Field operation framework

Abstract

A method may include receiving log data of different types of log records; identifying a portion of the log data corresponding to the formation type; defining a combination of the portion of log data corresponding to the formation type; implementing a machine learning model that generates scores for the combination, where each of the scores indicates the ability of each of the combinations to predict one or more target log records therein as selected from different types of log records; and on the basis of the ranking of the scores, outputting the combination with the highest ranking corresponding to the stratum type in the combinations.

Description

Related Applications

[0001] This application claims priority to and the benefit of U.S. Provisional Application Serial No. 63 / 471,050, filed June 5, 2023, which is incorporated by reference herein in its entirety. BACKGROUND

[0002] A reservoir can be a subterranean formation that can be at least partially characterized by its porosity and fluid permeability. As an example, a reservoir can be a portion of a basin, such as a sedimentary basin. A basin can be a depression in which sediments accumulate (e.g., caused by plate tectonic activity, subsidence, etc.). As an example, a petroleum system can develop within a basin in the case of a source rock in combination with appropriate burial depth and duration of burial, which can form a reservoir that includes hydrocarbon fluids (e.g., oil, natural gas, etc.). Various operations can be performed at a site to access such hydrocarbon fluids and / or to produce such hydrocarbon fluids. For example, consider an equipment operation in which equipment can be controlled to perform one or more operations (e.g., logging, drilling, etc.). In such an example, control can be based at least in part on properties of rock in which a wellbore is drilled, which can be completed to form a well for producing fluids from and / or injecting fluids into a reservoir. Although hydrocarbon fluid reservoirs are mentioned as examples, reservoirs including water and brine can be evaluated, for example, for one or more purposes such as, for example, carbon storage (e.g., sequestration), water production or storage, geothermal production or storage, extraction of metals from brine, etc. SUMMARY

[0003] A method can include receiving well log data of different types of well logs, identifying a portion of the well log data corresponding to a formation type, defining combinations of the portion of the well log data corresponding to the formation type, implementing a machine learning model that generates scores for the combinations, where each of the scores indicates an ability of each of the combinations to predict one or more target well logs selected from the different types of well logs therein, and based on a ranking of the scores, outputting a highest ranked one of the combinations corresponding to the formation type. A system can include one or more processors, a memory accessible to at least one of the processors, processor-executable instructions stored in the memory and executable to instruct the system to receive well log data of different types of well logs, identify a portion of the well log data corresponding to a formation type, define combinations of the portion of the well log data corresponding to the formation type, implement a machine learning model that generates scores for the combinations, where each of the scores indicates an ability of each of the combinations to predict one or more target well logs selected from the different types of well logs therein, and based on a ranking of the scores, output a highest ranked one of the combinations corresponding to the formation type. One or more computer-readable storage media can include processor-executable instructions for instructing a computing system to receive well log data of different types of well logs, identify a portion of the well log data corresponding to a formation type, define combinations of the portion of the well log data corresponding to the formation type, implement a machine learning model that generates scores for the combinations, where each of the scores indicates an ability of each of the combinations to predict one or more target well logs selected from the different types of well logs therein, and based on a ranking of the scores, output a highest ranked one of the combinations corresponding to the formation type. Various other apparatuses, systems, methods, etc. are also disclosed.

[0004] This summary is provided to introduce a selection of concepts, further described below in the detailed description. This summary does not identify key or essential features of the claimed subject matter and is not intended as an assistance in limiting the scope of claimed subject matter. BRIEF DESCRIPTION OF DRAWINGS

[0005] The features and advantages described can be better understood with reference to the following description together with the accompanying drawings.

[0006] Figure 1 An example system is shown that includes various framework components associated with one or more geologic environments; Figure 2 An example of a system is shown; Figure 3Examples of drilling equipment and examples of wellbore shapes are shown; Figure 4 Examples of systems are shown; Figure 5 Examples of a series of well logs are shown; Figure 6 Examples of a series of well logs are shown; Figure 7 Examples of workflows are shown; Figure 8 Examples of graphical user interfaces (GUIs) including examples of a series of well logs are shown; Figure 9 Examples of graphical user interfaces (GUIs) including examples of well log types evaluated by a model are shown; Figure 10 Examples of frameworks are shown; Figure 11 Examples of methods are illustrated; Figure 12 Examples of methods and examples of systems are shown; and Figure 13 Examples of computers and network equipment are shown. DETAILED DESCRIPTION

[0007] This description is not limiting, but is made to describe the general principles of implementing the described implementations. The scope of the described implementations is to be determined by the claims published.

[0008] Figure 1 Examples of systems 100 including a workspace framework 110 are shown, which can provide instantiation, rendering, interaction with, and the like, of graphical user interfaces (GUIs) 120. In Figure 1 Examples of GUIs 120 can include graphical controls for computing frameworks (e.g., applications) 121, projects 122, visualizations 123, one or more other features 124, data access 125, and data storage 126, in examples.

[0009] In Figure 1In the example of a particular geologic environment 150, the work space framework 110 can be customized for that particular geologic environment. For example, the geologic environment 150 can include multiple layers (e.g., stratification) including a reservoir 151 and can intersect a fault 153. The geologic environment 150 can be equipped with a variety of sensors, detectors, actuators, etc. In such an environment, various types of equipment, such as, for example, equipment 152 can include communication circuitry for optionally receiving and transmitting information with respect to one or more networks 155. Such information can include information associated with downhole equipment 154, which can be equipment for acquiring information, assisting resource recovery, etc. Other equipment 156 can be located at a location remote from a well site and include sensing, detecting, transmitting, or other circuitry. Such equipment can include storage and communication circuitry for storing and communicating data, instructions, etc. One or more satellites can be provided for communication, data acquisition, etc. purposes. For example, Figure 1 A satellite 170 is shown in communication with a network 155 that can be configured for communication, it is noted that a satellite can additionally or alternatively include circuitry for imaging (e.g., spatial imaging, spectral imaging, temporal imaging, radiation imaging, etc.).

[0010] Figure 1 The geologic environment 150 is also shown as optionally including equipment 157 and 158 associated with a well that includes a substantially horizontal portion that can intersect one or more fractures 159. For example, consider a well in a formation that can include natural fractures, artificial fractures (e.g., hydraulic fractures), or a combination of natural and artificial fractures. As an example, a laterally extending reservoir can be drilled. In such an example, there can be lateral variations in properties, stresses, etc., where assessment of such variations can assist in planning, operation, etc. to develop the laterally extending reservoir (e.g., via fracturing, injection, extraction, etc.). As an example, equipment 157 and / or 158 can include components, a system, multiple systems, etc. for fracturing, seismic sensing, seismic data analysis, assessing one or more fractures, etc.

[0011] In Figure 1 In the example of SLB, the GUI 120 shows some examples of computing frameworks, including DRILLPLAN, PETREL, TECHLOG, PETROMOD, ECLIPSE, and INTERSECT frameworks (SLB, Houston, Texas).

[0012] The DRILLPLAN framework is used for digital well construction planning and includes multiple features for automating repetitive tasks and validation workflows, enabling the rapid generation of quality-improved drilling programs (e.g., digital drilling plans, etc.) with consistency ensured.

[0013] The PETREL framework can be part of a DELFI cognitive exploration and production (E&P) environment (SLB, Houston, Texas, referred to as the DELFI environment) used in, for example, geoscience and geophysics to analyze subsurface data from fluids in reservoirs from exploration to production.

[0014] One or more types of frameworks can be implemented within or in operative coupling with a DELFI environment, which is a secure, cognitive, cloud-based collaboration environment that integrates data and workflows with digital technologies such as artificial intelligence (AI) and machine learning (ML). Such an environment can provide for operations involving one or more frameworks. The DELFI environment can be referred to as a DELFI framework, which can be one of a plurality of frameworks. The DELFI environment can include various other frameworks that can operate using one or more types of models (e.g., simulation models, etc.).

[0015] The TECHLOG framework can handle and process field and laboratory data for a variety of geological environments (e.g., deepwater exploration, shale, etc.). The TECHLOG framework can structure wellbore data for analysis, planning, etc.

[0016] The PIPESIM simulator includes a solver that can provide simulation results such as, for example, multiphase flow results (e.g., from reservoir to wellhead and beyond), tubing and surface facility performance, etc. The PIPESIM simulator can be integrated with, for example, the AVOCET production operations framework (SLB, Houston, Texas). The PIPESIM simulator can be an optimizer that can optimize one or more operational scenarios at least in part via simulation of physical phenomena.

[0017] The ECLIPSE framework provides a numerical solver for reservoir simulators to predict the dynamic behavior of various types of reservoirs and development scenarios.

[0018] The INTERSECT framework provides a high-resolution reservoir simulator for simulating geologic features and quantifying uncertainties, e.g., by creating production scenarios, and, with an integrated precise model of surface facilities and field operations, the INTERSECT framework can produce results that can be continuously updated through real-time data exchange (e.g., from the field, such as from one or more types of data acquisition equipment that can acquire data during one or more types of field operations, etc.). The INTERSECT framework can provide completion configurations for complex wells, where such configurations can be built at the field; detailed chemical enhanced oil recovery (EOR) recipes can be provided, where such recipes can be implemented at the field; applications of steam injection and other thermal EOR techniques can be analyzed for advanced production control in terms of reservoir coupling and flexible field management, as well as flexibility to write custom solutions for improved modeling and field management control. As with other example frameworks, the INTERSECT framework can be used as part of a DELFI environment, e.g., for rapid simulation of multiple concurrent cases.

[0019] The foregoing DELFI environment provides a workflow with various features for subsurface analysis, planning, building, and production, e.g., as shown in the workspace framework 110. As shown, output from the workspace framework 110 can be used to guide, control, etc., one or more processes in the geologic environment 150, and feedback 160 (e.g., acquired data regarding operating conditions, equipment conditions, environmental conditions, etc.) can be received via one or more interfaces in one or more forms. Figure 1

[0020] In examples, the visualization feature 123 can be implemented via the workspace framework 110, e.g., to perform tasks associated with one or more of a subsurface region, planning operations, building wells, and / or surface fluid networks, and production from reservoirs. Figure 1

[0021] The visualization feature can provide visualization of various earth models, properties, etc., in one or more dimensions. As an example, the visualization feature can include one or more control features for controlling equipment, which can include, e.g., field equipment that can perform one or more field operations. The workflow can utilize one or more frameworks to generate information that can be used to control one or more types of field equipment (e.g., drilling equipment, wireline logging equipment, fracturing equipment, etc.).

[0022] ​​With respect to reservoir models that can be suitable for use by a simulator, in view of acquisition of seismic data, such as via reflection seismology, the seismic data can be used in geophysics, e.g., to estimate properties of subsurface formations. The seismic data can be processed and interpreted, e.g., to better understand the composition, fluid content, extent, and geometry of subsurface rock. Such interpretation results can be used to plan, simulate, perform, etc., one or more operations for producing fluids from a reservoir (e.g., reservoir rock, etc.). Field acquisition equipment can be used to acquire seismic data, which can be in the form of traces, where a trace can include values organized with respect to time and / or depth (e.g., consider 1D, 2D, 3D, or 4D seismic data).

[0023] A model can be a simulated version of a geological environment, where a simulator can include a plurality of features for simulating physical phenomena in a geological environment based at least in part on a model or models. A simulator, such as a reservoir simulator, can simulate fluid flow in a geological environment based at least in part on a model that can be generated via a framework of receiving seismic data. A simulator can be a computerized system (e.g., a computing system) that can use one or more processors to execute instructions to solve a system of equations that describe physical phenomena subject to various constraints. While a number of simulators are shown in the example of FIG. 1, one or more other simulators can additionally or alternatively be used. Figure 1 A number of simulators are shown in the example of FIG. 1, but one or more other simulators can additionally or alternatively be used.

