a dynamic support system for wellbore positioning quality

CopyrightThinkTankMathsLtd2017

“ADynamicSupportSystemFor

WellborePositioningQualityControlWhileDrilling”

Dr.CyrilleMathis,Director&CSO,ThinkTankMaths(Edinburgh)

IOGPGeomatics/StatoilIndustryDay2017

26th April2017- Stavanger,Norway


Amathematicsresearchcompanythatcreatesparadigm-shiftingsolutions tohardproblemsinmanyfieldsthathaveimmenseoperationalcomplexity.

Tacklesproblemswhereoff-the-shelftechnology,systemsorsoftwaresolutionsdonotexistorareinsufficient.

TTM’ssolutionsimpactclients’currentbusinessmodelsandtechnologies.

Whatwedo…


• Complexdatasets

• Dataandanalysisgobeyonduseofspreadsheetortraditionalstats/maths– justtoomanyfactorstoanalyse

• Analysisrequiresnewwaysoflookingatdata-- multipleinter-relationships

• Neednewtoolsthatareintegratedintoexistingprocesses– mustfitwithinthesystemandcontextofhoworganisationsdobusiness

• Supportsdecision-makingtodrivebetteroutcomes

Whatdoourclientshaveincommon


DynamicSupportSystemforWellborePositioningQualityControlOverview

• PathConstruction- Mathematicallyunsound- Pooruseofsensorinformation

• QualityControl- GrosslyincorrectmeasurementscanpassQC- Subjectiveinterpretation- Analternativesourceofinformation

• UncertaintyModelling- Inadequatemathematics- Refinementcannotsolveproblems- Impactsallaspectsoffielddevelopment


DynamicSupportSystemforWellborePositioningQualityControlPathConstruction

Minimumcurvatureandbeyond:l Developedmoregeneralapproachesl Exploitinformation(‘signatures’)inMWDandLWDsensorsl CreateafoundationtoincorporateQualityControlandUncertainty

Left Minimumcurvature(MC)method. Middle Example well sections by MC vsnovel TTM methods.

Right Well paths generated by MC and novel TTMalgorithm. Despite being equally consistent withdata they disagree on the well position.


Impact of magnetic corrections and quality control reassessed• Possible to fail good readings• Possible to pass bad readings• Passing QC does not imply improved precision in azimuth

DynamicSupportSystemforWellborePositioningQualityControlQualityControl

Left Comparison of large set of magnetic readings which passindustry QC (black) vs failed industry QC (red), showing nostatistical improvement in azimuthal estimate.

Right Comparison of large set of magnetic readings which have beenunprocessed (black) vs processed by industry MagCorr (red),showing reduction in quality due to MagCorr modification.


LWD-MWD structures inform magnetic processing• Tailoring magnetic refinement methods to environment• Generate model of environment which may not require IIFR

DynamicSupportSystemforWellborePositioningQualityControlMagneticEnvironmentandIIFR

Figure Comparison of azimuthal estimates from high resolution gyro (orange), raw QC-passing magnetic(black), MagCorr corrected (red) and TTM methods TTM-1 (green) and TTM-2 (blue). MagCorrshows poor performance, often worse than raw. TTM methods give significantly improvedaccuracy, including in the case of TTM-2, which does not use any IIFR data to process magneticreadings.


DynamicSupportSystemforWellborePositioningQualityControlComplexEnvironments

Extraction of structure within LWD readings beyond “geological layers”Identification of structures or “signatures” guided by utility

Analysis must be able to handle practical challengesl Industry's frequent use of different toolsl Potential mis-calibrationl Multiple tool runs

Figure LWD surveys ordered by Measured Depth (MD) and labelled by colour basedon hidden structures extracted from LWD dataset. Complex mixing showsstructures do not follow MD ordering.


