Reach Becomes Fashion-able,” World Oil 215, no. 6 (June 1994): 27.

3. For discussion of the challenges of modeling andinterpreting while drilling and wireline electromag-netic tool response in deviated and horizontal wells:Betts P, Blount C, Broman B, Clark B, Hibbard L,Louis A and Oosthoek P: “Acquiring and InterpretingLogs in Horizontal Wells,” Oilfield Review 2, no. 3(July 1990): 34-51.Anderson B, Bryant I, Lüling M, Spies B and Helbig K:“Oilfield Anisotropy: Its Origins and Electrical Char-acteristics,” Oilfield Review 6, no. 4 (October 1994):48-56.Anderson B: “The Analysis of Some Unsolved Induc-tion Interpretation Problems Using Computer Model-ing,” Transactions of the SPWLA 27th Annual Log-ging Symposium, Houston, Texas, USA, June 9-13,1986, paper II.Anderson B, Minerbo G, Oristaglio M, Barber T,Freedman B and Shray F: “Modeling ElectromagneticTool Response,” Oilfield Review 4, no. 3 (July 1992):22-32.Allen DF and Lüling M: “Integration of WirelineResistivity Data with Dual Depth of Investigation 2-MHz MWD Resistivity Data,” Transactions of theSPWLA 30th Annual Logging Symposium, Denver,Colorado, USA, June 11-14, 1989, paper C.

4. White J: “Geological Steering Assists Cost EffectiveExploitation of Marginal Reserves,” paper SPE 30362,presented at the Offshore Europe Conference,Aberdeen, Scotland, September 5-8, 1995.

5. The attenuation resistivity probes roughly twice asdeep as the phase-shift resistivity. Absolute depths ofinvestigation vary with background resistivity,decreasing with decreasing resistivity. For more onthe subject, see Allen and Lüling, reference 3.

6. Karst is a type of topography formed on carbonaterocks by dissolution.

7. Magner TN and McKay WI: “PNG’s Kutubu Project:Lessons in the First 100 Million Barrels,” paper SPE28784, presented at the SPE Asia Pacific Oil & GasConference, Melbourne, Australia, November 7-10,1994.

nModeled andacquired logs in aNorth Sea horizon-tal well. Modelingof CDR Compen-sated Dual Resistiv-ity tool andgamma ray logs(top) shows that the high-resistivitycalcite-cementedzone at the base of the pay shouldbe detectablewhile drilling. Thedeep-sensing atten-uation resistivitylog (purple) crossesover the shallow-reading phase-shiftlog (brown) as thehigh-resistivity zoneis approached.Acquired logs (bottom) show asimilar feature, asthe well wassteered to avoid the calcite-cemented zone.

Gam

ma

Ray

, AP

I 200160

120

80

400

XX20

XX44

XX68

XX92

X116

X140True

ver

tical

dep

th, f

t

4400 4800 5600 6400 7200

Distance along the section, ft

Res

istiv

ity, o

hm-m

1

10

100

5200 6000 6800

Modeled

Phase shiftAttenuation

200Distance along the section, ft

Res

istiv

ityoh

m-m

10

51000

X70

Dep

th, f

t

X75

X80X85

may be too close to the current gas-oil con-tact and would not be perforated. The Aleneis drilled with mud weights in the range of 12to 14 ppg, while the current reservoir pres-sures in the Toro are in the 4.5 to 5.5 ppgequivalent range. To prevent lost circulationproblems and possible loss of the hole, it isnecessary to identify the top of the Toro cas-ing point before penetrating more than 1.5 to3 m [5 to 10 ft] of the sandstone.

An accurate predictive model of the Toroanticlinal geometry resulting from recogni-tion of overlying stratigraphic markers whiledrilling—as well as the ability to determinethe structural attitude of these layers—increases the probability for a successfullanding phase. With INFORM processing, a

Winter 1995

model of the stratigraphic interval above thetarget can be built using well logs and dip-meter data from nearby wells along withgeological structure models developed forthe planned horizontal well. LWD responsesfor the potential range of structural dipswithin a particular area of the anticlinal foldcan be simulated. (For a description of howthe INFORM system works, see “INFORMIngenuity,” page 52.)

As the well course builds to horizontal,the geosteering specialist and geologist cor-relate major stratigraphic LWD markers andestimate the structural dip of a stratigraphicunit in the plane of the well course by opti-mizing the match between the LWD curvesand the model log curves. The calculatedstructural dip estimates are compared tothose in the geologist’s predicted fold geom-etry cross-sectional model. The new dips arethen used to correct the Toro subsurfacestructure model and revise the top Toro tar-get coordinates.

