On-site pump equipment system

Abstract

A method may include receiving an input including time series data from a pump device at a wellsite, wherein the wellsite includes a wellbore in contact with a fluid reservoir; processing the input using the trained first machine learning model as an anomaly detector to generate an output; and processing the input and output using the trained second machine learning model to predict a survival function of the pump device.

Description

[0001] Related Applications

[0002] This application claims priority to and the benefit of U.S. Provisional Application Serial No. 63 / 358,189, filed on July 4, 2022, the entire contents of which are incorporated herein by reference. Background Art

[0003] A reservoir can be an underground formation that can be characterized at least in part 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, sedimentation, etc.). As an example, in the presence of source rocks in combination with appropriate burial depth and duration, a petroleum system can develop within a basin that can form a reservoir comprising hydrocarbon fluids (e.g., oil, gas, etc.). Various operations can be performed on site to obtain such hydrocarbon fluids and / or produce such hydrocarbon fluids. For example, consider equipment operations, in which equipment can be controlled to perform one or more operations. Summary of the invention

[0004] A method may include receiving an input including time series data from a pump device at a wellsite, wherein the wellsite includes a wellbore in contact with a fluid reservoir; processing the input using a trained first machine learning model as an anomaly detector to generate an output; and processing the input and the output using a trained second machine learning model to predict a survival function of the pump device. A system may include a processor; a memory accessible to the processor; and processor-executable instructions stored in the memory to instruct the system to: receive an input including time series data from a pump device at a wellsite, wherein the wellsite includes a wellbore in contact with a fluid reservoir; processing the input using a trained first machine learning model as an anomaly detector to generate an output; and processing the input and the output using a trained second machine learning model to predict a survival function of the pump device. One or more computer-readable storage media may include processor-executable instructions to instruct a wellsite computing system to: receive an input including time series data from a pump device at a wellsite, wherein the wellsite includes a wellbore in contact with a fluid reservoir; processing the input using a trained first machine learning model as an anomaly detector to generate an output; and processing the input and the output using a trained second machine learning model to predict a survival function of the pump device. Various other devices, systems, methods, etc. are also disclosed.

[0005] This Summary is provided to introduce a selection of concepts that are further described below in the Detailed Description. This Summary is not intended to identify key or essential features of the claimed subject matter, nor is it intended to be used to help limit the scope of the claimed subject matter. BRIEF DESCRIPTION OF THE DRAWINGS

[0006] The features and advantages of the described embodiments may be more readily understood by referring to the following description taken in conjunction with the accompanying drawings.

[0007] Figure 1 An example system including various framework components associated with one or more geological environments is shown;

[0008] Figure 2 Examples of devices, examples of networks, and examples of systems are shown;

[0009] Figure 3 An example of a device is shown;

[0010] Figure 4 An example of an electric submersible pump system is shown;

[0011] Figure 5 An example of a system is shown;

[0012] Figure 6 An example of a system is shown;

[0013] Figure 7 An example of a method is shown;

[0014] Figure 8 An example of a graphical user interface illustrating a plot of time series data;

[0015] Fig. 9 An example of a system is shown;

[0016] Fig.10 An example of a machine learning model architecture is shown;

[0017] Fig.11 An example of a machine learning tree model is shown;

[0018] Fig.12 An example of a drawing is shown;

[0019] Fig.13 An example of a system is shown;

[0020] Fig.14 An example of a system is shown;

[0021] Fig.15 Examples of methods and examples of systems are shown;

[0022] Fig.16 An example of a graphical user interface is shown;

[0023] Fig.17 An example of a graphical user interface is shown;

[0024] Fig.18 An example of a graphical user interface is shown;

[0025] Fig.19 shows examples of computers and network equipment; and

[0026] Fig. 20 Example components of systems and networked systems are shown. DETAILED DESCRIPTION

[0027] This description should not be taken in a limiting sense but is made merely for the purpose of describing the general principles of the embodiments. Reference should be made to the issued claims to determine the scope of the described embodiments.

[0028] Figure 1 An example of a system 100 is shown that includes a workspace framework 110 that can provide instantiation, rendering, interaction with, etc., a graphical user interface (GUI) 120. Figure 1 In the example of , GUI 120 may include graphical controls for computing framework (eg, application) 121 , project 122 , visualization 123 , one or more other features 124 , data access 125 , and data storage 126 .

[0029] exist Figure 1 In the example of, the workspace framework 110 may be customized for a particular geological environment, such as the exemplary geological environment 150. For example, the geological environment 150 may include layers (e.g., strata) that include a reservoir 151 and may intersect a fault 153. As an example, the geological environment 150 may be equipped with various sensors, detectors, predictors, actuators, and the like. For example, the device 152 may include communication circuits for receiving and sending information with respect to one or more networks 155. Such information may include information associated with downhole equipment 154, which may be equipment for acquiring information, assisting in resource recovery, and the like. Other equipment 156 may be located away from the well site and include sensing, detection, prediction, transmission, or other circuits. Such equipment may include storage and communication circuits to store and transmit data, instructions, and the like. As an example, one or more satellites may be provided for purposes of communication, data acquisition, and the like. For example, Figure 1 A satellite is shown in communication with network 155, which may be configured for communications, noting that the satellite may additionally or alternatively include circuitry for imaging (e.g., spatial, spectral, temporal, radiometric, etc.).

[0030] Figure 1The geological environment 150 is also shown as optionally including equipment 157 and 158 associated with a well, the well including a substantially horizontal portion that may intersect one or more fractures 159. For example, consider a well in a shale formation, which may include natural fractures, artificial fractures (e.g., hydraulic fractures), or a combination of natural fractures and artificial fractures. As an example, a well may be drilled for a laterally extending reservoir. In such an example, there may be lateral changes in properties, stresses, etc., where an assessment of such changes may help plan, operate, etc. to develop a laterally extending reservoir (e.g., via fracturing, injection, extraction, etc.). As an example, equipment 157 and / or 158 may include components, a system, multiple systems, etc. for fracturing, seismic sensing, analysis of seismic data, assessment of one or more fractures, etc.

[0031] exist Figure 1 In the example of , the GUI 120 shows some examples of computing frameworks, including DRILLPLAN, PETREL, TECHLOG, PETROMOD, ECLIPSE, and INTERSECT frameworks (SLB, Houston, Texas).

[0032] The DRILLPLAN framework provides digital well construction planning and includes features for automation of repetitive tasks and validation workflows, enabling rapid generation of drilling programs (eg, digital drill plans, etc.) of improved quality while ensuring consistency.

[0033] The PETREL framework can be part of the DELFI cognitive exploration and production (E&P) environment (SLB, Houston, TX, referred to as the DELFI environment) for geoscience and geoengineering, for example, to analyze subsurface data from exploration to production of fluids from reservoirs.

[0034] One or more types of frameworks may be implemented within or operably coupled to a DELFI environment, which is a secure, cognitive, cloud-based collaborative environment that integrates data and workflows with digital technologies, such as artificial intelligence (AI) and machine learning (ML). As an example, such an environment may provide operations involving one or more frameworks. A DELFI environment may be referred to as a DELFI framework, which may be a framework within a framework. As an example, a DELFI environment may include various other frameworks, which may include, for example, one or more types of models (e.g., simulation models, etc.).

[0035] The TECHLOG framework can process and process field and laboratory data from various geological environments (e.g., deepwater exploration, shale, etc.). The TECHLOG framework can structure wellbore data for analysis, planning, etc.

[0036] The PIPESIM simulator includes a solver that can provide simulation results, such as multiphase flow results (e.g., from the reservoir to the wellhead and beyond the wellhead, etc.), flowline and surface facility performance, etc. The PIPESIM simulator can be integrated, for example, with the AVOCET Production Operations Framework (SLB, Houston, Texas). As an example, one or more reservoirs can be simulated with respect to one or more enhanced recovery techniques (e.g., considering thermal processes such as steam assisted gravity drainage (SAGD), etc.). As an example, the PIPESIM simulator can be an optimizer that can optimize one or more operating scenarios at least in part via simulation of physical phenomena.

[0037] The ECLIPSE framework provides numerical solutions for reservoir simulators (eg, as a computational framework) for quickly and accurately predicting the dynamic behavior of various types of reservoirs and development scenarios.

[0038] The INTERSECT framework provides a high-resolution reservoir simulator for simulating detailed geological features and quantifying uncertainties, for example, by creating accurate production scenarios, and by integrating accurate models of surface facilities and field operations, the INTERSECT framework can produce reliable results that can be continuously updated through real-time data exchange (e.g., from one or more types of data acquisition equipment in the field, which 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 in the field, can provide detailed chemical enhanced oil recovery (EOR) recipes, where such recipes can be implemented in the field, can analyze the application of steam injection and other thermal EOR technologies for implementation in the field, advanced production control in terms of reservoir coupling and flexible field management, and the flexibility to write custom solutions for improved modeling and field management control. As with the other example frameworks, the INTERSECT framework can be used as part of the DELFI cognitive E&P environment, for example, for rapid simulation of multiple concurrent situations. For example, a workflow can utilize one or more of the DELFI on-demand reservoir simulation features.

[0039] The DELFI environment described above provides various features related to the workflow of subsurface analysis, planning, construction, and production, as shown, for example, in the workspace frame 110. Figure 1 As shown, output from the workspace framework 110 may be used to direct, control, etc., one or more processes in the geological environment 150, and feedback 160 may be received in one or more forms via one or more interfaces (e.g., acquired data regarding operating conditions, equipment conditions, environmental conditions, etc.).

[0040] As an example, the workflow may proceed to a geological and geophysical "(G&G") service provider, which may generate well trajectories, which may involve executing one or more G&G software packages.

[0041] exist Figure 1 In the example of , visualization features 123 can be implemented via workspace framework 110, for example, to perform tasks associated with one or more of a subsurface region, planning operations, constructing wells and / or surface fluid networks, and producing from a reservoir.

[0042] As an example, the visualization process can implement one or more of the various features that can be suitable for one or more web applications. For example, the template can involve the use of JAVASCRIPT object notation format (JSON) and / or one or more other languages / formats. As an example, the framework can include one or more converters. For example, consider a JSON to PYTHON converter and / or a PYTHON to JSON converter. In such a method, one or more features of a framework that can be available in a language can be accessed via a converter. For example, consider an APACHE SPARK framework that can include features available in a specific language, where a converter can convert code in another language to the specific language so that one or more features can be utilized. As an example, a production site can include various types of equipment, can operate with various frameworks, etc., where one or more languages ​​can be utilized. In such an example, a converter can provide feature flexibility and / or compatibility.

[0043] As an example, a visualization feature may provide visualization of various earth models, attributes, and the like in one or more dimensions. As an example, a visualization feature may provide rendering of information in multiple dimensions, which may optionally include multi-resolution rendering. In this example, the information being rendered may be associated with one or more frameworks and / or one or more data stores. As an example, a visualization feature may include one or more control features for controlling a device, which may include, for example, a field device that may perform one or more field operations. As an example, a workflow may utilize one or more frameworks to generate information that may be used to control one or more types of field devices (e.g., drilling equipment, cable equipment, fracturing equipment, etc.).

