Surrogate model creation using generative model frameworks

Abstract

A method for creating a surrogate model using a generative artificial intelligence (AI) model framework includes performing a simulation of a facility using a simulation model to produce a simulation output. The facility may be used to process oil and / or gas. The method may also include receiving first data related to equipment in the facility. The method may also include generating a plurality of surrogate models based upon the simulation output and the first data. The method may also include performing simulations using the surrogate models to produce surrogate outputs that predict the simulation output. The method may also include generating a recommendation based upon the surrogate outputs.

Description

Attorney Docket No.: IS23.0617-WO-PCTSURROGATE MODEL CREATION USING GENERATIVE MODEL FRAMEWORKSCross-Reference to Related Applications

[0001] This application claims priority to U.S. Provisional Patent Application No. 63 / 691,425, filed on September 6, 2024, which is incorporated by reference herein.Background

[0002] A surrogate model is a statistical model that is used to approximate the output of a complex physics-driven computer-simulation model. These models are often used in engineering design, optimization, scenario analysis, insight generation, etc. By using a surrogate model, engineers may quickly and easily explore the design space and identify optimal solutions. Surrogate model creation is challenging due to model complexity, data scarcity, model bias, and model observability / non-uniqueness. In addition, surrogate model creation is often time- consuming and involves expert knowledge. Therefore, what is needed is an improved system and method for surrogate model creation (e.g., using generative model frameworks).Summary

[0003] A method for creating a surrogate model using a generative artificial intelligence (Al) model framework is disclosed. The method includes performing a simulation of a facility using a simulation model to produce a simulation output. The facility may be used to process oil and / or gas. The method may also include receiving first data related to equipment in the facility. The method may also include generating a plurality of surrogate models based upon the simulation output and the first data. The method may also include performing simulations using the surrogate models to produce surrogate outputs that predict the simulation output. The method may also include generating a recommendation based upon the surrogate outputs.

[0004] A computing system is also disclosed. The computing system includes one or more processors and a memory system. The memory system includes one or more non-transitory computer-readable media storing instructions that, when executed by at least one of the one or more processors, cause the computing system to perform operations. The operations include performing a simulation of a facility using a simulation model to produce a simulation output. The facility is used to process oil and / or gas. The operations also include receiving first data related toAttorney Docket No.: IS23.0617-WO-PCT equipment in the facility. The first data is measured by one or more sensors. The operations also include generating a plurality of surrogate models based upon the simulation output and the first data. Each of the surrogate models is generated and trained by iteratively: generating the surrogate model using a first agent; generating training data; training the surrogate model to produce a trained surrogate model based upon the training data; and evaluating the trained surrogate model until it converges. The operations also include performing simulations using the surrogate models to produce surrogate outputs that predict the simulation output. The operations also include generating a recommendation based upon the surrogate outputs.

[0005] A non-transitory computer-readable medium is also disclosed. The medium stores instructions that, when executed by one or more processors of a computing system, cause the computing system to perform operations. The operations include performing a simulation of a facility using a simulation model to produce a simulation output. The facility is used to process oil and / or gas. The operations also include receiving first data related to equipment in the facility. The first data is measured by one or more sensors. The operations also include generating a plurality of surrogate models based upon the simulation output and the first data. Each of the surrogate models is generated and trained by iteratively: generating the surrogate model using a first agent; generating training data; training the surrogate model to produce a trained surrogate model based upon the training data; and evaluating the trained surrogate model until it converges. The operations also include performing simulations using the surrogate models to produce surrogate outputs that predict the simulation output. The operations also include generating a recommendation based upon the surrogate outputs.

[0006] It will be appreciated that this summary is intended merely to introduce some aspects of the present methods, systems, and media, which are more fully described and / or claimed below. Accordingly, this summary is not intended to be limiting.Brief Description of the Drawings

[0007] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments of the present teachings and together with the description, serve to explain the principles of the present teachings. In the figures:

[0008] Figure 1 illustrates an example of a system that includes various management components to manage various aspects of a geologic environment, according to an embodiment.Attorney Docket No.: IS23.0617-WO-PCT

[0009] Figure 2 illustrates a facility used to process oil and / or gas, according to an embodiment.

[0010] Figure 3 illustrates a system for creating a surrogate model using a generative artificial intelligence (Al) model framework, according to an embodiment.

[0011] Figure 4 illustrates an application of the system, according to an embodiment.

