ASPE49061Societyd PetroleumEnginearsAmiication ofIntemolationIn-FieldReferencingtoRemote OffshoreLocationsH.S~Williamson,SPE. BPExdoration, P.A.Gurden, SPE, Baker~uqhes INTEQ. D.J.Kerridae, BritishGeolo~ical Survev-.and G.Shiells, SPE, Baker HughesINTEQCopydght iWS, SWkty of PefmleumEngineers, inc.Thispaper waspreparedforpresenlafional the iWS SPE Annual Technical Conference andExhibition held in NewOrfeans, Louisiana, 27-SOSeptember 1998,Thispaper wasselectedforpresentationbyanSPE ProgramCommitea IollowingreviewofInformationcontained inanabrdract aubmited bythe author(a). Contents ot the paper, aspresentwl, havenot keenreviewedby theSocietyofPetroleumEngineemandaresubject tooorrecfion bythe author(s). The material, as presented, dces not necessarily reflect anyposition of theSociety01 PetroleumEnginews, ihnc.fficem, ormembers. PapempresentedatSPE meetings amsubject to publication reviewby Edtorial Committees G+the Sociely 01PetroleumEngineers. Electronicreproduction, dktribution, or storageof anypadof thispaperfor commemial purposeswithoul thewrittenconsent of theSccietyofPetroleumEngineers ispmhibfled Permission toreproduceinprint isrestrictedtoanabslracf of not more then300wordq illustrations may not be copied. The ebatracf must contain conspicuousacknowledgment of whereandby whomthepaper waspresented Wtie Librarian, SPE, P.O.Sox S3S8Ss, Richardson, TX750SS-282S, U. S.A., fax 01-972.952434S5.AbstractA method for modelling the crustrd magnetic field vector fkomtotal intensity datahasbeenused to determinethemagneticfield snapshot re@red for Interpolation In-Field Referencing(IIIR), Themethod has been validated in anumber of ways,including comparison of magnetic andgyroscopic survey datain three UK fields,IntroductionThe objective of the wellbore surveyor is simply stateddeliveryof wellbore position to the required accuracyatminimum cost to the operation. The technique of InterpolationIn-HeldReferencing(IIFR) described by Russell et a1102hasenabled survey accuracy previously only obtainable horngyroscopic systems to be achieved with standard MeasurementWhile Drilling (MWD) tools.The diftkultyof making accurate magnetic measurementsat seahas until reeentlylimitedtheoffshoreapplicationofIIFR to near-shore locations where the gradient of the erustalfield was knownto beslight. Thecurrentauthorshave beenable to surmount this diftlculty through innovative use of totalfielddat&typicallyacquired with an airborne survey. Bymaking certain simplifying assumptions, this data may be usedto deduee mathematically the direction of the crustal field,Operators are increasingly inviting directional chillingcontractors to share in project goats andto provide assurancethat their design activities and operations are technicallysound. In developing aneffective IIFR survey service withinthistype of relationship, the directionaldrillingcontractoristasked with the co-ordination of data flowbetween as many asfour organisations. A casual approach to this task willinevitably lead to confusion and disillusionment of thecustomer. For the examples described in this paper,responsibilityfor the implementationand operationof theservice was taken by an autonomous surveymanagementgroupwithinthedirectional drillingcompany. Inthis way,timelydelivery of theIIFRdata totheoperationwaskeptcentral to the targer positioning objectives of the wetl.TheGeomagnetic FieldThe geomagneticfieldvector, B, maybe specifiedbyitsdeclination D (the angle between true north and the horizontalprojection of the field vector, measured positive eastwards), itsdipangleI (theanglebetweenthehorizontal andthefieldvector, measured positive downwards), and its intensity F. Thefield can also be described in terms of its orthogonalcomponents X (north), Y(east) and Z (vertically downwards).X, Y,Z and Fare usually expressed in units of nanoteslas, nT.At any Wation near the Earths surface, B may beexpressed as the vector sum of the main field generated in theEarths core (&), the crust.alfield from beat reeks (E) and acombined disturbance field due to electrical currents flowing inthe upper atmosphere and magnetosphere, and currentsinduced in the sea and in the ground (BJ:B. accounts for approximately 98% of the field strength at theEarthssurface, anditsstrengthanddirection vary relativelyslowlywithtime. In the North Seathe rate of change istypically some tens of nT per year in F, and a few arc-minutesper year in direction. B. may be regarded as a static fiekl onlyvarying over geological timescates. Incontrast, B~ fluctuatesontimescales of minutes tohours. Duringseveremagneticstorms the intensity of B~ may vary by afew thousand nTatNorthSea latitudes, andit cantake any direction, leadingtovariations in the direction of B of several degrees.The normal practice of drilling surveyors in anatysing datacollected by magnetic survey tools has been to obtain estimatesofthegeomagnetic field strengthrmd directionat adrillinglocation from a global model of the geomagnetic field sueb asthe British Geological SurveyGlobal Geomagnetic Model(BGGM). However,global models are designedtoprovide3872 H.S.WIIJJAMSON, P. A. GURDEN, D, J. KERRIDGE, GSHIELLS SPE 49(X1estimatesof B. only. The contributions of BcandBdareeffectivelyerrors when B. alone is taken as an estimate of thelocal field, B.Ideally, magnetometers would be situated at a drillingsiteto measure the local strength and direction of Bcontinuouslyand with high accuracy. This has rarely proved to be a rerdisticproposition. The techniqueof IIFRhas beendevelopedtoprovide a practical approximation to the ideal. IIFR combinesa local one-off absolute measurement of the geomagnetic fieldwithcontinuousmeasurements madeat oneormoreremotemagneticobservatories toestimatethelocal vrdues ofB, Ineffect a virtualmagnetic observatory is run at the site, takingadvantage of the stringent quatity control procedures applied atthe remoteobservatories,Russell ef d and Shiells ef alzdescrhxl an application of IIFR where, txxause of theproximity of the drilling site to the shore, heat absoluteobservations of declination, dip angle and total intensity couldbe made using standard land survey methods.Magnetic FieldModelling withAeromagnetic DataAeromagneticsurveyshave been carried out in oil explorationareas for manyyears toprovide data to assist in defining sub-surface geology. Theaeromagnetic datasets typicatly consistofclosely-spaced spotmeasurementsof total intensitymadeusing an absolute