Multi-level learning scheme for calibrating wellbore trajectory models for directional drilling

Abstract

A multi-level learning scheme method and system for real time autonomous calibration of wellbore trajectory models for use in directional drilling, further comprising positioning one more sensors downhole in a geological formation; performing surveys and continuous measurements of the geological formation based on the one or more sensors; determining a first wellbore trajectory based on the surveys, a second wellbore trajectory based on the continuous measurements, and a predicted wellbore trajectory based on a wellbore trajectory model; determining errors between the predicted wellbore trajectory and the first wellbore trajectory and second wellbore trajectory and calibrating the wellbore trajectory model based on the errors.

Description

Docket Number: 164.2018-IPM-102337 V1 USMULTI-LEVEL LEARNING SCHEME FOR CALIBRATING WELLBORETRAJECTORY MODELS FOR DIRECTIONAL DRILLINGFIELD OF USE

[0001] The disclosure generally relates to the field of hydrocarbon production, andmore particularly to real time autonomous calibration of wellbore trajectory models foruse in directional drilling.BACKGROUND

[0002] Drilling for hydrocarbons, such as oil and gas, typically involves operation ofa drilling tool at underground depths that can reach down to thousands of feet below thesurface. Such remote distances of the downhole drilling tool, combined withunpredictable downhole operating conditions and vibrational drilling disturbances,creates numerous challenges in accurately steering the drilling tool to reach a set target.Sensors, located at or near a bottom hole assembly (BHA), detect various conditionsrelated to the drilling, such as position and angle of the drilling tool, characteristics of therock formation, pressure, temperature, acoustics, and / or radiation. Such sensormeasurement data is typically transmitted to the surface, where human operators analyzethe data to steer the downhole drilling tool to reach a set target.BRIEF DESCRIPTION OF THE DRAWINGS

[0003] Aspects of the disclosure may be better understood by referencing theaccompanying drawings.

[0004] FIG. 1 is a block diagram for calibration of a wellbore trajectory model.

[0005] FIG. 2 is a flow chart of functions associated with defining control input forcalibration of the wellbore trajectory model.

[0006] FIG. 3 is another flow chart of functions associated with defining the controlinput for calibration of the wellbore trajectory model.1Docket Number: 164.2018-IPM-102337 V1 US

[0007] FIG. 4 is a detailed block diagram of the self-learning controller whichprovides the control input into the trajectory model.

[0008] FIG. 5 shows results of the SLL and FLL learning loops.

[0009] FIG. 6 is a schematic diagram of a well system associated with calibration ofwellbore trajectory models.

[0010] FIG. 7 is a block diagram of apparatus for calibration of wellbore trajectorymodels.

[0011] The drawings are for purpose of illustrating example embodiments, but it isunderstood that the inventions are not limited to the arrangements and instrumentalitiesshown in the drawings.DESCRIPTION

[0012] The description that follows includes example systems, methods, techniques,and program flows that embody aspects / embodiments of the disclosure. However, it isunderstood that this disclosure may be practiced without these specific details. Forinstance, this disclosure refers to real time autonomous calibration of wellbore trajectorymodels for use in directional drilling in illustrative examples. In other instances, well-known instruction instances, protocols, structures and techniques have not been shown indetail in order not to obfuscate the description.Overview

[0013] Wellbore trajectory models are used to determine a trajectory of a wellboredrilled by a drilling tool such as mud motors and rotary steerable systems. The wellboretrajectory output by the model indicates curvature, attitude and / or position information ofthe wellbore which is used to determine whether the wellbore will reach a target in thegeological formation.

[0014] Embodiments described herein are directed to calibration of the wellboretrajectory model which predicts the wellbore trajectory of a wellbore being drilled either2Docket Number: 164.2018-IPM-102337 V1 USonshore or offshore. The calibration is based on a: (i) slow-learning loop (SLL) and (ii)fast-learning loop (FLL). SLL involves taking stationary surveys of the geologicalformation during drilling to determine a wellbore trajectory and FLL involves takingcontinuous measurements of the geological formation to determine a wellbore trajectory.The difference between SLL and FLL is that the stationary surveys are taken lessfrequently than continuous measurements (e.g., surveys can be taken every 30, 45 or 90feet, while continuous measurements taken via intelligent drilling tools or MWD toolscan be obtained as fast as every 0.5 feet). In this regard, the SLL captures the long-termexternal effects on wellbore trajectory (formation tendencies, wear on tool components)and the FLL captures the short-term and instantaneous effects (rock interface changes,faults) on the wellbore trajectory.