[0012] Figure 5 illustrates a flowchart of a method for creating a surrogate model using a generative artificial intelligence (Al) model framework, according to an embodiment.

[0013] Figures 6A-6F illustrate an Al-agent-driven workflow for building surrogate models using a think / act / observe loop on a simulated multi-well production network, according to an embodiment. This includes data preparation, space-filling and adaptive sampling, physics- informed training, active learning, uncertainty quantification, sensitivity analysis and deployment.

[0014] Figure 7 illustrates a schematic view of a computing system for performing at least a portion of the method(s) described herein, according to an embodiment.Detailed Description

[0015] Reference will now be made in detail to embodiments, examples of which are illustrated in the accompanying drawings and figures. In the following detailed description, numerous specific details are set forth in order to provide a thorough understanding of the present disclosure. However, it will be apparent to one of ordinary skill in the art that the present disclosure may be practiced without these specific details. In other instances, well-known methods, procedures, components, circuits, and networks have not been described in detail so as not to unnecessarily obscure aspects of the embodiments.

[0016] It will also be understood that, although the terms first, second, etc. may be used herein to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another. For example, a first object or step could be termed a second object or step, and, similarly, a second object or step could be termed a first object or step, without departing from the scope of the present disclosure. The first object or step, and the second object or step, are both, objects or steps, respectively, but they are not to be considered the same object or step.

[0017] The terminology used in the description herein is for the purpose of describing particular embodiments and is not intended to be limiting. As used in this description and the appended claims, the singular forms “a,” “an” and “the” are intended to include the plural forms as well,Attorney Docket No.: 1S23.0617-WO-PCT unless the context clearly indicates otherwise. It will also be understood that the term “and / or” as used herein refers to and encompasses any possible combinations of one or more of the associated listed items. It will be further understood that the terms “includes,” “including,” “comprises” and / or “comprising,” when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. Further, as used herein, the term “if’ may be construed to mean “when” or “upon” or “in response to determining” or “in response to detecting,” depending on the context.

[0018] Attention is now directed to processing procedures, methods, techniques, and workflows that are in accordance with some embodiments. Some operations in the processing procedures, methods, techniques, and workflows disclosed herein may be combined and / or the order of some operations may be changed.System Overview

[0019] Figure 1 illustrates an example of a system 100 that includes various management components 110 to manage various aspects of a geologic environment 150 (e.g., an environment that includes a sedimentary basin, a reservoir 151, one or more faults 153-1, one or more geobodies 153-2, etc.). For example, the management components 110 may allow for direct or indirect management of sensing, drilling, injecting, extracting, etc., with respect to the geologic environment 150. In turn, further information about the geologic environment 150 may become available as feedback 160 (e.g., optionally as input to one or more of the management components 110).

[0020] In the example of Figure 1, the management components 110 include a seismic data component 112, an additional information component 114 (e.g., well / logging data), a processing component 116, a simulation component 120, an attribute component 130, an analysis / visualization component 142 and a workflow component 144. In operation, seismic data and other information provided per the components 112 and 114 may be input to the simulation component 120.

[0021] In an example embodiment, the simulation component 120 may rely on entities 122. Entities 122 may include earth entities or geological objects such as wells, surfaces, bodies, reservoirs, etc. In the system 100, the entities 122 can include virtual representations of actualAttorney Docket No.: 1S23.0617-WO-PCT physical entities that are reconstructed for purposes of simulation. The entities 122 may include entities based on data acquired via sensing, observation, etc. (e.g., the seismic data 112 and other information 114). An entity may be characterized by one or more properties (e.g., a geometrical pillar grid entity of an earth model may be characterized by a porosity property). Such properties may represent one or more measurements (e.g., acquired data), calculations, etc.

[0022] In an example embodiment, the simulation component 120 may operate in conjunction with a software framework such as an object-based framework. In such a framework, entities may include entities based on pre-defined classes to facilitate modeling and simulation. A commercially available example of an object-based framework is the MICROSOFT® .NET® framework (Redmond, Washington), which provides a set of extensible object classes. In the .NET® framework, an object class encapsulates a module of reusable code and associated data structures. Object classes can be used to instantiate object instances for use in by a program, script, etc. For example, borehole classes may define objects for representing boreholes based on well data.