[0024] Figure 2 An example of a system 200 that can be operatively coupled to one or more databases, data streams, etc., is shown. For example, one or more field equipment, laboratory equipment, computing equipment (e.g., local and / or remote computing equipment), etc., can provide and / or generate data that can be used in system 200.

[0025] As shown, system 200 can include a geology / geophysics data block 210, a surface model block 220 (e.g., for one or more structural models), a volume module block 230, an application block 240, a numerical processing block 250, and an operational decision block 260. As shown, system 200 can include a reservoir model block 270, a simulation block 280, and a production block 290. Figure 2As shown in the example, geological / geophysical data block 210 may include data 212 from the well top or borehole, data 214 from seismic interpretation, data from outcrop interpretation, and optionally data from geological knowledge. As an example, geological / geophysical data block 210 may include data from digital images, which may include digital images of cores, cuttings, caves, outcrops, etc. Regarding ground model block 220, it can provide the ability to create, edit, etc., one or more ground models based on, for example, fault planes 222, ground planes 224, and optionally topological relationships 226. Regarding volumetric model block 230, it can provide the ability to create, edit, etc., one or more volumetric models based on, for example, boundary representations 232 (e.g., for forming watertight models), structured meshes 234, and unstructured meshes 236.

[0026] like Figure 2 As shown in the example, system 200 can allow the implementation of one or more workflows, such as where data from data block 210 is used to create, edit, or otherwise generate one or more ground models of ground model block 220, which can then be used to create, edit, or otherwise generate one or more volumetric models of volumetric model block 230. Figure 2 As indicated by the example, ground model block 220 can provide one or more structural models that can be input into application block 240. For example, such a structural model can be provided to one or more applications, optionally without performing one or more processes of volume model block 230 (e.g., for the purpose of numerical processing via numerical processing block 250). Thus, system 200 can be adapted to one or more workflows for structural modeling (e.g., optionally without performing numerical processing according to numerical processing block 250).

[0027] Regarding application block 240, it may include multiple applications such as well prediction application 242, reserve calculation application 244, and well stability assessment application 246. Regarding numerical processing block 250, it may include a seismic velocity modeling process 251 and subsequent seismic processing 252, a facies and petrophysical property interpolation process 253 and subsequent flow simulation 254, and a geomechanical simulation process 255 and subsequent geochemical simulation 256. As indicated, as an example, the workflow may proceed from volumetric model block 230 to numerical processing block 250, and then to application block 240 and / or operational decision block 260. As another example, the workflow may proceed from surface model block 220 to application block 240, and then to operational decision block 260 (e.g., consider an application using a structural model for operation).

[0028] exist Figure 2In the example of FIG. 2, the operational decision block 260 can include a seismic survey design process 261, a well rate adjustment process 252, a well trajectory planning process 263, a well completion planning process 264, and processes for one or more prospect areas, e.g., to decide whether to explore, develop, abandon, etc. the prospect area.

[0029] Referring again to the data block 210, the well top or borehole data 212 can include spatial positioning and optionally surface dip of an interface between two geological formations or a subsurface discontinuity such as a geological fault; the seismic interpretation data 214 can include a collection of points, lines, or surface patches interpreted from seismic reflection data and representing an interface between media (e.g., geological formations where seismic wave velocity is different) or subsurface discontinuities; the outcrop interpretation data 216 can include a collection of lines or points optionally associated with measured dips representing a boundary between geological formations or geological faults as interpreted on the surface of the Earth; and the geological knowledge data 218 can include knowledge of, e.g., the paleostructure and depositional evolution of the area.

[0030] Regarding the structural model, it can be, e.g., a collection of grid-like or mesh-like surfaces representing one or more interfaces between geological formations (e.g., horizons) or subsurface mechanical discontinuities (fault planes). As an example, the structural model can include some information about the topological relationships between the surfaces (e.g., fault A truncates fault B, fault B intersects fault C, etc.).

[0031] Regarding the facies and rock physics property interpolation 253, it can include an assessment of rock types and their rock physics properties (e.g., porosity, permeability), e.g., optionally in areas not sampled by well logs or coring. As an example, such interpolation can be constrained by interpretations from well logs and core data and prior geological knowledge.

[0032] Regarding the various applications of the application block 240, the well prognosis application 242 can include predicting the type and properties of geological formations that a drill bit can encounter and the locations where such rocks can be encountered (e.g., prior to conducting drilling); the reserves calculation application 244 can include assessing the total amount of hydrocarbons or mineral material present in a subsurface environment (e.g., and estimating the fraction that can be recoverable given a set of economic and technical constraints); and the well stability assessment application 246 can include estimating the risk of a drilled well or a well to be drilled collapsing or being damaged due to subsurface stresses.

[0033] With respect to the operational decision block 260, a seismic survey design process 261 can include deciding where to place seismic sources and receivers to optimize the coverage and quality of collected seismic information while minimizing acquisition costs; a well rate adjustment process 262 can include controlling injection and production well scheduling and rates (e.g., to maximize recovery and production); a well trajectory planning process 263 can include designing well trajectories to maximize potential recovery and production while minimizing drilling risks and costs; a well trajectory planning process 264 can include selecting appropriate well tubulars, casings, and completions (e.g., to meet expected production or injection targets in specified reservoir formations); and a prospect process 265 can include making decisions in the context of exploration to continue exploration, start production, or abandon a prospect (e.g., based on a comprehensive assessment of technical and financial risks versus expected returns).

[0034] The system 200 can include and / or be operatively coupled to systems such as the system 100 of Figure 1 The workspace framework 110 can provide instantiation, rendering, interaction with, etc., of the graphical user interface (GUI) 120 to perform one or more actions with respect to the system 200, for example. In such examples, access to one or more frameworks (e.g., DRILLPLAN, PETREL, TECHLOG, PIPESIM, ECLIPSE, INTERSECT, etc.) can be provided. The one or more frameworks can provide for geophysical data acquisition as in block 210, structural modeling as in block 220, volumetric modeling as in block 230, running applications as in block 240, numerical processing as in block 250, making operational decisions as in block 260, etc.

[0035] As an example, the system 200 can provide for monitoring data, which can include geophysical data for each geophysical data block 210. In various examples, the geophysical data can be acquired during one or more operations. For example, consider acquiring geophysical data via downhole equipment and / or surface equipment during a drilling operation. As an example, the operational decision block 260 can include the ability to monitor, analyze, etc., such data for the purpose of making one or more operational decisions, which can include controlling equipment, modifying operations, modifying plans, etc. In such examples, data can be fed into the system 200 at one or more points at which data quality can be of particular interest. For example, data quality can be characterized by one or more metrics, where data quality can provide an indication as to trust, probability, etc., which can be closely related to making operational decisions and / or making other decisions.

[0036] Figure 3An example of a wellsite system 300 (e.g., at a wellsite that can be located on land or offshore) is shown. As shown, the wellsite system 300 can include a mud tank 301 for storing mud and other materials (e.g., where the mud can be a drilling fluid), a suction line 303 that acts as an inlet for a mud pump 304 to pump mud from the mud tank 301 so that the mud flows to a vibrating hose 306, a drawworks 307 for hoisting one or more drilling wires 312, a riser 308 that receives mud from the vibrating hose 306, a kelly hose 309 that receives mud from the riser 308, one or more goosenecks 310, a traveling block 311, a crown block 313 for carrying the traveling block 311 via the one or more drilling wires 312, a derrick 314, a kelly 318 or top drive 340, a kelly drive bushing 319, a rotary table 320, a rig floor 321, a bell nipple 322, one or more blowout preventers (BOPs) 323, a drill string 325, a drill bit 326, a casing head 327, and a flow line 328 that carries mud and other materials to, for example, the mud tank 301.

[0037] In Figure 3 example systems, a wellbore 332 is formed in a subterranean formation 330 by rotary drilling; it should be noted that various example embodiments can also use one or more directional drilling techniques, equipment, etc.

[0038] As Figure 3 shown in the example, the drill string 325 is suspended within the wellbore 332 and has a drill string assembly 350 that includes the drill bit 326 at a lower end thereof. As an example, the drill string assembly 350 can be a bottom hole assembly (BHA).

[0039] The wellsite system 300 can provide for operation of the drill string 325 and other operations. As shown, the wellsite system 300 includes the traveling block 311 and the derrick 314 positioned above the wellbore 332. As mentioned, the wellsite system 300 can include the rotary table 320, with the drill string 325 passing through an opening in the rotary table 320.

[0040] As Figure 3As shown by way of example, the wellsite system 300 can include kelly 318 and associated components, or top drive 340 and associated components. With respect to the kelly example, the kelly 318 can be a square or hexagonal metal / alloy rod with holes drilled therein that act as a flow path for mud. The kelly 318 can be used to transfer rotational motion from the rotary table 320 to the drill string 325 via a kelly drive bushing 319 while allowing the drill string 325 to be lowered or raised during rotation. The kelly 318 can pass through the kelly drive bushing 319, which can be driven by the rotary table 320. As an example, the rotary table 320 can include a square bushing that is operatively coupled to the kelly drive bushing 319 such that rotation of the rotary table 320 can turn the kelly drive bushing 319 and thus the kelly 318. The kelly drive bushing 319 can include an interior profile that matches an exterior profile of the kelly 318 (e.g., square, hexagonal, etc.); however, the kelly drive bushing has slightly larger dimensions so that the kelly 318 can move freely up and down within the kelly drive bushing 319.

[0041] With respect to the top drive example, the top drive 340 can provide the functionality performed by the kelly and rotary table. The top drive 340 can turn the drill string 325. As an example, the top drive 340 can include one or more motors (e.g., electric and / or hydraulic motors) that are connected with appropriate gearing to a short pipe section known as a hollow shaft, which in turn can be screwed into a saver sub or the drill string 325 itself. The top drive 340 can be suspended from the traveling block 311, so the rotary mechanism can travel freely up and down along the derrick 314. As an example, the top drive 340 can allow drilling to be performed with more stand of pipe than the kelly / rotary table approach.

[0042] In Figure 3 As an example, the mud tank 301 can store mud, which can be one or more types of drilling fluid. As an example, a wellbore can be drilled to produce fluid, inject fluid, or both (e.g., hydrocarbons, minerals, water, etc.).

[0043] In Figure 3In the example of FIG. 3, drill string 325 (e.g., including one or more downhole tools) can be comprised of a series of pipes that are threadably connected together to form a long pipe with a drill bit 326 at a lower end. As drill string 325 is run into the wellbore for drilling, at some point before or coincident with drilling, mud can be pumped by pump 304 from mud tank 301 (e.g., or other source) via lines 306, 308, and 309 to a port of kelly 318, or for example, to a port of top drive 340. The mud can then flow via a passage (e.g., or passages) in drill string 325 and out a port located on drill bit 326 (see, e.g., directional arrows). As the mud exits drill string 325 via the port in drill bit 326, it can then circulate up through an annular region (e.g., open hole, casing, etc.) between an outer surface of drill string 325 and a surrounding wall, as indicated by directional arrows. In this manner, the mud lubricates drill bit 326 and carries thermal energy (e.g., frictional energy or other energy) and formation cuttings to the surface, where the mud can be returned to mud tank 301 after being processed to remove cuttings and other material, for example, for recirculation.