[0015] The wellbore trajectory determined by the SLL and FLL are used to calibratethe trajectory model so that the curvature, attitude, and / or position error calculated by adifference between the model-predicted wellbore trajectory and wellbore trajectory basedon the SLL stationary surveys and FLL continuous directional measurements isminimized. The wellbore is drilled, the determined FLL and SLL wellbore trajectory arecompared to a model-predicted wellbore trajectory, and control inputs are determinedbased on the calculated wellbore trajectory and model-predicted wellbore trajectory andone or more of steering ratio (SR), tool face (TF) information, and distance of the modelpredicted wellbore trajectory. The trajectory model is calibrated based on the controlinputs. In this regard, the calibration allows for the trajectory model to be calibrated andtested in real time, for instance while drilling through an unknown formation.

[0016] The description that follows includes example systems, apparatuses, andmethods that embody aspects of the disclosure. However, it is understood that thisdisclosure may be practiced without these specific details. In other instances, well-known instruction instances, structures and techniques have not been shown in detail inorder not to obfuscate the description.Example Illustrations3Docket Number: 164.2018-IPM-102337 V1 US

[0017] FIG. 1 is a block diagram 100 for calibration of a wellbore trajectory model.The block diagram 100 includes a self-learning controller 102 and a wellbore trajectorymodel 104, sensors 106, and a drilling system 108.

[0018] The wellbore trajectory model 104 is used to determine a trajectory of awellbore drilled by a drilling tool. The output by the wellbore trajectory model 104indicates curvature, attitude and / or position information of the wellbore. This output isused to determine whether a wellbore will reach a target in the geological formation. Thedrilling tool can take many forms including drilling an offshore or onshore well, with 2Dor 3D trajectories, and with mud motors, rotary steerable systems, or any drilling tool. Itcan be applied either on the surface or downhole in the drilling tool or another BHAcomponent that communicates with the measuring while drilling (MWD) tools and thedrilling tool.

[0019] The self-learning controller 102 receives one or more inputs. The one or moreinputs includes one or more of steering inputs, a model output, stationary surveys, andcontinuous measurements. The steering inputs may be a steering direction of the drillingtool including a toolface and steering ratio. The model output may be the wellboretrajectory output by the wellbore trajectory model at some earlier time. The stationarysurvey and continuous measurements may be measurements of the formation taken viaone or more sensors 106 positioned downhole and / or at a surface of the geologicalformation. The difference between the surveys and continuous measurements are that thestationary surveys are taken less frequently than continuous measurements (e.g., surveyscan be taken every 30, 45 or 90 feet, while continuous measurements taken via intelligentdrilling tools or measuring while drilling (MWD) tools can be obtained as fast as every0.5 feet). The self-learning controller 102 provides control inputs to the wellboretrajectory model 104 to facilitate the calibration of the wellbore trajectory model 104based on the one or more inputs. The calibrated trajectory model 104 may then beprovided to a drilling system 108 having the drilling tools. The calibrated trajectorymodel 104 is used to control drilling of the geological formation to form a wellborewhich reaches a target in the geological formation.4Docket Number: 164.2018-IPM-102337 V1 US

[0020] The self-learning controller 102 may have a slow-learning loop (SLL) andfast-learning loop (FLL). The SLL uses the stationary surveys of the geological formationand wellbore trajectory model 104 to determine a wellbore trajectory while the FLL usesthe continuous measurements of the geological formation during drilling and wellboretrajectory model 104 to determine a wellbore trajectory. The SLL captures the long-termexternal effects on wellbore trajectory (formation tendencies, wear on tool components)and the FLL captures the short-term and instantaneous effects (rock interface changes,faults) on wellbore trajectory.

[0021] An influence of SLL on the calibration of the trajectory model 104 is slowlychanging while influence of FLL on the calibration is more dynamic than SLL. SinceSLL uses stationary surveys and the accuracy and precision of stationary surveys arebetter than the continuous surveys (taken via intelligent drilling tools or MWD tools),SLL provides a more reliable calibration to the trajectory model 104. However, since itupdates less frequently than FLL, SLL just by itself would not provide a sufficientlyquick calibration for the trajectory model, for instance when an unexpected formationchange happens. For such scenarios, FLL is quite useful, although the continuousmeasurements may be noisy. To prevent reaction to noise, the calibration effect comingfrom the FLL can be picked sufficiently smaller than the calibration effect coming fromthe SLL.