[0023] In the example of Figure 1, the simulation component 120 may process information to conform to one or more attributes specified by the attribute component 130, which may include a library of attributes. Such processing may occur prior to input to the simulation component 120 (e.g., consider the processing component 116). As an example, the simulation component 120 may perform operations on input information based on one or more attributes specified by the attribute component 130. In an example embodiment, the simulation component 120 may construct one or more models of the geologic environment 150, which may be relied on to simulate behavior of the geologic environment 150 (e.g., responsive to one or more acts, whether natural or artificial). In the example of Figure 1, the analysis / visualization component 142 may allow for interaction with a model or model-based results (e.g., simulation results, etc.). As an example, output from the simulation component 120 may be input to one or more other workflows, as indicated by a workflow component 144.

[0024] As an example, the simulation component 120 may include one or more features of a simulator such as the ECLIPSE™ reservoir simulator (SLB, Houston Texas), the INTERSECT™ reservoir simulator (SLB, Houston Texas), etc. As an example, a simulation component, a simulator, etc. may include features to implement one or more meshless techniques (e.g., to solve one or more equations, etc.). As an example, a reservoir or reservoirs may be simulated withAttorney Docket No.: 1S23.0617-WO-PCT respect to one or more enhanced recovery techniques (e.g., consider a thermal process such as SAGD, etc.).

[0025] As an example, the simulation component 120 may include one or more features of a simulator such as SYMMETRY™ software (SLB, Houston, Texas). More particularly, SYMMETRY™ may process workflows in a single integrated environment with accurate thermodynamic fluid representation and consistent modeling across multiple disciplines including process, production, and HSE. The simulator integrates steady-state and transient (e.g., dynamic) analyses that can be tailored for each domain. This approach enables users to optimize processes in upstream, midstream, and downstream sectors while maximizing profits and minimizing capital expenditures. It may also help reduce emissions, energy consumption, and waste.

[0026] As an example, the simulation component 120 may include one or more features of a simulator such as PIPESIM™ (SLB, Houston, Texas). More particularly, PIPESIM™ is steadystate multiphase flow simulator that incorporates the three areas of flow modeling: multiphase flow, heat transfer and fluid behavior.

[0027] As an example, the simulation component 120 may include one or more features of a simulator such as OLGA™ (SLB, Houston, Texas). More particularly, OLGA™ is a dynamic multiphase flow simulator that models transient flow (e.g., time-dependent behaviors) to maximize production potential. Transient modeling is a component for feasibility studies and field development design. Dynamic simulation is useful in deep water and is used in both offshore and onshore developments to investigate transient behavior in pipelines and wellbores. Transient simulation with the OLGA™ simulator provides an added dimension to steady-state analysis by predicting system dynamics, such as time-varying changes in flow rates, fluid compositions, temperature, solids deposition, and operational changes.

[0028] In an example embodiment, the management components 110 may include features of a commercially available framework such as the PETREL® seismic to simulation software framework (SLB, Houston, Texas). The PETREL® framework provides components that allow for optimization of exploration and development operations. The PETREL® framework includes seismic to simulation software components that can output information for use in increasing reservoir performance, for example, by improving asset team productivity. Through use of such a framework, various professionals (e.g., geophysicists, geologists, and reservoir engineers) can develop collaborative workflows and integrate operations to streamline processes. Such aAttorney Docket No.: 1S23.0617-WO-PCT framework may be considered an application and may be considered a data-driven application (e.g., where data is input for purposes of modeling, simulating, etc.).

[0029] In an example embodiment, various aspects of the management components 110 may include add-ons or plug-ins that operate according to specifications of a framework environment. For example, a commercially available framework environment marketed as the OCEAN® framework environment (SLB, Houston, Texas) allows for integration of add-ons (or plug-ins) into a PETREL® framework workflow. The OCEAN® framework environment leverages .NET® tools (Microsoft Corporation, Redmond, Washington) and offers stable, user-friendly interfaces for efficient development. In an example embodiment, various components may be implemented as add-ons (or plug-ins) that conform to and operate according to specifications of a framework environment (e.g., according to application programming interface (API) specifications, etc.).

[0030] Figure 1 also shows an example of a framework 170 that includes a model simulation layer 180 along with a framework services layer 190, a framework core layer 195 and a modules layer 175. The framework 170 may include the commercially available OCEAN® framework where the model simulation layer 180 is the commercially available PETREL® model-centric software package that hosts OCEAN® framework applications. In an example embodiment, the PETREL® software may be considered a data-driven application. The PETREL® software can include a framework for model building and visualization.