During the planning for the first well, IHT-1, gamma ray and resistivity logs from threenearby wells were used to create a modelfor computing CDR responses for the fullrange of possible structural dip magnitudesalong potential well trajectories. Theresponses were stored in a relative angledata base. The programmed well course wasoblique to the strike of the Toro in this areaof the Iagifu anticline, and was designed tobe horizontal 15 m above the oil-water con-tact. This entry point is depth-constrained bythe predicted oil-water contact level, andlaterally constrained by the projected posi-tion of the Toro entry point, determined byprojecting the Toro structural dip away fromwell control points higher on the anticlinalflank. The kick-off depth and deviation

49

50

Oil-water contact

TVD target level

Initialwellplan

Revisedwell plan

Steeper dipTop Toro

target

Initialstructure

model

Toro sand

Kick-offpoints

12.2620.0923.9529.1356.8258.8960.2961.5462.8364.2666.1068.1470.6773.4180.23

True

ver

tical

dep

th, f

t

7600

7800

8000

8200

8400

8600

8800

9000400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 2600 2800

Section departure, ft

GR, API

Top Toro Ain IHT-1

Top Toro Ain IHT-1A

■■Kick-off points and build rates varying with entry point positionand structural dip. In this case, the location of the crest of theanticline was known, but the dip of the flanks was not. For an initial well position, a steeper dip requires a shallower kick-offpoint and slower build rate to reach the entry point at the correctlocation.

■■Cross-sectional model of the Chevron Niugini IHT-1 well trajectory. This 200-layer model was built using logs from threenearby wells with the INFORM Integrated Forward Modeling system. Formation dips were constrained by the matchbetween modeled logs and logs acquired while drilling. Well IHT-1 is in green, its gamma ray reading in magenta, andnearby Well IHT-1A in blue. A plan view of the two trajectories is inset. Layers are colored according to gamma ray values,with light colors indicating low readings, dark colors indicating high values. Water is indicated with blue-green shading.

deviation angle build rate depend on know-ing this entry position (left).

During the drilling of well IHT-1, a com-puter structure model with sections of 6°and 8° apparent dip was constructed withthe INFORM system, using data transmittedvia satellite link (above).8 The stratigraphichorizon boundaries, dip magnitude and truevertical depth of each section was deter-mined from the match between the mea-sured CDR logs and the modeled logs (nextpage, left). This match is consistent down tothe Toro, indicating the structural dip modelis a good representation of the actual Torosubsurface structure.

Typically the CDR tool, producing charac-teristic horns at high-angle bed boundaries,is run to land wells. For the IHT-1 well bot-tomhole assembly configurations, however,this tool is located 18 m [60 ft] behind thebit. To precisely locate the 95/8-in. casingsetting depth at the top Toro, the last bit tripis run with the GeoSteering tool, an instru-mented steerable downhole motor with tworesistivity sensors.

Oilfield Review

51Winter 1995

Dep

th, f

t

8100

Top of Toro A

B3 @ TVD8166

J1 @ TVD8467

A1 @ TVD8606

A2 @ TVD8683

OWC @8741 TVD

8500

8900

9300

10100

10500

ohm-mAPI0 150 2 200

Measured CDRGamma Ray

MeasuredGeoSteeringGamma Ray

Modeled CDRGamma Ray

ModeledGeoSteeringGamma Ray

Measured CDRAttenuation

Measured RBIT

Modeled CDRAttenuation

Modeled RBIT

nModeled and measured gamma ray,GeoSteering and CDR logs from theChevron Niugini IHT-1 well. Log signaturesthrough marker beds coincide with thoseat marker depths predicted before drilling.

8. Dip is called apparent when it is measured in a dif-ferent direction from that of the steepest dip.

9. Bonner S, Burgess T, Clark B, Decker D, Orban J,Prevedel B, Lüling M, and White J: “Measurements atthe Bit: A New Generation of MWD Tools,” OilfieldReview 5, no. 2/3 (April/July 1993): 44-54.

nMeasured gamma ray and bit and arc resistivities showing an apparent depth shiftas the GeoSteering tool slides across the top reservoir interface. The shift can beexplained by the counteracting effects of the high inclination of the trajectory and adip in the interface. Insets show simulated bit and arc resistivity logs for models withthree dips—6°, 8° and 10°. An 8° dip fits both resistivities.

2

2

0.2

0.2

2000

2000

200

200

ohm-m

ohm-m

ohm-m

ohm-m

RBIT measured

Arc-up measured

Dep

th, f

t

9540

9560

9580

9600

95200 100API

Measured Gamma Ray

Modeled Gamma Ray

0 API 100

RBIT modeled

Arc-up modeled

10° dip

8° dip

6° dip

10° dip

8° dip

6° dipBitresistivity

Arc resistivity

Gamma ray

The primary purpose of the GeoSteeringtool is to drill the horizontal drainhole andconfirm that the well is above the oil-watercontact in each sand. Normally GeoSteer-ing tool data are not acquired in the upper6 to 9 m [20 to 30 ft] of the Toro, until thetool signal receiver clears casing. Becauseof mechanical problems, the 95/8-in. casingin IHT-1 ended 27 m [90 ft] above the Toro.This allowed the GeoSteering tool toacquire data across the shale-sandstoneresistivity contrast at the top of the Toro.

The GeoSteering tool’s two resistivity sen-sors measure different rock volumes: thetoroid at the bit is hemispherically focusedaround the bit and the bit box; the arc-shaped electrode is azimuthally focused ina 120° arc perpendicular to the tool axis.9

The gamma ray sensor is also azimuthallyfocused with back shielding. While rotarydrilling, the arc and gamma ray sweep thefull 360° of borehole wall. When the motoris sliding, however—in this example, build-ing angle with zero tool face—the bit resis-tivity measures omnidirectionally, while thearc measures upward and the gamma raymeasures downward.