[0022] The wellbore trajectory determined by the SLL and FLL are used to calibratethe trajectory model so that the curvature, attitude, and / or position error calculated by adifference between the model-predicted wellbore trajectory and wellbore trajectory basedon the SLL stationary surveys and FLL continuous directional measurements isminimized. The wellbore is drilled, the determined FLL and SLL wellbore trajectory arecompared to a model-predicted wellbore trajectory, and the trajectory model is calibratedbased on the error between the calculated wellbore trajectory and model-predictedwellbore trajectory and one or more of steering ratio (SR), tool face (TF) information,and distance of the model predicted wellbore trajectory. The calibration may be a controlinput which is provided to the wellbore trajectory model 104.5Docket Number: 164.2018-IPM-102337 V1 US

[0023] In some examples, the SLL and FLL can use a multiplicity of historical (past)continuous measurements and surveys along with the most recent ones to determine thecontrol input. In other examples, either the FLL and SLL can be used as standalonecalibration methodology in case measurements are not available or unreliable.

[0024] Additionally, or alternatively, the update rate of FLL can theoretically beselected as fast as the continuous measurement updates; however, in practice there maybe no need for such high update rate. In that case FLL can be ran at a slower speed thanthe continuous measurement updates. In some examples, FLL can have additionalinput(s) that indicate the properties of the drilled rock, such as MSE, gamma-ray,resistivity, etc. sensor outputs. This could be used to take the formation effect intoaccount while calibrating the trajectory model. This information may be used to mapcalibration parameters from offset wells or to reset the self-learning process. In someexamples, the FLL can have additional input(s) that indicate drilling parameters, such asweight on bit, torque on bit, or RPM obtained from surface measurements, downholemeasurements, or estimated from the trajectory model. These could be taken into accountwhile calibrating the trajectory model.

[0025] In some cases, the trajectory model can be a probabilistic model built based onthe previously drilled wells in the same area / basin. In this case, the self-learningcontroller 102 can be used to calibrate the probabilistic model parameters such as themean and variance (model uncertainty). The probabilistic trajectory model can beconstructed from the results of a calibration model in a same area / basin or for a given setof tools, bit, etc. across different areas / basins. The probabilistic model can be used toprovide a better estimate of borehole position (TVD) by reducing propagation of sensoruncertainty due to knowledge of the area / basin.

[0026] FIG. 2 is a flow chart 200 of functions associated with the SLL in defining thecontrol input for calibration of the wellbore trajectory model.

[0027] At 202, a stationary survey which identities one or more of a curvature,attitude and / or position of a wellbore is received. At 204, the trajectory model is runbased on a past survey and the most recent survey to predict the curvature, attitude and / or6Docket Number: 164.2018-IPM-102337 V1 USposition of the wellbore at the most recent survey. At 206, an error is determinedbetween the predicted curvature, attitude and / or position of the wellbore and thecurvature, attitude and / or position of the wellbore measured / calculated at the most recentsurvey. At 208, a control input is determined as a function of this error, steering ratio(SR) and / or toolface (TF) information along a distance that the trajectory model was ran.At 210, the control input is provided to the model for calibration. The calibrated model isthen used for drilling operations.

[0028] FIG. 3 is a flow chart 300 of functions associated with the FLL in defining thecontrol input for calibration of the wellbore trajectory model.

[0029] At 302, continuous directional measurements (e.g., attitude) is received. At304, a sensor fusion technique combines the continuous directional measurements and amost recent survey to calculate a representative curvature, attitude and / or position of thewellbore trajectory. The fusion may involve filtering and / or weighting the survey andcontinuous measurements to determine the representative curvature, attitude and / orposition of the wellbore trajectory. At 306, the trajectory model is run between the mostrecent survey to the depth or the instant at which the most recent continuousmeasurement was received, to predict the curvature, attitude and / or position of thewellbore at the depth at which the most recent continuous measurement was received. At308, the error is determined between the model predicted wellbore trajectory, e.g.,curvature, attitude and / or position of the wellbore, and the calculated representativecurvature, attitude and / or position of the wellbore. At 310, a control input is determinedas a function of one or more of this error, steering ratio (SR) and / or toolface (TF)information along the distance the model was ran. At 312, the control input is provided tothe model for calibration. The calibrated model is then used for drilling operations.