[0031] As an example, a framework may include features for implementing one or more mesh generation techniques. For example, a framework may include an input component for receipt of information from interpretation of seismic data, one or more attributes based at least in part on seismic data, log data, image data, etc. Such a framework may include a mesh generation component that processes input information, optionally in conjunction with other information, to generate a mesh.

[0032] In the example of Figure 1, the model simulation layer 180 may provide domain objects 182, act as a data source 184, provide for rendering 186 and provide for various user interfaces 188. Rendering 186 may provide a graphical environment in which applications can display their data while the user interfaces 188 may provide a common look and feel for application user interface components.

[0033] As an example, the domain objects 182 can include entity objects, property objects and optionally other objects. Entity objects may be used to geometrically represent wells, surfaces,Attorney Docket No.: 1S23.0617-WO-PCT bodies, reservoirs, etc., while property objects may be used to provide property values as well as data versions and display parameters. For example, an entity object may represent a well where a property object provides log information as well as version information and display information (e.g., to display the well as part of a model).

[0034] In the example of Figure 1, data may be stored in one or more data sources (or data stores, generally physical data storage devices), which may be at the same or different physical sites and accessible via one or more networks. The model simulation layer 180 may be configured to model projects. As such, a particular project may be stored where stored project information may include inputs, models, results and cases. Thus, upon completion of a modeling session, a user may store a project. At a later time, the project can be accessed and restored using the model simulation layer 180, which can recreate instances of the relevant domain objects.

[0035] In the example of Figure 1, the geologic environment 150 may include layers (e.g., stratification) that include a reservoir 151 and one or more other features such as the fault 153-1, the geobody 153-2, etc. As an example, the geologic environment 150 may be outfitted with any of a variety of sensors, detectors, actuators, etc. For example, equipment 152 may include communication circuitry to receive and to transmit information with respect to one or more networks 155. Such information may include information associated with downhole equipment 154, which may be equipment to acquire information, to assist with resource recovery, etc. Other equipment 156 may be located remote from a well site and include sensing, detecting, emitting or other circuitry. Such equipment may include storage and communication circuitry to store and to communicate data, instructions, etc. As an example, one or more satellites may be provided for purposes of communications, data acquisition, etc. For example, Figure 1 shows a satellite in communication with the network 155 that may be configured for communications, noting that the satellite may additionally or instead include circuitry for imagery (e.g., spatial, spectral, temporal, radiometric, etc.).

[0036] Figure 1 also shows the geologic environment 150 as optionally including equipment 157 and 158 associated with a well that includes a substantially horizontal portion that may intersect with one or more fractures 159. For example, consider a well in a shale formation that may include natural fractures, artificial fractures (e.g., hydraulic fractures) or a combination of natural and artificial fractures. As an example, a well may be drilled for a reservoir that is laterally extensive. In such an example, lateral variations in properties, stresses, etc. may exist where anAttorney Docket No.: 1S23.0617-WO-PCT assessment of such variations may assist with planning, operations, etc. to develop a laterally extensive reservoir (e.g., via fracturing, injecting, extracting, etc.). As an example, the equipment 157 and / or 158 may include components, a system, systems, etc. for fracturing, seismic sensing, analysis of seismic data, assessment of one or more fractures, etc.

[0037] As mentioned, the system 100 may be used to perform one or more workflows. A workflow may be a process that includes a number of worksteps. A workstep may operate on data, for example, to create new data, to update existing data, etc. As an example, a may operate on one or more inputs and create one or more results, for example, based on one or more algorithms. As an example, a system may include a workflow editor for creation, editing, executing, etc. of a workflow. In such an example, the workflow editor may provide for selection of one or more predefined worksteps, one or more customized worksteps, etc. As an example, a workflow may be a workflow implementable in the PETREL® software, for example, that operates on seismic data, seismic attribute(s), etc. As an example, a workflow may be a process implementable in the OCEAN® framework. As an example, a workflow may include one or more worksteps that access a module such as a plug-in (e.g., external executable code, etc.).Surrogate Model Creation Using Generative Model Frameworks

[0038] The present disclosure describes systems and methods for creating surrogate models using generative models and / or large language models (LLMs). The uniqueness of the framework is that it thinks through the surrogate model building (e.g., the way a subject matter expert does) and automates this process.