Across the top of the Toro Sandstone, thebit resistivity, arc resistivity and gamma rayhave an apparent depth mismatch (above).The unit above the Toro entry point isAlene shale with true resistivity, Rt, of 6 to10 ohm-m, while directly below it is a sec-tion of tight, calcite-cemented Upper Toro

(continued on page 56)

52 Oilfield Review

INFORM Ingenuity

■■Log squaring, the first step in building a formation model for integrated forward modeling. In this example,an FMI Fullbore Formation MicroImager log from an offset well (track 1) guides manual layer placement forfine-tuning an automated log squaring result. The resulting squared log (red, track 2) is compared with themeasured resistivity log (blue, track 2). Modeled and measured logs are compared in track 3.

TopDepth

BottomDepth Value0.2 0

0.2 0

Input

Squared

Computed

Input

0.2 0

0.2 01.20

1220

1212.85

1213.35

1213.85

1214.40

1215.1

1215.8

1216.33

1216.83

1217.41

1217.71

1218.57

1219.03

1219.63

1220.15

1220.67

1221.06

1221.46

1221.72

1222.20

1213.35

1213.85

1214.40

1215.1

1215.8

1216.33

1216.83

1217.41

1217.71

1218.57

1219.03

1219.63

1220.15

1220.67

1221.06

1221.46

1221.72

1222.20

1223.02

0.067

0.135

0.009

0.069

0.015

0.001

0.031

0.101

0.095

0.100

0.072

0.063

0.077

0.066

0.099

0.076

0.051

0.069

0.103

The INFORM system allows the analyst to con-

struct a detailed model of the geometry and petro-

physical properties of the layers that have been or

will be penetrated by the well. Then tool

responses along the well trajectory through that

model are simulated.

Building the Model

Building the 2D petrophysical description of the

prospect from offset well logs and production data,

geologic maps and cross sections is the most time-

consuming part of the INFORM process. The tool

response calculation can be run over a coffee or

lunch break. Once the basic geometric framework

is built, additional information—improved esti-

mates of structure or stratigraphy from reprocessed

seismic data, more detailed correlation studies,

logs from the pilot hole—can be incorporated.

Logs and other petrophysical data from offset or

pilot wells provide the foundation for the INFORM

model, defining layer thicknesses and properties.

In a process called log squaring, bed boundaries

are determined from inflection points on the logs,

and the average layer properties are extracted

from the log values. INFORM modeling offers a

combination of automatic and interactive, or man-

ual, tools for log squaring (right). The squared log

layers are then stretched or squeezed to fit the

expected model at the location of the horizontal

well. Squared logs from other wells may indicate

lateral facies changes in the model.

Geologic maps and a cross section of the

prospect, oriented along the proposed drainhole,

provide the dip along the plane of the well, the

number and throw of faults, and additional informa-

tion about lateral facies changes. With this infor-

mation the analyst can subdivide the model into a

small number of simple blocks, with faults, dip

changes and other lateral variations as boundaries.

The beds, their thicknesses and petrophysical

properties are represented in the INFORM method-

ology as a “layer column”—a table that contains

all parameters describing one block of the 2D

model. Properties include gamma ray (GR) in API

units, horizontal and vertical resistivities (Rh and

Rv), bulk density (RHOB), photoelectric factor (Pe),

neutron porosity (NPHI) and sonic transit time

(DT). For simple formations, in which the forma-

tion is laterally homogeneous, only one layer col-

umn is necessary. For more complex cases, a

series of layer columns is required. This method-

ology separates the analysis of formation proper-

ties and bed thicknesses—stratigraphy—from for-

mation dip and depth—structure. Once layer

columns are constructed, it is a simple matter to

rotate coordinates to change dip or translate

columns to introduce faults.

Finally, the geometry of the well trajectory is

input so relative angles can be computed.

Relative Angle Data Base

As the well deviates with depth, in addition to vari-

ations in formation properties, the distance and

the angle between the tool and layer boundaries

change, affecting logging tool response. The angle

between the tool axis and the normal to the layer

is called the relative angle. Modeling tool

response through these continuous changes

requires an efficient code, one that is both fast and

accurate. The simplest technique is to approxi-

mate the actual well trajectory with a small num-

ber of straight-line modeling runs. This is fast, but

not accurate. Another method, which computes

tool response at every point on the trajectory, is

accurate, but too slow for geosteering purposes.

A compromise was proposed by Martin Lüling

while at Schlumberger LWD Engineering in Sugar

Land, Texas, USA. For a given trajectory, the

INFORM technique computes relative angles at

53Winter 1995

Recommendedangles

3840

3850

3860

3870

86 88 90Relative angle, degree

True

ver

tical

dep

th, f

t

843880

0 45 65 70 75 80 85 90

Shale 1

Shale 2

Oil sand

Watersand

Shale 3

Relative angle, degree

1. For an example of the technique: McCann D,Kashikar S, Austin J, Woodhams R and Siddiqui S:“Geologic Steering Keeps Horizontal Well on Target,”World Oil 215, no. 5 (May 1994): 37-43.

■■A relative angle plot—a graph of the relative angles at which the given trajectoryintersects bed boundaries. This relative angle plot corresponds to the formationmodel shown on page 63.