instrument. Conventionally, the datacollected arecorrected for time-varying fields by reference tomeasnrementa madebya magnetometeroperatedat a basestation. An estimate of the mainfield isthenremoved, oftenusing a global geomagnetic field model, andthe residuals arepresented in the form of crustal anomaly maps,A number ofanthors45,have considered the problem of using total intensitydata sets to estimate the vector components of the crustal field,but this work has received little attention to date because of thelimited value perceived for geological interpretations, There is,however, clearly considerable value in using these methods toextend the application of IIFRoffshore.The Method: Assumptions and Limitations. Themathematical formulation of the method is given in theAppendix, A number of assumptions are made,1.2.3.The crustal field vector, intheregionwhere the dataarecollected,maybeexpressedas the gradient of a scalarpotential, Asaeromagneticobservations aremadeintheair, where there areno electrical currentsflowing andnomagnetised materiat this assumption is vatid.Thescalar potentialis harmonic, i.e. it satisfies Laplacesequation. This follows ftomthefurther assumptionthatmagnetic monopoles do not exist.The anomaly in field intensity is, to a god approximation,the component of the crustat field vector in the direction ofthe local main field vector. This w be shown to be a goodassumption provided thestrengthoftheanomalyfieldismuch less than that of the main field. (If the field intensityis about 50000 nT and the anomaly is 500 nT the errorin4.5.making this approximation is about 2,5 nT.)Theanomalyinfieldintensityisharmonic. This istrueprovided the direction of the main field is constant over thearea of the aeromagnetic survey.The data points are collectedon a horizontal surface.Aeromagnetic surveys offshore are flown at a well-definedaltitude, approximating to a horizontal surface.The final two assumptions limit the total areafor which datamay be analysed, but in practical terms analysis of data over anareaof about 50 km by 50 kmis generally adequate andthedeviations from the exact requirements of these assumptions isthen small. An implicit assumption is that an accuraterepresentationof thegeomagnetic mainfield isavailableforthe survey area. In estimating the crustat field vectorcomponents no assumptions are made about the geometriczdorgeophysical propertiesofthesources ofthecrustai anomalyfieldsuchastheir dimensions, orientations, ordirectionsofmagnetisation. However,in practice the magnetic anomalydata should be examined carefully and local geologicalinformationused toas.wss thelikelihood ofthepresenceofsignificant shallow magneticsources, Insome circumstances,where such sources exisf, extrapolating the crustal componentvaluescomputedat the aircraft altitude to the subsurfaceenvironment may lead to significant errors.Validation of the Technique. The successof the aeromagneticdata transformation technique has been tested using bothsyntheticandreal data. Fig. 1illustrates the results of asynthetic test. The left-hand diagrams show calculatedrmomrdyfields in F, X, 1and Z due to a dipole source assumedto be magnetised in a direction representative of the North Searegion. ThecentrediagramsshowtheX, Y andZanomalyfields computed from the F anomrdy data using thetransformationmethod. The right-hand diagrams showtheerrorsinthetransformed values. Theanomalyvalues areofthe order of 100 nT, the errors are only a few nT.The tests with reddata were made for two areas on land inthe UKover whichaeromagneticsurveys hadbeenflown67.The transformation technique was applied to the aeromagneticdataand theresults comprwed withvectorcomponent datameasured atanetwork of sites ontheground. Inoneoftheareas (approximately 40 km by 50 km) the anomaly values intotat intensity were relatively small, with a range of about150nT. Ground observations were made at19 sites in the area andthe root mean square differences between the transformed andobserved valueswere0,05 indeclinationand0.02 indipangle. The second area provided a more challenging test. Thearea(approximately 50 km by 50 km)contained ananomalywith a range of about 750 nT - greater thanobserved in mostNorthSeaareas. Inaddition, therequirement for dataonahorizontal surfacewasnot well satisfiedbecausethe flightlines followed the undulating terrain, The transformationtechnique produced estimates of theanomalies indeclinationand dip angle with ranges over the area of about 1.4 and 0.5388SPE 49001 APPMCAT)ON OF lNTERPOLATtON IN-FIELD REFERENCING TOREMOTE OFFSHORE LOCATIONS 3respectively. Thercmt meansquaredifferencestwtweenthetransformed andobserved values at36 sites intheareawere0.16 in declination and 0,08 in dip angle. Fig. 2ashows thetotal intensityanomaliesover the area and theobservationpoints on the ground. Figs. 2b and 2Cboth showthedeclinationanomaliesoverthearea. Fig. 2bwascalculatedusing the transformation method, Fig. 2Cwas calculated fromthe measurements at the observation poin[s. The closesimilarity between these two figures is a striking confirmationof thevalidity of thetransformationmethod. Itisalso worthnotingthecharacteristicpatternofthedeclinationanomaly.The field is Ixmt in towards the (positive) total field anomaly,generatingapositive declinationanomaly to thewest, andanegative anomaly to the east.Data requirements for offshore mapping. In medemaeromagnetic surveyscarried out for exploration purposesoffshore, dataaretypically collected at anattitudeofabout80 m with aflight-line separation of 500 m. Caesium vapourmagnetometers with an accuracyof atwut 0.1 nT in fieldintensity are routinely used, typically sampled at about 10 Hz.Similarmagnetometersareoperatedat abasestation. Goodquatitypositional andtiminginformationisavailableusingGPS. Compensation methods areused toeliminatefromthedatathe magnetic interference caused by the aircraft, Forthepurposesofgatheringdataforuseinestimatingthecrustalcontribution to the local vector magnetic field for use in IIFRaflight-lineseparation of2kmwill besuftlcientunlessthearea is unusuatlymagneticallycomplex, If the main flightlinesarenorth-souththenaset of east-west linesshouldbeflown ata separation of about 6 kmto provide data atcross-over points. The accuracies in magnetic field, navigationat andtiming data whichare routinelyachievedin aeromagneticsurveys aremorethanadequate forIIFR. Toallowaccuratedata processing it is recommended that the base stationmagnetometer