[0044] Although in Figure 1Several simulators are shown in the examples, but one or more other simulators may be utilized in addition or alternatively. For example, consider the VISAGE geomechanical simulator (SLB, Houston, Texas) or the PETROMOD simulator (SLB, Houston, Texas), etc. The VISAGE simulator includes a finite element numerical solver that can provide simulation results, such as compaction and settlement of geological environments, well and completion integrity in geological environments, cap rock and fault-seal integrity in geological environments, fracture behavior in geological environments, heat recovery in geological environments, CO2 processing, etc. The PETROMOD framework provides petroleum system modeling capabilities that can combine one or more of seismic, well, and geological information to model the evolution of sedimentary basins. The PETROMOD framework can predict whether and how a reservoir will be filled with hydrocarbons, including the source and time of hydrocarbon generation, migration routes, quantity, and type of hydrocarbons under underground or surface conditions. The MANGROVE simulator (SLB, Houston, Texas) provides optimization of production enhancement designs (e.g., production enhancement treatment operations, such as hydraulic fracturing) in a reservoir-centric environment. The MANGROVE framework can combine scientific and experimental work to predict geomechanical propagation of hydraulic fractures, reactivation of natural fractures, etc., as well as production predictions within a 3D reservoir model (e.g., production from the drainage zone of the reservoir, where fluids move to and / or from the well via one or more types of fractures). The MANGROVE framework can provide results related to the heterogeneous interactions between hydraulic fracture networks and natural fracture networks (which can help optimize the number and location of fracture treatment stages (e.g., stimulation treatments)), such as those related to improving perforation efficiency and recovery.

[0045] Figure 2 An example of a geological environment 210 (geological environment 210 includes reservoirs 211-1 and 211-2, which may be faulted due to faults 212-1 and 212-2), an example of a network of devices 230, an enlarged view of a portion of the network of devices 230, referred to as network 240, and an example of a system 250 are shown. Figure 2 Some examples of offshore equipment 214 for oil and gas operations associated with reservoir 211-2 and onshore equipment 216 for oil and gas operations associated with reservoir 211-1 are shown. Figure 2 In the example of FIG. 2 , geological environment 210 may include fluids such as oil (o), water (w), and gas (g), which may be layered in reservoirs 211 - 1 and 211 - 2 .

[0046] exist Figure 2In the example of , equipment 214 and 216 may include one or more of drilling equipment, cable equipment, production equipment, etc. For example, consider equipment 214 to include a drilling rig that can drill into the formation to reach a reservoir target where a well can be completed for hydrocarbon production. As an example, equipment 216 may include production equipment such as a wellhead, valves, pump equipment, gas processing equipment, etc. As an example, Figure 1 One or more features of the system 100 may be used for operations in the geological environment 210. For example, consider utilizing a drilling or well planning framework, a drilling execution framework, a production framework, etc. to plan, execute, etc. one or more drilling operations, production operations, etc.

[0047] exist Figure 2 In FIG. 2 , network 240 may be an example of a relatively small production system network. As shown, network 240 forms a somewhat tree-like structure, where flow lines represent branches (eg, segments) and nodes represent nodes. Figure 2 As shown, network 240 provides for the transport of fluids (e.g., oil, water, and / or gas) from well locations along flow lines interconnected at nodes, ultimately to a central processing facility (CPF). In the case where the fluids include solids (e.g., sand, etc.), one or more pieces of equipment may provide solid removal, collection, etc.

[0048] exist Figure 2 In the example of , various portions of network 240 may include pipelines. For example, consider a perspective view of a geological environment including two pipelines, which may be a pipeline to Man1 and a pipeline to Man3 in network 240, where Man1, Man2, and Man3 are manifolds. Figure 2 In the example of FIG. 2 , various portions of network 230 may include one or more pumps, which may include one or more surface pumps and / or underground pumps.

[0049] like Figure 2As shown, the example system 250 includes one or more information storage devices 252, one or more computers 254, one or more networks 260, and instructions 270 (e.g., organized as one or more sets of instructions). With respect to the one or more computers 254, each computer may include one or more processors (e.g., or processing cores) 256 and a memory 258 for storing, for example, instructions 270 (e.g., one or more sets of instructions) that can be executed by at least one of the one or more processors. As an example, the computer may include one or more network interfaces (e.g., wired or wireless), one or more graphics cards, a display interface (e.g., wired or wireless), etc. As an example, images such as surface images (e.g., satellite, geological, geophysical, etc.) may be stored, processed, transmitted, etc. As an example, the data may include SAR data, GPS data, etc., and may be stored, for example, in one or more of the storage devices 252. As an example, the information that may be stored in one or more of the storage devices 252 may include information about the device, the location of the device, the orientation of the device, fluid properties, etc.

[0050] As an example, the instructions 270 may include instructions (e.g., stored in the memory 258) executable by at least one of the one or more processors 256 to instruct the system 250 to perform various actions. As an example, the system 250 may be configured such that the instructions 270 provide for establishing, for example, a network modeling may be performed (see, e.g., Figure 1 As an example, one or more sets of instructions may be used to perform one or more methods, techniques, etc., and the one or more sets of instructions may be, for example, Figure 2 Instructions 270.

[0051] As an example, Figure 2 The various graphics in may be part of a graphical user interface (GUI) that may be generated using executable instructions that may be executed locally and / or remotely using local and / or remote display devices (e.g., mobile devices, workstations, etc.).

[0052] Figure 3 Examples of equipment 310, 330, 350, and 370 that can be used to move fluids in the field are shown. As shown, equipment 310 can include a gas lift device, equipment 330 can include a rod pump device, equipment 350 can include an electric submersible pump (ESP) device, and equipment 370 can include a progressive cavity pump (PCP) device. As an example, the pump can be a compressor (e.g., a centrifugal compressor, a reciprocating compressor, etc.). As described above, Figure 2The network 230 may include one or more pumps, which may include one or more types of devices 310, 330, 350, and 370 and / or one or more other types of devices.

[0053] exist Figure 3 In the example, devices 310, 330, 350, and 370 may be artificial lift devices, wherein one or more controllers 312, 332, 352, and 372 may be included in and / or operably coupled to devices 310, 330, 350, and 370. In such an example, one or more features of system 250 may be included in and / or operably coupled to one or more controllers 312, 332, 352, and 372.

[0054] Artificial lift devices can add energy to the fluid column in a wellbore with the goal of activating and / or increasing the production of the well. Artificial lift systems can utilize a range of operating principles (e.g., rod pumping, gas lift, electric submersible pumps, etc.). Thus, artificial lift devices can operate by utilizing one or more resources (e.g., fuel, electricity, gas, etc.).

[0055] Gas lift is an artificial lift method in which gas is injected into the production tubing to reduce the hydrostatic pressure of the fluid column. The resulting reduction in bottomhole pressure allows reservoir fluids to enter the wellbore at a higher rate. In gas lift, the injected gas may be transported down the tubing-casing annulus and into the production system through a series of gas lift valves. In this method, the gas lift valve position, operating pressure, and gas injection rate may be operating parameters (e.g., determined by specific well conditions, etc.).

[0056] A sucker rod pump is an artificial lift pumping system that uses a surface power source to drive a downhole pump assembly. For example, a beam and crank assembly can generate reciprocating motion in a sucker rod string connected to the downhole pump assembly. In such an example, the pump can include a plunger and a valve assembly to convert the reciprocating motion into vertical fluid motion. As an example, electricity and / or fuel can be used to drive a sucker rod pump. For example, the prime mover of the sucker rod pump can be an electric motor or an internal combustion engine.

[0057] ESP is an artificial lift system that utilizes an electrically driven downhole pumping system. In such an example, the pump may include a stepped centrifugal pump section that may be specifically configured to suit the production and wellbore characteristics of a given application. ESP systems may offer flexibility in a range of sizes and output flow rates.

[0058] The PCP is a sucker rod type pumping unit using a rotor and a stator. In this method, the rotation of the rod with the help of an electric motor at the surface causes the fluid contained in the cavity to flow upward. The PCP can be called a rotary positive displacement unit.

[0059] A compressor may be a mechanical device for increasing the pressure of a compressible fluid (e.g., gas, steam, etc.). A compressor may increase the pressure of a fluid and reduce the volume of the fluid to aid in fluid delivery. Compressors are used in the oil and gas industry for applications such as, for example, gas lift, fluid collection, fluid processing operations, transmission and distribution systems, reinjection of fluids (e.g., for pressure maintenance, etc.), reducing the volume of a fluid for storage or shipment via tanker, etc.

[0060] exist Figure 3 In an example, one or more sensors may be included. For example, consider a meter coupled to the downhole end of the ESP, wherein signals from the sensors of the meter may be transmitted to surface equipment using a power cable and / or a dedicated meter cable. For example, consider a PHOENIX meter (SLB, Houston, Texas), which includes sensors that may measure intake pressure, temperature, oil temperature, winding temperature, vibration, current leakage, and / or pump discharge pressure. The meter may be operably coupled to a controller, which may, for example, provide control of ESP derotation, ESP sanding, ESP flux, and ESP gas-related issues. For example, during operation in the presence of sand (e.g., suspended solids, etc.), sand may accumulate in one or more stages of the ESP, wherein a control scheme may be used to remove at least a portion of the sand from the ESP.

[0061] As an example, a PCP may be suitable for production in a well characterized by high viscosity fluids and high sand flushing, where the PCP has some sand lifting capability. However, sand may accumulate, where a control scheme may be utilized to clear at least a portion of the sand from the PCP.

[0062] As an example, a rod pump can be operated as a stroke pump to release sand and other materials. In such an example, in order to minimize damage to the plunger and barrel, a groove body plunger can be used to capture the sand and take the sand away from those components.

[0063] As an example, a gas lift device may be used in applications where abrasive materials (e.g., sand) may be present, and may be used in low production rate, high gas / oil ratio wells or deviated wellbores. As an example, a gas lift device such as a set mandrel may utilize a slide-line retrievable gas lift valve that may be pulled and replaced without disturbing the pipeline.

[0064] As an example, equipment may include water injection equipment. For example, consider an enhanced oil recovery (EOR) method, in which a small amount of surfactant is added to an aqueous fluid injected to purge the reservoir. In such an example, the presence of a surfactant reduces the interfacial tension between the oil phase and the aqueous phase, and can also change the wettability of the reservoir rock (for example, to improve oil recovery). In such an example, the movement of fluid (for example, oil and / or water) and / or the presence of a surfactant can carry the particles of the reservoir rock to one or more production wells, where such particles (for example, sand) can cause sand events, regardless of whether one or more of the one or more production wells include artificial lift equipment. As water injection becomes more common in the world, the increase of problems associated with sand (for example, sand flows into the production well) can be expected.

[0065] As an example, the apparatus may include one or more chokes, which may include a surface choke and / or a downhole choke. A choke is a device including an orifice that can be used to control the flow of a fluid through the orifice, for example, to control fluid flow, downstream system pressure, etc. Chokes may be used in a variety of configurations, including fixed and adjustable chokes. Adjustable chokes enable fluid flow and pressure parameters to be changed as desired (e.g., for process, production, etc.).

[0066] The adjustable choke includes a valve that can be adjusted to control well operation, for example, to reduce the pressure of the fluid from the high pressure in the closed wellbore to atmospheric pressure. The adjustable choke can be adjusted (e.g., fully open, partially open, or closed) to control the pressure drop. As an example, the adjustable choke can be manually adjustable or adjustable via a controller that can be integrated into the adjustable choke or operably coupled to the adjustable choke. The controller for the adjustable choke can be responsive to a locally generated and / or remotely generated signal.

[0067] A downhole choke or bottom hole choke may be a downhole device used to control fluid flow under downhole conditions. As an example, a downhole choke may be removable via a slide line intervention, where the downhole choke may be located in a landing sub in a tubing string. In some cases, a downhole choke may act as a flow regulator and participate in downhole pressure drop, which may help reduce the likelihood of hydrate issues.

[0068] Figure 4 An example of an ESP system 400 is shown, which includes an ESP 410 as an example of a device that may be placed in a geological environment. As an example, the ESP may be expected to function in the environment for an extended period of time (eg, optionally, on the order of years).

[0069] exist Figure 4In the example of FIG. 4 , an ESP system 400 includes a network 401, a well 403 disposed in a geological environment (e.g., with surface equipment, etc.), a power source 405, an ESP 410, a controller 430, a motor controller 450, and a variable speed drive (VSD) unit 470 (e.g., a surface control unit). The power source 405 can receive power from a power grid, an on-site generator (e.g., a natural gas-driven turbine), or other sources. The power source 405 can supply, for example, a voltage of about 4.16 kV.