■■The relative angle data base. The INFORM program computes the tool responseto each layer column at specified relative angles and stores them in a lookup table,or data base. The program then interpolates between stored values to output amodeled response along the wellbore. This figure represents one layer column.

every point along the well path (above). Next the

tool response to each layer column is computed at

specified relative angles and stored in a lookup

table (right). The program then interpolates

between tabled values to deliver a modeled

response at any desired sampling along the well-

bore. If the tool responses are found to be extra-

sensitive to changes in depth or angle, a more

finely sampled relative angle table may be con-

structed. The lookup procedure is rapid, so investi-

gation of multiple scenarios, involving changes to

the trajectory, formation dip or true vertical depth

is feasible.

Each output case can be stored for access by the

GeoSteering screen at the drillsite. Then modeled

logs and logs recorded while drilling can be com-

pared, helping to identify marker beds that guide

the well, and avoiding problem zones.1

The flexibility that comes with speed of compu-

tation allows testing of scenarios both before

drilling and after, for formation evaluation. A

54 Oilfield Review

■■Modeling the effect of slight changes in formation dip with the INFORM system. Uncertainty in formation dipcan cause failure to reach a horizontal well target. The first model (top) is designed to land the well in the payzone at 4330 ft TVD, then build angle to investigate other layers. Keeping the same trajectory but adding ahalf-degree to the formation dip (bottom), the pay zone is completely missed. The two models can be com-puted before drilling, using the same relative angle data base, and stored for access by the GeoSteering screento help real-time drilling decisions.

Gam

ma

Ray

, AP

IR

esis

tivity

, ohm

-mTr

ue v

ertic

al d

epth

, ft

150

0200

0.24200

4240

4280

4320

4360

4400500 1000 1500 2000


750 1250 1750

4050637075808590


GR, API

2. Lüling MG, Rosthal RA and Shray F: “Processing and Modeling 2-MHz Resistivity Tools in Dipping,Laminated, Anisotropic Formations,” Transactions ofthe SPWLA 35th Annual Logging Symposium, Tulsa,Oklahoma, USA, June 19-22, 1994, paper QQ.

Gam

ma

Ray

, AP

IR

esis

tivity

, ohm

-mTr

ue v

ertic

al d

epth

, ft

150

0200

0.24200

4240

4280

4320

4360

4400500 750 1000 1250 1500 1750 2000


Resistivity,ohm-m

0.150.24

0.400.500.600.750.901.051.202.002.206.00

0.30


predrilling planning example shows the effect of a

minor change in formation dip (left). With an

added half-degree of formation dip, the pay zone

at 4330 ft TVD is completely missed.

In a postjob formation evaluation example,

acquired CDR Compensated Dual Resistivity phase

shift and attenuation resistivities can be matched

best by introducing resistivity anisotropy—unequal

vertical and horizontal resistivities (next page). In

such formations the phase shift resistivity reads

higher than the attenuation measurement.2 The

INFORM software can simulate CDR curves in an

anisotropic formation, to distinguish its log

response from other phenomena that might have

similar signatures, such as nearby beds of con-

trasting resistivities.

INFORM Capabilities

The INFORM system is evolving rapidly. Currently

it can model responses for a variety of tools in sev-

eral environments. The catalog includes:

• 3D finite-element method laterolog codes for

wireline logs—ARI Azimuthal Resistivity Imager

and DLL Dual Laterolog Resistivity Logs; and

LWD tools—RAB Resistivity-at-the-Bit and

GeoSteering tool measurements in anisotropic

formations with dipping beds and invasion.

• 2D analytical induction codes for modeling wire-

line logs—AIT Array Induction Imager Tool mea-

surements and DIL Dual Induction Resistivity

Logs; and LWD logs—ARC5 Array Resistivity

Compensated and CDR logs. These can be mod-

eled in anisotropic formations with dipping beds.

• 2D sensitivity function density codes for model-

ing wireline measurements—IPL Integrated

Porosity Lithology and Litho-Density logs; and

LWD measurements—ADN Azimuthal Density

Neutron and CDN Compensated Density Neutron

tools in formations with dipping beds.

• 2D ray tracing sonic codes for modeling wireline

measurements—DSI Dipole Shear Sonic Imager

logs; and LWD measurements—ISONIC (IDEAL

sonic-while-drilling tool) logs in formations with

dipping beds.

• 1D convolution filter codes for modeling gamma

ray and neutron tools.

55Winter 1995

150

0

200

0.2

4310

4320

4330

4340

4350

43601000 1200 1400 1600 1800 2000


Tru

e ve

rtic

al d

epth

, ft

Res

istiv

ity, o

hm-m

Gam

ma

Ray

, AP

I

Modeled

Measured

Modeled phase shiftMeasured phase shiftModeled attenuationMeasured attenuation

0.240.300.400.460.470.480.600.750.901.2010.030.0

HorizontalResistivity,ohm-m

Modeled phase shiftMeasured phase shift

Modeled attenuationMeasured attenuation

200

2Res

istiv

ity, o

hm-m

20

200

2Res

istiv

ity, o

hm-m

20

200

2Res

istiv

ity, o

hm-m

20

2

200

Res

istiv

ity, o

hm-m

20

200

2Res

istiv

ity, o

hm-m

20

Rh = 10Rv = 10

Rh = 30Rv = 30

Rh = 60Rv = 60

Rh = 10Rv = 50

Rh = 10Rv = 90

nMatching modeled andmeasured CDR phase shiftand attenuation resistivitiesby introducing resistivityanisotropy. In formationswith resistivity anisotropy—Rv greater than Rh—theattenuation and phase shiftresistivities separate: thephase shift resistivity readshigher than the attenuationmeasurement. This phe-nomenon can be simulatedwith the INFORM software toproduce formation modelsconsistent with measuredlogs and offset wells. Thebest fit here is with Rv = 90ohm-m and Rh = 10 ohm-m.The model with Rv and Rh =60 ohm-m seems to fit theacquired logs well, but con-flicts with offset well informa-tion in the same formation.