[0044] In Figure 3 In the example of FIG. 3, the processed mud pumped by pump 304 into drill string 325 can form a mud cake lining the wellbore after exiting drill string 325, which can reduce friction between drill string 325 and the surrounding wall (e.g., hole, casing, etc.), among other things. The reduction in friction can facilitate advancement or retraction of drill string 325. During drilling operations, the entire drill string 325 can be pulled out of the wellbore and optionally replaced, for example, with a new or sharp drill bit, a smaller diameter drill string, etc. As mentioned, the act of pulling or replacing the drill string in the hole is referred to as tripping. Depending on the tripping direction, tripping can be referred to as tripping out or tripping out, or tripping in or tripping in.

[0045] As an example, consider tripping in, where after drill bit 326 of drill string 325 reaches the bottom of the wellbore, mud is pumped by pump 304 into a passage of drill string 325 for the purpose of lubricating drill bit 326 for drilling to enlarge the wellbore. As mentioned, the mud can be pumped by pump 304 into the passage of drill string 325, and after filling the passage, the mud can be used as a transmission medium for transmitting energy (e.g., energy that can encode information like in mud pulse telemetry). The properties of the mud can be used to determine how to transmit the pulses (e.g., pulse shape, energy loss, transmission time, etc.).

[0046] As an example, mud pulse telemetry equipment can include downhole devices configured to effectuate changes in mud pressure to produce a sound wave or multiple sound waves based on which information can be modulated. In such examples, information from downhole equipment (e.g., one or more modules of drill string 325) can be transmitted uphole to a surface device, which can relay such information to other equipment for processing, control, etc.

[0047] As an example, telemetry equipment can operate by transmitting energy through drill string 325 itself. For example, consider a signal generator that delivers an encoded energy signal to drill string 325, and a repeater that can receive such energy and relay it to further transmit the encoded energy signal (e.g., information, etc.).

[0048] As an example, drill string 325 can be outfitted with telemetry equipment 352 that includes a rotatable drive shaft, a turbine impeller mechanically coupled to the drive shaft such that mud can cause the turbine impeller to rotate, a modulator rotor mechanically coupled to the drive shaft such that rotation of the turbine impeller causes the modulator rotor to rotate, a modulator stator mounted adjacent or proximate to the modulator rotor such that rotation of the modulator rotor relative to the modulator stator produces pressure pulses in the mud, and an actuator for selectively braking rotation of the modulator rotor to modulate the pressure pulses. In such examples, an alternator can be coupled to the aforementioned drive shaft, where the alternator includes at least one stator winding electrically coupled to control circuitry to selectively short the at least one stator winding to electromagnetically brake the alternator and thereby selectively brake rotation of the modulator rotor to modulate the pressure pulses in the mud.

[0049] In examples of Figure 3 In examples of

[0050] Components 350 of the illustrated example include a logging-while-drilling (LWD) module 354, a measuring-while-drilling (MWD) module 356, an optional module 358, a rotary steerable system (RSS) and / or motor 360, and drill bit 326. Such components or modules can be referred to as tools, where a drill string can include multiple tools. Such components or modules can provide for generation of a well log, which may, for example, include one or more types of well logs.

[0051] With respect to RSS, it relates to techniques for directional drilling. Directional drilling involves drilling into the earth to form a deviated borehole, such that the trajectory of the borehole is not vertical; rather, the trajectory deviates from vertical along one or more portions of the borehole. As an example, consider a target that is located at a lateral distance from a surface location at which a rig can be fixed. In such an example, the drilling can begin from a vertical portion, and then deviate from vertical, such that the borehole is aimed at the target and ultimately reaches the target. Directional drilling can be implemented where the target is not reachable from a vertical position at the surface of the earth, where there are materials on the earth that can impede drilling or otherwise be harmful (e.g., consider salt domes, etc.), where the formation is laterally extensive (e.g., consider a relatively thin but laterally extensive reservoir), where multiple boreholes are to be drilled from a single surface borehole, where a relief well is desired, etc.

[0052] One approach to directional drilling involves a mud motor; however, mud motors can present some challenges depending on factors such as rate of penetration (ROP), weight transfer to the drill bit due to friction (e.g., weight on bit (WOB)), etc. A mud motor can be a positive displacement motor (PDM) that operates to drive a drill bit (e.g., during directional drilling, etc.). A PDM operates as drilling fluid is pumped through it, where the PDM converts the hydraulic power of the drilling fluid into mechanical power to cause the drill bit to rotate.

[0053] As an example, a PDM can operate in a combined rotary mode, where surface equipment is used to rotate the drill bit of the drill string by rotating the entire drill string (e.g., a rotary table, a top drive, etc.), and drilling fluid is used to rotate the drill bit of the drill string. In such an example, surface RPM (SRPM) can be determined by using the surface equipment, and downhole RPM of the mud motor can be determined using various factors related to flow of the drilling fluid, mud motor type, etc. As an example, in combined rotary mode, the drill bit RPM can be determined or estimated as the sum of SRPM and mud motor RPM, assuming that SRPM and mud motor RPM are in the same direction.

[0054] The LWD module 354 can be housed in a suitable type of drill collar and can contain one or more selected types of logging tools. It will also be appreciated that more than one LWD and / or MWD module can be employed. The LWD module can include capabilities for measuring, processing, and storing information, and for communicating with the surface equipment. In the illustrated example, the LWD module 354 can include a seismic measurement device.

[0055] The MWD module 356 can be housed in a suitable type of drill collar and can include one or more devices for measuring characteristics of the drill string 325 and drill bit 326. As an example, the MWD module 356 can include equipment for generating electrical power, for example, to power various components of the drill string 325. As an example, the MWD module 356 can include telemetry equipment 352, for example, in which a turbine impeller can generate electrical power from the flow of mud; it can be appreciated that other power sources and / or battery systems can be employed to power the various components. As an example, the MWD module 356 can include one or more of the following types of measuring devices: a weight-on-bit measuring device, a torque measuring device, a vibration measuring device, a shock measuring device, a stick slip measuring device, a direction measuring device, and an inclination measuring device.

[0056] Figure 3 Some examples of the types of wellbores that can be drilled are also shown. For example, consider a straight deviated wellbore 372, an S-shaped wellbore 374, a highly deviated wellbore 376, and a horizontal wellbore 378.

[0057] Drilling operations can include directional drilling, in which, for example, at least a portion of the well includes a curved axis. For example, consider a radius that defines a curvature, in which the inclination with respect to the vertical can vary until an angle of between about 30 degrees and about 60 degrees is reached, or, for example, an angle of about 90 degrees or possibly greater than about 90 degrees is reached.

[0058] Directional wells can include several shapes, in which each of the shapes can be intended to meet a particular operational requirement. As an example, once information is conveyed to a drilling engineer, the drilling process can be performed based on the information. As an example, the inclination and / or direction can be modified based on information received during the drilling process.

[0059] As explained, the system can be a steerable system and can include equipment for performing methods such as geosteering. The steerable system can include equipment located at the lower portion of the drill string, just above the drill bit, which can mount a bent sub. Above the directional drilling equipment, the drill string can include MWD equipment and / or LWD equipment, which MWD equipment provides real-time or near real-time data of interest (e.g., inclination, direction, pressure, temperature, actual weight on bit, torque stress, etc.). With respect to the latter, the LWD equipment can enable the transmission of various types of data of interest to the surface, including, for example, geologic data (e.g., gamma ray logs, resistivity, density, and sonic logs, etc.).

[0060] The coupling of sensors that provide information about the well trajectory path in real-time or near real-time with, for example, one or more logs that characterize the formation from a geologic point of view can allow for the implementation of geosteering methods. Such methods can include navigating the subsurface environment to follow a desired route to reach a desired target or targets.

[0061] The drill string can include an azimuthal density neutron (ADN) tool for measuring density and porosity, an MWD tool for measuring inclination, azimuth, and shock, a compensated dual resistivity (CDR) tool for measuring resistivity and gamma ray related phenomena, one or more variable gauge stabilizers, one or more bent sub joints, and a geosteering tool, which can include a motor and optionally equipment for measuring and / or responding to one or more of inclination, resistivity, and gamma ray related phenomena.

[0062] Geosteering can include intentional directional control of the wellbore based on downhole geologic logging measurements in a manner intended to keep the directional wellbore within a desired region, zone (e.g., a producing oil layer), etc. Geosteering can include directing the wellbore to keep the wellbore within a particular section of a reservoir, for example, to minimize gas breakthrough and / or water breakthrough, and for example, to maximize economic production from a well that includes the wellbore.

[0063] Referring again to Figure 3 , the wellsite system 300 can include one or more sensors 364 operatively coupled to the control and / or data acquisition system 362. As an example, a sensor or sensors can be at a surface location. As an example, a sensor or sensors can be at a downhole location. As an example, a sensor or sensors can be at one or more remote locations that are more than about one hundred meters from the wellsite system 300.

[0064] The system 300 can include one or more sensors 366 that can sense and / or transmit signals to a fluid conduit, such as a drilling fluid conduit (e.g., a drilling mud conduit). For example, in the system 300, the one or more sensors 366 can be operatively coupled to a mud flowed portion of the riser 308. As an example, a downhole tool can generate a pulse that can travel through mud and be sensed by one or more of the one or more sensors 366. In such an example, the downhole tool can include associated circuitry, such as, for example, encoding circuitry that can encode a signal, for example, to reduce requirements for transmission. Circuitry at the surface can include decoding circuitry for decoding encoded information transmitted at least in part via mud pulse telemetry. Circuitry at the surface can include encoder circuitry and / or decoder circuitry, and circuitry downhole can include encoder circuitry and / or decoder circuitry. As an example, the system 300 can include a transmitter that can generate a signal that can be transmitted downhole via mud (e.g., drilling fluid) as a transmission medium.

[0065] Figure 4An example of environment 401 is shown, which includes an underground portion 403 in which a drilling rig 410 is positioned at a ground location above borehole 420. Figure 4 In the example, various wireline logging service equipment can be operated to perform one or more wireline logging services, including, for example, acquiring data from one or more locations within borehole 420.

[0066] As an example, wireline logging tools and / or wireline logging services can provide data acquisition, data analysis, data-based determination, and data-based decision-making. Some examples of wireline logging data may include gamma rays (GR), spontaneous potential (SP), caliper diameter (CALI), shallow resistivity (LLS and ILD), deep resistivity (LLD and ILD), density (RHOB), neutron porosity (BPHI, TNPH, or NPHI), acoustic wave (DT), photoelectric field (PEF), dielectric constant, and conductivity.

[0067] exist Figure 4 In the example, borehole 420 includes drill rod 422, sleeve shoe 424, cable side entry connector (CSES) 423, wet connector adapter 426, and open-eye section 428. As an example, borehole 420 can be a vertical borehole or an off-center borehole, wherein one or more portions of the borehole can be vertical, and one or more portions of the borehole can be off-center, including substantially horizontal.

[0068] exist Figure 4 In the example, CSES 423 includes a cable clamp 425, a packer sealing assembly 427, and a check valve 429. These components provide insertion of a logging cable 430, which includes a portion 432 extending outside the drill pipe 422 to be inserted into the drill pipe 422, such that at least a portion 434 of the logging cable extends inside the drill pipe 422. Figure 4 In the example, the logging cable 430 extends through the casing shoe 424 and the wet connection adapter 426 and enters the open hole section 428 to reach the logging string 440.

[0069] like Figure 4 As shown in the example, the logging vehicle 450 (e.g., a cable logging service vehicle) can deploy the logging cable 430 under the control of the system 460. Figure 4As shown by way of example, system 460 can include one or more processors 462, a memory 464 operatively coupled to at least one of the one or more processors 462, instructions 466 that can be stored, for example, in the memory 464, and one or more interfaces 468. As an example, system 460 can include one or more processor-readable media including processor-executable instructions that are executable by at least one of the one or more processors 462 to cause system 460 to control one or more aspects of equipment of the well column 440 and / or the logging truck 450. In such examples, the memory 464 can be or can include the one or more processor-readable media, where the processor-executable instructions can be or can include the instructions. As an example, the processor-readable media can be a computer- readable storage medium that is not a signal and is not a carrier wave.