[0030] FIG. 4 is a detailed block diagram 400 of the self-learning controller whichprovides the control input into the trajectory model. The block diagram showsinteractions between the trajectory model, sensor fusion logic, and control laws for SLLand FLL. The trajectory model is shown, ( , ), which is a function of the steeringinputs (toolface and steering ratio), , and the learning control input . Its output is the7Docket Number: 164.2018-IPM-102337 V1 USwellbore trajectory, ̂, which can include curvature, attitude and / or position. The sensorfusion logic is shown as ( ,Θ ) which calculates a curvature, attitude and / orposition at the depth (or instant) of the continuous measurement, , as a function ofcurvature, attitude and / or position of the survey, , and the continuous attitudemeasurement Θ . The SLL control law is shown as ( , ) which calculates thecontrol input for SLL, , based on the inputs and , which are the error betweenthe wellbore trajectory predicted by the model and the wellbore trajectory predicted bythe survey measurement, and the steering input throughout the interval where error iscalculated, respectively. The FLL control law is shown as ( , ) which calculatesthe control input for FLL, , based on the inputs and , which are the errorbetween the wellbore trajectory predicted by the model and the wellbore trajectorypredicted by the survey measurement, and the steering input throughout the intervalwhere error is calculated, respectively.

[0031] Both and are historical (spatially delayed) control inputs where thedelays correspond to the distances between the bit and the sensor locations of the surveymeasurement system and continuous measurement system. is the superposition of theindividual control inputs from SLL and FLL.

[0032] The sensor fusion may consist of one or more of the following in order todenoise and smoothen the continuous measurements: Finite Impulse Response (FIR)filter, Infinite Impulse Response (IIR) filter, Gaussian Process Regression (GPR) model.There can be more than one continuous measurement that are fed to the sensor fusionlogic alongside the survey data. Noise and offset in the measurements can beestimated / computed to improve the sensor fusion algorithm.

[0033] The SLL and FLL can be designed such that the calibration to the trajectorymodel can be based on multiple terms where these terms are determined by a weightingfunction. At least one term can target calibrating the model parameters, while at least oneother parameter can target compensating for biases that are not modeled such asformation tendencies, anisotropy, faults, pad, bit, bearing wear etc. which are valuable forgeosteering and evaluating whether a given target can be successfully reached or for8Docket Number: 164.2018-IPM-102337 V1 USestimating an ability of a tool to follow a well plan. The well plan is typically preparedbefore a lateral drilling begins in a geological formation. The well plan defines steeringdecisions for a drilling tool so that a smooth wellbore trajectory is produced to a target ina geological formation. The following two examples provide additional details ofweighting functions associated with the calibration.Example 1:

[0034] Consider that a trajectory model only captures curvature, where the drillingtool’s maximum curvature generation capability is represented by κMax: ̂= ( , )=κMax + ( , )

[0035] Here, =SRcos(TF), where SR (0<SR<1) and TF represent the steering ratioand toolface, respectively. ( , ) is a nonlinear calibration logic. The logic for SLL andFLL can be chosen as an array of two elements (weights) one proportional over theinterval the error was calculated and one proportional to 1− : = [ 1− ] = [ 1− ]

[0036] Here, and are constant gains which could be considered as proportionalgains for SLL and FLL. Then, given = + , the nonlinear calibration logic, ( , ),can be picked as ( , )= [ 1] = =κMain +κBias

[0037] whereκMain= + κBias= (1− ) + (1− )

[0038] As a result, the calibrated model becomes ̂= ( , )=(κMax+κMain) +κBias

[0039] κMain calibrates the model parameter κMax, whereas κBias compensates for the un-modeled effects on steering.9Docket Number: 164.2018-IPM-102337 V1 USExample 2:

[0040] In this example, the trajectory model, control laws for SLL and FLL, and thesensor fusion logic are selected as in Table 1.

[0041] Table 1 – functional relation, dynamics, details, and array size:

[0042] Nomenclature for the variables in Table 1 is provided in Table 2.