IntelligentWellborePositioningSystemPrototype


SiteCom WITSML Server“Real-time”

Visualisation ArchiveCSV

Internal ArchiveBespoke format

“Static data” ServerNominal Field ValuesBGGM + IFR + IIFRBGS

Deliverable Server ( + Client )

Sensor Data from BHAMWD + LWDDrilling Company

Deliverable ClientsBrowser

External formats requiring client/libraries for real-time processing

Bespoke proprietary format for internal use only

Simple CSV format for later or external visualisation

“Drill Simulator”Historical Surveys

CSV, LAS

Historical survey formats for off-line processing

IntelligentWellborePositioningSystem- Workflow


IntelligentWellborePositioningSystem- EngineWorkflow

Real-timeSiteComWITSML

StaticTIDE

HistoricalArchivedCSV/LAS

MagneticProcessingMethods

PathConstructionMethods

TTMData

ProcessingEngine Positional

UncertaintyMethods

TTMIntelligentDecisioningEngine

AndAlgorithmSelection PositionalEstimate

Uncertainty

QCOutputs

MagneticSignatures

GeologicalInsight

Inputs


PrototypeUserInterfaceExampleOutputs


A)WellnameB)Controlbuttons– Show/Hide“Panels”,Show“Help”,ExitVisualisationC)Graph– Plotof1,2or3userselectedoutputvariablesforentirewellD)Table– ColumneddataforuserselectedvariablesE)Panels– Displayedmessages/informationaboutselectedsurveysanddisplayoptionsforgraphandtableF)Informationbuttons– Displaywelloverview,ExporttabledatatoCSVandExportgraphtoPNG

A B

C D EF

PrototypeUserInterfaceVisualisationScreenDescription


UserInterfaceGraph/TableDisplayOptions

.


UserInterfaceWellandSurveyInformationMessages

Additional information can be obtained on the well overall (right top) and agiven survey (left bottom) and is displayed in the “Messages” panel.

The overall well data, via the “Well Overview” button, provides a shortsummary of the length of each run and density of surveys within each runup to that point in time.

Information on individual surveys is displayed when a survey’s row in theTable section is clicked. All available information for the survey will bedisplayed in the “Messages” panel regardless of which data types areshown in the Table.


TrustedReasoningArchitecture(TRA)


SiteCom WITSML Server“Real-time”

Visualisation ArchiveCSV

Internal ArchiveBespoke format

“Static data” ServerNominal Field ValuesBGGM + IFR + IIFR

BGS TRA Server ( + Client )

Sensor Data from BHAMWD + LWD

Drilling Company

TRA ClientsBrowser

“Drill Simulator”Historical Surveys

CSV, LAS

TrustedReasoningArchitecture(TRA)EnabledSystemWorkflow

TrustedReasoningArchitecture

(TRA)


DecisionSupportSystem– situationawarenessTrustedReasoningArchitecture(TRA)

In order to train and refine the system it is necessary to use TTM’s Trusted Reasoning Architecture (TRA).

Capabilities and insight gained within the highly demanding military domain (“Command and Control” andunmanned aircraft) can be transferred to the Oil and Gas industry in order to provide an intelligent driller decisionsupport system with “man-in-the-loop”.

“DrillbyWire”systemwithnumerouspotentialbenefits:• Continualupdatingofscenariosandexplanatoryoutputs• Automatedcontrolofsteeringanddrillparameters(WoB,mudpressure,RoP etc)– wellstabilityprediction• Refinedqualitycontrol- eg distinguishingbetweenbrokensensors,beingoff-planandsensorinterference• Automaticcataloguingofnewevents(topreventundesiredsituationsfromreoccurring)


TrustedReasoningArchitectureAtmosphericsandAnomalies

Improved communication of information essentiall “Atmospherics” drawn from multiple sensorsl Explainable actions lead to increased trustl Avoid overwhelming operator with irrelevant or

“raw” data


FutureDevelopments

Collision Avoidance Procedure

Through the continued development and refinement of magnetic field processing and novel LWD directionalitymethods identification of signals indicative of collision risks :

• Offset active wells• Ghost wells• Broken or stuck equipment left in an abandoned and plugged well• Unknown obstructions, man-made or natural, which could pose a risk to the drill

Well Planning

Automated Directional Drilling Support

Leakage Detection and Mud Losses

Well Stability While Drilling (Preventing “Troubles” While Drilling)

Deliberate wellbore interception (relief well planning)

Geology Insight

Subsidence Prediction


Misused trendytechniques…

• Statisticalanalysis(Analytics)• Neuralnetworks• CFD(computationalfluiddynamics)• Deconvolutiontechniques• MonteCarlosimulations

…andflawed thinking

• Systemover-simplification• Assumedlinearity• Bayesianinference• Gaussiandistribution• Mechanicalfunctionality• Directcorrelation• Artificialequilibria,…


A New Perspective

23


Thankyou

Contact:AngelaMathis,[email protected]

ThinkTankMathsLtdEdinburgh,UK

www.thinktankmaths.com

a dynamic support system for wellbore positioning quality

Documents