[0070] As shown, well 403 includes a wellhead, which may include a choke (e.g., a choke valve). For example, well 403 may include a choke valve to control various operations, such as reducing the pressure of the fluid from the high pressure in the closed wellbore to atmospheric pressure. The wellhead may include one or more sensors, such as a temperature sensor, a pressure sensor, a solid sensor, etc. As an example, the wellhead may include a temperature sensor and a pressure sensor.

[0071] Regarding ESP 410 , it is shown as including cable 411 (e.g., or cable), pump 412 , gas handling features 413 , pump inlet 414 , motor 415 , one or more sensors 416 (e.g., temperature, pressure, strain, current leakage, vibration, etc.) and protector 417 .

[0072] As an example, the ESP may include a REDA HOTLINE high temperature ESP motor. As an example, the ESP motor may include a three-phase squirrel cage with two-pole induction. As an example, the ESP motor may include steel stator laminations, which may help focus magnetic forces on the rotor, for example, to help reduce energy losses. As an example, the stator windings may include copper and insulators.

[0073] As an example, one or more sensors 416 of ESP 410 may be part of a digital downhole monitoring system. For example, consider the PHOENIX MULTISENSOR XT150 system (SLB, Houston, Texas). The monitoring system may include a base unit that is operably coupled to an ESP motor (see, e.g., motor 415), e.g., directly, via a motor-base crossover, etc. As an example, such a base unit (e.g., a base meter) may measure intake pressure, intake temperature, motor oil temperature, motor winding temperature, vibration, current leakage, etc. As an example, the base unit may transmit information via a power cable that provides power to the ESP motor, and may also receive power via such a cable.

[0074] like Figure 4As shown in the example of , one or more sensors 416 may include circuitry 460. As an example, such circuitry 460 may include one or more processors and a memory that may store processor-executable instructions. As an example, such instructions may include instructions for one or more monitoring and / or control features. As an example, circuitry 460 may be used as an edge device and / or a portion of an edge device (see, e.g. Figure 5 ).

[0075] As an example, a remote unit may be provided that may be located at the pump discharge (e.g., at the end opposite the pump inlet 414). As an example, the base unit and remote unit may measure inlet and discharge pressures on a pump (see, e.g., pump 412) in combination, e.g., for analyzing a pump curve. As an example, an alarm may be set for one or more parameters (e.g., measurements, parameters based on measurements, etc.).

[0076] In the case where the system includes a base unit and a remote unit (such as those of the PHOENIX MULTISENSOR XT150 system), the units may be linked via electrical wires. Such an arrangement provides power from the base unit to the remote unit and allows communication between the base unit and the remote unit (e.g., at least the transmission of information from the remote unit to the base unit). As an example, the remote unit is powered via a wired interface to the base unit so that one or more sensors of the remote unit can sense a physical phenomenon. In such an example, the remote unit may then transmit the sensed information to the base unit, which in turn may transmit such information to the ground unit via a power cable configured to provide power to the ESP motor.

[0077] exist Figure 4 In the example of FIG. 4 , well 403 may include one or more well sensors 420, such as an OPTICLINE sensor or a WELLWATCHER BRITEBLUE sensor (SLB, Houston, Texas). Such sensors are fiber-optic based and can provide real-time sensing of temperature, for example, in SAGD or other operations. Figure 1 As shown in the example of , the well may include a relatively horizontal section. Such a section may collect heavy oil that is heated in response to steam injection. Temperature measurements along the length of the well may provide feedback, for example, to understand the downhole conditions of the ESP. The well sensor may extend to a considerable distance in the well and may extend beyond the location of the ESP.

[0078] exist Figure 4In the example, controller 430 may include one or more interfaces, for example, for receiving, transmitting, or receiving and transmitting information with motor controller 450, VSD unit 470, power source 405 (e.g., gas fuel turbine generator, power company, etc.), network 401, equipment in well 403, equipment in another well, etc.

[0079] As an example, the controller 430 may include features of an ESP motor controller and optionally replace the ESP motor controller 450. For example, the controller 430 may include features of an INSTRUCT motor controller (SLB, Houston, Texas) and / or features of a UNICONN motor controller (SLB, Houston, Texas), which may be connected to a SCADA system, an ESPWATCHER monitoring system (SLB, Houston, Texas), a LIFTWATCHER system (SLB, Houston, Texas), a LIFTIQ system (SLB, Houston, Texas), etc. The UNICONN motor controller and / or the INSTRUCT motor controller may perform some control and data acquisition tasks for an ESP, surface pump, or other monitoring well. As an example, the motor controller may interface with the above-mentioned PHOENIX monitoring system, for example, to access pressure, temperature, and vibration data and various protection parameters and to provide DC power to downhole sensors. As an example, the motor controller may interface with a fixed speed drive (FSD) controller or a VSD unit (e.g., such as VSD unit 470).

[0080] For FSD controllers, the motor controller can monitor ESP system three-phase current, three-phase surface voltage, power supply voltage and frequency, ESP spin frequency and leg grounding, power factor and motor load. For VSD units, the motor controller can monitor VSD output current, ESP operating current, VSD output voltage, power supply voltage, VSD input and VSD output power, VSD output frequency, drive load, motor load, three-phase ESP operating current, three-phase VSD input or output voltage, ESP spin frequency and leg grounding.

[0081] exist Figure 4 In the example of , the ESP motor controller 450 includes various modules to handle, for example, virtual flow estimation, de-swirl of ESP, sanding of ESP, flux of ESP, gas problem of ESP, emulsion presence, emulsion formation, etc. The motor controller 450 may include any of the various features, additionally, alternatively, etc.

[0082] exist Figure 4In an example of a VSD unit 470, the VSD unit 470 may be a low voltage drive (LVD) unit, a medium voltage drive (MVD) unit, or other type of unit (e.g., a high voltage drive that can provide a voltage in excess of about 4.16 kV). As an example, the VSD unit 470 may receive power having a voltage of about 4.16 kV and control a motor to a load having a voltage from about 0 V to about 4.16 kV. The VSD unit 470 may include a control circuit, such as a SPEEDSTAR MVD control circuit (SLB, Houston, Texas).

[0083] Figure 5 An example of a system 500 and an example of an architecture 501 are shown, where the system 500 may include various local components that may communicate with one or more remote components. Figure 5 As shown in the example of , architecture 501 can provide one or more security components 502, one or more machine learning models 503, data 504, objects 505, predictive techniques 506 (e.g., identification, detection, prediction, etc.), analytical techniques 507, and outputs 508. As an example, system 500 can be operably coupled to one or more pumps, which can include one or more ESPs. As an example, system 500 can operate as a controller, motor controller, etc., and / or provide information to a controller, motor controller, etc.

[0084] As shown, the system 500 may include a power source 513 (e.g., solar, generator, battery, grid, etc.) that may provide power to the edge framework gateway 510, which may include one or more computing cores 512 and one or more media interfaces 514, which may, for example, receive a computer-readable medium 540, which may include one or more data structures, such as an operating system (OS) image 542, a framework 544, and data 546. In such an example, the OS image 542 may enable one or more of the one or more cores 512 to establish an operating system environment suitable for executing one or more applications. For example, the framework 544 may be an application suitable for execution in an operating system established in the edge framework gateway 510.

[0085] exist Figure 5 In the example of , edge framework gateway 510 ("EF") may include one or more types of interfaces suitable for receiving and / or sending information. For example, consider one or more wireless interfaces that may provide local communications at a site, such as to one or more local devices (which may include device 532, device 534, and device 536) and / or to one or more remote sites 552 and 554. In such an example, fewer or more devices may be included.

[0086] As mentioned, Figure 4 Circuit 460 of one or more sensors 416 of the example can be used as an edge device and / or a part of an edge device. As an example, circuit 460 can include and / or host a framework such as framework 544. As an example, circuit 460 can include and / or host containerized instructions (see, e.g. Fig.13 As an example, the circuit 460 may be operably coupled to one or more pieces of ground equipment, such as, for example Figure 5 The edge framework gateway 510 of FIG. As an example, the ESP can be equipped with its own edge computing resources, which can operate at least partially downhole to monitor and / or control the ESP. In various examples, one or more downhole sensors can acquire one or more pressures, one or more temperatures, drive frequencies, etc., which can be inputs to one or more models, monitoring and / or control components, etc.

[0087] As an example, devices 532, 534, and 536 may include one or more types of devices, such as Figure 3 Device 310, device 330, device 350, and device 370. As an example, the devices may include non-artificial lift devices and / or artificial lift devices.

[0088] As an example, EF 510 may be installed at a site that is some distance from a city, town, etc. In such an example, EF 510 may be accessed via a satellite communication network and / or one or more other networks that may transmit, receive, etc. data, control instructions, etc.

[0089] As an example, one or more pieces of equipment at a site may be locally and / or remotely controllable. For example, a local controller may be an edge framework-based controller that may issue control instructions to a local device via a local network, and a remote controller may be a cloud-based controller or other type of remote controller that may issue control instructions to a local device via one or more networks that reach outside of the site. As an example, a site may include features for implementing local and / or remote control. As an example, a controller may include an architecture such as a supervisory control and data acquisition (SCADA) architecture.

[0090] Satellite communications tend to be slower and more expensive than other types of electronic communications due to factors such as distance, equipment, deployment, and maintenance. For well sites that do not have other forms of communications, satellite communications may be limiting in one or more aspects. For example, where the controller is to operate in real-time or near real-time, a cloud-based control approach may introduce too much latency.

[0091] like Figure 5As shown in the example of , EF 510 can be deployed where it can operate locally with one or more devices 532, 534, and 536. As an example, EF 510 can include switching and / or communication capabilities, such as for information transmission between devices.

[0092] As needed, communication may occur from time to time between EF 510 and one or more remote sites 552, 554, etc., which may be done via satellite communications where delays and costs are tolerable. As an example, CRM 540 may be a removable drive that may be brought to the site via one or more modes of transportation. For example, consider airdrops, people via helicopters, airplanes, ships, etc.

[0093] As about Figure 5 As explained, the EF may be executed within a gateway, such as an AGORA gateway (e.g., considering one or more processors, memory, etc., which may be deployed as a "box" that may be locally powered and may communicate locally with other devices via one or more interfaces). As an example, one or more pieces of equipment may include computing resources that may be similar to, or more or less than, the computing resources of an AGORA gateway. As an example, an AGORA gateway may be a network device with various networking capabilities.

[0094] As an example, the gateway may include one or more features of an AGORA gateway (e.g., v.202, v.402, etc.) and / or another gateway. For example, consider features such as an INTEL ATOM E3930 or E3950 dual-core with DRAM and eMMC and / or SSD. Such a gateway may include a trusted platform module (TPM), which may provide secure and measured boot support (e.g., via hashing, etc.). The gateway may include one or more interfaces (e.g., Ethernet, RS485 / 422, RS232, etc.). Regarding power, the gateway may consume less than about 100W (e.g., consider less than 10W or less than 20W). As an example, the gateway may include an operating system (e.g., consider LINUX DEBIAN LTS or another operating system). As an example, the gateway may include a cellular interface (e.g., 4G LTE, 5G, etc. with a global modem / GPS). As an example, the gateway may include a WIFI interface (e.g., 802.11a / b / g / n). As examples, the gateway may operate using AC 100-240V, 50 / 60Hz, or 24VDC.Regarding size, consider a gateway having a protective box with dimensions of approximately 10 inches by 8 inches by 4 inches (eg, 25 cm by 20.3 cm by 10.1 cm).