[0043] The trajectory model output consists of wellbore curvature and attitude: ̂=[ Θ] . The matrices A, B and C are functions of the drilling tool geometry, drill-bit’s sidecutting efficiency, WOB, hole overgauge, and tool response characteristics, among otherparameters.

[0044] The SLL and FLL are selected to distribute the trajectory model error betweena control term and a bias term as a function of the steering ratio and duty cycle used alonga section of the wellbore where error was calculated. A simplistic sensor fusion10Docket Number: 164.2018-IPM-102337 V1 UStechnique is also selected that utilizes survey measurements and continuous measurementto extrapolate the states forward.

[0045] The SLL and FLL are shown with a proportional logic in the aforementionedcases. In other examples, any combination of proportional, integral and derivative logiccan be designed depending on the scenario. In some examples, the logic gains can also beextended to vectors to allow more flexibility to tune between bias effects versus Kmain.

[0046] The calibration and biases provided by the SLL and FLL can be used in otherways as well other than for calibrating the trajectory model. If ran real-time, thecalibration and bias could be used to characterize the drilling tool’s steering performancesuch as the curvature generation performance or toolface offset. If ran as a post-job tool,the calibration and bias could be used to learn about the un-modeled steering tendenciesin the drilled basin, which can be used to design the next job in the same area. Forexample, the unaccounted for (external, un-modeled, or un-predictable) effects on thewellbore trajectory can be identified (e.g., formation tendencies, anisotropy, faults, pad,bit, bearing wear etc.), which are valuable for geosteering and evaluating whether a giventarget can be successfully reached or for directional drilling to estimate the ability of atool to follow a well plan. The calibration and bias may be used in other ways as well.

[0047] FIG. 5 shows results of the SLL and FLL learning loops for the systemdescribed in Table 1. At every survey, the trajectory model is used to project the modeltrajectory till the next survey with three settings: (i) without calibration, (ii) with SLL and(iii) with SLL and FLL. SLL used the steering inputs between the previous survey to thecurrent survey and the calibrated trajectory model is then used to project to the nextsurvey. FLL frequency was set to 10 ft, so every 10 ft it used the data from the last surveyto the most recent continuous measurement. The projection results for three settings arecompared with the actual wellbore trajectory indicated by the survey measurements andthe resulting inclination errors (scaled per 100 ft due to varying survey intervals) for acurve section and tangent section of the wellbore trajectory. The calibrated wellboretrajectory model with SLL and FLL significantly reduces the trajectory model projection11Docket Number: 164.2018-IPM-102337 V1 USerror compared to the wellbore trajectory model calibrated with only SLL (calibratedmodel with SLL) or model which is not calibrated (uncalibrated model).

[0048] FIG. 6 is a schematic diagram of a well system 600 associated with calibrationof wellbore trajectory models in which above the functions may be performed. The wellsystem 600 includes a drill bit 602 disposed on a drill string 604 of the well system 600for drilling a wellbore 606 in a subsurface formation 608. While wellbore 606 is shownextending generally vertically into the subsurface formation 608, the principles describedherein are also applicable to wellbores that extend at an angle through the subsurfaceformation 608, such as horizontal and slanted wellbores. For example, the wellbore canbe angled vertically followed by a low inclination angle, high inclination angle orhorizontal placement of the well. It should further be noted that a land-based operation isdepicted, but those skilled in the art will readily recognize that the principles describedherein are equally applicable to subsea operations that employ floating or sea-basedplatforms and rigs, without departing from the scope of the disclosure.

[0049] The well system 600 may further includes a drilling platform 610 thatsupports a derrick 612 having a traveling block 614 for raising and lowering drill string604. Drill string 604 may include, but is not limited to, drill pipe and coiled tubing, asgenerally known to those skilled in the art. A kelly 616 may support drill string 604 as itmay be lowered through a rotary table 618. The drill bit 602 may crush or cut rock, beattached to the distal end of drill string 604 and be driven either be a downhole motorand / or via rotation of drill string 604 from the surface 620. Without limitation, drill bit602 may include, roller cone bits, PDC bits, natural diamond bits, any hole openers,reamers, coring bits, and the like. As drill bit 602 rotates, it may create and extendwellbore 606 that penetrates various subterranean formations. A pump 622 may circulatedrilling fluid through a feed pipe 624 to kelly 616, downhole through interior of drillstring 604, through orifices in drill bit 602, back to surface 620 via annulus 626surrounding drill string 604, and into a retention pit 628.