[0039] In one embodiment, the system and method may select multiple model surrogates to approximate the complex simulator. In another embodiment, the system and method may (e.g., optimally) generate training data, train the models, and evaluate the model iteratively until convergence. One feature of the system is sampling a predetermined subspace and generating an optimal dataset using a chain of thinking process.

[0040] In another embodiment, a generative agent evaluates the sample coverage of the space spanned by the parameter space and adaptively samples using a chain of experiment process. In yet another embodiment, the generative agent determines the complexity and nonlinearity of the surface to be modelled and samples the space iteratively. In yet another embodiment, two generative agents act as a generator and a critic to generate the samples and score the usefulnessAttorney Docket No.: 1S23.0617-WO-PCT of the same in adapting in each iteration. In yet another embodiment, the critic agent acts as a feedback agent and thus improves the sample generation process using reinforcement learning with a critic feedback process. In yet another embodiment, the generative agent and critic agent interactively plan to build the (e.g., best) surrogate model based on the data generated in the previous step.Al Agent Approach

[0041] The Al agent thinks before acting (i.e., analyzes the problem first). The Al agent also learns from experience and improves with each task. The Al agent also adapts strategies based upon previous strategies that have worked (or not worked).The Re-Act Pattern Example (A Simulated Well Network)

[0042] An example may include the following:• Think: high-dimensional coupled system; need to understand structure• Act: generate analysis code (e.g., identify well coupling patterns)• Observe: wells 1-4 are coupled; others independent• Think: I can exploit this structure for efficient sampling• Act: generate adaptive samples (e.g., method = expected_improvement focus = coupled_wells)• Observe: good coverage achieved with 70% fewer samplesDomain-Agnostic Tools (Pre-Built)

[0043] The method may use advanced sampling strategies (e.g., LHS, sobol, adaptive). The method may also use a machine-learning (ML) model suite (e.g., neural networks, Gaussian processes, ensembles). The method may also use statistical analysis (e.g., sensitivity, uncertainty, convergence). The method may also use optimization algorithms (e.g., Bayesian, evolutionary, meta-learning).Domain-Specific Analysis (Code Generation)Attorney Docket No.: IS23.0617-WO-PCT

[0044] The method may use a custom data analysis for any domain. The method may also use problem-specific pre-processing. The method may also use domain constraints and physics. The method may also use specialized visualizations.Experience Database

[0045] The agent may get “smarter” by analyzing problem characterization data, data requirements and sampling history, and model performance trajectory. The agent may also analyze architecture specifications, hyperparameter configurations, training strategies, and uncertainty quantification methods.Exemplary Facility

[0046] Figure 2 illustrates a facility 200 used to process oil and / or gas, according to an embodiment. The facility 200 may include an upstream facility, a midstream facility, or a downstream facility. The facility 200 may include equipment 210 therein. The equipment 210 may include one or more pumps, heat exchangers, compressors, separators, control and / or choke valves, gas-lift manifolds, dehydrators, scrubbers, reboilers, distillation / flash columns, storage tanks, flare systems, or a combination thereof. The facility 200 may also include one or more sensors 220 that are configured to measure parameters related to the equipment 210 and / or the processes being performed within the facility 200.Exemplary System

[0047] Figure 3 illustrates a system 300 for creating a surrogate model using a generative artificial intelligence (Al) model framework, according to an embodiment. The system 300 may also include an experience database 305 that stores prior problems, sampling history, model performance, and prior lessons learned for meta-analysis. The system 300 may also include a model repository 310, which may be or include a versioned store or database for trained surrogates and configurations. The system 300 may also include an agent 315, which may serve as a generative / critique planner that decides what to sample / train next. The system 300 may also include a sandbox 320, which serves as an execution environment for generated analysis and simulation code. The system 300 may also include one or more simulators 325 that are high- fidelity engines (e.g., PIPESIM™, OLGA™, SYMMETRY™, INTERSECT™) used to generateAttorney Docket No.: IS23.0617-WO-PCT data. The system 300 may also include a GPU compute 330, which may be or include accelerators used to train Al models. The system 300 may also include task constraints 335 that are operational and / or physical limits and business goals. The system 300 may also include a model 340, which is a current candidate surrogate under training and / or evaluation. The system 300 may also include empirical knowledge 345, which is a set of domain rules and / or correlations injected as priors or loss terms.Exemplary Application

