美国新型钻完井技术概述与发展建议.pdf

doi:10.11911/syztjs.2023032引用格式：SusanSmithNash.美国新型钻完井技术概述与发展建议 J.石油钻探技术，2023,51（4）：192-197.SusanSmithNash.AbrieflookatnewtechniquesandtechnologiesfordrillingandcompletingintheUnitedStatesanddevelopingsuggestionsJ.PetroleumDrillingTechniques，2023,51(4)：192-197.美国新型钻完井技术概述与发展建议SusanSmithNash（美国石油地质学家协会（AAPG）创新与新兴科学技术部，俄克拉何马州诺曼市，美国，73071）摘要:美国的钻完井技术在自动化、环境合规和风险规避方面取得了引人注目的突破。为探索和评估钻完井技术在上述领域中的重大进展，通过梳理回顾美国钻完井技术突破，阐明了当前技术需求，提出了解决方案并指出了其未来发展的方向。值得注意的是，数据分析在上述技术中扮演了重要角色，基于云的解决方案也被广泛应用。然而，这些技术也面临着共同的挑战，即需要快速适应排放检测、危险监测和减少碳排放等不断发展的需求。面对这种挑战，需要充分利用现有数据库，并赋予其新的功能和目标。关键词:美国油气；钻井；完井；数据分析；发展建议中图分类号:TE22文献标志码:A文章编号:10010890（2023）04019206A Brief Look at New Techniques and Technologies for Drilling and Completing in theUnited States with Suggestions and RecommendationsSusan Smith Nash(Innovation and Emerging Science and Technology,AAPG,Norman,Oklahoma,73071,U.S)Abstract:Technological developments that specifically address automation,compliance with environmentalregulations,andhazardavoidancehaveacceleratedmorequicklythanotherdrillingandcompletiontechnologiesintheUnitedStates.Thisbriefarticleprovidesareviewofareasexperiencingsomeofthemostdramaticadvances,whiledescribingtheneedandaddressingthesolutionsastheyarenow,andhowtheycanbedevelopedinthefuture.Oneaspectthatallthenewtechnologieshaveincommonisanenhanceduseofdataanalyticsandinmanycases,cloud-basedsolutions.Achallengethatallhaveisaneedtobeabletoquicklyaccommodaterapidlyevolvingrequirementsforemissionsdetection,hazardmonitoring,andareducedcarbonfootprint.Manysolutionstothechallengesrequiretheabilitytorepurposeexistingdatabasesandusethemfornewpurposes.Key words:oilandgasintheU.S.;drilling;wellcompletion;dataanalytics;technologydevelopmentrecommendationsAcceleratingataremarkablepace,technologicaladvancementsintheUnitedStateshaveledtoremarkablebreakthroughsinautomation,environmentalcompliance,andhazardavoidancewithinthedrillingandcompletiondomain.Thisworkaimstoexploreandassessthemostsignificantstridesintheseareas,whilealsosheddinglightonthecurrentneedsandproposedsolutions,aswellastheirpotentialforfuturedevelopment.Notably,acommonthreadamongthesecutting-edgetechnologiesliesintheirrelianceondataanalyticsandcloud-basedsolutions,revolutionizingtheireffectiveness.However,akeychallengefacedbyalltheseinnovationsisthenecessitytoadaptswiftlytoevolvingdemands,includingimprovingemissionsdetection,monitoringhazards,andachievingreducedcarbonfootprint.Manyofthesolutionstothesechallengeshingeontheabilitytorepurposeexistingdatabasesfornewapplications.Thisbriefoverviewshowsdramaticadvancement收稿日期:2023-01-02；改回日期:2023-04-02。作者简介:SusanSmithNash(1958),female,Norman(Oklahoma,U.S.).SheholdsaB.SinGeology(PetroleumGeologyTrack)fromtheUniversityofOklahomain1981,M.A.inEnglishfromtheUniversityofOklahomain1989,Ph.D.inEnglishfromtheUniversityofOklahomain1996.Herrecentinitiativesincludeapplicationsofanalytics.第51卷第4期石油钻探技术Vol.51No.42023年7月PETROLEUMDRILLINGTECHNIQUESJul.,2023thefollowingareas:automatedgeosteering,HSE-friendlydrillingfluids,drillingproblempredictivefailuremodels,remotesurveillanceandautomationforenhanceddrillingoperations,suchasroboticscuttingsgatheringandanalysis,andinnovativesolutionsforwaterissues,suchasthestepsneededtobeabletousebrackishwaterinsteadoffreshwater.1NovelDrillingTechnologiesintheU.S.1.1 Automated