is operated for atleast 24 hoursatthetime ofthe survey, not just while the survey is being flown.Survey DataProcessingThere areatleast five options of applyingbeat field modelsandobservatory datato magnetic wellbore surveys which cank deseribed as IIFR. The simpler methods give the most rapidturn-around time and require the least complex techniealassurance. The more advanced methods offer the potentiatofgreater accuracy, but atthe expense of greater computationaland logistical complexity. Typieatly, an operation will need toselect two correction methods: one for instantaneousapplication at the rig site to enable drilling ahead, and one forproduction of definitive wellboreposition data. Only where themagnetic observatory data is available atthe rigsite, togetherWithappropriate software and expertise, ean definitive data begeneratedintruereal time. The mainoptions for azimuthcorrection are as follows.Option l-Correction for crustrd field declination. Themagneticdeclination at thedrill site asdetermined fromtheaeromagnetic data is used in preference to the value predictedby amainfield modeltocorrectmagneticsurvey azimuths.Sincethisvalue representsasnap-shotintime, itsaccuracywill quickly degrade unless it is corrected for secular (ie. long-term) variation inthefield, Thismay be done by calculatingthedeclinationanomaly(ie. thedifference betweenthetruevalue and the main field value) at the time of the observations,and adding this valuetoall subsequent main fieldmodelpredictions.Option 2-Correction for instantaneous fielddeclination. Abetter estimate for declinationmay be obtained by addinginthe short-termvariations detectedat one or more nearbyobservatories. This will largely eliminate the effects ofmagnetic disturbances, Where observatory data are notavailable in real-time, this correction will typicallybe made byshore-based staff, delayingdeliveryof the final data tothedrilling operation by a day or two.Option 3-Correction for crustal field declination anddrillstring magnetic interference. The limitations ofestablished magnetic interference correction algorithms toerrors in the assumed values of magnetic field strength and dipangle are well known89.Hitherto, this has restrictedtheirapplication offshore. Where values for magnetic field strengthanddipanglemoreaccuratethanthose provided by amainfield model areavailable, these limitations aremitigated, butnot eliminated. Thepractical effect isthat thesecorrectionsmay be applied with confidence to surveys over a greater rangeof hole orientations, with only those withinafew degrees ofhorizontal east-westbeingexcluded. Estimates of magneticfield strength and dip angle at the drill site should be made asin Option 1 and updated for each well drilled.Option 4-Correction for instantaneous field declinationand drillstring magnetic interference. Acombination ofOptions2and3, Estimatesof instantaneousmagneticfieldstrengthanddipangleareusedasinputsinanestablishedinterference correction rdgorithm, The resulting magneticazimuths are corrected for instantaneous declination.OptionS-Correction fortoolsensor errors, fieldvariationandinterference using near real-time data. Corrections formagnetic survey azimuthswhich use multiple survey stationsto track tool sensor perfcmnanee as well as externalinterference are now entering widespread uselOl 1. Theireffectiveness isgreatlyenhancedbytheavailabilityofreal-time estimates of the magnetic field, Indeed, the increased useof IIFR has encouraged their development. Generallyspeaking, these methods have yet to be successfullyautomated,and their application remains the preserve of oftke-basedexperts within the directional drilling companies.For the operational examples included in this paper,correction Options 3 and 5 were used for real-time rig site dataand definitive positional data respectively.3894 H. S. WILLIAMSON, P. A. GURDEN, D. J. KERRIDGE, G. SHIELLS SPE 49061Operational Sequence ofanOffshore IIFRServiceThe implementation of the technique can be divided intotwoparts :-G Acquisition and validation of the crustal anomaly data,. Useof IIFRin dailydrilling operations. This includesapplying the improveddeclinationdata, quality controlmonitoring, decidingwhich,if any,correction techniquesare to be applied, and delivering the data.Acquisition of Aeromagnetic Data. Prior tocommissioningthe survey, it is important to obtain as much existing magneticdata as possible surrounding the field, This will help determinethe character of the magnetic gradients andcrustal anomaliesin the area. The scaleand locationof these features willdetermine the extent, sample frequencyand potential benefit ofany new aeromagnetic data. For an isolated field in the NorthSea covering a10 km by 10 km area, an aeromagnetic surveyof an area 50 kmby 50 km would typically be required.The potential for errors of hundreds of metres in thegeographical locationof magneticanomalies shownonoldairborne surveysmust be recognised. Use of such data for IIFRwouldrequire significant relaxation of surveyperformancemodels to allow for these errors.Wh.h atypical flight height of about 80 metres above sealevel for aeromagnetic data acquisition, it is important that allonshore and offshore personnel in the area are informed whenand where the survey is to be undertaken.Computation of magnetic field values. Once the dataacquisition is complete, a new digital magnetic field map maybe calculated (see Appendix). This will showanomaliesrelative to a particular main field model. The exact co-ordinatesofeachstructureor wellheadwithintheareaareinterpolated on the new magnetic anomaly mapto determinethecrustal offset relativetothismodel. Shouldanupdatedmain fieldmodel be availableat the time of the drillingoperations, newoffsets for each drilling facilitymust becalculated. Computingthelocal magneticfieldas anoffsetfroma main field model allows long-termsecular fieldvariations to be properly allowed for in the IfFR correction.For simplicity, thesamecomputed magneticfieldvalues(those for the wellhead) are used throughout the length of eachwell. The aeromagnetic map must k reviewed to ensure thatthe variation of the crustal field along each wellbore is not inexcess of the values which have been incorporated into the toolperformance model.Validation of magnetic field values. Sincethey arenot theresult of direct observation andrely on thevalidity of severalassumptions, the magnetic field values obtained fromtheanomalymapmust