nModeled andlogged gammaray and GeoSteer-ing tool responsefor the ChevronNiugini IHT-1Awell.

Modeled Gamma Ray

Logged Gamma Ray

API

API0 150

0 150 ohm-m

ohm-m2 2000

2 2000

Mea

sure

dde

pth,

ft

9800

10,000

10,200

10,400

10,600

10,800

11,000

11,200

11,400

Modeled GeoSteeringBit Resistivity

Logged GeoSteeringBit Resistivity

Top of Toro AGeostop

sandstone with an Rt of 210 ohm-m, whichoverlies porous sandstone. As the nearly hori-zontal GeoSteering tool slid across the dip-ping interface, the downward lookinggamma ray was the first to register the tightsand, while the upward looking arc resistivitywas the last. Modeling after the job with theINFORM program shows that this mismatchcan be explained by the counteracting effectsof the high inclination in the trajectory andan 8° apparent dip at the top Toro.

Unexpectedly, IHT-1 entered the dippingToro reservoir beneath a present-day oil-water contact at 8741 ft true vertical depth(TVD). The contact was apparently 15 to 18 m [50 to 60 ft] shallower than predicted,probably due to pressure depletion of theupper Toro reservoir in this area of the field.The bit resistivity gave an immediate indica-tion of water-saturated Toro. The plannedtrajectory was modified to build angle togreater than 90 degrees in an upward trajec-tory, crossing the oil-water contact fromunderneath. During drilling in the mid-Toro,the well encountered lost fluid circulationproblems, possibly at a fault or fracturezone. With sudden unloading of the bore-hole, collapse occurred in the unstableshale openhole section above the Toro, andthe hole was lost.

IHT-1A, a sidetrack designed to take a par-allel well path, was planned using the struc-tural attitude data and oil-water contactinformation from IHT-1. A short 30.5-m[100-ft], 81/2-in. pilot hole was drilled at theend of the buildup section with theGeoSteering tool to “geostop” exactly on theshale-sandstone reservoir boundary (right).This hole was enlarged, and the 95/8-in. cas-ing set just on the reservoir top. As expected,dips were close to those in IHT-1, and thewell was landed within the Toro oil leg asplanned, 15 m above the present-day oil-water contact. It continued for 427 m [1400 ft]across the three main Toro reservoir sand-stone members (next page, top). The wellwas completed as an oil well, producingmore than 10,000 stock tank barrels of oilper day, at solution GOR.

56

Another well, IHT-2, on the same struc-ture, encountered 55° dips, much steeperthan the 22° anticipated (next page, bottom).These were successfully modeled with theINFORM program and the well path modi-fied to hit the target.

After Drilling, Model AgainOnce drilled and logged, horizontal wellscontinue to pose challenges in visualizationand formation evaluation. Log simulationcan help verify a formation model or thelocation of a well in space, to use for futuredevelopment planning and quality control.More importantly, modeling helps untangle

true formation properties such as formationfluid resistivity, Rt, and water saturation, Sw,from the melange of shallow and deepresponses of while-drilling and wirelinetools. The INFORM program takes an inte-grated approach to log simulation for forma-tion evaluation by modeling a wide range oftool measurements simultaneously.

In the Gulf of Mexico, Lee Lehtonen atMobil Exploration and Producing in NewOrleans, Louisiana, USA tested simulationto validate the model of a horizontal well

Oilfield Review

57Winter 1995

8400Tr

ue v

ertic

al d

epth

, ft

8500

8600

8700

8800

8900

90002000 2200 2400 2600 2800 3000 3200 3400 3600 3800


12.2620.0923.9529.1356.8258.8960.2961.5462.8364.2666.1068.1470.6773.4180.23

GR, API

0 200 400 600 800 1000 1200 1400 1600 1800 2000


True

ver

tical

dep

th, f

t

7000

7100

7200

7300

7400

7500

7600

7700

7800

7900

8000

8100

8200

8300

Ga 12.9018.2323.2140.26101.99105.81108.68110.83113.43116.03120.92123.96126.50129.68136.79

GR, API

■■Cross-sectional model encountered by the IHT-1A well. The well was steered to penetrate the reservoir above the oil-watercontact in the top sand, and to thread through the three main sands of the reservoir. Layer color indicates gamma ray value.

■■Cross section for Well IHT-2, the second well in the Chevron Iagifu-Hedina horizontal development program. The CDRlogs were matched with dip panels from 55 to 24°. This provided an approximate top Toro landing point for trajectorycontrol. The exact landing point was determined by “geostopping” with the bit resistivity from the GeoSteering tool inthe last trip. Each dip panel in this example has the same true stratigraphic thickness. The true vertical thickness ofeach panel will vary with dip. The depth of each panel is adjusted to keep the formations continuous at the panelintersections along the well path. The gamma ray log is shown in magenta.