[0095] As an example, a gateway can be part of a drone. For example, consider a mobile gateway that can take off and land, where it can land to operably connect with a device, thereby providing control of such a device. In such an example, the device may include a landing pad. For example, a drone can be guided to a landing pad, where the drone can interact with the device to control the device. As an example, a wellhead may include a landing pad, where the wellhead may include one or more sensors (e.g., temperature and pressure), and where the mobile gateway may include features for generating fluid flow values ​​using information from one or more sensors. In such an example, the mobile gateway may issue one or more control instructions (e.g., to a throttle, a pump, etc.).

[0096] As an example, the gateway itself may include one or more cameras so that the gateway can record conditions. For example, consider a motion detection camera that can detect the presence of an object. In such an example, an image of the object and / or an analysis (e.g., image recognition) thereof may be sent (e.g., via a satellite communication link) so that risk can be assessed at a site remote from the gateway.

[0097] As an example, the gateway may include one or more accelerometers, gyroscopes, etc. As an example, the gateway may include circuitry that may perform seismic sensing indicative of ground movement. Such circuitry may be suitable for detecting and recording device movement and / or movement of the gateway itself.

[0098] As explained, the gateway may include features that enhance its operation at a remote site that may be far from a city, town, etc., such that travel to the site and / or communication with equipment at the site is problematic and / or expensive. As explained, the gateway may include an operating system and a memory that may store one or more types of applications that may be executed in the operating system environment. Such applications may include one or more security applications, one or more control applications, one or more simulation applications, etc.

[0099] As an example, various types of data may be available, for example, considering real-time data and ad hoc data from devices. In various examples, data from sources connected to the gateway may be real-time, ad hoc data, episodic data, etc. As an example, laboratory test data may be available, which may be used to fine-tune one or more models (e.g., locally, etc.). As an example, data from a framework (such as the AVOCET framework) may be utilized, where the results and / or its data may be sent to the edge. As an example, one or more types of ad hoc data may be stored in a database and sent to the edge.

[0100] With respect to real-time data, it may include data acquired via one or more sensors at a site and then transmitted after acquisition to, for example, a framework, which may be local, remote, or partially local and partially remote. Such transmission may be as a stream (e.g., streaming data) and / or as a batch. With respect to batches, a buffer may be utilized in which a certain amount of data may be stored and then transmitted as a batch. In various instances, a sampling rate or sampling frequency may be used to characterize real-time data. For example, consider 1 Hz as a sampling frequency sufficient to track various types of physical phenomena that may occur during well operations. As an example, the sensor and / or framework may provide adjustment of sampling (e.g., at the sensor and / or at the framework). In various instances, data from multiple sensors may be at the same sampling rate or at one or more sampling rates.

[0101] As explained, various systems may operate in a local manner, optionally without access to a network such as the Internet. For example, a site may be relatively remote, with satellite communications present as the primary communication mode, which may be expensive and / or low bandwidth. In such a scenario, security may resort to local features rather than remote features, such as a remote authentication server.

[0102] The authentication server can provide a network service that applications use to authenticate credentials, which can be or include an account name and password of a user (e.g., a person and / or a machine). When a client submits one or more valid credentials to the authentication server, the authentication server can generate an encrypted ticket that the client can then use to access one or more services.

[0103] exist Figure 5 In the example of , edge framework 544 can be an edge-enabled data processing framework. As an example, such a framework can include features that perform one or more of the following tasks: real-time data cleansing to synchronize information from existing well metering (e.g., wellhead, pipeline, flow, ESP, etc.); real-time execution of one or more machine learning (including self-learning) models (e.g., one or more ML models that can identify one or more problems, etc.); and communication of control set points to controllers (e.g., actuatable valves, etc.) and / or one or more other pieces of equipment. As mentioned, the edge framework can use downhole circuits (see, e.g., Figure 4 The downhole circuit may be deployed in the form of a circuit 460), which may be a downhole circuit operably connected to a surface circuit, etc.

[0104] System 500 can be part of an infrastructure used as a secure gateway to transmit monitoring to an operator's monitoring station or its own monitoring platform. The presence of such a gateway can also support operators to introduce one or more additional IIOT (Industrial Internet of Things) implementations.

[0105] As an example, Figure 3 and Figure 4 One or more of the controllers may include or provide access to one or more frameworks, applications, etc. As an example, Figure 3 and Figure 4 One or more of the controllers may include Figure 5 One or more features of system 500.

[0106] As explained, ESP can be implemented at the site for pumping fluids, whether for injection or for production. For example, ESP can be used in stimulation treatments to inject fluids including various chemicals, and ESP can be used as an artificial lift technique to assist in producing fluids from a reservoir.

[0107] Since ESP finds various uses in various environments, knowledge about operation, performance, etc. can be spread among the various domains, each of which may have its own experts.

[0108] As an example, the system may provide failure predictions and / or operational life estimates for one or more pumps for one or more purposes, which may include control and / or predictive health monitoring (PHM).

[0109] Pump systems are prone to failures for a variety of reasons that can cause significant downtime and impact production. As an example, a method can include predicting potential failures in advance by using a sequence of complex ML models. In such an example, the model can learn the pump behavior during normal operation and provide a probability curve that can be used to issue one or more types of signals (e.g., alarms, controls, etc.). For example, when a precursor to a failure can be detected, the signal can alert a controller and / or one or more operators. This operational life prediction can be observed over time, along with appropriate thresholds, to allow initial proactive actions to be taken to mitigate the failure.

[0110] With respect to the ESP, failures may be caused by a variety of reasons, including operating conditions such as pump wear, no flow, pipeline leaks, gas disturbances that may or may not exhibit a predefined signal signature. ESP failures may also be caused by sudden and unexpected conditions, such as startup and shutdown cycles that cause a lot of stress, and electrical failures that do not exhibit a signature. The ESP can be connected to a cable, which can have a considerable length (e.g., hundreds of meters, thousands of meters). The cable can include various conductors, insulation, and strength members, and can include armor or another material for protection. In various instances, the cable can support the ESP, while in other instances, the ESP can be supported in another way. As explained, the cable can provide multi-phase power for the operation of the electric motor, and in the case of a meter equipped with one or more sensors, the cable can provide power for the operation of the meter and the transmission of acquired sensor data. Various types of faults can be cable related (e.g., ground faults, etc.).

[0111] Regarding the method of fault detection, it can operate based on a set of assumptions, such as: a) a specific behavior is exhibited by the signal during the duration of the fault; b) there is consistency in the signature of each fault event; and c) there is no variability in the signature (e.g., one signature corresponds to all faults). In addition, since this behavior is expected during the fault event, this method can be limited to detecting faults in real time at best.

[0112] As mentioned, the system can provide failure prediction. For example, consider a system that can use deep learning techniques to identify and learn potential precursors to failure events and use them to predict pump failure using an integrated approach.

[0113] Figure 6 An example of a system 600 is shown, which includes components for inference data acquisition 610, data preprocessing 620, optional data visualization 625, and fault prediction 630, which can be implemented using a containerized application programming interface (API) 632 for accessing a trained ML model 634. As shown, an edge application architecture can be utilized, in which one or more suitable languages ​​(e.g., JSON, etc.) can be implemented. Figure 6 In the example of , the system 600 can pre-process the data, where the API call can be in the form of a JSON request, where a JSON response can be received. Figure 6 In the example of , system 600 may include one or more edge applications and one or more types of components (e.g., containerized, etc.) that can be accessed using one or more APIs.

[0114] Figure 7Examples of methods 710, 730, and 750 are shown. As shown, method 710 may include a data preprocessing block 711, a drill-down event labeling block 712 that may utilize one or more subject matter experts (SMEs), an evaluation block 713 for evaluating event signatures, a feature engineering block 714 for determining suitable features, a model building block 715 that may be iterated, a testing block 716 for testing the trained model of the model building block 715, and a packaging and deployment block 717 that may provide containerized packaging and deployment, for example, to an edge device.

[0115] As shown, the method 730 includes a data acquisition block 731 for acquiring suitable frequency time series signal data, such as pressure, temperature, current, etc., which can come from one or more ESP systems (e.g., or one or more other pump systems); a pre-processing block 732 for handling outliers, inconsistencies, frequencies, and other quality control issues; and a fault event marking block 733, which can be assisted by (SME), such as causing the data to be marked to mark the start time and end time of the fault event. In such an example, the marking can be binary, where 0 indicates no event and 1 indicates the presence of a fault event. The fault event can be identified by an indicator signal such as motor frequency, flat lining, etc. As shown, the method 730 can also include an evaluation block 734 for evaluating event signatures, a feature engineering block 735 for feature engineering, a model building block 736 for model generation, a model adjustment block 737, and a packaging and deployment block 738.

[0116] In various regions in the data before and after the fault event, regions of abnormal behavior can be identified. As an example, the data can be divided into regions of normal behavior and abnormal behavior, where the underlying assumption is that there are no precursors to faults in regions of normal behavior. This approach provides for creating a data set suitable for ML model training and testing, so that the ML model can be trained for inputs of normal behavior data to replicate the normal behavior data. It can be expected that such a trained ML model will be able to closely replicate the inputs when the pump behaves normally, and experience errors when attempting to replicate the inputs when the pump behaves abnormally because the ML model is not trained on data representing abnormal behavior (e.g., abnormal behavior).

[0117] Regarding the cause of abnormal behavior, consider one or more physical phenomena regarding sanding, gas entrainment, bearings, shafts, motor windings, electrical insulation, etc. For example, the cause can be equipment and / or environmental based. As an example, consider sanding caused by environmental sand, which may cause increased stress on the pump equipment, which may increase temperature, increase wear on components, increase torque requirements, reduce shaft stability, etc. As another example, consider temperature, such as motor temperature, which may depend on flow rate, temperature of the flowing fluid, energy used to drive the motor, etc. In such an example, there may be a relationship between the heat generated due to pumping and / or friction and the heat removed due to the flowing fluid. As an example, the system can provide for issuing commands to control the pump equipment to resolve one or more problems (e.g., sanding, temperature, etc.).

[0118] As explained, the pumping equipment may include various components, which may be mechanical and / or electrical, and may be located above ground and / or underground. With respect to ESP, the cables may be a source of failure, for example, where short circuits may occur due to stress, wear, etc. (note, for example, that ESP cables may be hundreds or thousands of meters in length). Again, referring to Figure 8 , the various plots of GUI 800 show some types of data that can indicate one or more types of abnormal events that can be detected using one or more types of ML models. As an example, such as Fig. 9 The system 900 of system may include a control component that may operate using the output of component 920 and / or component 930. For example, consider issuing one or more control instructions to a pump device to reduce the risk of encountering an event that may result in an undesirable reduction in the operating life of the pump device and / or may help resolve an emerging problem (e.g., increased grit levels, increased motor temperature, etc.).

[0119] As an example, an autoencoder ML model including an encoder portion and a decoder portion can be utilized. An autoencoder ML model can be described in which the encoder portion maps an input to a code, and in which the decoder portion maps the code to a reconstruction of the input. The autoencoder ML model can be a feedforward non-recursive neural network (e.g., similar to a single-layer perceptron) that participates in a multilayer perceptron (MLP), for example, employing an input layer and an output layer connected by one or more hidden layers. The output layer can include the same number of nodes (neurons) as the input layer. As explained, the autoencoder ML model can reconstruct its input (e.g., by minimizing the difference between the input and the output), rather than predicting a target value Y given an input X. As an example, an autoencoder ML model can be trained using unsupervised learning. As explained, data can be acquired and processed so that the data represents the normal behavior of a pump (e.g., a pump system), wherein once available, such processed data can be used to train the ML model in an unsupervised manner.

[0120] As explained, the ML model can simulate normal behavior, where the ML model is built using an autoencoder-decoder model architecture. As explained, the output of such an ML model can be expected to be the same as the input (e.g., the original high-frequency signal). Such an ML model can be used to calculate deviations in the input signal by providing more complete time series data as input. For example, consider a method involving calculating the absolute difference between the input and output (e.g., reconstruction error).