bevalidatedbyindependent techniquesprior to use. This is typically done by comparing IIFRcorrectedMWDsurveyswithhigh accuracyinertial gradedownholegyrosurveys.There are three variations on thismethod, all of which have been used to good effeccGGGRun gyros in the fiist one or two wells drilled after themagnetic mapping exerciseIf gyro surveys have been taken in previous wells, and ifsuitable MWD data are available, re-process the data andcompare results retrospectively.CompareIIFRMWDresultswith results in-holereferencedtoa gy&copictoo112. This validationmethod was used in BPsWytchFarmfield.Application of IIFRdata. At a basic level, the application ofthe IIFR technique is simple, The rig drills ahead correcting itssurveys for declination only. These are replaced with a batch ofIIFR corrected surveys as and when required (Fig. 3).Knowledge of the crustal anomaly atthe drillsite benefitsthe real-time survey dataintwo ways. First, amore accuratedeclination value may be used for correcting data at the rig. Toavoidconfusion,a singlevalueisnormallyadoptedfor thewhole well, but for some extended reach or deep wells use ofmore than one value may bejustified, Second, improvedestimates of total field strength and dip angle allow better real-time magnetic data quality control to be applied on the rig.All MWD surveys acquired on the rig are compiled into afile and sent to the processing centre, typicallyon a daily basis,Each survey must be time stamped using a time basesynchronous with the magnetic observatory data. Exceptduringmagneticdisturbances,particularlyat highlatitudes,synchronisation accurate to +/- 1 minute is suftlcient, and mayke achieved by the MWD/ Survey engineers using a broadcastradiosignal. For quality control purposes itis important thatthe six-axis accelerometer and magnetometer data be suppliedfor full analysis using triaxial bias and scale factor errorevaluation techniques.Frequency of Data Delivery. Priorto commencing theIIFRservice, it is important to determine the appropriate frequencyof data delivery to the rig, Thishelps establish the techniquewithin theoperational sequenmand enablesfit-for-purposecommunicationinfrastructureto beinstalledandtested.Thedecision depends on operational requirements and is inevitablya compromisebetween:GGGkeepingthe rigappraised of the best estimate of thewellposition at all timesavoiding the proliferation of datasets with differentprocessing appliedminimizing operational cost390SPE 4=> APPL}CAT\ON OF INTERPOLATION IN-FIELD REFERENCING TOREMOTE OFFSHORE LOCATIONS5During times of relative magnetic calm, the field variationscmbe adequately monitored on a daily basis, and delivery oftheupdated wellbore position may be left to theendof eachhole section. In periwis of high magnetic disturbance, or whenknowledge of wellboreposition is criticrd, near real-timeupdates will be required to maintain surveyperformancewithinthepre-determinedbounds.IftheprimaryreasonforapplyingIIFRis for futurewell collisionavoidance, or forimproved reservoir mapping, prmessing need not beundertaken until drillingis complete. This of course removestheadvantageof improvedconfidencein the while-drillingwell position.Experience has shown that monitoring magnetic variationsdaily andreactingto problems asthey cccurisasatisfactorycompromise. This allows update to the wellbore position dailyor at casing point dependent on circumstances,Service Management andCommunications. Acquisition anddelivery of IIFR datainvolves uptofour organisations. Theoperator, thedirectionaldrillingcompany, theMWDsurveycompany andthe providers of themagneticobservatory dataall play an important role. Eftlcient management of the serviceis a non-triviat exercise in communication. In particular, rapidand robust verbal and digital data links are critical to success.Transmissionof dataeither byhand-entryor faxhas kenshowninpractice tobe inadequateexcept as a temporarymeasure.The applications of IIFRdiscussed in this paper areassigned to a survey management service operatingindependently within a directional drilling company. Thesurvey management service is given full responsibility for theset up, monitoring and control of the survey service. They alsoprovide an operational position to whom all questionsregarding theapplicationandimplicationsof IIFRmaybeaddressed.Results fromUsingIIFROffshoreBPis currentlyapplyingIIFRinthreeoffshorefieldsoverwhich the magnetic field has been mapped using aeromagneticdata.. The Foinaven and Schiehallion fields are neighbors intheWest of Shetlandprovince. Magneticvariationdata forthese fields is supplied by the BGS observatory at Lerwick intheShetlandislands, 165 kmtotheeast. AndrewisintheCentral North Sea. Its IIFR service uses magnetic datainterpolated between Lerwick, 275 km to the northwest, andthe BGSobservatory at Eskdalemuir, 415kmtothe southwest. Fig. 4 shows locations of the fields and observatories.Magnetic Anomaly Maps. The aeromagneticdata for theWest of Shetlandfields was partof alarger datasetacquiredfor exploration purposes in 1993and 1994, The cost ofreprocessingwasthereforetheonlyexpensetothedrillingoperation. Processing the data(see Appendix) revealed a250nT total field anomaly centred some 5 km to the south of thefields. Angular anomalies are small over both fields. Thecomputed values for declination and dip anomaly at theFoinaven B and Schiehallion Central drillingsites are atllessthan O.1O.The anomalies surrounding the Andrew field are illustratedin Figs. 5a, 5band5c, The data from which these maps werecomputedwereacquiredin 1997. Theaeromagneticsurveycoverage was extendedtothewest (not shown) to covertheAlba and Britanniafields, operated by Chevron. This reducedthe capital cost of data acquisition to both Operators,Fig. 5a shows the totat intensity anomalies over an area 40km by 52 km containing the Andrewplatform. A large feature,of intensityjust over 400nT can be seen centred some 23 km tothe south east of the platform. The declination anomaly (Fig.5b) shows the characteristic positive-negativepattern describedabove. The dip angle anomaty(Fig, 5c) is less obviouslyrelated to the total field anomaly, but it is worth noting that thecentre of the main feature, where the dip angle anomalyexceeds0.2, does not coincide with the centre of the totat fieldanomaty. It happens that the direction (but not the strength) ofthe magnetic field at the Andrew platform is very close to thefield predicted by the BGGMmain field model. The