58

0.310

0.50

mile

km

N

Dis

tanc

e, ft 1500

2000

1000

500

00 500 1000 1500 2000 2500 3000 3500 4000

Plan View

Offsetwell

Pilot well

Trapping fault

Fault 4

Fault 3Fault 2

Fault 1

Horizontal well

nPlan view of a Gulf of Mexico horizontal well trajectorythrough a compartmentalized gas reservoir. In the seis-mic amplitude plot inset above, yellow and red indicatehydrocarbon extent. Amplitudes are bounded to thenorth by a large sealing fault. Breaks in color continuityhighlight additional faulting. The drilling plan called forintersecting each of these compartments.

XX90

X120

X130

X140

X150

X160

X170

7300 7800 8300 8800 9300 9800

Fault 2 Fault 3 Fault 4

X100

X110

Top B

Tight zone

Top A

Top B

Top A

Top A

Dep

th, f

t

nCross section of the faults and formations penetrated by the well.

designed to tap multiple compartments in afaulted reservoir (left).10 The horizontal wellwas to traverse four fault blocks (below). Payin the first and fourth blocks would be iso-lated by enough shale to allow setting exter-nal casing packers. In this case, INFORMmodeling showed how LWD porosity logscould be used to distinguish a change in for-mation properties associated with faultingfrom changes encountered in a new strati-graphic layer.

The ADN Azimuthal Density Neutron toolmeasures—while drilling—bulk density,ultrasonic standoff, photoelectric factor andneutron porosity.11 Magnetometers continu-ously measure tool orientation, and resultsare distributed into readings above, belowand to each side of the borehole (next page,bottom). This allows discrimination of theorientation of planes of porosity and densitydiscontinuity in the formation.

In the Mobil well, CDR and ADN datawere recorded into memory while drilling,and data were brought uphole with each bitchange. These logs were compared withlogs simulated using a formation modelbuilt from the known structure and pilotwell logs. During the fifth bit run, the den-sity tool encountered a shale-sand contact(next page, top). Examination of the densityporosity logs shows that the average andbottom quadrant curves both detect theinterface at the same measured depth,XX340 ft. Comparing the acquired and sim-ulated logs shows the contact can be mod-eled as a fault separating shale from sand.

During the seventh bit run, the wellencountered a shale-sand interface beforecrossing the next fault. As the tool enteredthe sand, the bottom quadrant densityporosity saw the sand, while the average ofall four quadrants still indicated a shale

10. Prilliman JD, Allen DF and Lehtonen LR: “HorizontalWell Placement and Petrophysical Evaluation UsingLWD,” paper SPE 30549, presented at the 70th SPEAnnual Technical Conference and Exhibition, Dal-las, Texas, USA, October 22-25, 1995.

11. Holenka J, Best D, Evans M, Kurkoski P and SloanW: “Azimuthal Porosity While Drilling,” Transac-tions of the SPWLA 36th Annual Logging Sympo-sium, Paris, France, June 26-29, 1995, paper BB.

Oilfield Review

59Winter 1995

Ultrasonic Caliper

Gamma Ray

Rate of Penetration5 Ft Average

True Vertical Depth

Resistivity AttenuationDeep

Resistivity Phase Shift

Bottom Quadrant

Average DensityPorosity

Neutron Porosity

XX400

XX300

Dep

th, f

t80 18

0 150

500 0

XX60 XX10

in.

GAPI

ft/hr

ft

0.2

0.2

200

200

ohm-m

ohm-m

60

60

60

0

0

0

p.u.

p.u.

p.u.

XX350

Gam

ma

Ray

, AP

I 10084

68

52

3620

Res

istiv

ity, o

hm-m

1

10

100

1000

RH

OB

, gm

/cm

3

1.651.85

2.05

2.25

2.452.65

XX10

XX12

XX14

XX16

XX18

XX20

True

ver

tical

dep

th, f

t

X200 X250 X300 X350 X400

Di t l th ti ftFa

ult

2

Measured

Modeled

Modeled Measured

Modeled

Measured

Left Right

Bottom

Top

nThe ADN Azimuthal Density Neutrontool reading above, below and to eachside of the borehole while drilling. Mag-netometers continuously measure ADNtool orientation, and processing groupsbulk density, ultrasonic standoff, photo-electric factor and neutron porosity mea-surements into quadrants.

nLogs acquired while drilling through a fault in the Mobil offshore Louisiana well. The density porosity logs show that the averageand bottom quadrant curves detect the interface at the same measured depth, XX340 ft (left). The difference between the bottomand average density porosity in the zone XX340 ft to XX370 ft is due to vertical segregation of invading mud filtrate. The logs can bemodeled by a nearly horizontal well intersecting a vertical fault separating shale from sand (right).

60

nLogs recordedwhile grazing ashale-sand inter-face. As the toolentered the sand,the bottom quad-rant density poros-ity saw the sand,while the averageof all four quad-rants still indicateda shale zone (top).As the well cutdeeper, the aver-age and bottomquadrant readingscame together 40 ft[12 m] farther on.The logs are consis-tent with a modelthat includes agently dippingsand (bottom).