[0048] Figure 4 illustrates an application of the system 300, according to an embodiment. The system 300 may perform an initial sampling (e.g., Latin hypercube for space-fdling design). The system 300 may then run simulations in a facility or plant. The system 300 may then train a model (e.g., a neural network). The system 300 may then evaluate (e.g., check accuracy such as R2or MAE). The system 300 may then implement adaptive sampling (e.g., Monte Carlo uncertainty). The system 300 may then export a model for deployment.Surrogate Model for Simulated Well Network

[0049] Illustrative inputs may include:• wellhead pressure (per well)• choke valve positions (per well)• gas lift injection rates• water cut percentage (per well)• gas-oil ratio (GOR) (per well)• reservoir pressure (per well)• productivity index (per well)• pipeline roughness coefficients• ambient temperature• separator pressure• total system back-pressure

[0050] Illustrative outputs may include:• oil production rate (per well and total)• gas production rate (per well and total)Attorney Docket No.: IS23.0617-WO-PCT• water production rate (per well and total)• bottomhole pressure (per well)• pressure at junction points• separator inlet pressure• flow regime indicators• erosional velocity ratio• liquid holdup in pipelinesExemplary Method

[0051] Figure 5 illustrates a flowchart of a method 500 for creating one or more surrogate models using a generative artificial intelligence (Al) model framework, according to an embodiment. An illustrative order of the method 500 is provided below; however, one or more portions of the method 500 may be performed in a different order, simultaneously, repeated, or omitted. At least a portion of the method 500 may be performed by a computing system.

[0052] The method 500 may include performing a simulation of a facility 200 using a simulation model, as at 505. The facility 200 may be used to process oil and / or gas. In an example, the simulation may include multiphase flow in a coupled well and pipeline network, separator-train pressure, temperature dynamics, or compressor station surge / throughput control, amongst others.

[0053] The method 500 may also include receiving first data related to equipment 210 in the facility 200, as at 510. The first data may be measured by one or more sensors 220. In an example, the first data may be or include temperature measurements, pressure measurements, vibration measurements, composition measurements, flow-rate measurements, valve or choke position, motor current or power draw, differential-pressure across equipment, acoustic and sand-rate indicators, level measurements, or a combination thereof. As mentioned above, the equipment 210 may include one or more pumps, heat exchangers, compressors, separators, control and / or choke valves, gas-lift manifolds, dehydrators, scrubbers, reboilers, distillation / flash columns, storage tanks, flare systems, or a combination thereof.

[0054] The method 500 may also include generating a plurality of surrogate models based upon the simulation and / or the first data, as at 515. The surrogate models may be configured to generate a plurality of surrogate outputs that predict a simulation output of the simulation model. For example, each surrogate model may generate a different surrogate output. Each of the surrogateAttorney Docket No.: IS23.0617-WO-PCT models may be generated by generating the surrogate model using a first agent. The first agent may implement a generative Al model. The surrogate models may also be generated by generating training data based upon (1) simulation queries at points selected by the first agent, (2) synthetic samples from the generative Al model, and / or (3) historical data from the one or more sensors 220. In an example, the queries may be or include what-if simulation queries varying choke setpoints, lift-gas rates, separator pressures, compressor speed, or ambient temperature. The queries may also or instead include design-of-experiments queries and / or uncertainty-driven queries targeting high- variance regions. In another embodiment, the queries may also or instead include operatingenvelope extrema, historical setpoints / excursions, and / or regions of high surrogate uncertainty. The surrogate models may also be generated by training the surrogate model to produce a trained surrogate model based upon the training data. The surrogate models may also be generated by evaluating the trained surrogate model until it converges. The trained surrogate model may be evaluated by the first agent and a second agent that critiques the trained surrogate model. The foregoing steps may be performed iteratively.

[0055] The method 500 may also include performing simulations using the surrogate models to produce the surrogate outputs, as at 520. The surrogate outputs may be predicted by the trained surrogate models faster than the simulation output is predicted by the simulation model. Performing the simulation may include forward-simulating the surrogate outputs, optimizing the surrogate outputs, forecasting the surrogate outputs, or a combination thereof.

[0056] The method 500 may also include generating a recommendation based upon the surrogate outputs, as at 525. The recommendation may be to repair the equipment 210, replace the equipment 210, operate the equipment 210 at a different speed, operate the equipment 210 at a different pressure, operate the equipment 210 at a different temperature, re-tune PID loops, adjust gas-lift injection rates, change chemical -injection setpoints, schedule heat-exchanger cleaning, reroute flows or change pump / compressor sequencing, preheat lines to avoid hydrates, or temporarily shut-in or switch artificial-lift mode.