GeosteeringTraditionalgeosteeringisverylabor-intensiveandrequireshumanoversightandadependenceonlogging-while-drilling(LWD)data.Generallyspeaking,onlytwoorthreewellscanbemonitoredatatime.However,anautomatedgeosteeringalgorithmhasbeendevelopedtomakeitpossibletosimultaneouslymonitornumerouswells.Suchamovewillsignificantlyreducetheneedforhumangeosteeringexperts.Theprocessinvolvesautomatedgeo-correlationsusingmachinelearning.Themostsuccessfulcasesareinwellsthathavegeologythatdoesnothavemajorfaultsormajordipanglechanges.Thedrillingcontractor,Helmerich&Payne,workingwiththeoperator,SabineOil&Gas,testedthealgorithminwellsintheHaynesvilleShaleintheU.S.,withpositiveresults1.AsshowninFig.1,thealgorithminvolvesadvancedLWDfiltering,faultdetection,correlation,trackingofmultipleinterpretationswithassociatedprobabilitiesandvisualizationusingnovelstratigraphicmisfitheatmaps2.PredictedLWDcorrelationType logcorrelationProbability(marginals)Most probablestructure(blue line)Fig.1 Algorithm tested in wells in the Haynesville ShaleInaddition,automatedgeosteeringhasbeendevelopedbyFactorTechnologiesusingdeeplearningBayesianmodels.CompaniessuchasRogii,withitsStarSteertechnology,andZoneVu,ageosteeringtechnologyfromUbiterra.Theprimarychallengeshavetodowithmanagingthemultipleinterpretivepossibilities,andclassifying,ranking,andassessingthegoodnessoffit34.Smartautomationandroboticshaveevolvedrapidlyduringpandemicandpost-pandemictimesastheneedforefficiency,cost-cutting,andemissionsdetectionandcontrolarebeingrequirebylegislationandshareholderaction.ExamplesofcompaniesinvolvedinsuchincludeDiversifiedWellLogging(DWL),BakerHughes,andHalliburton.Technologiesincludingmud/cuttingssampleanalysis,hazardavoidance,bitpressuremonitoringandcontrol,machinelearning,torqueavoidance,automatedgeosteering,wellboretrajectorycontrol,etc.canbeusedinareaswithasignificantrisktopersonnel,suchasdeepwaterplatforms.Theycanbeusedinapplicationsthathelpimproveenvironmentalsafety,suchasmethaneemissionsmonitoring,pluggingleaksinsubseapipelines,etc.Thefocusmustbegiventoensurethatcustomizedsolutionsaredevelopedforeachusecasesothatthetruegainsthesetechnologiescanbring,eitherincostreductions,revenuemaximization,orenhancedsafety,arerealized.Thefullpromiseofautomationisyettoberealized.Dataintegrationproblems,poorsamplingfromsensors,dataharmonizationproblems,andflawedalgorithmsmustbeovercometoavoidpoorresults.1.2 Machine Learning for Drilling HazardsIdentificationTheneedtomeasureandmonitorgeopressureandthegeomechanicalstressregimesindrillingandcompletionisofgreatimportanceinultra-deepdrilling,wherepressures,stresses,andtemperaturescanbehigh,leadingtopotentialfailuresinboreholestabilityandtheinabilitytodesignaneffectivehydraulicfracturingprogram.Differenttypesofsensorshavebeendevelopedforthosepurposes.Forexample,fiberopticssensors,whileexpensive,havebecomeveryimportantinboreholestabilityandinmonitoringofwellsduringdrilling,completion,andinstimulation.Specifically,distributedacousticsensors(DAS)canbeimplemented.AsshowninFig.2,Optasensehasdevelopedanoveldatastreamingsolutionforwellboredigitalization.Itallowsremoteoperationsaswellasearlyhazarddetection,suchasvibrationdetection.Machinelearningisusedto第51卷第4期SusanSmithNash.美国新型钻完井技术概述与发展建议193implementalgorithmsthatdetectpatterns,andthensubsequentlyidentifythem.Distributedtemperature(DTS)anddistributedacoustic(DAS)fiberopticsensingarealsonowcommonlyusedaskeyreservoirsurveillancetools.Thisworkshowsthebenefitofcontinuousdownholemonitoringduringthelifetimeofawell.Inonestudy,fiberopticcableswerepermanentlyinstalledinadoubletinjector/monitorwellsystemaspartofaCO2controlledreleasedexperimentattheIn-SituLaboratoryinWesternAustralia.Duringthecompletionandinjectionoperationsvariousplannedandunplannedevents(mudcirculation,cementing,drilling,wirelinelogging,gasandwaterflows)occurred.TheeventsweremonitoredfromsurfacetoreservoirwithDTSandDASfiberopticcables.TheDTSwasrecordedcontinuouslydatastartingduringwellcompletionthroughoutthelifetimeofthewellswhileDASwasrecordedatspecificpointsintime,mostlyassociatedwithboreholetime-lapseseismicacquisitions5.Fig.3demonstrateshowthreedifferentapproachestodataanalyticarecurrentlybeingusedforspecificusesindrillingandcompletions.ThedatawarehouseissomethingthatMicrosoftAzureandSnowflakespecializein.TheDataLakeapproachisonethathasbeenusedbybothAmazonWebServicesandGoogle.TheDataLakehouseisahybridapproachthathasbeendevelopedbyDatabricks.Data