Gam

ma

Ray

, AP

I 150120

90

60

300

Res

istiv

ity, o

hm-m

0.2

20

200

2000

RH

OB

, gm

/cm

3

1.51.7

1.9

2.1

2.32.5

XX00

XX10

XX20

XX30

XX40

XX50

True

ver

tical

dep

th, f

t

X100 X150 X200 X250 X300


Faul

t 3

MeasuredModeled

Clean sand

Bottomof hole

Topof hole

Ultrasonic Caliper

Gamma Ray

Rate of Penetration5-ft Average

True Vertical Depth

Resistivity AttenuationDeep

Resistivity Phase Shift

Bottom Quadrant

Average DensityPorosity

Neutron Porosity

X200

X100

Dep

th, f

t

80 18

0 150

500 0

XX60 XX10

in.

API

ft/hr

ft

0.2

0.2

200

200

ohm-m

ohm-m

60

60

60

0

0

0

p.u.

p.u.

p.u.

Bottomquadrantreads sand

zone (left). As the well cut deeper, the aver-age and bottom quadrant readings cametogether 40 ft [12 m] beyond the first indica-tion of the shale. Simulation with theINFORM system indicates the log responsescan be explained by a slightly dipping, 5-ftclean sand. Modeling the azimuthally sensi-tive response of the ADN tool allowed theorientation of the interface to be verified,constraining the subsurface structure andstratigraphy.

In another well offshore Gulf of Mexico,the Amoco team of Patricia Hall, KeerthiMcIntosh, Jessie Lopes and Gerard Simmsenlisted the INFORM system to add con-straints to the structural interpretation afterdrilling. The reservoir structure had beenmapped from log and 3D surface seismicdata, but the scarcity of wells in the southernblock of the reservoir left uncertainties instructural detail. In particular, the location ofthe crest of the targeted anticlinal featurewas poorly constrained on early maps (nextpage, top).

The horizontal well ran under the crest ofthe structure, and gamma ray and ARC5Array Resistivity Compensated logs wereacquired in memory while drilling.12 Afterdrilling, several structural and stratigraphicmodels were input to the INFORM programto determine the one that best explained therecorded logs. In this way, the relationshipbetween the well and the formations couldbe visualized, and completion and produc-tion strategies weighed.

A preliminary attempt to model the struc-ture as a simple anticline gave disappointingresults. Enhancements to the model—in theform of minor faults near the crest of theanticline and a facies change on the far sideof the structure—produce simulated logsthat begin to mimic some of the complexityof the acquired logs (next page, bottom).

In addition to evaluating LWD logs for for-mation properties, the INFORM method canalso be used to extract petrophysical proper-ties from wireline tools conveyed bydrillpipe or coiled tubing. Bob Dennis at theSchlumberger Wireline Division in Muscat,Oman has interpreted Rt from AIT ArrayInduction Imager Tool responses in some ofOman’s horizontal wells.

In Oman, horizontal wells make up morethan 80% of the wells drilled per year. The

Oilfield Review

12. For information on the ARC5 Array Resistivity Compen-sated tool: Bonner SD, Tabanou JR, Wu PT, Seydoux JP,Moriarty KA, Seal BK, Kwok EY and KuchenbeckerMW: “New 2-MHz Multiarray Borehole-CompensatedResistivity Tool Developed for MWD in Slim Holes,”paper SPE 30547, presented at the 70th SPE AnnualTechnical Conference and Exhibition, Dallas, Texas,USA, October 22-25, 1995.

61Winter 1995

A5A9

nReservoir structure mapped before and after the Amoco horizontal well. The structure mapped from log and 3D surface seismicdata (left) was refined with data from the horizontal well (right).

8320

8328

8336

8344

8352

8360True

ver

tical

dep

th, f

t

1800 2000 2200 2400


0.1

1000

Res

istiv

ityG

amm

a R

ay

150120

90

60

300 Modeled

Measured

AP

Ioh

m-m

Modeled

Measured

0.301.002.0020.00

Resistivity, ohm-m

8320

8328

8336

8344

8352

8360True

ver

tical

dep

th, f

t

1800 2000 2200 2400


0.1

1000

Res

istiv

ityG

amm

a R

ay150120

90

60

300

Modeled

Measured

AP

Ioh

m-m

Modeled

Measured

1.0015.0040.00

Resistivity, ohm-m

nStructural models input to the INFORM program to determine which best explained the recorded logs. Early attempts (left) to simulta-neously model the ARC5 resistivity and gamma ray curves were unsuccessful using an oversimplified formation model. A more com-plex model with minor faults and a lateral facies change (right) begins to produce modeled logs that better match the measured logs.

B11ST1 B11ST2

62 Oilfield Review

aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa a aaaaaaaaaaaaaaaaaaDepth, m Actual Sw (50%)

AF10 AF20 AF30 AF60 AF90

Sw Sensed by AF60

Actual Sw (10%)

Sw Sensed by AF60

100

200

300

400

ohm-m2.0

AITResistivity Distance from Interface, m

7.5 6.0 4.5 3 1.5 0 100Water Saturation

% 050

Shale2 ohm-m

Limestone50 ohm-m

nTranslating errorsin Rt into errors inSw. AIT resistivitieswith five depths ofinvestigation (lefttrack) diverge asthe low-resistivityshale is approach-ed (middle track).Water saturationscalculated fromthe AF60 curve areplotted for twocases (right track).The error in Swincreases as Sw increases.