[0070] Figure 4 Also shown is a battery 470, which can be operatively coupled to system 460, for example, to power system 460. As an example, battery 470 can be a backup battery that operates when another power source is not available to power system 460 (e.g., via a generator of the wireline logging truck 450, a separate generator, a power line, etc.). As an example, battery 470 can be operatively coupled to a network, which can be a cloud network. As an example, battery 470 can include smart battery circuitry and can be operatively coupled to one or more pieces of equipment via an SMBus or other type of bus.

[0071] As an example, system 460 can be operatively coupled to a client tier 480. In Figure 4 As an example, client tier 480 can include features that allow access and interaction via one or more private networks 482, one or more mobile platforms and / or mobile networks 484, and via a “cloud” 486, which can be considered to include distributed equipment forming a network, such as a network of multiple networks. As an example, system 460 can include circuitry to establish multiple connections (e.g., sessions). As an example, the connections can be via one or more types of networks. As an example, the connections can be client-server type connections, where system 460 operates as a server in a client-server architecture. For example, clients can log into system 460, where multiple clients can optionally be handled simultaneously.

[0072] Although Figure 4Examples show the system 460 associated with the logging truck 450, but one or more features of the system 460 can be included in a downhole assembly, which can be a wireline assembly and / or a LWD assembly. In such a manner, various calculations can be performed downhole, where their results can be optionally transmitted to the surface (e.g., to the logging truck 450, etc.) using one or more telemetry techniques and / or technologies (e.g., mud pulse telemetry, wireline logging, etc.).

[0073] As an example, the tool can include one or more features of an ORA platform (SLB, Houston, Texas). The ORA platform includes various tool options, which include metrology options (e.g., various types of sensors that can be disposed in a sensor array, etc.). For example, consider a tool that includes a fluid in situ scanner that can measure one or more of density and viscosity, resistivity, and full spectrum viscosity. As an example, the tool can include one or more pressure sensors (e.g., quartz pressure sensors, etc.) and / or one or more temperature sensors. As an example, the tool can include one or more sensors to measure volume fractions, composition, color, etc. of oil, water, and gas. With respect to composition sensing, consider sensing C1 through C6 or C 6+ (e.g., with less than about 6 weight percent uncertainty), and, for example, sensing CO2. With respect to fluid density, consider a range from about 0.01 to 2.0 g / cm 3 With respect to fluid viscosity, consider a range from 0.1 to 300 cP. With respect to color, consider optical density as a measurement result. With respect to optical measurements, for example, the tool can include a spectrophotometer, fluorometer, etc.

[0074] Figure 5 An example is shown of a series of well logs 500 as acquired during a drilling operation. In Figure 5In the example of the log records 500, the log records include: a depth log record (e.g., measured depth), which can have a span of 0 ft to 4,500 ft; a bit position (BPOS) log record, which can have a span of 110 ft to 0 ft; a hook load (HKLD) log record, which can have a span of 250 klbf to 0 klbf; a standpipe pressure (SPPA) log record, which can have a span of 0 psi to 4,000 psi; an RPM log record, which can have a span of 0 c / min to 90 c / min; a rig state log record with connections and run time, which can have a span of 0 min to 16 min; a surface torque (TQA) log record, which can have a span of 0 kft.lbf to 21,821.02 kft.lbf; and a surface weight on bit (SWOB) log record, which can have a span of 0 klbf to 50 klbf. Such log records can be acquired for various sections of a well, which can intersect and / or be disposed within one or more formation types. With respect to log records associated with drilling, consider one or more of the following: a bit depth (DBTM), a WOB, a measured depth (MD or DMEA), a flow rate of mud in the borehole (FLWI), an RPM, a surface torque (STOR), a standpipe pressure (SPPA), a hook load (HKLD), a bit position (BPOS), a bit size (BS), a hole size, etc. As an example, depth can be provided in one or more ways, e.g., with respect to a casing, an open hole, a measured depth (MD), a true vertical depth, a bit true measured depth (DBTM), a hole- eye true measured depth (DMEA), etc.

[0075] As an example, a formation type can be characterized by rock physics properties. As an example, the log records 500 can be related to a drill bit interaction with a formation, and, for example, to a drill string interaction with a formation (e.g., accounting for friction between the drill string and the wellbore). During drilling, a drill bit can break rock of a formation, where an interaction between the drill bit and the rock can be characterized by a physical parameter such as, for example, torque, which can be measured at the surface (e.g., at a rig) and / or downhole (e.g., by one or more downhole sensors).

[0076] As an example, a metric known as mechanical specific energy (MSE) can be determined, which is the energy required to remove a unit volume of rock. To achieve optimal drilling efficiency, a goal can be to minimize MSE and maximize rate of penetration (ROP). To control MSE, a driller can control weight on bit (WOB), torque (TQA), ROP, and revolutions per minute of the bit (RPM).

[0077] As an example, drilling operations can be performed, at least in part, using a controller, which can be automated, semi-automated, etc. As an example, automation can be available at one or more levels, where more or less, depending on the level, a human can be in the loop (HITL). As an example, a controller can switch levels from one level to another level depending on feedback, performance, etc.

[0078] As an example, an adaptive drilling system can enable a driller to input a relatively high ROP setpoint, where the system performs adaptive actions, which can reduce the time and effort spent on adjustments, etc. As an example, the system can provide an ROP averaging feature, which can assist in dynamically adjusting ROP setpoints and limits based on one or more factors, which can include, for example, a current well profile, which includes one or more of weight on bit, top drive torque (e.g., surface torque), and differential pressure.

[0079] During drilling, torque and drag (T&D) can refer to the effects of the geometry and other aspects that a wellbore can have on the rotation and the sticking out of a drill string. T&D can vary depending on the drilling mode. Consider, for example, a sliding mode and a rotary mode. In a sliding mode, the drill string can or can not be whipping, and the torque can be low, however, the axial drag can be high, and sticking can occur. Sticking is when a section of the drill string buckles within the wellbore, and can prevent force transmission to the drill bit or BHA. In a rotary mode, the drill string is rotated (e.g., in a single rotational direction) at a rate that tends to reduce drag to a relatively very low level. In a rotary mode, sticking can not necessarily occur, however, the torque can be relatively high.

[0080] Other aspects of T&D can include the maximum drill string weight available to the drill bit, drill string buckling (sticking), coefficient of friction, and the maximum available torque for the drill bit.

[0081] As explained, torque can be a useful measure during drilling. As shown in an example log 500, the log can be a torque log, which can be a surface torque log as related to one or more mechanisms (e.g., top drive, rotary table, etc.). As an example, the torque log can include data that can be related to data in one or more other logs. As such, the torque log can be used to estimate or predict data, behavior, etc., in one or more other parameters. Figure 5

[0082] An example is shown of a series of logs 600 as acquired during logging and / or drilling operations (e.g., consider LWD, etc.). As an example, the logs can include a torque log, which can be a surface torque log as related to one or more mechanisms (e.g., top drive, rotary table, etc.). As an example, the torque log can include data that can be related to data in one or more other logs. As such, the torque log can be used to estimate or predict data, behavior, etc., in one or more other parameters. Figure 6 Figure 6 ​In this example, the log record 600 is shown with respect to measured depth in meters (MD) over a range of approximately 100 m MD and includes a neutron (NEU) log record spanning 0.45 cubic feet to -0.15 cubic feet, a density correction (DENC) log record spanning -0.8 g / cm 3 to 0.2 g / cm 3 , a density (DEN) log record spanning 1.95 g / cm 3 to 2.95 g / cm 3 , a slowness (DT) log record (e.g., sonic log record) spanning 240 ms / ft to 40 ms / ft, a resistivity (RES) log record spanning 0.2 ohm.m to 2000 ohm.m, and a gamma ray (GR) law spanning 0 gAPI to 150 gAPI.

[0083] In Figure 6 this example, the log record 600 can be acquired using one or more techniques, which can include one or more techniques involving emitted energy and received energy, where the received energy is dependent on properties of the formation and / or the borehole wall. In various cases, mud (e.g., drilling fluid) can line the borehole such that energy interacts with the mud, where such energy can also interact with the formation behind the mud. As the properties of the mud can be dependent on the drilling technology being implemented, the mud properties can vary. For example, consider oil-based mud, water-based mud, synthetic mud, etc. As an example, the salinity of the mud can vary, where the log record can be dependent on or otherwise influenced by the mud salinity. With respect to mud, mud filtrate can alter the measurements. Filtrate is the liquid that passes through the filter cake from the slurry held on the filter medium driven by the pressure differential, it should be noted that dynamic or static filtration can produce filtrate. Mud filtrate can seep into the formation and can drive formation fluid movement, which can alter one or more physical properties of the formation (e.g., the formation and the formation fluid).

[0084] As an example, a framework such as, for example, the TECHLOG framework (SLB, Houston, Texas) can be used to acquire, evaluate, alter, etc., one or more log records, which can include drilling log records as in Figure 5 and as in Figure 6petrophysical well logs. As an example, the TECHLOG framework can provide treatment of well logs such as bit size (BS), caliper (CALI), gamma ray (GR), shallow resistivity (RES SLW), medium resistivity (RES MED), deep resistivity (RES DEP), density (DEN), density correction (DENC), neutron porosity (NEU), sonic (DT), photoelectric well logs (PEF), and lithology (LITH-Petrel). Such well logs can be used to calculate one or more of volume, porosity, water saturation, and permeability of one or more types of rock (e.g., shale, etc.).

[0085] While various well logs are described with respect to depth, such as, for example, measured depth, one or more well logs can be described with respect to time. In either case, the well data can be series data, such as depth series data and / or time series data.

[0086] As an example, the framework can be a computing framework suitable for performing one or more workflows. For example, consider a framework for selecting optimal well logs for rock property interpretation and / or drilling interpretation using a combination of ranking and machine learning (ML) approaches. As an example, such a framework can provide a data preparation automation process for selecting optimal candidate well logs for a set of measurement types.

[0087] As explained, well data (e.g., well measurements) can describe a physics system involving interaction with one or more types of formations. As an example, the framework can implement a selection method where well logs are grouped by their measurement types and based on a ranking matrix of available well log candidates that match the measurement types in the well. Such a matrix can be constructed in a way that allows for a unique occurrence of each well measurement type per scenario. As an example, ML model based prediction of target well logs can score each well log combination from the matrix and push forward the well logs with the highest positive score. As an example, the ML model based approach can utilize one or more target well logs as can be selected from different types of well logs. As an example, a target well log of well logs of drilling operations can be a ROP well log. As an example, a target well log of well logs of petrophysical measurements can be a density well log. As an example, in the ranking matrix, the target well logs can be structured as the last type of well log (e.g., in the last column, etc.).

[0088] Table 1. Example ranking matrix The above example arrangement matrix considers various scenarios, which can be for particular types of formations, etc. The formation type can be selected as a basis to link the well log data to the physical reality. In such a manner, a comparison (e.g., correlation) can be performed to determine whether one or more well logs can predict one or more other well logs.

[0089] As an example, the framework can implement one or more techniques. As explained, the framework can implement an arrangement and ML manner, which can be accompanied by one or more other manners, which can include one or more existing manners based on physics, where a quality of each individual measurement is assessed in accordance with a user-defined set of rules or based on physics-based criteria.