[0121] By using the trained ML model, a reconstruction error can be generated, where a higher error can be an indicator of an input that is behaving abnormally. With respect to feature engineering, the reconstruction error can be used to generate features to build another ML model. For example, consider using a random survival forest model, which can provide predictions as a survival curve for each timestamp. Given the survival curve, the remaining useful life (RUL) of the pump can be assessed.

[0122] Random Survival Forest (RSF) is an ensemble of tree-based learners. RSF ensures that individual trees are decorrelated by: 1) building each tree on a different bootstrap sample of the original training data, and 2) evaluating the splitting criteria on a randomly selected subset of features and thresholds at each node. A prediction can be formed by aggregating the predictions of the individual trees in the ensemble. RSF can be built with many independent decision trees, where each decision tree can receive a subset of random samples and randomly select a subset of variables at each split in the tree for prediction, and where the final prediction can be the average of the predictions of each individual tree.

[0123] RSF can be used, for example, to provide predictions for diseases such as breast cancer. Consider a data set of 686 women and 8 prognostic factors: 1. age, 2. estrogen receptor (estrec), 3. whether hormone therapy is administered (horTh), 4. menopausal status (menostat), 5. number of positive lymph nodes (pnodes), 6. progesterone receptor (progrec), 7. tumor size (tsize), 8. tumor grade (Tgrade). In such an example, the goal may be to predict recurrence-free survival time. The method may include loading the data and converting it to numerical values, then splitting the data 75 / 25 for training and testing. With respect to training, one or more of several splitting criteria may be utilized (e.g., log-rank test, etc.). Once trained, the RSF may be tested using test data.

[0124] With respect to making predictions, a sample can be pulled down from each tree in the forest until it reaches a terminal node. The data in each terminal node can be used to estimate survival and cumulative hazard functions nonparametrically, for example, using Kaplan-Meier and Nelson-Aalen estimators, respectively. As an example, a risk score can be calculated that represents the expected number of events for a particular terminal node. For example, an ensemble prediction can be generated by averaging the trees in the forest. As an example, a method can include generating a predicted survival function that can show differences over certain time periods.

[0125] Reference again Figure 7 , method 750 includes an input channel block 751, an anomaly detector block 752 for health indicator (HI) modeling, an output channel block 753, a reconstruction error block 754, a feature generation block 755, a survival model block 756 for RUL modeling, and a survival probability curve block 757.

[0126] exist Figure 7 In an example of , method 750 may include iterative modeling that may be performed by hyperparameter optimization of a model for use in a case where an inference pipeline may be built. As an example, the model may be packaged as a standalone container (e.g., a DOCKER container, etc.) that may reside on a physical gateway (e.g., an edge device). Figure 6 As explained in the example of , the containerized model set can provide interaction with one or more applications on the edge to provide a survival probability curve of the signal, which can be at each timestamp for each well, which can be visualized on one or more devices.

[0127] Figure 8 An example of a graphical user interface (GUI) 800 is shown, which includes plots relative to time within a period of one year, where the plots may include a fault area plot, a flowing bottom hole pressure (FLP) plot, a motor current plot, a layer interval plot, a label plot, a motor frequency plot, a pressure plot, another pressure plot, a pump status plot, a remote terminal unit phase (RTU_PHS) plot, a temperature plot, another temperature plot, a motor voltage plot, and a working horsepower plot. In the GUI 800, instances where the motor current and motor frequency drop to zero are indicated, each with a corresponding fault area. In addition, plots of pump status and RTU_PHS indicate such instances. Various data channels can provide signatures in the fault area, for example, before zero current and / or zero voltage.

[0128] Fig. 9An example of a system 900 for executing a workflow using various components is shown. As shown, the workflow group of components may include a data evaluation component 910 that may provide for cleaning and preprocessing data, evaluating and mapping fault records, reconciling data frequencies, performing analysis of operating life data (e.g., statistical analysis, etc.), etc.; an anomaly detector component 920 that may provide pump metrics (e.g., motor temperature, differential pressure, etc.), modeling framework support for one or more additional metrics and / or one or more surrogate models, and generation of pump anomaly events for one or more types of pumps that may be used as input to one or more remaining useful life (RUL) models, etc.; and a remaining useful life (RUL) component 930 that may provide for receiving inputs (e.g., residual modes, operating life, and / or one or more other characteristics), a survival model framework for use with dynamic data for one or more types of pumps, extensibility of one or more types of pump events for one or more types of pumps, calibration with fault data and specific anomaly model settings, etc.

[0129] As an example, component 910 can provide an evaluation of pump high frequency data, an evaluation of component level fault records, and a reconciliation of data frequencies; component 920 can provide for building a normal behavior model (e.g., autoencoder, clustering, isolation forest, etc.) and using the normal behavior model to calculate deviations as input features; and component 930 can provide for calculating input features using both normal model deviations and input features from one or more pumps and metadata, and for building one or more RUL models (e.g., time-dependent Cox models, random survival forests as survival curves, etc.).

[0130] In system 900, examples of inputs, histories of one or more pumps used for modeling, and examples of outputs are shown. Regarding inputs, consider Figure 8 With respect to history, consider the history of the pump from installation to failure, where various types of data, actions, etc. occur over time; and with respect to output, consider the operating life (RL) relative to current, estimated, etc., which can be accompanied by a probability relative to a number of days (e.g., 30, 90, 180, etc.), as well as one or more plots such as a plot of survival probability versus operating life in days.

[0131] As an example, system 900 can be used to access historical data for one or more pumps (e.g., a pump system, etc.) so that a complete or more complete history can be established. Fig. 9As shown, the history of the pump can extend from the installation date to the failure or end-of-life date. As an example, one or more ML models can be trained using a complete or more complete data set about the pump, which can be accessed from one or more databases, local memory of the pump or pump system, etc. For example, consider that the ML model trained using historical data is a digital twin of the pump or pump system, so that the trained ML model can be used to test or predict the behavior of the pump or pump system given specific inputs (e.g., operating parameters, environmental conditions, etc.). As an example, the digital twin can be a predictive model for predicting normal expected behavior, which can be used in combination with anomaly detection models and / or survival models, for example, to facilitate field operations, which can include controlling one or more pumps in the field. As an example, a digital twin can be used to generate scenarios that can be intended to improve RUL and / or one or more performance goals, where, for example, in response to an unexpected predicted RUL, one of the scenarios can be selected for implementation in an effort to improve RUL and / or performance. As an example, the digital twin model can be an autoencoder model or another type of model. As an example, the digital twin model can be part of an anomaly detection model, where, for example, the deviation between the actual behavior and the predicted behavior of the digital twin model can be the basis for anomaly detection. As an example, an anomaly detection model may have one or more uses in a system (eg, framework, etc.) for the purpose of managing, monitoring, controlling, etc., one or more pumps.

[0132] As about Figure 6 As explained, the system can be implemented on an edge device, which can include containerized components accessible via one or more APIs, for example, to receive calls from one or more applications (see, for example, system 600). As an example, the edge device can be a controller that can issue one or more control signals to control the operation of a pump, for example, in response to a fault prediction, a survival issue, etc.

[0133] Fig.10An example of an ML model architecture 1000 including an encoder or compressor and a decoder or decompressor is shown. As shown, the ML model architecture 1000 can reduce the dimensionality of a time series input to a compressed latent space representation (e.g., a feature space) that can be used for one or more purposes. However, as mentioned, when decoding or decompressing, the compressed latent space representation can be processed to generate a time series output that can be compared with the time series input. The ML model architecture 1000 can be an input reconstructor, where the input and its machine reconstruction (referred to as output) can be compared. Typically, when the input and output compare favorably, the ML model is accurate, however, as explained, in the case where they are not, this can indicate that the ML model is not properly trained or is not trained on a particular input. As explained, by training the ML model using normal behavior data, it may not be able to properly reconstruct abnormal behavior data (e.g., anomalous behavior data), which can thus be an indication of the presence of abnormal (e.g., abnormal) behavior. As explained, the reconstruction error can be used as a metric to quantitatively and / or qualitatively evaluate the ability and inability of the trained ML model to reconstruct the input, which can be used as a proxy for the presence of abnormal (e.g., abnormal) behavior.

[0134] As an example, one or more other types of models may be utilized, additionally or alternatively, for the purpose of anomaly detection. For example, consider an unsupervised clustering model, an unsupervised isolation forest model, etc. As an example, the type of model may be selected based on the available data. For example, in cases where the amount of data is sufficient to train an autoencoder, an autoencoder may be utilized; however, in cases where the data is insufficient to adequately train the autoencoder, another type of model that requires less data may be selected. As an example, such as Fig. 9 The system 900 of the system can automatically select a model for anomaly detection based at least in part on the amount of available data and / or the quality of the available data. As an example, the system may include one or more components for data enhancement and / or supplementation to increase the amount of effective data. As an example, when more data becomes available, the system may optionally use multiple model types to switch model types. For example, consider running a lightweight model for anomaly prediction while waiting for additional data, and then start training a heavier weight model for anomaly prediction. In such an example, multiple models (e.g., a collection) can be executed at least partially simultaneously to provide anomaly detection, where the output of the model can be evaluated. In such an example, one model may be better at detecting one type of anomaly, while another model may be better at detecting another type of anomaly, or, for example, one model may be good at detecting anomalies, so that the model is selected for use when one or more other models are closed. As an example, the framework can be dynamic, for example, in a data-dependent manner intended to improve anomaly detection, with respect to model selection, model switching, and the like.

[0135] As an example, an unsupervised clustering model can implement a k-means approach, where the number of clusters k can be optimized as a hyperparameter, for example, using an elbow technique or other suitable technique. The elbow technique can use a heuristic to determine the number of clusters in a data set, for example, by plotting the variation in explanation as a function of the number of clusters and selecting the elbow as the number of clusters to use in the plotted curve. Such an approach can also be used to select the number of parameters in one or more other types of data-driven models (e.g., the number of principal components used to describe a data set). With respect to anomaly detection, clustering can highlight (e.g., identify) anomalies in the data. For example, outliers that are considered inappropriate for one or more clusters and / or one or more clusters that may be associated with anomalies may include, for example, relatively few members.

[0136] As an example, an unsupervised isolation forest model can be used to detect anomalies directly using isolation (e.g., how far is a data point from the rest of the data). Such a method can run with a linear time complexity similar to distance-related models (such as k-nearest neighbors (KNN)), which can also be used for anomaly detection. Isolation forests can provide a basis for the properties of outliers, such as there will be few outliers and outliers will be different from non-outliers in characteristics. Isolation forests can introduce a set of binary trees that recursively generate partitions by randomly selecting features and then randomly selecting split values ​​for the features. The partitioning process can continue until it separates the data point from the rest of the sample. In an isolation forest, it can be expected that outliers require fewer partitions on average to obtain isolation compared to normal samples. Each data point can then receive a score based on how easily they are isolated after multiple rounds, so that data points with abnormal scores can be detected as abnormal.

[0137] As explained, RSF may be utilized in a system where a workflow may include computing input features using both normal model deviations and input features from pump data and metadata, and using RSF to build a RUL model for survival curve generation.

[0138] As an example, a time-dependent Cox model can be used for survival purposes (e.g., RUL). The time-dependent Cox model can be a time-dependent Cox regression model (TDCM) that quantifies the effect of repeated measurements of covariates in the analysis of time to event data. As an example, one or more of a pooled logistic regression model (PLRM), a cross-sectional pooled model (CSPM), a Kaplan-Meier survival model, and a log-rank test model can be implemented (e.g., for output, for comparison, for additional output, etc.). As an example, a survival model that takes into account (e.g., statistically) the time of measuring a time-dependent covariate can provide more reliable estimates compared to an unadjusted method.