[0050] Drill bit 602 may be just one piece of the drill string 604 that may include oneor more drill collars 630 and one or a plurality of logging tools 634 such as logging-12Docket Number: 164.2018-IPM-102337 V1 USwhile-drilling (LWD) or measuring-while-drilling (MWD) tools for measuring,processing, and storing information such as the survey measurements and / or continuousmeasurements used to calibrate model 650 which takes the form of the wellboretrajectory model. It will also be understood that more than one logging tool 634 (i.e., oneor more LWD and / or MWD module) can be employed. Logging tool 634 may bearranged to communicate with a computing system 636. Computing system 636 mayinclude a processing unit 638, a monitor 640, an input device 642 (e.g., keyboard, mouse,etc.), computer media (e.g., optical disks, magnetic disks), and / or the wellbore trajectorymodel 650. The computing system 636 may store code representative of the self-learningcontroller described herein for real time autonomous calibration of wellbore trajectorymodels for use in directional drilling. In this regard, computing system 636 may act as adata acquisition system and possibly a data processing system that analyzes informationfrom logging tool 634. Any suitable technique may be used for transmitting signals fromlogging tool 634 to the computing system 636 residing on the surface 620. As illustrated,a communication link 644 (which may be wired or wireless, for example) may beprovided that may transmit data from logging tool 634 to the computing system 636.Communication link 644 may implement one or more of various known drillingtelemetry techniques such as mud-pulse, acoustic, electromagnetic, etc.

[0051] FIG. 7 is a block diagram of apparatus 700 (e.g., the computing system) forcalibration of wellbore trajectory models. The apparatus 700 may be located at a surfaceof a formation or downhole. In the case that the apparatus 700 is downhole, the apparatus700 may be rugged, unobtrusive, can withstand the temperatures and pressures in situ atthe wellbore.

[0052] The apparatus 700 includes a processor 702 (possibly including multipleprocessors, multiple cores, multiple nodes, and / or implementing multi-threading, etc.).The apparatus 700 includes memory 704. The memory 704 may be system memory (e.g.,one or more of cache, SRAM, DRAM, zero capacitor RAM, Twin Transistor RAM,eDRAM, EDO RAM, DDR RAM, EEPROM, NRAM, RRAM, SONOS, PRAM, etc.) or13Docket Number: 164.2018-IPM-102337 V1 USany one or more of the above already described possible realizations of machine-readablemedia.

[0053] The apparatus 700 may also include a persistent data storage 706. Thepersistent data storage 706 can be a hard disk drive, such as magnetic storage device.The apparatus 700 also includes a bus 708 (e.g., PCI, ISA, PCI-Express, NuBus, etc.).Coupled to the bus 708 is a network interface 710 which facilitates communication withthe logging tool. The network interface 710 may receive the continuous measurementsand surveys from the logging tool which is stored in the persistent data storage 706. Thepersistent data storage 706 may provide and / or access one or more of the continuousmeasurements and / or surveys and other data. The apparatus 700 may have the self-learning controller 712 and wellbore trajectory model 714 to perform real timeautonomous calibration of wellbore trajectory models for use in directional drilling basedon the SLL and FLL. The network interface 710 may send control input to actuatorsassociated with steering the drill bit based on the calibration process to reach a target inthe geological formation.

[0054] Further, the apparatus 700 may comprise a display 716. The display 716 maycomprise a computer screen or other visual device. The display 716 may indicatewhether a target can be reached and / or a trajectory of the wellbore. In the case that theapparatus 700 is located downhole, the display 716 may not be coupled to the bus 708,and instead the network interface 710 may be used to provide the feasibility informationto the display located on the surface. Additionally, the display 716 may convey alerts718. The formation processing module 712 may generate the alerts 718 relating whetherthe target is reachable. The alerts 718 may be visual in nature but they may alsocomprise audible alerts output by an audio output device (e.g., speaker).