[0090] Various issues can exist with well log data. For example, historical well log data can be altered by one or more workflows, where one or more altered versions of the well log data are stored to a data store. In such an example, the original version can not be available, such that the available versions are “children” derived from the original version, which can be directly or indirectly derived via one or more intermediate versions. As an example, the framework can provide access to well logs from one or more data stores, where the framework can process the well logs to generate output well logs that are properly acceptable for one or more workflows, which can include, for example, one or more workflows involving machine learning, where the output well logs are usable for one or more of training a machine learning model (ML model) and / or testing the ML model.

[0091] As an example, a physical system can refer to a geological formation and / or a drilling system, where the physical system can be characterized by measurements (e.g., well log data). As an example, the measurements can be sensor readings. As mentioned, the measurements can be acquired during well logging and / or during drilling, where well logging measurements can include wireline logging measurements, coiled tubing logging measurements, etc., and where drilling measurements can be acquired using a rig control system. In various cases, a combination of measurement types can help characterize the physical system.

[0092] With respect to candidates for measurement types, there can be multiple versions of measurements, where, for example, some versions can have been altered (e.g., adjusted for time and / or depth shifts, environmental effects, mud, etc.) and / or where the adjustments are a result of human interpretation.

[0093] Figure 7 An example of a workflow 700 is shown, which can be executed at least in part by the framework. In Figure 7In the example of FIG. 7, workflow 700 can include an access block 710 for accessing well log data, a clean-up block 720 for cleaning up the accessed well log data, a well log coverage block 730 for selecting well log data that covers a particular formation (e.g., or depth, time, etc.), a well log preparation block 740 for preparing the well log data, a well log matrix and ranking block 750 for constructing a permutation matrix and ranking well logs with respect to comparison, and an output block 760 for outputting one or more workflows of well log data.

[0094] As an example, workflow 700 can include building cleaned-up datasets assigned with appropriate well log measurement types, which can be controlled by zone of interest (ZOI) selection (e.g., formation type, etc.); defining a selection process for suitable family-driven well log selection by well log coverage threshold definition within a ZOI; defining a permutation matrix that lists possible combinations of available well logs in a project in a way that allows for a single occurrence per family per scenario; creating a scenario-based well log selection process from available wells within a project and their corresponding datasets, and scoring measurement candidates for their ability to predict each other’s accuracy in a machine learning fashion (e.g., a process that can assign a score to each permutation of the permutation matrix); ranking results of permutation matrix implementations for best well log combination; and retaining the best permutation implementation, where the process can be repeated for one or more additional ZOIs (e.g., one or more additional geologic formations, etc.). Such a method can be self-cleaning, where problematic well logs (e.g., negative values such as “-999” that can be assigned to missing data in the TECHLOG framework) can be automatically excluded from the selection process because such well logs can yield relatively low scores and thus have low rankings.

[0095] As an example, the framework can process well logs to address one or more issues, which can include, for example, depth misalignment issues. For example, if some of the measurement type candidates are depth misaligned, this can result in worse cross-correlation estimates (e.g., depth misaligned candidates can be lower than non-depth misaligned candidates). With respect to another example, consider a scenario where the entire candidate set is depth misaligned, which can result in inability to differentiate scores. With respect to yet another example, consider bad measurements and / or corrections. In such an example, if some measurement candidates include bad readings (e.g., do not conform to physical reality or are otherwise misleading), while others have been corrected, this can affect cross-correlation. As another example, consider inappropriately assigned measurement types. In such an example, if some candidates are inappropriately assigned to a measurement type, the physics system will not be well described, and these candidates will have lower cross-correlation scores.

[0096] As an example, the framework can provide tools for performing tasks such as... Figure 7 The workflow 700 describes the characteristics of one or more workflows. This approach may be applicable when the measurement result types have some expected physical cross-correlation. For example, some form of cross-correlation can be expected between compression slowness, shear slowness, gamma rays, neutron porosity, and volume density. Conversely, if there is insufficient physical relationship between the measurement results, this approach may not be able to select the best candidate for the measurement result type.

[0097] Figure 8 An example of a graphical user interface (GUI) 800 comprising a series of logging records, including depth logging records, density logging records, gamma-ray logging records, sonic logging records, and a predicted logging record (labeled density_mw_14) in feet. Regarding the predicted logging records, density is highlighted as a predicted logging record (e.g., the target logging record) that appropriately matches the density logging record. Other underlying logging records are also shown, which can be prepared for purposes such as ML model-based analysis (e.g., using logistic regression, etc.). Thus, for a specific formation type selected along the depth (see depth logging records), gamma-ray and sonic data can predict density data (e.g., density, gamma-ray, and sonic data are appropriately correlated). Regarding the score, it can indicate how well the combination of logging records (e.g., logging data) predicts the target logging record (e.g., target logging data). Figure 8 In the example, various versions of each measurement result can each produce different predictions of the true measurement result (e.g., see the density highlighted in red on the last trajectory). As an example, a framework can be implemented to identify the best version of each measurement result (e.g., well logging data) that will produce the most accurate prediction.

[0098] Figure 9 An example of a graphical user interface (GUI) 900 that can be rendered by a framework is shown. Figure 9 In the example, GUI 900 includes results from Principal Component Analysis (PCA). PCA can be implemented to assess correlations in data by linearly transforming it to a new coordinate system, such that the coordinate orientation (e.g., principal components) provides the maximum variation in the data. PCA can be applied as a linear decomposition technique, transforming the set of variables into principal components, which are equivalent sets of the transformed variables. Principal components are orthogonal (independent of each other) and can be ordered by the explained variance. Correlation circles can be generated as a visualization that helps convey the degree of correlation between the original variables and two principal components (typically the first two principal components, e.g., PC1 and PC2). Figure 9In the example, the projection of the variable plot is directed toward the first two principal components, where the table includes correlation values, where a value of one (value 1) indicates a perfectly positive correlation, a value of negative 1 indicates a perfectly negative (inverse) correlation, and values ​​close to zero indicate a very weak correlation, which, according to a threshold, can be considered to indicate no correlation.

[0099] exist Figure 9 In the example, a GUI 900 can be generated and rendered to indicate how a variable (e.g., well log type) is relevant or irrelevant (e.g., by using PCA as model selection). Figure 9 The results show that the combined model has a reasonably good ability to distinguish between "good" and "bad" versions of combinations of bulk density, compressibility slowness, and neutron porosity, but exhibits weaker discriminative power for gamma rays. While gamma rays provide some correlation, their discriminative power is not as strong as... Figure 9 Other variables in the example.

[0100] As an example, one approach could utilize, such as Figure 9 The results from the examples can be used to adjust or otherwise select the type of logging record to use in the combination. For example, consider removing gamma rays from the group, so that the group is redefined as volume density, compressibility slowness, and neutron porosity.

[0101] although Figure 9 The examples in the text involve rock-related logging, but consider scenarios where drilling-related logging can be evaluated. In such scenarios, variables such as riser pressure (SPP or SPPA) can be substantially orthogonal to one or more other drilling-related variables because the fluid pressure in the rig riser may be relatively unrelated to other variables that more directly characterize the interaction between the drill bit and the formation (e.g., rock). In drilling, a metric known as mechanical specific energy (MSE) can be used to represent drilling efficiency. MSE can be defined as the energy required to remove a unit volume of rock. In various cases, to achieve optimal drilling efficiency, the driller (e.g., man and / or machine) may aim to minimize MSE and maximize ROP, for example, by controlling one or more of the following: weight on bit (WOB), torque, ROP, and rotational speed per minute (RPM). Therefore, ROP, WOB, torque (e.g., TQA), RPM, and energy can be expected to exhibit some degree of correlation (e.g., cross-correlation).

[0102] like Figure 9 As shown in the example, correlation circles (e.g., variable factor plots) can be used to plot component values. Although a single plot is shown, more than one plot can be generated and / or shown (e.g., for a factor plane, which can be a vector space consisting of the intersection of two principal components in the principal components).

[0103] Table 2 below provides some processing conditions or processing issues and indications about acceptability and some examples of reasons why it is acceptable or not acceptable.

[0104] Table 2. Example processing conditions or issues.

[0105] With respect to ML models, consider a framework that can implement a relatively lightweight ML model such as logistic regression (LR). LR is a statistical model (e.g., also known as a logit model) that can be used for classification and predictive analytics. LR estimates the probability of an event occurring, such as voting or not voting, based on a given set of independent variable data. Since the outcome is a probability, the dependent variable is bounded between 0 and 1. In logistic regression, a logit transformation can be applied to the odds, which is the probability of success divided by the probability of failure.

[0106] As mentioned, LR can be used for classification. As an example, the framework can implement an LR classifier. For example, consider the scikit-learn LR classifier, also known as logit and MaxEnt (see, e.g., sklearn.linear_model.LogisticRegression). In the multiclass case, if the “multi_class” option is set to “ovr”, the training algorithm can use a one- versus-rest (OvR) scheme, and if the “multi_class” option is set to “multinomial”, cross-entropy loss is used; note that the “multinomial” option is supported by the “lbfgs”, “sag”, “saga”, and “newton-cg” solvers. In the scikit-learn framework, the LR class implements regularized logistic regression using the ‘liblinear’ library, ‘newton-cg’, ‘sag’, ‘saga’, and ‘lbfgs’ solvers. Note that regularization is applied by default. It can handle both dense and sparse input. The implementation can use C-ordered arrays or CSR matrices containing 64-bit floating point numbers for best performance; note that other input formats can be converted (and copied). In the scikit-learn framework, the “newton-cg”, “sag”, and “lbfgs” solvers support L2 regularization with the original formula, or no regularization. The “liblinear” solver supports both L1 and L2 regularization with the dual formulation only for the L2 penalty term. Elastic-Net regularization is supported by the “saga” solver.

[0107] In the scikit-learn framework, LR is implemented as a linear model that is used for classification rather than regression in terms of scikit-learn / ML nomenclature. Logistic regression is also referred to in the literature as logit regression, MaxEnt, or log-linear classifier. In this model, a logistic function is used to model the probability of the possible outcome describing a single trial. As explained, the scikit-learn implementation of LR can fit binary, one-vs-rest, or multinomial logistic regression with optional or Elastic-Net regularization. Regularization is applied by default, which is common in machine learning but not in statistics. Another advantage of regularization is that it improves numerical stability. No regularization is equivalent to setting the parameter C to a very high value. LR is a special case of the generalized linear model (GLM) with binomial / Bernoulli conditional distribution and logit link. It can be used as a classifier by applying a threshold (default 0.5) to the numeric output of logistic regression (i.e., the predicted probability). This is how it can be implemented in scikit-learn such that it expects a classification target, making the LR method a classifier.

[0108] As an example, LR can be implemented using an input dataset to create a predictive model of a result variable. For example, consider an input dataset of well logging data of one or more types of well logs, which can create a predictive model of a result variable that can be for different types of well logs. As an example, LR can be implemented in a multinomial fashion. Multinomial LR can be implemented as a classification technique that generalizes logistic regression to multiclass problems (e.g., with more than two possible discrete outcomes). For example, consider a model that can be used to predict the probability of different possible outcomes of a dependent variable of a categorical distribution given a set of independent variables (e.g., real-valued, binary-valued, categorical-valued, etc.). As an example, the scikit-learn framework can implement multinomial classification. For example, for multi_class problems, if multi_class is set to “multinomial,” a softmax function is used to find the predicted probability for each class; otherwise, a one-vs-rest (OVR) approach can be used (e.g., assuming each class is positive, a logistic function is used to compute the probability for each class, and these values are normalized across all classes).

[0109] As explained, the number of permutations can be quite large. As an example, consider more than 1 million permutations to be evaluated. As an example of the equation that shows how the number of permutations can increase, for a more general case, consider the following equation: where n is the total number of objects, and kto select an object (e.g., n a number of elements of a set. k a number of elements of a set.