[0139] Fig.11 Some examples of trees 1100 in RSF are shown, which may differ in one or more aspects. As explained, outputs from multiple trees may be used to generate RUL outputs; note that one or more other techniques may be utilized (eg, time-dependent Cox models, etc.).

[0140] Fig.12 Examples of plots 1210, 1230, and 1250 are shown. Specifically, plot 1210 shows actual values ​​versus predicted values ​​over a period of approximately 300 minutes. In plot 1230, the threshold is indicated by a dashed line, and the predicted survival function with respect to time is indicated by a solid line. Over time, additional data may be received and plots 1210 and 1230 updated. For example, multi-channel data may be received as input to a trained first ML model to generate an output, wherein the output may be compared to the input to generate a reconstruction error, wherein a trained second ML model may utilize such reconstruction error to generate a survival function. Fig.12 In the example of , the inputs may be appropriately windowed and updated based on time intervals, events, etc. For example, consider a 300 minute window with a time interval of approximately 5 minutes. In such an approach, one or more thresholds may be utilized with respect to the probability of survival relative to time to determine a future time at which there is a reduced probability of survival, wherein one or more actions may be invoked between the current time and the future time, e.g., in an effort to increase the probability of survival, mitigate the effects of a failure, prepare for service, etc. With respect to plot 1250, it shows the relationship between the probability of survival and the operating life of the pump in days. As shown, as the number of days progresses, the probability of survival decreases. As explained, such a plot may be generated by using, for example, a graph such as Fig. 9 The system 900 is generated by the workflow of components 910, 920 and 930.

[0141] As an example, a method may include predicting how long a pump system will survive by using a model trained using normal behavior data, where an error in reconstructing the normal behavior data by the trained model may indicate abnormal (e.g., abnormal) behavior, and where a metric related to the reconstruction error may be utilized by another model that may generate a survival function with respect to time.

[0142] Regarding input, it may include multi-channel input, for example, considering one or more of the following channels: current, voltage, intake pressure, exhaust pressure, wellhead pressure, flow rate, temperature, etc.

[0143] As explained, an autoencoder is an example of a type of ML model that can be trained to mimic an input, where the input can be normal behavior data that can be sorted from abnormal behavior data in one or more data sets for the purpose of training. As an example, an area several days before a fault and several days after a fault can be considered to include an abnormal or abnormal behavior area. For example, for an ESP, 15 days before a fault and 5 days after a fault are considered to be a fault area that can contain one or more signatures (e.g., future, current, and past) of a fault. Such one or more signatures can be identified in the reconstruction error of an autoencoder, where a marker can be used to provide the time between the signature and the fault. Such labeled data can be used to train another type of ML model, such as an RSF model. As explained, when a problem is developing, the reconstruction error may be relatively high in a region, where the reconstruction error can be generated on a channel-by-channel basis and / or on one or more other bases (e.g., as a total error of a sum, average, etc.). The RSF model can be used to predict the remaining useful life (RUL) of a pump device, for example, via a survival function relative to time. As explained, although autoencoders and RSF are mentioned, one or more other types of models can be used for anomaly detection and / or RUL prediction.

[0144] As mentioned, supervised training can involve labeling, so that training is performed using labeled data. As explained, labels can involve the time between signatures and failures (see e.g. Figure 8 Given labeled data, the RSF can be trained to predict the time to failure of one or more signatures, which can be signatures of one or more features (e.g., model features, etc.). As explained, the survival function can be scaled from zero to one for probability relative to time, so that the output of the trained RSF can be a curve. By using a threshold (e.g., a probability value less than 1), the future time when the curve crosses the threshold can be identified. For example, the model can predict that within four days, the probability of the pump equipment surviving will be significantly reduced.

[0145] Referring to plot 1230, the survival function spans a 15-day period (e.g., over 20,000 minutes), where plot 1210 spans a 300-minute period (e.g., 5 hours), which can be updated every 5 minutes. In plot 1230, a threshold of 0.75 is used, and the survival function indicates that within the next 15 days, the probability of survival is greater than 0.75. However, if a deviation occurs between the input and the reconstructed input (e.g., output) in plot 1210, the survival function can be generated to include a probability of less than or equal to 0.75 within the next 15 days. In such an example, the future time at which the curve crosses the threshold can be used to trigger an alarm, control action, etc., so that one or more actions can be taken to address the reduced probability of survival before the future time. As explained, the survival function can evolve over time so that an operator and / or controller can determine whether a risk still exists and / or whether one or more actions taken can mitigate the risk.

[0146] The multi-model approach for predicting survival (e.g., remaining useful life, etc.) operates beyond just real-time fault detection and improves the ability to solve one or more problems. As explained, the system can provide predictions of pump equipment behavior in advance by estimating the operating life over time. Such a system can provide predictive health management for one or more groups of pump equipment and allow rapid maintenance to improve production, reduce indirect costs of equipment replacement and save SME inspection time. In various instances, non-productive time (NPT) can be scheduled and / or reduced. For example, consider the amount of time it may take to run into and / or run out of an ESP from a wellbore. In this example, if running into (e.g., running in) and / or running out (e.g., pulling out) can be planned to be at least partially consistent with one or more other types of operations (e.g., NPT or non-NPT), then NPT and / or resource production and / or resource utilization can be reduced. In contrast, unexpected, unpredictable events that require pulling out an ESP at the site can introduce a large amount of NPT because the equipment may not be available at the site, and resource production (e.g., time to get equipment to the site, pull out a failed ESP, and run an operational ESP) can be reduced for an extended period of time.

[0147] As explained, the system can be local, such as an edge device system that can run in real time with one or more edge application computing to make predictions multiple days in the future. Such a system can provide an accurate assessment of the expected failure in the pump system. As an example, the system can be deployed locally and / or on the cloud (e.g., as a SaaS product, etc.).

[0148] As an example, the system may utilize sophisticated ML techniques that do not necessarily require the presence of identifiable, consistent precursors to failure. As an example, the system may be lightweight (e.g., IoT or edge device) with fast response times.

[0149] Fig.13 An example of a system 1300 is shown that can include one or more containerized ML models 1310 hosted within an application programming interface (API) wrapper 1320 within a container 1330. In such an example, the containerized ML model 1310 can be deployed in the field, for example, using an edge computing device 1350 having a predictor 1352 and an edge application 1354 component. In such an example, the edge application component 1354 can provide processing of the data so that the data or data derived therefrom is in a form suitable for ingestion by the predictor 1352 (e.g., appropriate ML model input, etc.).

[0150] In the example experiment, a data science framework (Dataiku) and a container framework (Docker) were implemented. The container framework provides the construction of a software unit that packages executable code and its dependencies so that applications can be quickly and reliably executed from one computing environment to another. As an example, a container can be an image, which is a lightweight, independent executable software package that includes code, runtime, system tools, system libraries, and settings. The container image becomes a "container" at runtime, for example, when running on a suitable engine (e.g., the Docker engine for the Docker container image). As an example, the edge implementation can utilize a framework, such as, for example, a lightweight machine learning framework, such as the Tensorflow Lite (TFL) framework (Google LLC, Mountain View, California).

[0151] As an example, a model inference pipeline can be built inside a container, wrapped around an Asynchronous Server Gateway Interface (ASGI) based API technology to allow real-time requests and responses (see e.g. Fig.13 As an example, a pump signal may be received in real time as expected by the ML model to produce as an output, which may be utilized by another ML model to generate a survival function. This approach allows for continuous monitoring and / or control and parallel computation on multiple gateways.

[0152] Fig.14An example of a system 1400 is shown that includes a data source 1404 that can provide real-time data, such as from one or more sensors deployed in the field as part of or otherwise associated with a pump, an instance of a predictor 1452, an instance of an edge application 1454, and a monitoring and / or control component 1460. As shown, the predictor 1452 can receive an API call (e.g., an API request) issued by the edge application 1454, and wherein the predictor 1452 acts in response to the API call to generate an API response directed to the edge application 1454. As shown, the monitoring and / or control component 1460 can be operably coupled to the edge application 1454 for transmitting data, issuing one or more alarms, issuing control commands, etc. As explained, the edge application 1454 can provide data processing, data export, etc., which can be suitable for generating appropriate inputs for the predictor 1452.

[0153] exist Fig.14 In the example of , examples of ML models 1453 are shown, which may include an anomaly detector ML model and a RULML model. Although an autoencoder and a tree are shown, as mentioned, one or more other types of models may be utilized in addition or alternatively. As explained, data may be used as input and / or may be processed to generate input to one or more ML models, where the one or more ML models may provide one or more outputs. As an example, one or more ML models may output probabilities as predictions, etc., which may be tracked over a period of time or over multiple calls, where if the output is consistent over such a period of time over multiple calls, a final determination may be made, which may then trigger the issuance of an alarm and / or control instructions.

[0154] Fig.14 The example system 1400 can be scalable, customizable, and independent, and provide reliability and robustness when initially deployed to one or more edge computing devices. Such a system can provide real-time results for multiple pumps in multiple wells, thereby reducing downtime for various wells.

[0155] As explained, one or more features of the system may be associated with equipment that may be deployed downhole. For example, Figure 4 The circuit 460 may include Figure 5 One or more features of the edge device 510, which may, for example, provide hosting Fig.141452 and edge applications 1454; note that multiple predictors and / or applications may be hosted (e.g., in whole and / or in part). As an example, the ESP itself may be an "intelligent" ESP that contains one or more executable models (e.g., ML models, etc.) that may provide monitoring and / or control of the ESP. As explained, the circuitry may be carried by a meter that may be mounted to an assembly including a pump and a motor, wherein one or more sensors of the meter may provide input to the monitoring and / or control circuitry.

[0156] As an example, systems, methods, etc. can utilize one or more machine learning features, which can be implemented using one or more machine learning models. Regarding the type of machine learning model, consider one or more of a support vector machine (SVM) model, a k-nearest neighbor (KNN) model, an ensemble classifier model, a neural network (NN) model, etc. As an example, the 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, bootstrap aggregation, adaptive boosting (AdaBoost), stacked generalization, a gradient boosted regression tree, 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., a ridge regression, a least absolute shrinkage and selection operator, an elastic net, a least angle regression), a rule system model (e.g., a cubic, a one rule, a zero rule, repeated incremental pruning to produce error reduction), a regression model (e.g., a linear regression, an ordinary least squares regression, a stepwise regression, a multivariate adaptive regression spline, a local estimation scatter plot smoothing, a logistic regression, etc.) , Bayesian models (e.g., naive Bayes, average dependency estimator, Bayesian belief network, Gaussian naive Bayes, multinomial naive Bayes, Bayesian network), decision tree models (e.g., classification and regression trees, iterative bisection3, C4.5, C5.0, chi-square automatic interaction detection, decision stump, conditional decision tree, M5), dimensionality reduction models (e.g., principal component analysis, partial least squares regression, Sammon map, multidimensional scaling, projection pursuit, principal component regression, partial least squares discriminant analysis, mixture discriminant analysis, quadratic discriminant analysis, regularized discriminant analysis, flexible discriminant analysis, linear discriminant analysis, etc.), instance models (e.g., k-nearest neighbors, learning vector quantization, self-organizing map, local weighted learning, etc.), clustering models (e.g., k-means, k-median, expectation maximization, hierarchical clustering, etc.), etc.

[0157] As an example, a computing framework with a library, a toolbox, etc. can be used to build a machine model, such as the computing framework of the MATLAB framework (MathWorks, Natick, Massachusetts). The MATLAB framework includes a toolbox that provides supervised and unsupervised machine learning algorithms, including support vector machines (SVM), boosting and set decision trees, k- nearest neighbors (KNN), k- means, k- centers, hierarchical clustering, Gaussian mixture models, and hidden Markov models. Another MATLAB framework toolbox is a deep learning toolbox (DLT), which provides a framework for designing and implementing deep neural networks with algorithms, pre-trained models, and applications. DLT provides convolutional neural networks (ConvNet, CNN) and long short-term memory (LSTM) networks to perform classification and regression on images, time series, and text data. DLT includes the features of using custom training loops, shared weights, and automatic differentiation to build network architectures such as generative adversarial networks (GANs) and Siamese networks. DLT provides model exchanges for various other frameworks.