magneticeffect of the platform itself, although clearly visible in the rawaeromagnetic data, was removed numerically at an early stagein the processing.Comparison of IIFR vs. Gyro and Inertial Surveys. Allthree fieldsinthis studyhave twowellsin whicha highaccuracy gyroscopicor inertial tool and IIFR-processed MWDdata are simultaneously available. Fig. 6shows the results ofthedatacomparisons for all sixwells. Thewhite circleandbarsshow themeanandstandard deviation of thedifferencebetween the gyro/inertiat survey azimuth and the MWDazimuthcorrected onlyformainfield declinationThemainfield model used was the current BGGM. The black square andbars compare the gyrdinertial survey azimuthwith the MWDazimuths after full IIFRcorrection (Gption 5),For all sixwells, applicationof IIFRproducesa meandifference ktween MWD andthe gydinertial survey oflessthan 0.3. For five of the six wells, this represents a significantimprovement, There is little effect on the standard deviation ofthe difference. Thisis probably dueto the(random)residualerrors leftafter applicationof themagneticdataprmessingalgorithm. At these accuracy levels the gyrdinertial survey canno longer he regarded as the truth. They must be treated, likethe IIFR data, as an approximation to it.There is no evidence that the IIFR technique is correcting asystematic azimuthdifference common to all wells inanyofthe three tields. This is unsurprising given the size of the localdeclination anomalies in comparison to other sources of error.SomeTechnical Issues Affecting IIFRDaily andMagnetic StormVariations. There aretwo mainsources of magnetic field which vary on timescales of minutesto hours; the regular daily variation and magnetic stormvariations. The daily variation has a fundamental peria.i of 24hours and has a very similar structure over the whole region of3916 H. S. WILLIAMSON, P. A. GURDEN, D. J, KERRIDGE, Q. SHIELLS SPE 49061the British Isles. Its typical range,which varies with latitude,theseasons andwith the1l-year solar cycle, is afew tens ofnTintotal intensity, approximately 0.2 indeclinationandabaut 0.03 indipangle. Thephaseof thedailyvariationdepends on local time.Magnetic storm variations, on the other hand, areessentially simultaneous over large regions and their amplitudedepends on geomagnetic latitude. In the area of the North Sea,the amplitude of variations maybeof the order of a fewthousandnanoteslasinfieldintensityandafew degreesindirection during severe magnetic storms.Tnrbitt and Chtrk*3 compareddata for 1991fromthepermanent magnetic observatories at Lerwick, Dombas(IWrway)and Brorfelde (Denmark). They useddata fromtick and Brorfeldeto studylongitudinal effects. Theyfoundthat onquiet daysthecross-comelations betweenthedata sets peaked at lags consistent with the difference in locattime between thetwo sites (about 51 minutes).Ondisturbeddays the cross-correlationspeaked at zero lag, showing that thedisturbance fields affected both sites simultaneously andcaused similar field perturbations.Change ofMain FieldModel. Thetransformation techniquedescribed inthispaperestimates thelocal magneticfield bythe combination of a main field model and aeromagnetic data.The crustal fieldvalueswhichare derivedare apparentvalues consistent with the main field model used in theanalysis. Whena newmain fieldmodel is adopted, thesecrustat field values are adjusted to be consistent with thenewmodel, so there is no jump in the local field estimates, Strictly,thewhole crustal field analysis should be repeatedusingthenew mainfield model, but thiswill have very littleeffect onthe results, Any discrepancy arisingfollowing model revisionwill be a result of differences in the secular variation models.This will be a small effect over a few years.IIFRat High Latitude. Duringperiodsofsevere magneticdisturbance there are intense electrical currents flowing in theauroral regions, at a height of about 100 km, and the strengthand the pattern of the currents can change rapidly. As a result,the ability to use magnetic observatorydata to estimatevariationsat remotelocations, possible overa few hundredkilometres in the UKregion, will decrease to tens ofkilornetres at high-latitude locations close to the auroralelectrojet currents. Becauseof the increased rate of fieldvariation, efforts should be made to improve thesynchronisation between theMWD system andthemagneticobservatory to better than 1 minute.Induced Currents. Short termgeomagneticfieldvariationsmay be viewedas electromagnetic waves which propagate intotheEarthgeneratinginducedcurrentsastheyareabsorbed.The characteristic length scale for absorption, the skin depth,depends on the electrical conductivity of thesolid earth(andthesea) andthefrequency of thevariations. Inshallowseawater only short period variations will induce significant392currents (theskindepthinseawater for variations withaperiod of 10 seconds is about 1 km). This means that there willhe effects duringsevere magnetic disturbances, but in generalthe uncertainties due to fluctuations in the source currents willbe more significant than the induced currents, This alsodemonstrates that there is, on most days, very littlescreeningoffieldvariationsbythesea. Themost significant inducedcurrent effects are expected to k in coastal areas where thereisacontrast inconductivity betweenthelandandthesea.Experience at Liverpool Bay12hasshown that theabilitytointerpolate accurately falls slightly during disturbances.Impact of IIFRonSurveying UncertaintyFuII apprwiationof the impact of HFRon survey performancedepends on understandingthedifference between uncertaintyreduction and error reduction. Just as highintensity magneticstorms are relatively infrequent events, so high intensityCrustal anomalies are relatively sparsely scatteredgeographically. As a consequence, IIFR corrected survey datawill frequentlyshowonlyslight differencesfromthe samedataset processedconventional y. Anatural reaction is toquestionthevalue ofthetechnique, butconsideration oftheassociated survey uncertainties shows this reaction to bemistaken. Since both high-intensity storms and crustalanomaliesareessentially randomandunpredictableintheiroccurrence, ampleallowancemust bemade for themina-priori estimates of survey uncertainty for services which do notattempt to eliminate t.km. This justifies the apparentlypessimisticvaluesfor magnetic fielduncertaintywhicharecurrently in use. A technique which can be shown to eliminatemost of the