[0055] The flowcharts are provided to aid in understanding the illustrations and arenot to be used to limit scope of the claims. The flowcharts depict example operations thatcan vary within the scope of the claims. Additional operations may be performed; feweroperations may be performed; the operations may be performed in parallel; and the14Docket Number: 164.2018-IPM-102337 V1 USoperations may be performed in a different order. It will be understood that each block ofthe flowchart illustrations and / or block diagrams, and combinations of blocks in theflowchart illustrations and / or block diagrams, can be implemented by program code. Theprogram code may be provided to a processor of a general purpose computer, specialpurpose computer, or other programmable machine or apparatus.

[0056] As will be appreciated, aspects of the disclosure may be embodied as asystem, method or program code / instructions stored in one or more machine-readablemedia. Accordingly, aspects may take the form of hardware, software (includingfirmware, resident software, micro-code, etc.), or a combination of software andhardware aspects that may all generally be referred to herein as a “circuit,” “module” or“system.” The functionality presented as individual modules / units in the exampleillustrations can be organized differently in accordance with any one of platform(operating system and / or hardware), application ecosystem, interfaces, programmerpreferences, programming language, administrator preferences, etc.

[0057] Any combination of one or more machine readable medium(s) may beutilized. The machine readable medium may be a machine readable signal medium or amachine readable storage medium. A machine readable storage medium may be, forexample, but not limited to, a system, apparatus, or device, that employs any one of orcombination of electronic, magnetic, optical, electromagnetic, infrared, or semiconductortechnology to store program code. More specific examples (a non-exhaustive list) of themachine readable storage medium would include the following: a portable computerdiskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), anerasable programmable read-only memory (EPROM or Flash memory), a portablecompact disc read-only memory (CD-ROM), an optical storage device, a magneticstorage device, or any suitable combination of the foregoing. In the context of thisdocument, a machine readable storage medium may be any tangible medium that cancontain, or store a program for use by or in connection with an instruction executionsystem, apparatus, or device. A machine readable storage medium is not a machinereadable signal medium.15Docket Number: 164.2018-IPM-102337 V1 US

[0058] A machine readable signal medium may include a propagated data signal withmachine readable program code embodied therein, for example, in baseband or as part ofa carrier wave. Such a propagated signal may take any of a variety of forms, including,but not limited to, electro-magnetic, optical, or any suitable combination thereof. Amachine readable signal medium may be any machine readable medium that is not amachine readable storage medium and that can communicate, propagate, or transport aprogram for use by or in connection with an instruction execution system, apparatus, ordevice.

[0059] Program code embodied on a machine readable medium may be transmittedusing any appropriate medium, including but not limited to wireless, wireline, opticalfiber cable, RF, etc., or any suitable combination of the foregoing.

[0060] Computer program code for carrying out operations for aspects of thedisclosure may be written in any combination of one or more programming languages,including an object oriented programming language such as the Java® programminglanguage, C++ or the like; a dynamic programming language such as Python; a scriptinglanguage such as Perl programming language or PowerShell script language; andconventional procedural programming languages, such as the "C" programming languageor similar programming languages. The program code may execute entirely on a stand-alone machine, may execute in a distributed manner across multiple machines, and mayexecute on one machine while providing results and or accepting input on anothermachine.

[0061] The program code / instructions may also be stored in a machine readablemedium that can direct a machine to function in a particular manner, such that theinstructions stored in the machine readable medium produce an article of manufactureincluding instructions which implement the function / act specified in the flowchart and / orblock diagram block or blocks.

[0062] While the aspects of the disclosure are described with reference to variousimplementations and exploitations, it will be understood that these aspects are illustrativeIn general, techniques for real time autonomous calibration of wellbore trajectory models16Docket Number: 164.2018-IPM-102337 V1 USfor use in directional drilling as described herein may be implemented with facilitiesconsistent with any hardware system or hardware systems. Many variations,modifications, additions, and improvements are possible.

[0063] Plural instances may be provided for components, operations or structuresdescribed herein as a single instance. Finally, boundaries between various components,operations and data stores are somewhat arbitrary, and particular operations are illustratedin the context of specific illustrative configurations. Other allocations of functionality areenvisioned and may fall within the scope of the disclosure. In general, structures andfunctionality presented as separate components in the example configurations may beimplemented as a combined structure or component. Similarly, structures andfunctionality presented as a single component may be implemented as separatecomponents. These and other variations, modifications, additions, and improvementsmay fall within the scope of the disclosure.