[0158] As an example, the system can utilize one or more recurrent neural networks (RNNs). One type of RNN is called long short-term memory (LSTM), which can be a unit or component (e.g., one or more units) that can be in one or more layers. The LSTM component can be an artificial neural network (ANN) that is designed to recognize patterns in data sequences (such as time series data). When provided with time series data, the LSTM considers time and sequence, so that the LSTM can include a time dimension. For example, consider utilizing one or more RNNs to process time data from one or more sources, optionally combined with spatial data. This approach can identify temporal patterns that can be used to make predictions (e.g., one or more patterns about future times, etc.).

[0159] As an example, the TENSORFLOW framework (Google LLC, Mountain View, California) can be implemented, which is an open source software library for data flow programming, including a symbolic math library, which 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 Berkeley AI Research (BAIR) (University of California, Berkeley, California). As another example, consider the SCIKIT platform (e.g., scikit-learn), which utilizes the PYTHON programming language. As an example, a framework such as the APOLLO AI framework can be utilized (APOLLO.AI GmbH, Germany). As an example, a framework such as the PYTORCH framework can be utilized (Facebook AI Research Lab (FAR), Facebook, Menlo Park, California).

[0160] As an example, a training method may include various actions that may be performed on a dataset to train an ML model. As an example, a dataset may be split into training data and test data, where the test data may be provided for evaluation. The method may include cross-validation of parameters and optimal parameters, which may be provided for model training.

[0161] The TENSORFLOW framework can be run on multiple CPUs and GPUs with optional CUDA (NVIDIA Corp., Santa Clara, California) and SYCL (The Khronos Group Inc., Beaverton, Oregon) extensions for general-purpose computing on graphics processing units (GPUs). TENSORFLOW is available for 64-bit LINUX, MACOS (Apple Inc., Cupertino, California), WINDOWS (Microsoft Corp., Redmond, Washington), and mobile computing platforms including Android (Google LLC, Mountain View, California) and iOS (Apple Inc.) operating systems.

[0162] TENSORFLOW computations can be represented as stateful data flow graphs; note that the name TENSORFLOW derives from the operations that this neural network performs on multidimensional arrays of data. Such arrays can be called "tensors".

[0163] As an example, the device and / or distributed device can utilize TENSORFLOW LITE (TFL) or another type of lightweight framework. TFL is a set of tools that implement on-device machine learning, where models can run on mobile devices, embedded devices, and IoT devices. TFL is optimized for on-device machine learning by addressing latency (no round-trip to the server), privacy (no individual data leaves the device), connectivity (requires Internet connectivity), size (simplified model and binary size), and power consumption (e.g., efficient inference and lack of network connection). TFL provides multi-platform support, covering ANDROID and iOS devices, embedded LINUX, and microcontrollers; diverse language support, including JAVA, SWIFT, Objective-C, C++, and PYTHON; and high performance, with hardware acceleration and model optimization. Machine learning tasks can include, for example, data processing, image classification, object detection, pose estimation, question answering, text classification, etc. on multiple platforms. As an example, Figure 5 The system 500 may utilize one or more features of the TFL framework.

[0164] Fig.15 An example of a method 1500 and an example of a system 1590 are shown. As shown, the method 1500 may include a receiving block 1510 for receiving an input including time series data from a pump device at a well site, wherein the well site includes a wellbore in contact with a fluid reservoir; a process block 1520 for processing the input using a trained first machine learning model as an anomaly detector to generate an output; and a process block 1530 for processing the input and the output using a trained second machine learning model to predict a survival function of the pump device.

[0165] exist Fig.15 Method 1500 is shown in conjunction with various computer readable media (CRM) blocks 1511, 1521, and 1531. Such blocks typically include instructions suitable for execution by one or more processors (or processor cores) to instruct a computing device or system to perform one or more actions. Although various frames are shown, a single medium may be configured with instructions to at least partially allow the various actions of method 1500 to be performed. As an example, a computer readable medium (CRM) may be a computer readable storage medium that is non-transitory and not a carrier wave. As an example, one or more of blocks 1511, 1521, and 1531 may be in the form of processor executable instructions.

[0166] exist Fig.15In an example of , a system 1590, which may be a wellsite system, may include one or more information storage devices 1591, one or more computers 1592, one or more networks 1595, and instructions 1596. With respect to the one or more computers 1592, each computer may include one or more processors (e.g., or processing cores) 1593 and a memory 1594 for storing, for example, instructions 1596 executable by at least one of the one or more processors 1593 (see, for example, blocks 1511, 1521, and 1531). As an example, a computer may include one or more network interfaces (e.g., wired or wireless), one or more graphics cards, a display interface (e.g., wired or wireless), and the like.

[0167] As an example, system 500, system 900, system 1400, etc. may include a memory that may store instructions (such as instructions of one or more of CRM blocks 1511, 1521, and 1531). As explained, the system may be operably coupled to a pump device in the field, where, for example, the system may receive data from the pump device (e.g., directly and / or indirectly), and issue one or more commands (e.g., control signals, etc.) to the pump device as appropriate to cause the pump device to take one or more actions. As explained, the actions may be intended to extend operating life, avoid anomalies, respond to the occurrence of anomalies, etc. In the field of ESP, as explained, anomalies may relate to the device and / or the environment, where one or more actions may address the device and / or the environment (e.g., considering scrub, flow, temperature, etc.).

[0168] Fig.16 An example of a graphical user interface (GUI) 1600 is shown which may include various graphical features such as, for example, a series of renderings of gauges 1610, a plot of pressure versus time 1610, and a plot of temperature versus time 1630. GUI 1600 may be a data analysis GUI that may be selected from a number of different GUIs (e.g., Overview, Data Analysis, Alerts, RL Analysis, Settings, etc.). As shown, gauges 1610 may include a motor temperature gauge, a differential pressure gauge, a drive frequency gauge, and a VSD voltage gauge. Fig.16In an example, GUI 1600 may be used for a specific pump installed at a site, such as an ESP. As an example, GUI 1600 may include a selector graphic for selecting a site, a pump, etc., wherein GUI 1600 may be updated accordingly. As an example, GUI 1600 may provide renderings for more than one pump. For example, consider a fleet of pumps used at one or more sites. In this method, an individual may provide supervision and make control decisions, which may be recommended. For example, consider a predicted event that may occur at a future time, where a recommendation may be to reduce the drive frequency to a specific value that may be indicated by a mark on a drive frequency meter. In such an example, an individual may interact with GUI 1600 to cause the VSD to automatically reduce the drive frequency to the recommended drive frequency (e.g., by clicking on a meter, etc.). In such a method, the graphic may be updated by the operation of the framework to determine whether the predicted event is avoided or extended to a suitable time (e.g., further in time than the original prediction). In such an example, the control of one or more pumps may be manual, automatic, semi-automatic, etc. For example, regarding the semi-automatic method, consider the "Accept" button of GUI 1600 that may be actuated by an individual. With respect to automated methods, consider one or more criteria that may be evaluated for the purpose of automatically adjusting one or more operating parameters of the pump in relation to one or more predicted events, predicted decreases and / or increases in RUL, etc.

[0169] exist Fig.16 In the example of FIG. 1 , plot 1620 includes intake pressure, wellhead pressure (WHP), discharge pressure, and differential pressure, while plot 1630 includes motor temperature, intake temperature, and wellhead temperature (WHT). As shown, pressures and temperatures may vary. In particular, there may be variations in WHT, which may include normal variations, where such variations may be expected to exceed variations in motor temperature and intake temperature; note that WHT may be expected to be lower than motor and intake temperatures, where motor temperature is highest.

[0170] Fig.17 An example of a GUI 1700 is shown that includes temperature information in plots 1712 and 1714 and pressure information in plots 1722 and 1724. Fig.17 In an example of , GUI 1700 can be an anomaly detection GUI (e.g., an anomaly detection dashboard). For example, anomaly detection can be performed through analysis of residuals, which can be based on differences between actual measurements and model predictions. In such an example, an anomaly can be identified when the residual exceeds a defined threshold (e.g., default, user-defined, automatic, etc.). As an example, the threshold can be defined based on a standard deviation from a rolling mean.

[0171] like Fig.17As shown, plots 1712 and 1722 show the sum or alarm of temperature and pressure, respectively, relative to the month of operation. In such a plot, an individual can quickly evaluate monthly performance based on two different metrics. With respect to plots 1714 and 1724, these plots show the residual motor temperature and residual pressure relative to time and dotted lines indicating upper and lower limits, which can be set using one or more statistical parameters (e.g., standard deviation, etc.). As with other GUIs, various features can be animated, color-coded, etc., where alarms can be issued as pop-ups, etc.

[0172] Fig.18 An example of a GUI 1800 for running a life analysis is shown, which may include one or more features for pump control. Fig.18 In the example of , GUI 1800 includes plot 1810 and graph 1830. As shown, plot 1810 includes a plot of survival probability versus pump operating life in days, where the survival probability is predicted using a suitable survival model. Fig.18 In the example of , a current curve is shown along with a selected curve in plot 1810, which may correspond to a selected "current" time (t) that may measure future times (e.g., 30 days, 90 days, 180 days, etc.). For example, an individual may select a scenario different from the current scenario to determine whether the pump operating life is acceptable to meet one or more site and / or equipment goals. With respect to graph 1830, these show predictive statistics for the current operating life, 20% estimated operating life, 10% estimated operating life, 30-day survival probability, 90-day survival probability, and 180-day survival probability, which may be color coded, e.g., becoming redder or more red as the probability decreases.

[0173] As explained, one or more GUIs can facilitate control of field equipment, including, for example, maintenance, tripping, etc., which may help improve field operations by reducing NPT, etc. As an example, a background process of the framework can be utilized to run various scenarios, where an optimal scenario can be generated that can meet one or more field objectives. In such an example, the optimal scenario (e.g., or a set of top-ranked scenarios) can be appropriately presented for review and acceptance, thereby changing the operation of one or more pumps at one or more fields.

[0174] As explained, the framework can be local at the site and / or can be remote from the site and operably coupled to equipment at the site.As explained, the framework can implement multiple models that can be driven by site data to evaluate and / or control the operation of one or more pumps in the site.

[0175] As an example, a method may include receiving an input including time series data from a pump device at a well site, wherein the well site includes a wellbore in contact with a fluid reservoir; processing the input using a trained first machine learning model as an anomaly detector to generate an output; and processing the input and the output using a trained second machine learning model to predict a survival function of the pump device. In such an example, the trained first machine learning model may be or include one or more of an autoencoder model, a clustering model, and a tree model.

[0176] As an example, a normal behavior dataset of a pump device can be used to train the trained first machine learning model, where, for example, the trained first machine learning model can be trained using unsupervised learning.

[0177] As an example, a method may include processing an input and an output by calculating a difference between the input and the output, where, for example, the time series data as the input includes time series data of multiple channels, and where the difference may include a difference for each of the multiple channels.

[0178] As an example, a survival function may be generated that indicates the probability of survival of a pump device over a number of days with respect to time.

[0179] As an example, a method utilizing a trained first machine learning model and a trained second machine learning model may include, for the trained second machine learning model, using the output of the trained first machine learning model to train a normal behavior and abnormal behavior dataset of a pump device. In such an example, supervised learning may be used to train the trained second machine learning model. For example, consider using a label indicating a signature and a time between a fault.