effects of magneticstorms andcrustalanomaliescan be assigned a much tighter uncertainty model, withconsequent benefits for overatl wellbore surveying costs.Magnetic field predictions based onmain field models. ABritish Geological Surveyreport14 commissionedby BakerHughes INTEQ estimated the magnitude of the various effectscausing the instantaneous magnetic field at a randomly chosentime andplaceto differfromthepredictionofamainfieldmodel. Although not a comprehensive study, the results can beusedtoderive robustestimates ofthetotal uncertaintyinatypical main field model prediction in the North Sea regionDeclination 0.5 perjieldDip angle 0,2 perjleldTotal Field Strength 130 nTperjieldThe characterization of these uncertainties, quotedat onestandard deviation, as per j7eld indicates that whenimplemented as error terms in a tool performance model, theyshould betreated asactingsystematically throughout afield.This is becausethe bulk of the uncertainty is attributable to thecrustal field, thespatial variations of whicharerather largerthanmost oil andgas fields. Whileopentochallenge, theauthors consider these values to be the best currently available.SPE 49061 APPLICATION OF INTERPOLATION IN-FIELD REFERENCING TOREMOTE OFFSHORE LOCATIONS 7Magnetic field predictions based onIIFR. Likely errorsintheestimates of theinstantaneousmagneticfield at thedrillsite may be split into errors in the combined main and crustatfields, B., and errors in the short-term variations, B4.Main Field plus Crustul Fiehi. The transformationtechniqueprovides an estimateof the combinedmain andcmstal tields, and any errors apply to the sumof thesecomponents. The UKland studies discussed above usedaeromagnetic data acquired in1962 and1956 respectively. Itis likely thatsimilarstudies with aeromagnetic dataacquiredwith modem equipment would show better agreement betweenthemeasuredandcalculatedfields. Theresultsof the landstudiesindicatethati particulmlyinDandI, errors in thetransformation method increase with the size of the anomaly ina roughlylinear fhshion. Usingthis assumption, individualuncertainties may be calculated for particular locations basedon the size of local anomalies. Given our knowledge of typicalanomalyintensityin the NorthSearegion, the 1standarddeviationuncertaintiesinthecomponents of themainfieldplus cmstal field vector at a typical drilling locationareestimatedas0.08inD, 0.025in1 and40uTinF. Thesevrdues do not include the possible effects of short wavelengthdown-hole anomalies which would not h detected byaeromagnetic surveys.Short-Term Variations. Anumber of studieshavebeencarried out using data Born the three UK magneticobservatoriesand observatories in Norwayand Denmark to testthe accuracy of the interpolation technique in bothmagnetically quiet and disturbed conditions. Barraclough andMacmiIlan*5examinedthedegreetowhichfieldvariationsobserved at Eskdalemuir Observatory (55,3N) could bereproduced using data from Lerwick Observatory (60.1N) andHartland Observatory (51.0N) over a selection of 83magnetically disturbed days during the three-year period1987-89. Setting thresholds of 5 arc-minutes (0.083) in declinationand dip angle and 50 nT in field intensity they foundagreements between theobserved andinterpolatedvatues for96.6%, 94.3% and 93.9% of the time, respectively. Given thatEskdatemuir ismorethan450kmfrombothLerwickandHartland, this result gave confidence in the ability tointerpolate between observatories even on disturlxxl days, Theintermittent, spikynatureof short-termmagnetic variationsmakes these values conservative as 2 standard deviationestimates of the errors in B~,These results are borne out by the success of theinterpolation technique for Liverpool Bayi2.Over a period of 4months the standarddeviationof differencesbetweendatahorn au on-site monitor and data interpolated fromEskdalemuirandHartlandObservatories was less thanO.O1in Eothdeclination and dip angle, and less than10 nT in fieldintensity. These resultsshow errorsless thanthose found byBarraclough andMacmillan15.Thisis beeause Liverpool Bayis at a comparatively low latitude and the distance over whichthe data were interpolated was less than in their study.Error Model Parameters. From the above data, bestestimates for the 1 standard deviation magnetic fielduncertainty when using IIFR in the North Sea region are:Declination 0.08 perjield 0.04 per surveyDip angle 0.025 perjield 0.04 per surveyField Strength 40 nT perjeld 25nTper surveyThe characterizationof the uncertaintiesdue toshort termvariations in the field as per surveyindicates that error modelterms arising horn themare to be treated as systematictx?tweenstations in the same survey, but not between differentsurveysin the samewell. This reflects the distinctlynon-random variations characteristic of some disturbance events, InthecaseofMWD holesection surveys, whicharenormallyacquiredover a periodof severat days, this is a distinctlyconservative assumption. Withelectronicmukishot surveys,which are acquired over a few hours, the assumption will stillbe conservative, but far less so.The above vatues are thought to be readily attainable for anHFR serviceset upandmanagedina waysimilar tothatdescribedinthis paper. Significant change ingeographicallocation, magnetic environment, or data acquisition andprocessing would necessitatea review.Impact onTotal MWDError Budget. In most applications,total azimuth uncertainty of an MWD measurement isdominated by two sources of error: declination uncertainty anddrillstring magnetic interference. The figuresabovesuggestthat use of HFR will typically reduce the impact of declinationuncertainty by a factor of about five. If no attempt is made tocorrect for drillstring magnetic interference, IIFR will have noeffect on this error source, However, the improved knowledgeof the total fieldstrengthand dip angle affordedbyIIFRsignificantlyreduces the errors inherent in the correctionalgorithms currently in use. Used with care,these correctionsor their successors will replacetheerror duetouncorrectedinterference with a residual error several times smaller.Both declinationuncertainty andtheimpactof cirilIstringinterference arestrongly dependentongeographical locationaud/or hole direction. This precludes generalization about theoverall reduction in azimuth uncertainty available through useof IIFR. Nevertheless, two statements can be made withreasonable confidence. Careful use of IIFR:G enables substantial reductions