[0064] As used herein, the term “or” is inclusive unless otherwise explicitly noted.Thus, the phrase “at least one of A, B, or C” is satisfied by any element from the set {A,B, C} or any combination thereof, including multiples of any element.17Docket Number: 164.2018-IPM-102337 V1 USREPRESENTATIVE CLAIMS:1. A method comprising:positioning one more sensors downhole in a geological formation;performing surveys and continuous measurements of the geological formationbased on the one or more sensors;determining a first wellbore trajectory based on the surveys, a second wellboretrajectory based on the continuous measurements, and a predicted wellbore trajectorybased on a wellbore trajectory model;determining errors between the predicted wellbore trajectory and the firstwellbore trajectory and second wellbore trajectory; andcalibrating the wellbore trajectory model based on the errors.2. The method of claim 1, further comprising drilling a wellbore in the geologicalformation based on the wellbore trajectory model.3. The method of claim 1, wherein determining the second wellbore trajectorycomprises filtering or weighting the survey and continuous measurements.4. The method of claim 1, wherein calibrating the wellbore trajectory modelcomprises determining a product of a model parameter and a control input andadding a bias parameter to the product.5. The method of claim 4, wherein the control input is based on a steering ratio, toolface, and the errors.6. A system comprising:sensors for performing surveys and continuous measurements of thegeological formation;a wellbore trajectory model;a processor; and18Docket Number: 164.2018-IPM-102337 V1 USa machine-readable medium having instructions stored thereon that areexecutable by the processor to:determine a first wellbore trajectory based on the surveys, a secondwellbore trajectory based on the continuous measurements, and a predictedwellbore trajectory based on the wellbore trajectory model;determine errors between the predicted wellbore trajectory and the firstwellbore trajectory and second wellbore trajectory; andcalibrate the wellbore trajectory model based on the errors.7. The system of claim 6, further comprising instructions stored on the machine-readable medium and executable by the processor for drilling a wellbore in thegeological formation based on the wellbore trajectory model.8. The system of claim 6, wherein the instructions to determine the second wellboretrajectory comprises instructions to filter or weight the survey and continuousmeasurements.9. The system of claim 6, wherein the instructions to calibrate the wellbore trajectorymodel comprises instructions to determine a product of a model parameter and acontrol input and adding a bias parameter to the product.10. The system of claim 9, wherein the control input is based on a steering ratio, toolface, and the errors.191 / 72 / 7

Claims

REPRESENTATIVE CLAIMS:

1. A method comprising:positioning one more sensors downhole in a geological formation; performing surveys and continuous measurements of the geological formation based on the one or more sensors;determining a first wellbore trajectory based on the surveys, a second wellbore trajectory based on the continuous measurements, and a predicted wellbore trajectory based on a wellbore trajectory model;determining errors between the predicted wellbore trajectory and the first wellbore trajectory and second wellbore trajectory; andcalibrating the wellbore trajectory model based on the errors.

2. The method of claim 1, further comprising drilling a wellbore in the geological formation based on the wellbore trajectory model.

3. The method of claim 1, wherein determining the second wellbore trajectory comprises filtering or weighting the survey and continuous measurements.4.The method of claim 1, wherein calibrating the wellbore trajectory model comprises determining a product of a model parameter and a control input and adding a bias parameter to the product.

5. The method of claim 4, wherein the control input is based on a steering ratio, tool face, and the errors.

6. A system comprising:sensors for performing surveys and continuous measurements of the geological formation;a wellbore trajectory model;a processor; and a machine-readable medium having instructions stored thereon that are executable by the processor to:determine a first wellbore trajectory based on the surveys, a second wellbore trajectory based on the continuous measurements, and a predicted wellbore trajectory based on the wellbore trajectory model;determine errors between the predicted wellbore trajectory and the first wellbore trajectory and second wellbore trajectory; andcalibrate the wellbore trajectory model based on the errors.7.The system of claim 6, further comprising instructions stored on the machinereadable medium and executable by the processor for drilling a wellbore in the geological formation based on the wellbore trajectory model.

8. The system of claim 6, wherein the instructions to determine the second wellbore trajectory comprises instructions to filter or weight the survey and continuous measurements.

9. The system of claim 6, wherein the instructions to calibrate the wellbore trajectory model comprises instructions to determine a product of a model parameter and a control input and adding a bias parameter to the product.

10. The system of claim 9, wherein the control input is based on a steering ratio, tool face, and the errors.