[0180] As an example, the trained machine learning model may include a decision tree. For example, consider a decision tree that is part of a random survival forest (RSF or RSF model). As an example, the trained machine learning model may include a time-dependent Cox model.

[0181] As an example, a method may be implemented using a computing device at a well site, wherein the computing device receives an input, processes the input to generate an output, and processes the input and the output to generate a survival function. In such an example, the computing device may include an application, an application programming interface, and an executable containerized data structure for the trained first machine learning model and the trained second machine learning model, wherein the application accesses the executable containerized data structure using the application programming interface.

[0182] As an example, a pumping apparatus may include an electric submersible pump disposed in a wellbore and a surface control unit (e.g., a VSD unit, etc.), wherein at least a portion of the time series data is received by a computing device from one or more sensors connected to the electric submersible pump (e.g., consider a downhole meter including multiple sensors).

[0183] As an example, the input may correspond to a time window greater than 30 minutes, where the input is updated, for example, according to a time interval, where the time interval is greater than 30 seconds and less than 30 minutes. In such an example, the survival function (e.g., prediction) may be updated according to the time interval. As an example, a method may include using a threshold to determine a future day when a pump device may fail. As an example, a method may include calculating the remaining useful life of a pump device based at least in part on a predicted survival function.

[0184] As an example, a method may include adjusting one or more operating parameters of a pump device based at least in part on a predicted survival function of the pump device. In such an example, the adjustment may be intended to extend the remaining service life of the pump device. As an example, a method may include adjusting one or more operating parameters of a pump device based at least in part on a predicted survival function of the pump device and at least in part on a digital twin of the pump device, which predicts the performance of the pump device in response to the realization of one or more operating parameters. For example, the digital twin may be a machine learning model, which may be a dynamic model that is learned using data that may include online real-time data, so that the digital twin can predict the behavior of the pump device. This method can be run as a background process and used to generate one or more control strategies to meet one or more goals, which may be about the production of resources, the operating life of the pump device, the reduction of NPT, etc. As explained, the pump device can be operated (e.g., controlled) in a manner that can help reduce NPT for the purpose of scheduling maintenance, repair, replacement, etc. This method can utilize one or more digital twins of one or more pumps (e.g., pump devices, pump systems, etc.).

[0185] As an example, a system may include a processor; a memory accessible to the processor; and processor-executable instructions stored in the memory for instructing the system to: receive an input comprising time series data from a pump device at a well site, wherein the well site comprises a wellbore in contact with a fluid reservoir; process the input using a trained first machine learning model as an anomaly detector to generate an output; and process the input and output using a trained second machine learning model to predict a survival function of the pump device.

[0186] As an example, one or more computer-readable storage media may include processor-executable instructions to instruct a wellsite computing system to: receive input including time series data from a pump device at a wellsite, wherein the wellsite includes a wellbore in contact with a fluid reservoir; process the input using a trained first machine learning model as an anomaly detector to generate an output; and process the input and the output using a trained second machine learning model to predict a survival function of the pump device.

[0187] As an example, a computer program product may include one or more computer-readable storage media, which may include processor-executable instructions to instruct a computing system to perform one or more methods and / or one or more parts of a method. The various example methods may be performed in various combinations.

[0188] In some embodiments, one or more methods may be performed by a computing system. Fig.19 An example of system 1900 is shown, which may include one or more computing systems 1901-1, 1901-2, 1901-3 and 1901-4, and the one or more computing systems may be operably connected via one or more networks 1909, which may include wired and / or wireless networks; note that one or more other features 1908 may be included in system 1900.

[0189] By way of example, a system may include a single computer system or an arrangement of distributed computer systems. Fig.19 In the example of , computer system 1901-1 may include one or more modules 1902, which may be or include processor-executable instructions, for example, executable to perform various tasks (e.g., receiving information, requesting information, processing information, simulating, outputting information, etc.).

[0190] As an example, the modules may be executed independently or in coordination with one or more processors 1904, which are operably coupled to one or more storage media 1906 (e.g., via wire, wireless, etc.). As an example, one or more of the one or more processors 1904 may be operably coupled to at least one of the one or more network interfaces 1907. In such an example, the computer system 1901-1 may, for example, send and / or receive information via one or more networks 1909 (e.g., consider one or more of the Internet, a private network, a cellular network, a satellite network, etc.).

[0191] As an example, computer system 1901-1 can receive information from and / or send information to one or more other devices, which can be or include, for example, one or more of computer system 1901-2, etc. The device can be located at a physical location different from the physical location of computer system 1901-1. As an example, the location can be, for example, a processing facility location, a data center location (e.g., a server farm, etc.), a drilling rig location, a well site location, a downhole location, etc.

[0192] As an example, a processor may be or include a microprocessor, a microcontroller, a processor module or subsystem, a programmable integrated circuit, a programmable gate array, or another control or computing device.

[0193] As an example, storage media 1906 can be implemented as one or more computer-readable or machine-readable storage media. As an example, storage can be distributed within and / or across multiple internal and / or external enclosures of the computing system and / or additional computing systems.

[0194] As an example, the one or more storage media may include one or more different forms of memory, including semiconductor memory devices (such as dynamic or static random access memory (DRAM or SRAM)), erasable and programmable read-only memory (EPROM), electrically erasable and programmable read-only memory (EEPROM) and flash memory, magnetic disks (such as fixed disks, floppy disks, and removable disks), other magnetic media (including magnetic tape), optical media (such as compact disks (CDs) or digital video disks (DVDs), Blu-ray disks or other types of optical storage), or other types of storage devices.

[0195] As an example, one or more storage media may be located in the machine running the machine-readable instructions, or at a remote site from which the machine-readable instructions may be downloaded over a network for execution.

[0196] As examples, various components of a system (such as, for example, a computer system) may be implemented in hardware, software, or a combination of both hardware and software (eg, including firmware), including one or more signal processing and / or application specific integrated circuits.

[0197] As an example, a system may include a processing device that may be or include a general purpose processor or a dedicated chip (eg, or chipset) such as an ASIC, FPGA, PLD, or other suitable device.

[0198] Fig. 20Components of an example of a computing system 2000 and an example of a networked system 2010 with a network 2020 are shown. System 2000 includes one or more processors 2002, a memory and / or storage component 2004, one or more input and / or output devices 2006, and a bus 2008. In an example embodiment, instructions may be stored in one or more computer-readable media (e.g., memory / storage component 2004). Such instructions may be read by one or more processors (e.g., processor 2002) via a communication bus (e.g., bus 2008), which may be wired or wireless. One or more processors may execute such instructions to implement (completely or partially) one or more properties (e.g., as part of a method). A user may view output from a process and interact with the process via an I / O device (e.g., device 2006). In an example embodiment, a computer-readable medium may be a storage component, such as a physical memory storage device, e.g., a chip, a chip on a package, a memory card, etc. (e.g., a computer-readable storage medium).

[0199] In an example embodiment, the components may be distributed in, for example, a network system 2010. The network system 2010 includes components 2022-1, 2022-2, 2022-3, ... 2022-N. For example, component 2022-1 may include (multiple) processors 2002, and (multiple) component 2022-3 may include a memory accessible to (multiple) processors 2002. In addition, (multiple) component 2022-2 may include an I / O device for displaying and optionally interacting with the method. The network 2020 may be or include the Internet, an intranet, a cellular network, a satellite network, etc.

[0200] As an example, the device can be a mobile device including one or more network interfaces for information communication. For example, the mobile device may include a wireless network interface (e.g., operable via IEEE 802.11, ETSI GSM, Bluetooth, satellite, etc.). As an example, the mobile device may include components such as a main processor, a memory, a display, a display graphics circuit (e.g., optionally including touch and gesture circuits), a SIM slot, an audio / video circuit, a motion processing circuit (e.g., an accelerometer, a gyroscope), a wireless LAN circuit, a smart card circuit, a transmitter circuit, a GPS circuit, and a battery. As an example, the mobile device can be configured as a cellular phone, a tablet computer, etc. As an example, the method can be implemented using a mobile device (e.g., in whole or in part). As an example, the system may include one or more mobile devices.

[0201] As an example, the system may be a distributed environment, such as a so-called "cloud" environment, where various devices, components, etc. interact for the purpose of data storage, communication, computing, etc. As an example, a device or system may include one or more components for communicating information via one or more of the Internet (e.g., where communication occurs via one or more Internet protocols), a cellular network, a satellite network, etc. As an example, the method may be implemented in a distributed environment (e.g., in whole or in part as a cloud-based service).

[0202] Although only a few example embodiments have been described in detail above, those skilled in the art will readily appreciate that many modifications may be made in the example embodiments. Therefore, all such modifications are intended to be included within the scope of the present disclosure as defined in the appended claims. In the claims, means-plus-function clauses are intended to cover structures described herein as performing the described function, and to cover not only structural equivalents, but also equivalent structures. Thus, although a nail and a screw may not be structural equivalents because a nail employs a cylindrical surface to fasten wooden parts together, while a screw employs a helical surface, in the context of fastening wooden parts, a nail and a screw may be equivalent structures.

Claims

1. A method (1500), comprising: receiving input including time series data from a pump device at a wellsite, wherein the wellsite includes a wellbore in contact with a fluid reservoir (1510); processing the input using the trained first machine learning model as an anomaly detector to generate an output (1520); and The input and the output are processed using a trained second machine learning model to predict a survival function of the pump device (1530).

2. The method of claim 1, wherein the trained first machine learning model comprises one or more of an autoencoder model, a clustering model, and a tree model.

3. A method according to claim 1 or 2, wherein the trained first machine learning model is trained using a normal behavior dataset of the pump device, and optionally, wherein the trained first machine learning model is trained using unsupervised learning.

4. A method according to any preceding claim, wherein processing the input and the output comprises calculating a difference between the input and the output, optionally wherein the time series data comprises time series data for a plurality of channels, and wherein the difference comprises a difference for each of the plurality of channels.

5. A method according to any preceding claim, wherein the output of the trained first machine learning model is used to train the trained second machine learning model for the normal behavior and abnormal behavior data sets of the pump device.

6. A method according to any preceding claim, wherein the trained second machine learning model comprises a decision tree.

7. A method according to any preceding claim, wherein the trained second machine learning model comprises a time-dependent Cox model.

8. The method of any preceding claim, wherein a computing device at a wellsite receives the input, processes the input to generate the output, and processes the input and the output to generate the survival function.

9. A method according to any preceding claim, wherein the pump apparatus comprises an electric submersible pump disposed in a wellbore and a surface control unit, and wherein at least a portion of the time series data is received from one or more sensors coupled to the electric submersible pump.

10. A method according to any preceding claim, wherein the input corresponds to a time window greater than 30 minutes, wherein the input is updated according to a time interval, wherein the time interval is greater than 30 seconds and less than 30 minutes, and wherein the survival function is updated according to the time interval.

11. A method according to any preceding claim, comprising adjusting one or more operating parameters of the pump device based at least in part on the predicted survival function of the pump device, optionally wherein the adjustment extends the remaining useful life of the pump device.

12. A method according to any preceding claim, wherein adjusting is based at least in part on a digital twin of the pump device, the digital twin predicting performance of the pump device in response to achievement of one or more operating parameters.

13. A method according to any preceding claim, comprising utilizing a remaining useful life of the pump device based at least in part on the survival function.

14. A system (1590), comprising: Processor(1593); a memory (1594) accessible to the processor; and Processor executable instructions (1596), stored in the memory, to instruct the system to: receiving input including time series data from a pump device at a wellsite, wherein the wellsite includes a wellbore in contact with a fluid reservoir (1511); processing the input using the trained first machine learning model as an anomaly detector to generate an output (1521); and The input and the output are processed using a trained second machine learning model to predict a survival function of the pump device (1531).

15. A computer program product comprising computer executable instructions for instructing a computing system to perform the method according to any one of claims 1 to 13.

Images