in azimuth uncertaintycompared with conventional MWD surveying. Theevidence presented in this papersuggests that the averageazimuth error over a well can generally be reduced to lessthan 0.3.G greatly reduces the dependency of azimuthuncertainty onexternal factors, such as the presence of magneticanomalies. Thisbuildsconfidenceintheuseof thetoolperformance model for quantitative risk-based decisions,Impact of IIFRonWell PositioningThe development of HFRfromits fwst ground-br~lngapplication into anestablished survey service hasenabled the3938 H. S. WILLIAMSON, P. A. GURDEN, D. J. KERRIDGE, G. SHIELLS SPE 49081benefits forecast in a previous paperl to lx substantiallyreatiied. These can be summarised under four headings.Survey Program Optimisation. Gyroscopic surveys aregenemlly no longer required in the deeper hole sections. Theyare retained at shallowdepths where external magneticinterference may be expected. Electronic multishot surveys areno longer required for end-of-section quality control, They addlittle extra assurance if the MWD tool performance hasbeenanalysed using one of the latest magnetic processingalgorithms.Near Real-Tbne Quality Control. The new magneticprocessing algorithms highlight the deterioration in MWD toolperformance characteristic of imminent failure. Changingoutsuch a todat the next convenient opportunity may save a trip.Near ReaI-Thne Definitive Survey. Theeffect of IIFR on thedirectional drillingoperationhasbeenlikenedtogettingagyrosurveyeveryday(an unusuatly reliableone). Missedtargets andcorrection runscaused by misleadingsurveysareelimimted.Interpretation of Well Position. Thereductioninpositionuncertaintyafforded byIIFRreducesthenumber of seismiclines against which a match with the well logs may need to besought. The increased confidence in absolute well position alsoreducesthetemptationto runextrasurveys whenthewellsposition relative to the existing reservoir model is in doubt.Future DevelopmentsAs nse of IIFR becomes widespread, itwill beincumbent onthe directional drilling industry to develop operationsprocedures, dataformats andperformance models which are,asfar aspossible,standardisedacrosscompanies. This willimprove theOperatorsconfidence inthetechnique, andthevalue they derive from it.Integral to the type of IIFR service described here is the useof the new generation of magnetic survey processingalgorithms. These methods are not claimed to be 100%accurate, nor will they produce good surveys fromfimdamentally poor quality data, but they are highly indicativeoferroneoussurveys. The sizeof the residurderrors tobeexpected after prcxessing requires deeper investigation beforethe somewhat conservative performance estimates currently inuse can be tightened. This will be the subject of furtherpublication.Conclusions1. Mathematical analysis of aeromagnetic total field dataprovides anaccurate andcost-effective means of mappingthecrnstrd fieldvector whichisparticularlysuitableforoffshore locations.2. IIFRcombined with recently develo@ magnetic dataprocessing methcds can deliver MWDsurveys to anaccuracy normally associated only with high-accuracysurvey systems,3. Provision of an effective HFR service requires a significanteffort of co-ordination and surveying expertise.Communication systems, data processing methods and datadelivery schedules must be tailored to the individualoperation.NomenclatureA(u,v) Fourier transform of the magnetic field potentialb unit vector in direction of magnetic fieldB magnetic field vectorBOOMBritish Geological Survey Globat Geomagnetic ModelC(u,v) Fourier transformof the total intensity anomalyvaluesD magnetic declinationF total magnetic field strengthAF total intensity anomalyI magnetic dip angleIIFR interpolation in-field referencingMWD measurement while drillingu wave number in x-directionv wave number in y-directionv magnetic field potentialfunction relating A(u,v) and C(n,v):,x north directionY,y east directionz, z down directionSubscriptso excludingexternal field sourcesdue to crustat anomaly: due to external field sourcesm due to main fieldAcknowledgmentsTheauthors wishtothank BPExploration, Baker HughesINTEQand the Director of the British Geological Survey(NERC) for permission to publish this paper.A number of the studies that established the basis of IIFRweresupportedbySperry-SunDrilling Services, whoalsomanage the IIFR service for BP Exploration on the Foinavenfield.References1, Russell, J,P., Shiells, G. and Kerridge, D.J.: Reduction ofWell-Bore Positional Uncertainty Through Application of aNew Geomagnetic In-Field Referencing Technique, paperSPE30452 presented at the1995 SPEAnnual TechnicalConference and Exhibition, Dallas, TX, Oct. 22-25.2. Shiells, G., D. J. Kerridge and J. P. Russell, 1996.Reduction of Wellbore Positionat Lhcertaintywith a newGeomagnetic Referencing Technique.IPT (March1996)~43-2@,3. Rixse, M.R. and Thorogood, J.L.: Case Study - Building aSystemin a Service Company to Assure Technical Integrityand Institutionalize Organizational Learning, paper394SPE 49061 APPLICATION OF INTERPOLATION IN-FIELD REFERENCING TOREMOTE OFFSHORE LOCATIONS 94.5.6.7.8.9.IADC/SPE 35067 presented at the 1996 IADC/SPEDrilling Conference, NewOrleans, LA, Mar. 12-16.LeMout21,J.-L., Lelev~ a&omagn&iquedelaFrance.Calcul des composantes du champ ?Ipartirdes mesures delintensit&Annales deGkophysique, 26,2,229-258.,1970Lourenco, J.S., and Morrison, H,F., Vector magneticanomalies derived from measurements of a singlecomponent of the field Geophysics, 38, 359-369.,1973S., and Barraclough, D.R., Areviewofmethods for deriving allelements of thecrustatmagneticfield from total intensity observations, British GeologicalSurvey TechnicaJReport WM/90/18C, 1990.Macmillan, S., and Barraclough, D.R., Transformations oftotal intensity aeromagnetic data in Norfolk, BritishGeological SurveyTechnical Report WM/90/25C, 1990.Nljden Twilhaar, G.D,, Accurate Magnetic Surveying in asingle NMDC: Analysis of magnetic interferencecorrection, paper SPE 16530 presented at Offshore Europe87, Aberdeen, UK, Sep. 8-11.Brooks, A.G,,andWilson, H., AnImproved Met.hl forComputing Wellbore Position Unc&aint y and itsApplication to Collision and Target Intersection ProbabilityAnalysisfl paper SPE36863presentedat the 1996 SPEEuropean PetroleumConference, Mihm, Itaty, Oct. 22-24.10. McElhinney, G,, Sognnes R., and Smith, R,, CaseHistories Demonstrate a New Method for Well AvoidanceandRelief Well Drilling, paper 37667presentedat the1997SPE/L4DCDrilling Conference, Amsterdam, TheNetherlands, Mar. 4-6.11.Brooks, A.G., Gurden, P.A., and Noy, K,, PracticatApplication of a Multiple-SurveyMagnetic CorrectionAlgorithm, paperSPE 49060 presented at the1998 SPEAnnual Technical Conference and Exhibition, NewOrleans, LA, Sep. 27-30.12.Thorogood, J.L., andKnott, D,R.,Surveying TechniquesWithaSolid-StateMagneticMultishot Device,SPEDESept. 