warehouseData lakeData lakehouseBIReportsBIReportsData lakeData lakeDataScienceMachine LearningBI ReportsDataScienceMetadata andgovernance layerMachine LearningData warehousesData warehousesETLETLStructured dataStructured,Semi-structured and Unstructured dataStructured,Semi-structuredand Unstructured dataFig.3 Three different approaches to data analyticIntermsofsafety,beingabletodetectandquantifygaskicksduringdrillingandcompletionsisvital.Itisonewaytopreventblowouts.Again,distributedsensingtechniques,acoustic(DAS)andtemperature(DTS),canmakeitpossibletohavereal-timecommunicationofthesemultiphaseflowevents.Gatheringthedataisnotenough,however.Itisimportanttobeabletomanageitinawaythatyieldsrequltsquickly.Thefirststepistoputallthedatainadatawarehouse,whereitiskeptinthecloudandaccessible,althoughnotclassified.Thedatalakeiswheretheanalyticscantakeplace.Forexample,identifyingandvalidatingeventsignatures(fingerprinting)inthesesensingtechnologiescanhelpoperatorsdecidehowtointerpretthesedatastreams.Itispossiblebecausethistypeofanalyticsinvolvesworkingwithstructured,semi-structured,andunstructureddata.Thedatalakehousecanworkquicklyacrossallkindsofdataandcannotonlyidentifypatternsandclassifyevents,butcanalsopredictwhenothereventswillhappen,suchasequipmentfailure,drillbitsticking,seismicity,andmore.Further,afull-scaleanalysiscanlettheteamaccuratelyinterprettheevent,giventhecomplexitiesinthefluidmechanicsandgasdynamics.1.3 Drill Pipe Failure and Sticking Prediction ModelsBeingabletopredictthefailurefordrillpipecanincreaseproductivityandreducecosts,particularlyindeepwateroffshoreoperations.Braziliandrillingcontractor,Ocyan,workedwiththeartificialintelligencetechnologycompany,RIOAnalyticstodevelopapredictivefailuremodel.Itwasthenimplementedtomanageandcontroltheuseofdrillpipeinoffshoredrillingplatforms.Todevelopthealgorithm,thefirststepinvolvedidentifyingthefactorsthatcausedrillpipefatigue,includingstressduringdrilling,corrosionpitting,thinningofthepipe,abrasivewear,mechanicaldamagefrombadhandling.Althoughthepipeisinspected,andsensorsaresometimesused,theproblemissparsedataandthetimingofdetectionofconditionsoflikelyfailure.Thepredictivealgorithmcanimprovethehistoricalpracticesofcollectingexcelspreadsheetsthatincludeoperationalandhistoricalmaintenancedata,rotatinghours,drilledmeters,andaggregatingdata,whichtendtobefullofgaps68.StreamingBatchBronzeD a t aq u a l i t ySilverGoldAI&ReportingRawCleanUse case driven dataiiiFig.2 Novel data streaming solution for wellbore digitaliz-ation194石油钻探技术2023年7月Further,wellboreinstabilityisoftencausedbycavingsandincreasedvolumeofcuttingsandotherrockfragments.Cavingsoccurastherocksundergospallingandbreakage,whichcanbeexplainedthroughgeomechanicalmodels.Thefailedrockpredictionmodelincludesdatafromoffsetwellsdata,hollowcylindertestsconductedoncoresfromtheformationofinterest,whichreflectthequantityandsizeofcavings,bothfrommechanicaldisturbancefromthebit,plusspalling910.Theresultwasa3Dporo-elasto-plasticfiniteelementmodel(FEM)whichallowedforderiskingoperationsbypredictingwellboreinstability,spalling,cavings,andthusreducingthelikelihoodofstuckpipe.AlBahranietalsmodelisathree-dimensionalnumericalmodelandusesauniquecavingsvolumedeterminationmethodstodeterminethedownholepressurenecessarytoavoidtheproductionofcavingsandthuswellboreinstabilityandpotentialsticking9.Initialmudweightorbottomholepressureisinsufficientforboreholestability,resultingincavingsthatdramaticallyincreasethemudvolumeintheannulus.Theeffectoftheadditionalvolumeisawellboreenlargementandincreaseddiameter.Theresultisanincreaseinmudweight,pressurewiththecuttingsandcavingsloading.Hereiswherepredictionscomeinanddifferentchangesinmudweight,rheologicalproperties,righydraulicscanbemadetopreventfurthercavings.Theinterventionscanstabilizethewellbore.1.4 