[0110] To make the framework practical in performing such evaluations, an LR model can be implemented that can be considered a relatively lightweight model. As an example, the framework can include learning on a portion of data and performing a blind test on another portion of the data to evaluate the predictive effect that can be achieved by the model. As explained, a score can be generated as an indication of the working effect of the model (e.g., its predictive ability). As explained, the physical system can tie together well logs such that a certain amount of meaningful correlation can be expected for at least some implementations (e.g., arrangements). As explained, the LR can be implemented as a relatively computationally fast ML model to determine whether various well logs can build an acceptable model, with the LR approach being applied to terms of an arrangement matrix.

[0111] As an example, the workflow can provide identifying a best term in an arrangement matrix with respect to predictability of a well log, and then retaining that term (e.g., well log arrangement) for one or more purposes, which can include, for example, applying the arrangement to one or more additional formation types, if appropriate. As an example, the workflow can operate relatively independently of each formation type, such that the best term is different between at least two different formation types. For example, one arrangement can be best for one formation type, while another arrangement can be best for another formation type.

[0112] As explained, well logs can be selected with respect to types of formations that can be in a field (e.g., basin) in which multiple wells have been drilled; thus, there can be well logs for multiple wells, where the well logs include well log data corresponding to one or more types of formations. As explained, the well log data can be altered, for example, via adjustments that can occur during one or more workflows; thus, for an original well log, there can be multiple versions of the original well log stored in a data store. As explained, an arrangement matrix can include hundreds of terms, thousands of terms, tens of thousands of terms, hundreds of thousands of terms, to more than a million terms. Thus, to evaluate terms in an arrangement matrix, the framework can provide implementing one or more relatively lightweight techniques to speed up scoring. In various examples, a work or project can include evaluating well logs for more than one formation type, where there is an arrangement matrix for each formation type.

[0113] As explained, the framework can provide scores for a plurality of items in the permutation matrix (see, e.g., the scenarios in Table 1), which can be all of the number of items. Such scores can indicate how well the model works, e.g., how well the log data correlates for learning. As explained, the ML model can be trained using one portion of the log data and then tested using another portion of the log data. The framework can include features for accessing one or more types of models, where, e.g., the model library can include a logistic regression model that can include various model parameters that can be appropriately selected. With respect to drilling logs, the number of log types can be relatively limited as compared to the number of rock physics log types. For example, the drilling log types can be on the order of 20 or less; whereas the rock physics log types can be more, which can be greater than 30, 40, 50, 100, etc.

[0114] The output from the framework can be used for one or more purposes. For example, consider using the output for interpretation, machine learning, control, etc. As explained, the log data can include rock physics logs and / or drilling logs. With respect to control of drilling, consider using the output to determine one or more parameters for drilling in a particular type of formation, where the drilling can be for a new well in a field, where the output of the framework can be based on a compensated well in the field.

[0115] As an example, the output of the framework can be a best set of measurements to use for drilling. In such an example, consider measurements that appropriately correlate RPM with ROP, where the RPM measurements can be used to drill into a type of formation to provide a desired ROP (e.g., a desired, expected ROP). As explained, the physical system can include drilling equipment and a formation to be drilled. Thus, in the case where the RPM log data correlates with the observed ROP log data, RPM can be a predictor for ROP. While RPM is mentioned in the foregoing example, one or more other types of logs can additionally or alternatively be used. In various cases, torque log data can be used alone or in combination with other one or more types of log data, as torque can characterize the physical system including the drilling equipment and the formation to be drilled. As an example, the framework can evaluate the log data to determine whether torque log data is present. In such an example, in the case where torque log data is absent, a series of logs can be excluded (e.g., because predicting ROP can be challenging without torque knowledge). A ROP prediction model that does not utilize torque can be characterized by a low correlation score.

[0116] As an example, the method can evaluate well logging data in one or more data stores of one or more physical systems. In such an example, the framework can be provided access to the one or more data stores to determine which data therein is optimal in terms of scenarios definable in the arrangement matrix. The output of the optimal well logging data can inform the data owner as to which data is suitable for use at a certain level of confidence, and for example, which data can not be suitable for use and can be deleted from the one or more data stores if desired. As an example, the optimal well logging data can be suboptimal for one or more purposes. In such an example, the optimal well logging data can be subjected to one or more processes to improve the data. For example, consider a human-in-the-loop (HITL) approach in which the identified optimal well logging data is subjected to a HITL interpretation, which can provide adjustments to at least a portion of the optimal well logging data. While reference is made to a HITL approach, one or more other approaches, optionally machine-based and automated, can be used. As an example, the framework can generate metrics, which can include optimal well logging data metrics and / or metrics for other well logging data. As an example, the framework can generate metrics regarding duplicates, data lineage, etc.

[0117] As an example, based on one or more metrics, a data handling practice can be determined. For example, consider retaining well logging data that can have been generated from suboptimal measurements. In such an example, one or more forensic techniques can be applied to identify how and / or why such a data handling practice occurred. In turn, one or more workflows can be revised, redrafted, etc. As an example, the framework can generate one or more family trees of well logging data as part of data forensic features. As an example, the framework can provide a determination of which interpretations can have been derived based on below-standard measurements. In such an example, where the framework identifies optimal well logging data, the optimal well logging data can be used to repeat one or more workflows (e.g., interpretation workflows). As an example, improved interpretations can be propagated to other workflows to improve their results, decisions, etc.

[0118] As an example, the framework can provide a field-wide evaluation of well logging data of a field to determine whether suboptimal (e.g., below-standard) well logging data was used in one or more workflows, which can have had an adverse impact on decisions, field operations, etc. Such an approach can be used as a sanity check and / or field optimization. For example, consider a field in which production can have declined in a manner that deviates from predictions after a number of years. In such an example, the framework can evaluate underlying well logging data to determine whether optimal well logging data was used, and as explained, optionally to identify where substandard well logging data was used and possibly propagated.

[0119] As an example, the framework can be arranged to operate as part of a background process, which can be relatively continuous or periodic. For instance, a field may be scheduled for reassessment every three years. In such an example, prior to the reassessment, the framework can evaluate well logging data in one or more data stores used in the field and generate output regarding whether optimal well logging data was used in previous assessments and / or previous reassessments. In such an example, the framework can indicate which well logging data is optimal for the reassessment, allowing the reassessment to be performed in a more accurate manner.

[0120] Figure 10 An example of framework 1010 is shown, which can access and evaluate field data 1004 to generate output 1008, which can be used for one or more purposes, such as control, machine learning, forensics, etc. Figure 10 As illustrated, framework 1010 may include a data cleaner component 1020, an overlay selector component 1030 (e.g., for ZOI, etc.), a logging record preparer component 1040 (e.g., for preparing logging records for ML, etc.), and a logging record matrix and metric component 1050. As shown, the logging record matrix and metric component 1050 may be operatively coupled to one or more ML libraries 1006. For example, consider an ML library that includes one or more relatively lightweight ML models that can evaluate items (e.g., scenarios, etc.) in a permutation matrix. As explained, a logistic regression (LR) model may be used to provide a score as a metric. As explained, the score may be used to rank to determine the best logging data (e.g., field data) present in one or more data stores. As an example, the framework may provide archiving, compression, tagging, deletion, etc., of data that can be considered poorly ranked (e.g., to save space, time, etc. for future data projects).

[0121] Figure 11An example of the method 1100 is shown that includes performing a field operation 1102, where a log record 1110 is generated during performance of the field operation (e.g., wireline logging, drilling, etc.). In such examples, the field operation 1102 can be numerous in number and can be performed over a period of time that can be weeks, months, years, etc. As an example, the log record 1110 can be stored in one or more databases, which can be public, private, etc. As an example, the databases can be service provider databases, well operator databases, etc. In various cases, a well operator can request a service provider to perform a field operation, which can involve logging that generates a log record. Such logging can be directed to solving a particular problem that is closely related to well development, well production, etc. Once the problem is solved, the log record can be stored to one or more databases, e.g., archived, as the particular problem has been solved such that the service provider moves on to one or more other customers to solve their problems. In such a manner, log records can simply sit idle over time without being used for any particular purpose, and the number of such idle log records can increase substantially. In various cases, the number of log records can be vast for a field that includes hundreds or thousands of wells. However, deriving value from these log records can be a relatively insurmountable task, especially with respect to time and / or resources for review, quality control, alignment, unit conversion, etc.

[0122] As explained, a framework such as, for example, the framework 1010 can provide for extracting log records that are somewhat non-inconsistent. As explained, correlation in a broad sense is a measure of association between variables. As explained, certain variables can be expected to have some association with one another and, thus, be considered consistent or somewhat non-inconsistent. For example, density, gamma, and sonic log records in an example can be expected to be consistent such that, within such a set of log records, log records that fail to adequately provide for predictability can be considered inconsistent when considered in different combinations. Such inconsistent log records can be considered problematic (e.g., of low quality, etc.) and, thus, can be excluded from other log records that provide sufficient predictability. In such a manner, log records can be effectively classified with respect to predictability as an indicator of consistency, or otherwise, as an indicator of non-inconsistency (e.g., as can be determined relatively, numerically, etc. via one or more criteria). Figure 8 As explained, a framework such as, for example, the framework 1010 can provide for extracting log records that are somewhat non-inconsistent. As explained, correlation in a broad sense is a measure of association between variables. As explained, certain variables can be expected to have some association with one another and, thus, be considered consistent or somewhat non-inconsistent. For example, density, gamma, and sonic log records in an example can be expected to be consistent such that, within such a set of log records, log records that fail to adequately provide for predictability can be considered inconsistent when considered in different combinations. Such inconsistent log records can be considered problematic (e.g., of low quality, etc.) and, thus, can be excluded from other log records that provide sufficient predictability. In such a manner, log records can be effectively classified with respect to predictability as an indicator of consistency, or otherwise, as an indicator of non-inconsistency (e.g., as can be determined relatively, numerically, etc. via one or more criteria).

[0123] Referring again to Figure 11In the context of well logging records 1110, the framework can filter such records (which can number in the thousands, tens of thousands, or more) and easily identify sufficiently consistent (or sufficiently inconsistent) records, making those identified records usable for one or more purposes. In this way, the framework can help extract values ​​from well logging records that might otherwise lie idle in one or more databases. As explained, well logging can be a complex process that consumes time and resources, making the acquisition of records from such wells potentially expensive. As an example, a framework can be implemented to help extract values ​​from such expenditures.

[0124] like Figure 11 As shown in method 1100, well logging records 1110 can undergo a well logging record type classification process. For example, consider the process of accessing individual well logging records and determining which types of well logging records might be within a well logging chart. In some cases, a well logging chart may include one type of well logging record, while in other cases, it may include more than one type. Figure 11 As shown, well log record 1110 can be subdivided into various types, which can be considered as candidate types 1130-1, 1130-2, 1130-3, ..., 1130-N. For example, once the well log record type is known, method 1100 can consider a threshold as a break regarding the type for further evaluation purposes. For example, if a well log record type is rare compared to the other various types, that type can be excluded from the candidate types because it may cause some limitations within the generation of permutations (e.g., candidate sets).

[0125] As an example, the method may consider depth or depth range, which may rely on information about where one or more strata might be located in the subsurface environment, such as stratum tops and / or other markers. For instance, well logs can be evaluated with respect to depth (e.g., one or more zones of interest, etc.), which corresponds to known depths or expected depths of a rock type, etc. Where rock strata may be inclined within the site, the depth of the boundary of that rock strata can vary spatially, which may depend on the surface location, etc. (e.g., considering surface x and y coordinates or latitude and longitude, etc.).