1990, p. 209-214.13.Turbitt, C, W. andT, D. G. Clark. The use of Lerwickvariometer measurements to estimate magneticdisturbances over the North Sea. British GeologicalSurvey Technical Report WM/94/21C, 1994.14.Macmillan, S., Firth, M.D., Clarke, E., Clark, T.D,G. andBarraclough, D.R., Error estimates for geomagnetic fieldvalues computedfromthe BGGM, British GeologicalSurvey Technical report WM193t28C, 1993.15.Barraclough, D,R. and S. Macmillan. Correctionofmagneticsurvey observationsfortheeffects ofmagneticdisturbance. British Geological Survey Technical ReportWM/90/22C, 1990.AppendixTransformation Methodof CalculatingCrustal FieldVector fromTotal Intensity DataSupposethat B. is the geomagneticfieldvectorat a pointwhere an aeromagnetic spot measurement of field intensity ismade, andassume, tosimplify thealgebra, that thereisno395contribution from external field sources. If B~is the main fieldand B. is the crustal field vector at the sample point, thenBO=BW+BCand the measurement made by the magnetometer is[Bo[,The total intensity anomaly AF is defined asAF=lBO[-lB~lwhere IB~ I maybeestimatedusingaglobal geomagneticfield model. The total intensity anomaly is not the magnitudeof the crustal field becauseAF=lB~+BC[-lB,nl #lBelHowever, ifthecrustal fieldissmall comparedtothemainfield, i.e. [BmI>> \ BC\, it follows that) 1AF=(Bm.Be / Bm= BC.b~= XCX~ I Fm+lCY~I F~ +ZcZ,~ I Fm. . ..(A-l)where BC=(Xc, k,, ZJand b. = (X#F., YJF., Z#F.) is theunit vectorinthedirectionof B~. Thus the total intensityanomaty is the magnitude of the component of the crustal fieldvector in the direction of the main field vector.Astheaeromagneticdata are collectedin a source-fleeregionthecrustal field maybewrittenasthegradient ofascalar potential satisfying Laplaces equation, soBC= -VVC and V%c= oFromthis it follows that thecomponentsofB. alsosatisfyLaplaces equation,If b. is constant for the area of the survey thenV2AF = V2(b~.BC) = b~. V2BC = OHence the total intensity anomaly values satisfy Laplacesequation, A general solution to Laplaces equation for thecrustat field potentiat V, in cartesian coordinates (x - north, y -east, z- vertically down) maybewritten as{V=(x, y,z) = ] J exp 2ti(m +,7) +2nz(u2 +1,2)2}A(u. r) d~4dv--.-whereuandvare wavenumtwrs inthexandydirectionsrespectively, and A(u,v) is the Fourier transform of VC(X,Y,O),If the altitude at which the aeromagnetic data werecollected istaken as z =O, thenasimilar expression may be10 HS.WILLIAMSON, P. A. GURDEN, D. J. KERRIDGE, G. SHIELLS SPE 49061used to describe the total intensity anomaly values.-AF(x,y,O)=~~exp{2?ri(ux +vy)}C(u,v) dudv . . ..(A-2)----whereC(Wv) is theFourier transformof the total intensityanomaly values. The components of the crustal field in the x, yand z directions on the surface z = Oare then given byX.k,o=+z=o=-d 7i uexp{2@P+vy) }A( uv) ~dv.-.0 -80Z(X,Y,O)=~=o=-2zj ~ivexp{2ti((x+*y)}A(u,v)&dvz -co--Inserting these expressions into Eq. A-1 and then using Eq. A-2 gives a relationship between A(u,v) and C(u,v):A(u,v) = C(u, v)/2zwwherew= i(uXjJFm + vYJFjjJ + (U2+ v2)1nUFm.Hence, ffomthetotal intensityanomalyvaluesthefunctionC(u,v) maybe computed, then the function A(u,v) and, finalIy,the values of Xc, Ycand z.SynfheticDeta Transformed Defa Differencesmntour interval. 10nT contour interval =10nT contwr mtewal =1nlmmmGlrEIFig. 1Synthettctest of the transformation methodof magneticfletdmodelling. Thedirectionof the mainffeld vector is typical of:$+Y3ij&f!* ,--. GG270 230 290 3(X) 310 320Easiings (km)Fig. 2aUK land teat of the transformation methodof magnatlcflafd modafllng. Aeromagnetfctotal intensityanomalies(nT) andgroundobservation locations.396SPE 49081 APPLICATION OF INTERPOLATION IN-FIELD REFERENCING TOREMOTE OFFSHORE LOCATIONS 11270 280 280 300 310 320Eaetings(km)FfrJ. 2b UK land teat of the tranaformatfonmethodof meanettcfl~d modalling. Declinationanomalies(arcminutes) calculatedfrom~!eeo i:- 12to880 :,) s.. . . 0 ,P1 ,--- ,fn~,,, , !670-\,,, , : ,,!. .=,l:,, ,.am-l \i- t- 12\040,,270 280 280 350 310 320Eaetings(km)Take MWD Process QC, reporlsuweys + -Real timeusing and drillat rig Option 1 ahead(rig site)4Compilesurveysinto dailytile1Get dailyHProcessmagnetic usingvariation Optiondata 3,4or 5dRepxt resultsto rig (frequency Regulardependsonturn-aroundIoperationalI(office)requirements)Fig. 3Typical flow of MWDsurveydatein anIIFR operation6W 4w. . . . .. . . . ., ----- -- .,...-;,!1~ Foinaven ~ : ~00Schiehsllion.. . . . ..!. ..-. .,,,! ;)):,!,w,W;, b. . . . . . . . . . . ..1..,,;.!d1, T. . . . . . . . . . . . . . .a ;...*.0e, . ,!,,.,.Fig. 4 Location mapfor Foinaven, Schiehallion and Andrewffeldsand BGSmagneticobservatories at Lemvick and Eskdalemulr4,>! . .,,, ,;,$ ,,:.. , ... . . ...,,56N,(:!!!,,,:,,, ;,;,, ,,,,>,!,,, ,. .,. ,: *NFig. 2cUKland test of the traneformatfonmethodof magneticflefd mcdalling. Declinationanomaliea(arcminutes) from groundobservations.39712H. S. WILLIAMSON, P. A. GURDEN, D. J. KERRIDGE, G. SHIELLSSPE 49061J1 1 I 1 # I 1\1Ko0/----------* ----..~ ----.-~--------8*-*1,.~#-./,---,390 400 410 420UTMZorP331 Eestings (km)Fig. 5aTotal fieldanomalies(nT) for the Andrewfield andsurrounding area.390 400 410 420UTMZone31Eeetinge(km)1 I 1 i I I i 16460-------- e-----L,. 2..; 7------/ ------ -------0 .6450~.t.~- ,(,g-------- .,--1 #--.2-); 6440- ./~.------..;czlg!!3,------Andrew .o5p!atioml2~> 6430 0 4+3e68J12 106420ItI-- -2--------- , .- -I_041U ~ . t1 I I 1 1 I I Ir360 400 410 420UTMZone31Eastings(km)Fig. 5CMagneticdip angleanomaiiaa (nT) for the Andrewfieldan-dsurroundingarea.~ +1 ,5EFJ] T: +1 ,0g5= +05;: 1, , ;f -T- ;, :, :G o.gEI 1g .0,5 -$[- 1- C02AIIZS05-1.oO-(39w weys)(64aufveya)(39Wiwy$)A07g(3! surveys)S06-1.5- (21Wtveys)C03(50 Wlww)Foinaven Schiehallion AndrewFig. 6 Comparfeonof MWDva. gyroscopic reference azimufhsforsix wells. Uncorrected azimuths(white circleand bars) have onlytheBGGMdeclinationapplied. llFR+orrected azimuths(blackequareand bars)arecorrectedusing Option5 (aaetext).Fig. 5b Declinationanomalies(arcminutes) for the Andrewfieldandsurroundingarea.398