HSE-Friendly Drilling FluidsDrillingfluidshavebeendevelopedtobegreenandmoreeffectiveindevelopingahealthyandsafeworkingenvironmentfortheworkersandfortheecosystem.Further,environmentallawsandregulationsareincreasinglystrict,makingenvironmentallyhostileadditivesandchemicalsimpossibletouse11.Theindustryisacceleratingisprogresstowardeco-friendly,biodegradable,non-toxic,andHSE-friendlyadditivesandchemicals.Tofindaneffectivesubstitute,researchersdevelopeddrillingfluidsfromwastevegetableoilfromthefoodandcateringindustry,andthentestedthemonwellsdrilledbySaudiAramco.Theproductswereeffectiveinpractice12.Fig.4showsthedrillingfluidsdevelopedfromwastevegetableoil.Toassureanongoingsupply,partnershipswereestablishedwithsourcesofwastevegetableoil.Oil-basedmudsareused,particularlywhendrillingthroughsaltzonesoranywherethatwater-basedmudswouldresultinwashoutzonesandcavings.Tofindtherawmaterialsforoil-basedmudscanbecomplicated,particularlywheretherearesupplychainissues.Tothatend,studieshavebeenperformedtodetermineaprocessforconvertingwastevegetableoiltofattyacidsbybasehydrolysisreaction.Applicationofsynthesizedfattyacidsforwater-basedandoil-basedmudformulationaslubricants,emulsifiersandrheologymodifiersisalsoused.Halliburton,AESDrillingFluids,andBakerHughesarejustafewofthecompaniesthathavebeenmakingstridesindevelopingenvironmentallyfriendlydrillingfluidsforalowcarbonfootprint.Thecoefficientoffrictionreducingpotentialofwastevegetableoil-basedlubricantswasdemonstratedinaseriesofexperimentswhilealsoprovidingasolutionforthedisposalofwastecookingoil.1.5 Bits and Cutting Elements Failure PredictiveModelDrillbitdamagedetectionandfailurepredictionarevitalforprolongingthelifeofthebitandimprovingdrillingefficiency.Themethodologyrequiresidentifyingdrillbitcutters,quantifyingthedamagetoeachcutterthroughcutterimageanalysis,locationofthecutters,andtherootcauseofthedamage.Findingthecutters,gradingthecutters,andthencategorizingcutterlocationsaredone,andthenaclassificationforrootcause.Thecriticalfactorforsuccessisbasedonthequalityofbitimagesandthestandardization13.Further,successfulpredictivemodelingrequiresaclearflowchartforquantifyingcutterdamage.Fig.5providesaflowfortheprocessesusedtodevelopalgorithmsthatpredictthemostlikelyoutcomeandtheappearance.PerhapsthemostcriticalstephastodowithcharacterizingthecuttercontourastransmittedbytheAdditivesChemicaltransformation stagesProducts resultingfrom the processesStart with waste cooking oilAdd methanol and NaOHWaste vegetable oil(WVO)Physical treatmentEsterificationSeparationWashingDryingWaste vegetable oil esterTriglycerides and raw esterFig.4 Drilling fluids developed from waste vegetable oil第51卷第4期SusanSmithNash.美国新型钻完井技术概述与发展建议195sensors.Thenextstepsrequirealgorithmsthateliminatenoiseandthenbinarizetheresultingdata.Thedifferencebetweentheideal(original)shapeandthedamagedshapeprovidesawaytoquantifycutterdamage.Thefollowingdiagramillustratestheprocess:InputErodedilateBinarizationDenoisingDamagedshapeGetcontoursOutput87.303%Grate:1OriginalshapeFig.5 Prediction processAresearchinitiativewaslaunchedtoinvestigatenewtypesofcuttingelements.Theprojectwassuccessfulandyieldedaninnovativeconical-shapedpolycrystallinediamondelement(CDE).ThiselementhastwicethediamondthicknessofconventionalPDCcutters,resultinginhigherimpactstrengthandmoreresistancetowardabrasivewearbyapproximately25%.AnewbittypewasdesignedwiththeCDEsstrategicallyplacedacrossthebitfacefromgaugetothebitcenterutilizingFEA-basedmodelingsystem.NOV,EXIM,andothersarecompaniesthathavedevelope