[0126] As shown in the method 1100, the permutation computing process 1140 can provide for generating a set of candidates 1150, which can be referred to as a permutation. For example, where the candidate types include gamma ray, density, neutron porosity, compressional slowness, and shear slowness, each candidate type can include, for example, ten or more individual candidates of that candidate type; it should be noted that the number of individual candidates of various candidate types can reach one hundred or more. As can be appreciated, the number of permutations can become quite large (see also, for example, Table 1). Accordingly, a relatively fast (e.g., lightweight, relatively low computational and / or memory requirements) machine learning approach can be implemented, as each set of individual candidates is to be used for machine learning training, testing, and scoring.

[0127] As shown in the method 1100, the data of each set of candidates can be split for training and testing 1160, where such split data can be used to train, test, and score 1170 the predictive ability of each set of candidates. As shown, the sets of candidates can be ranked 1180 by their scores with respect to predictive ability.

[0128] As explained, the framework can provide for well log filtering such that well logs that do not provide sufficient predictive ability are considered to be somewhat inconsistent. As an example, poorly ranked sets of candidates can be evaluated to determine whether one or more particular well logs frequently appear in those sets of candidates. Well logs that frequently appear in poorly ranked sets of candidates can be considered to be inconsistent or detrimental to predictability when presented in various permutations (e.g., sets of candidates).

[0129] Figure 12 An example of the method 1200 and an example of the system 1290 are shown. As shown, the method 1200 can include a receiving block 1210 for receiving well log data of different types of well logs, an identifying block 1220 for identifying a portion of the well log data corresponding to a formation type, a defining block 1230 for defining a combination of the portion of the well log data corresponding to the formation type, an implementing block 1240 for implementing a machine learning model that generates scores for the combination, where each of the scores indicates an ability of each of the combinations in the combination to predict one or more target well logs selected from the different types of well logs therein, and an outputting block 1250 for outputting, based on a ranking of the scores, a highest ranked one of the combinations corresponding to the formation type. As an example, the method 1200 can be implemented at least in part using the framework of the framework 1000, such as, for example, Figure 10 Figure 12 The method 1200 can be implemented at least in part using the framework of the framework 1000, such as, for example,

[0130] The method 1200 can be implemented at least in part using the framework of the framework 1000, such as, for example, Figure 12 ​The various computer-readable medium (CRM) blocks 1211, 1221, 1231, 1241, and 1251 are shown in association with various computer-readable media (CRM) blocks 1211, 1221, 1231, 1241, and 1251. Such blocks typically include instructions suitable to be executed by one or more processors (or processor cores) to instruct a computing device or system to perform one or more actions. While various blocks are shown, a single medium can be configured with instructions to allow at least some of the various actions of the method 1200 to be performed, in part. As an example, a computer-readable medium (CRM) can be a non-transitory and non-carrier computer-readable storage medium. As an example, one or more of the blocks 1211, 1221, 1231, 1241, and 1251 can be in the form of processor-executable instructions.

[0131] In Figure 12 In an example, the system 1290 includes one or more information storage devices 1291, one or more computers 1292, one or more networks 1295, and instructions 1296. With respect to the one or more computers 1292, each computer can include one or more processors (e.g., or processor cores) 1293 and a memory 1294 to store instructions 1296 that are executable, for example, by at least one of the one or more processors 1293 (see, e.g., blocks 1211, 1221, 1231, 1241, and 1251). As an example, a computer can include one or more network interfaces (e.g., wired or wireless network interfaces), one or more graphics cards, display interfaces (e.g., wired or wireless display interfaces), and the like.

[0132] As explained, the framework enables efficient filtering of large amounts of work data, which can be or include well logging data. In various cases, service providers can be called upon to handle emergency matters, perform work and resolve issues quickly, and then later focus on learning from the data (if time permits). As the amount of work stacks up, is performed, and the like, the amount of data can gradually increase, making the task of sifting through the data and learning from it more arduous. As explained, the framework enables sifting through the data to find sets of inconsistent data (e.g., at a certain level, according to one or more criteria). As explained, the framework can be applied to data such as well logging data, which can be for formations (e.g., rock) and / or drilling operations.

[0133] As an example, the framework can improve planning of drilling operations. For example, consider a field in which wells have been drilled and logging records have been acquired. In planning a new well or additional operations on an existing well, the framework can access the logging records and efficiently filter the logging records to identify a set of logging records that can be used to improve the planning. For example, consider identifying a set of logging records that can be used to more accurately identify one or more types of formations, formation boundaries, etc. at a new well location (e.g., with respect to a wellbore trajectory for a new well at the location). As an example, where the identified set of logging records is relevant to drilling operation variables, the logging records can be used to plan drilling operations. For example, consider extracting drilling operation variables from the identified set(s) of logging records, where such drilling operation variables can be used in a digital drilling plan that can provide for automated and / or semi-automated drilling of a new well (e.g., or additional drilling of an existing well, etc.). For example, consider an automated drill rig as a type of controller, which can be programmed using information contained in the one or more identified sets of logging records with respect to drilling operation variables. In such an example, the automated drill rig can aim to drill more optimally, e.g., with reduced MSE and increased ROP.

[0134] With respect to some types of machine learning models that can be implemented for one or more purposes, consider one or more of, for example, a support vector machine (SVM) model, a k-nearest neighbors (KNN) model, an ensemble classifier model, a neural network (NN) model, etc. As examples, a machine learning model can be a deep learning model (e.g., a deep Boltzmann machine, a deep belief network, a convolutional neural network, a stacked autoencoder, etc.), an ensemble model (e.g., a random forest, a gradient boosting machine, a bootstrap aggregation, AdaBoost, stacked generalization, gradient boosting regression trees, etc.), a neural network model (e.g., a radial basis function network, a perceptron, back-propagation, a Hopfield network, etc.), a regularization model (e.g., ridge regression, least absolute shrinkage and selection operator, elastic net, least-angle regression), a rule system model (e.g., cubist, one-rule, zero-rule, repeated incremental pruning to produce an error-reduced, etc.), a regression model (e.g., linear regression, ordinary least squares regression, stepwise regression, multivariate adaptive regression splines, locally estimated scatterplot smoothing, logistic regression, etc.), a Bayesian model (e.g., naive Bayes, average dependency estimation, Bayesian belief network, Gaussian naive Bayes, multinomial naive Bayes, Bayesian network), a decision tree model (e.g., classification and regression tree, iterative dichotomiser 3, C4.5, C5.0, chi-squared automatic interaction detection, decision stump, conditional decision tree, M5), a dimensionality reduction model (e.g., principal component analysis (PCA), partial least squares regression, Sammon mapping, multidimensional scaling, projection pursuit, principal components regression, partial least squares discriminant analysis, mixture discriminant analysis, quadratic discriminant analysis, regularized discriminant analysis, flexible discriminant analysis, linear discriminant analysis, etc.), an instance model (e.g., k-nearest neighbors, learning vector quantization, self-organizing map, locally weighted learning, etc.), a clustering model (e.g., k-means, k-medians, expectation maximization, hierarchical clustering, etc.), etc.

[0135] As an example, machine models can be constructed using computing frameworks having libraries, toolboxes, and the like, such as, for example, those of the MATLAB framework (MathWorks, Inc., Natick, MA). The MATLAB framework includes toolboxes that provide supervised and unsupervised machine learning algorithms, including support vector machines (SVM), boosted and bagged decision trees, k-nearest neighbors (KNN), k-means, k-medoids, hierarchical clustering, Gaussian mixture models, and hidden Markov models. Another MATLAB framework toolbox is the deep learning toolbox (DLT), which provides a framework for designing and implementing deep neural networks with algorithms, pre-trained models, and apps. The DLT provides convolutional neural networks (ConvNets, CNNs) and long short-term memory (LSTM) networks to perform classification and regression on image, time-series, and text data. The DLT includes features for building network architectures such as generative adversarial networks (GANs) and twin networks using custom training loops, shared weights, and automatic differentiation. The DLT is used for model exchange with various other frameworks.

[0136] As an example, the TENSORFLOW framework (Google LLC, Mountain View, CA) can be implemented, which is an open-source software library including a symbolic math library that can be implemented for machine learning applications that can include neural networks. As an example, the CAFFE framework can be implemented, which is a DL framework developed by the Berkeley AI Research Center (BAIR) (University of California, Berkeley). As another example, consider the SCIKIT platform (e.g., the scikit-learn framework), which utilizes the PYTHON programming language. As an example, frameworks such as the APOLLO AI framework (APOLLO.AI GmbH, Germany) can be utilized. As an example, frameworks such as the PYTORCH framework (Facebook AI Research Lab (FAIR) of Facebook, Inc., Menlo Park, CA) can be utilized.

[0137] As an example, training methods can include various actions that can be performed on a dataset to train an ML model. As an example, a dataset can be divided into training data and test data, where the test data can provide an evaluation. One approach can include cross-validation of parameters and optimal parameters, which can be used for model training.

[0138] The TENSORFLOW framework can run on multiple CPUs and GPUs (with optional CUDA extensions for general computing on graphics processing units (GPUs) (NVIDIA Corporation, Santa Clara, CA) and SYCL extensions (Khronos Group, Inc., Beaverton, OR). TENSORFLOW can be used on 64-bit LINUX, MACOS (Apple Inc., Cupertino, CA), WINDOWS (Microsoft Corporation, Redmond, WA), as well as mobile computing platforms including shims based on ANDROID (Google LLC, Mountain View, CA) and IOS (Apple Inc.) operating systems.

[0139] TENSORFLOW computations can be expressed as stateful dataflow graphs; note that the name TENSORFLOW derives from the operations that such neural networks perform on multidimensional data arrays. Such arrays can be referred to as “tensors.”

[0140] As an example, the apparatus can utilize TENSORFLOW LITE (TFL) or another type of lightweight framework. TFL is a suite of tools that support machine learning on devices, where models can run on mobile, embedded, and IoT devices. TFL is optimized for machine learning on devices by addressing latency (no round trips to a server), privacy (no personal data leaves the device), connectivity (internet connection is needed), size (reducing model and binary size), and power (e.g., efficient inference and lack of network connectivity). TFL provides multi-platform support, covering ANDROID devices and iOS devices, embedded LINUX, and microcontrollers. TFL provides multi-language support, including JAVA, SWIFT, Objective-C, C++, and PYTHON. TFL provides high performance with hardware acceleration and model optimization.

[0141] As an example, TFL or other lightweight framework approaches can be implemented in the field, optionally within a downhole toolstring that can perform framework processes downhole, which can provide real-time decision making, control, etc.

[0142] As an example, a method can include receiving well log data of different types of well logs, identifying a portion of the well log data corresponding to a formation type, defining combinations of the portion of the well log data corresponding to the formation type, implementing a machine learning model that generates scores for the combinations, where each of the scores indicates an ability of each of the combinations to predict one or more target well logs selected from the different types of well logs, and outputting a highest ranked one of the combinations corresponding to the formation type based on a ranking of the scores. In such an example, the machine learning model can be or can include a logistic regression model.

[0143] As explained, a target well log can be selected from well logs in a set (e.g., a candidate set, which can be referred to as a permutation or combination), where the target well log can provide an evaluation of predictability (e.g., or consistency or lack of inconsistency) as an indicator of relevance. Once evaluated, a model having some demonstrated ability to predict the target well log can be valuable or not. As explained, the purpose can not be to generate a predictive model for further use, rather, the purpose can be to generate a predictive model as a means for evaluating consistency or lack of consistency in a set of well logs. As explained, density can be used as a target for formation-related well logs, rate of penetration (ROP) can be used as a target for drilling operation-related well logs, one or more targets can be utilized, etc. As explained, selecting one or more targets for prediction can provide an evaluation of consistency in a set of well logs (e.g., a combination or permutation).