1.003.004.505.506.007.009.0010.0011.0014.0017.0018.0019.0020.0022.00

Resistivity,ohm-m

4550

4575

4600

4625

1000 1500 2000 2500 3000


True

ver

tical

dep

th, f

t

500

4525

4650

nBuilding a resistivity profile from offset vertical well logs. Resistivitiesfrom a vertical well are extrapolated along a predicted dip to createthe initial model for the structure penetrated by the horizontal well.nModeled response of AIT curves

varying with distance to an adjacentbed approached at high angle.Within 3 m of a low-resistivity bed—acommon occurrence in horizontalwell trajectories—the AIT resistivitiesread significantly below actual Rtvalues. Shown are AF10 throughAF90, representing AIT 4-ft verticalresolution resistivities with 10-through 90-in. depths of investigation.

Dis

tanc

e fro

m in

terfa

ce, m

10

5

0

10

5

100101

AIT resistivities, ohm-m

AF10 AF20 AF30 AF60 AF90

0.2�

0.8 ohm-m

50 ohm-m

objective is to optimize oil recovery indeveloping and mature fields. A commontarget is the Shuaiba Limestone, in whichhorizontal wells are designed to run parallelto the reservoir top within 3 to 5 m [10 to16 ft] of the overlying Nahr Umr shale.

The Nahr Umr-Shuaiba interface, though auseful feature for steering horizontal wells,creates problems later when logs are inter-preted for Rt. For example, at the interfacebetween a 0.8-ohm-m shale and a 50-ohm-mlimestone, AIT resistivities can fall 50% belowactual Rt values when within 3 m of the inter-face (below). This suggests that in many hori-zontal wells, measured resistivities may bearlittle resemblance to true resistivities. How-ever, simulation can help arrive at true forma-tion properties: by testing several scenarios forcomparison with measured results, the forma-tion resistivity model can be found that bestexplains the real logs.13 This model can thenbe further evaluated for water saturation.

63Winter 1995

1.003.004.505.506.007.009.0010.0011.0014.0017.0018.0019.0020.0022.00

Resistivity,ohm-m

AP

IG

amm

a R

ay10080

60

40

200

Res

istiv

ity

1.0

100

4520

4540

4560

4580

4600

4620True

ver

tical

dep

th, f

t

500 1000 1500 2000 2500

Distance along the section, ft3000

1.0

100

Modeled

Measured

ModeledILM

Measured ILM

Rt

Measured ILDModeledILD

ohm

-mR

esis

tivity

ohm

-mnComparison of horizontalwell induction logs withthose simulated using theoriginal, predrilling, forma-tion model. The measuredlogs do not match the mod-eled logs, and show higherthan expected resistivities(bright green). The next itera-tion is to update the modelby increasing resistivities inthat zone.

13. Anderson BI, Barber TD and Lüling MG: “TheResponse of Induction Tools to Dipping,Anisotropic Formations,” Transactions of theSPWLA 36th Annual Logging Symposium, Paris,France, June 26-29, 1995, paper D.

Without the modeling step, errors in Rt translateinto large errors in Sw (previous page, top).

The first step in evaluating resistivity logsin a horizontal well is to build a resistivityprofile using an offset vertical well or a near-vertical pilot section of the horizontal well(previous page, right). The layers are charac-terized by their thickness and average petro-physical values and are entered into theINFORM program as the initial model. Geo-logical and structural knowledge of the fieldis used to provide the INFORM model withdip and azimuth information on the layers. Ifavailable, FMI Fullbore Formation MicroIm-ager and ARI Azimuthal Resistivity Imagerlogs are checked to confirm the bed geome-try and to identify fractures in the formations.The depth interval and relative anglesrequired for the forward modeling are deter-mined from the relative angle plot, andfinally the modeled resistivity is computed.

After the first modeling run, the simulatedlogs are compared with the actual logs(above). Iterations of the model are tested todetermine if the differences in resistivitiesare due to offsetting beds across a fault orchanges in the formation resistivity. Once amatch of the resistivities is obtained, the Rtsquare log—the resistivity model—is used todetermine water saturation, giving a betteranswer for Sw and providing guidance forreservoir management decisions.

Showing the WayIntegrated forward modeling for planning andevaluating horizontal wells is an evolvingtechnology. Presented here is a snapshotshowing the progress to date in answeringthe important questions about landing thewell, visualizing it once it is drilled, andassessing reservoir quality. As more peopletest the technique and gain experience in themethod, the scope of forward modeling inhorizontal wells will widen.

Improvements in the INFORM system areexpected to be in the form of modelingcodes for more tools, both LWD and wire-line tools. And as more measurementsbecome available while drilling, forwardmodels for their responses can be added.Plans call for 3D visualization and more 3Dtool response modeling to be able toinclude invasion and proximity effectssimultaneously. —LS

Seismic Tools for Reservoir Management - Oilfield...

Documents

Transcript of Seismic Tools for Reservoir Management - Oilfield...