[0144] As explained, a formation type can correspond to a particular depth or depths in a subsurface environment. For example, where well logs relate to physical properties of rock, well logs can be evaluated with respect to a particular rock layer, which can be defined by one or more boundaries (e.g., interfaces, etc.). As an example, where well logs relate to drilling operations, drilling operation variables can be set or adjusted to drill into and / or through a formation type. For example, drilling operation variables expected to drill into one rock type can be different from one or more drilling operation variables expected to drill into another rock type. In various examples, within a formation, drilling operation variables can be set or adjusted with respect to depth (e.g., measured depth, etc.). As an example, a formation type can be an indicator of depth, such that identifying a portion of well log data corresponding to the formation type can provide identifying the portion as corresponding to a depth (e.g., a particular depth range, etc.).

[0145] As an example, defining the combinations can include defining a permutation matrix. For example, consider a permutation matrix that includes dimensions for a combination (e.g., a scenario) and dimensions for different types of log records. As explained, a combination can be referred to as a permutation or a candidate set (e.g., a set of candidates, the candidate being a log record of a different log record type, etc.).

[0146] As an example, the log data can include different sets of log data for one or more of the different types of log records. As an example, the log data can include petrophysical log data and / or drilling operations log data. As an example, the drilling operations log data can include at least torque data.

[0147] As an example, the one or more target log records can include a density log record. As an example, the one or more target log records can include a rate of penetration log record.

[0148] As an example, the log data can characterize a physical system. For example, consider a physical system that includes one or more types of formations and one or more types of log sensors that emit one or more types of energy; or consider a physical system that includes one or more types of formations and at least a drill bit coupled to a drill string.

[0149] As an example, the different types of log records can include more than five different types of log records. As an example, the combinations can include more than one hundred combinations.

[0150] As an example, the log data can include multi-well log data. For example, consider that the multi-well log data is from more than five wells. As an example, the log data can include versions of log data. For example, consider a family tree of log data, where an original version is changed via one or more generations of changes. As an example, a method can provide for evaluating one or more sets of log records that can be part of a family tree of log data.

[0151] As an example, a method can include performing one or more of control of field equipment, machine learning, and data management based at least in part on the highest ranked combination. As an example, the data management can include data forensics. As an example, the data forensics can provide a basis for workflow evaluation, which can include workflow revision.

[0152] As an example, a system can include: one or more processors; memory accessible to at least one of the one or more processors; processor-executable instructions stored in the memory and executable to instruct the system to: receive well log data of different types of well logs; identify a portion of the well log data corresponding to a formation type; define combinations of the portion of the well log data corresponding to the formation type; implement a machine learning model that generates scores for the combinations, where each of the scores indicates an ability of each of the combinations to predict one or more target well logs selected from the different types of well logs therein; and based on a ranking of the scores, output a highest ranked one of the combinations corresponding to the formation type. As an example, one or more computer-readable storage media can include processor-executable instructions to instruct a computing system to: receive well log data of different types of well logs; identify a portion of the well log data corresponding to a formation type; define combinations of the portion of the well log data corresponding to the formation type; implement a machine learning model that generates scores for the combinations, where each of the scores indicates an ability of each of the combinations to predict one or more target well logs selected from the different types of well logs therein; and based on a ranking of the scores, output a highest ranked one of the combinations corresponding to the formation type.

[0153] As an example, a computer program product can include one or more computer- readable storage media comprising processor-executable instructions to instruct a computing system to perform one or more methods and / or one or more portions of a method.

[0154] In some embodiments, one or more methods can be performed by a computing system. Figure 13 An example of a system 1300 is shown, which can include one or more computing systems 1301-1, 1301-2, 1301-3, and 1301-4, which can be operatively coupled via one or more networks 1309, which can include wired and / or wireless networks. As shown, the system 1300 can include one or more other components 1308.

[0155] As an example, a system can include an individual computer system or a distributed computer system of arrangements. In Figure 13 In an example, the computing system 1301-1 can include one or more modules 1302, which can be or can include processor-executable instructions, for example, executable to perform various tasks (e.g., receive information, request information, process information, simulations, output information, etc.).

[0156] As an example, a module can execute independently or in coordination with one or more processors 1304 that are operatively coupled, e.g., via wired, wireless, etc., to one or more storage media 1306. As an example, one or more of the processors 1304 can be operatively coupled to at least one of the one or more network interfaces 1307; note that one or more other components 1308 can also be included. In such examples, the computer system 1301-1 may, for example, transmit and / or receive information via one or more networks 1309 (e.g., consider one or more of the Internet, a proprietary network, a cellular network, a satellite network, etc.).

[0157] As an example, the computer system 1301-1 can receive information and / or transmit information to one or more other devices, which can be or include, for example, one or more of the computer systems 1301-2, etc. The devices can be located at a different physical location from the physical location of the computer system 1301-1. As an example, the locations can be, for example, a processing facility location, a data center location (e.g., a server farm, etc.), a drilling rig location, a wellsite location, a downhole location, etc.

[0158] As an example, a processor can be or include a microprocessor, microcontroller, processor module or subsystem, programmable integrated circuit, programmable gate array, or another control or computing device.

[0159] As an example, the storage media 1306 can be implemented as one or more computer- or machine-readable storage media. As an example, storage can be distributed across multiple internal and / or external hard drives, and / or solid state drives, within and / or across a computing system and / or additional computing systems.

[0160] As an example, the one or more storage media can include one or more different forms of memory including semiconductor memory, such as dynamic or static random access memories (DRAMs or SRAMs), erasable programmable read only memories (EPROMs), electrically erasable programmable read only memories (EEPROMs) and flash memories; magnetic disks such as fixed, floppy and removable disks; other magnetic media including tape; optical media such as compact disks (CDs) or digital video disks (DVDs), BLUERAY disks, or other types of optical storage; or other types of storage devices.

[0161] As an example, one or more storage media can reside with a machine that runs the machine- readable instructions, or one or more storage media can be remote from a machine that runs the machine-readable instructions, and the machine-readable instructions can be downloaded from the remote storage media through a network for execution by the machine. As an example, various components of a system such as, for example, a computer system, can be implemented in hardware, software, or a combination of both hardware and software, e.g., including firmware.

[0162] As an example, a system can include a processing device, which can be or can include a general purpose processor or a special purpose chip (e.g., or chip set), such as an ASIC, an FPGA, a PLD, or other appropriate device.

[0163] As an example, an apparatus can be a mobile apparatus that includes one or more network interfaces for communicating information. For example, a mobile apparatus can include a wireless network interface (e.g., operable via IEEE 802.11, ETSI GSM, BLUETOOTH, satellite, etc.). As an example, a mobile apparatus can include components such as a main processor, memory, a display, display graphics circuitry (e.g., optionally including touch and gesture circuitry), a SIM slot, audio / video circuitry, motion processing circuitry (e.g., accelerometer, gyroscope), wireless LAN circuitry, smart card circuitry, transmitter circuitry, GPS circuitry, and a battery. As an example, a mobile apparatus can be configured as a cellular telephone, a tablet computer, etc. As an example, a method can be implemented (e.g., in whole or in part) using a mobile apparatus. As an example, a system can include one or more mobile apparatuses.

[0164] As an example, a system can be a distributed environment, e.g., a so-called “cloud” environment, in which various apparatuses, components, etc. interact for purposes of data storage, communication, computation, etc. As an example, an apparatus or system can include one or more components for communicating information via one or more of the Internet (e.g., where communication is via one or more Internet protocols), a cellular network, a satellite network, etc. As an example, a method can be implemented in a distributed environment (e.g., in whole or in part as a cloud-based service).

[0165] As an example, information can be input from a display (e.g., consider a touchscreen), output to a display, or both. As an example, information can be output to a projector, laser device, printer, etc., so that it can be viewed by information. As an example, information can be output stereographically or holographically. In regard to a printer, consider a 2D or 3D printer. As an example, a 3D printer can include one or more substances that can be output to build a 3D object. For example, data can be provided to a 3D printer to build a 3D representation of a subsurface formation. As an example, layers can be built in 3D (e.g., ground planes, etc.), geologic bodies built in 3D, etc. As an example, wellbores, fractures, etc. can be built in 3D (e.g., as positive structures, as negative structures, etc.).

[0166] While only a few example embodiments have been described in detail, it should be appreciated that numerous modifications are possible. Accordingly, it is intended that all such modifications are included within the scope of the disclosure as defined in the following claims in which the device plus function clause is intended to cover the structures described herein as performing the recited functionality, and not only structural equivalents, but also equivalent structures. Thus, although a nail and a screw can not be structural equivalents in that a nail employs a cylindrical surface to secure wooden parts together, whereas a screw employs a helical surface, in the environment of fastening wooden parts, a nail and a screw can be equivalent structures.

Claims

1. A method comprising: Receives logging data from different types of logging records; Identify the portion of the well logging data that corresponds to the formation type; Define the combination of the portion of the well logging data corresponding to the formation type; Implement a machine learning model that generates scores for the combinations, wherein each score indicates the ability of each combination to predict one or more target well logs, such as those selected from the different types of well logs; as well as Based on the ranking of the scores, output the highest-ranking combination among the combinations that corresponds to the stratigraphic type.

2. The method of claim 1, wherein the machine learning model includes a logistic regression model.

3. The method of claim 1, wherein defining the combination includes defining a permutation matrix.

4. The method of claim 3, wherein the permutation matrix includes dimensions for the combination and dimensions for the different types of logging records.

5. The method of claim 1, wherein the logging data includes a different set of logging data from one or more logging records of the different types.

6. The method of claim 1, wherein the logging data includes rock physical logging data.

7. The method of claim 1, wherein the logging data includes drilling operation logging data.

8. The method of claim 7, wherein the drilling operation logging data includes at least torque data.

9. The method of claim 1, wherein the one or more target logging records include density logging records.

10. The method of claim 1, wherein the one or more target logging records include drilling rate logging records.

11. The method of claim 1, wherein the logging data characterizes the physical system.

12. The method of claim 11, wherein the physical system comprises one or more types of formations and one or more types of well logging sensors transmitting one or more energies.

13. The method of claim 11, wherein the physical system comprises one or more types of formations and at least a drill bit coupled to a drill string.

14. The method of claim 1, wherein the different types of logging records include more than five different types of logging records.

15. The method of claim 1, wherein the combination comprises more than one hundred combinations.

16. The method of claim 1, wherein the logging data includes multi-well logging data.

17. The method of claim 16, wherein the multi-well logging data comes from more than five wells.

18. The method of claim 1, comprising performing one or more of the following: control of field equipment, machine learning, and data management, at least in part based on the highest-ranking combination.

19. A system comprising: One or more processors; The memory, wherein at least one of the one or more processors can access the memory; Processor-executable instructions, stored in the memory and executable to instruct the system: Receives logging data from different types of logging records; Identify the portion of the well logging data that corresponds to the formation type; Define the combination of the portion of the well logging data corresponding to the formation type; Implement a machine learning model that generates scores for the combinations, wherein each score indicates the ability of each combination to predict one or more target well logs, such as those selected from the different types of well logs; as well as Based on the ranking of the scores, output the highest-ranking combination among the combinations that corresponds to the stratigraphic type.

20. One or more computer-readable storage media, including processor-executable instructions for instructing a computing system: Receives logging data from different types of logging records; Identify the portion of the well logging data that corresponds to the formation type; Define the combination of the portion of the well logging data corresponding to the formation type; Implement a machine learning model that generates scores for the combinations, wherein each score indicates the ability of each combination to predict one or more target well logs, such as those selected from the different types of well logs; as well as Based on the ranking of the scores, output the highest-ranking combination among the combinations that corresponds to the stratigraphic type.

Images