[0057] The method 500 may also include displaying the simulation output, the surrogate outputs, and / or the recommendation, as at 530.

[0058] The method 500 may also include performing an action, as at 535. The action may be performed based upon and / or in response to the surrogate outputs and / or the recommendation, generating or transmitting a signal that instructs or causes an automated action to implement theAttorney Docket No.: IS23.0617-WO-PCT recommendation. The action may be or include generating and / or transmitting a signal (e g., using a computing system) that instructs or causes the recommendation to be implemented or a physical action to occur (e.g., at the facility / plant or a wellsite). For example, the action may be or include closed-loop control of the equipment 210 and / or a process that takes place at the facility / plant 200. The action may also or instead include performing the physical action to implement the recommendation.

[0059] In an embodiment, the method 500 may also include augmenting the training data with synthetic operating scenarios generated by the first agent, such that rare or safety-critical regimes that are under-represented in the first data are probabilistically up-sampled before training the surrogate models. In an example, the synthetic operating scenarios may include cold start-ups and ramp-ups, emergency shutdown and restart sequences, multi-well shut-in and restart combinations, rapid composition swings, and step-tests of control valves. This may also or instead include rare or safety-critical regimes such as compressor surge, relief-valve lift events, over-pressure, hydrate / wax formation risk, severe slugging, very low suction pressure, and high-temperature excursions.

[0060] In an embodiment, each surrogate model may be regularized by one or more physics-informed loss terms that enforce conservation of mass, energy, or momentum, thereby constraining the surrogate outputs of the surrogate models to remain physically plausible. The physics-informed loss terms may be or include (1) residual penalties enforcing conservation of mass, energy, or momentum, (2) constraint penalties for operational bounds on pressure, temperature, or flow, and / or (3) governing-equation residuals for ordinary or partial-differential equations. In an example, the residual penalties may include node-wise mass-balance residuals at junctions, momentum residual along pipelines, energy-balance residual across heaters or heat exchangers, or a combination thereof. In an example, the constraint penalties may include pressure below the safety limit, temperature below the material limits, flow within pump and / or compressor maps, minimum suction pressure, valve travel limits or rate-of-change limits, or a combination thereof. In an example, the governing-equation residuals may include steady multiphase pressuredrop and / or advection-conduction energy along pipelines.

[0061] In an embodiment, the method 500 may also include quantifying predictive uncertainty for each trained surrogate model via an ensemble, Bayesian neural -network, or Monte-CarloAttorney Docket No.: IS23.0617-WO-PCT dropout technique. The recommendation may include a confidence score derived from the quantified predictive uncertainty.

[0062] In an embodiment, the method 500 may also include an active-learning loop that: (1) automatically identifies regions where an uncertainty of the surrogate outputs of the surrogate models is greater than a threshold; (2) triggers additional high-fidelity simulations of the simulation model in the identified regions; and (3) retrains the surrogate models with outputs from the additional high-fidelity simulations and / or new data from the one or more sensors until a predefined accuracy threshold is met. In an example, the regions may include near operational boundaries (e.g., max pressure / temperature / flow), choke-transition regimes, high water-cut or high-GOR corners, multi-well shut-in patterns, low suction-pressure nodes, phase-envelope crossover, viscosity inflection zones, compressor surge lines, heat exchanger fouling, or a combination thereof.

[0063] In an embodiment, convergence of the trained surrogate model may be assessed within a sliding time window of the first data to detect non-stationary behavior in the facility. The failure to converge within the sliding time window automatically may invoke a model-drift alarm. In an example, the non-stationary behavior may include heat-exchanger fouling, reservoir depletion changing PI, sensor drift / recalibration offsets, equipment wear , pump / compressor efficiency loss, addition of new wells or tie-ins, fluid-composition drift, seasonal ambient temperature changes, or a combination thereof.

[0064] In an embodiment, the first agent may be or include a generative-adversarial-network (GAN) generator, and the second agent may be or include a GAN discriminator. The GAN discriminator may be configured to minimize a divergence metric between a distribution of the surrogate outputs from the surrogate models and a distribution of the simulation outputs from the simulation model. The divergence metric may be or include a Jensen-Shannon divergence, a Wasserstein-1 distance, a maximum mean discrepancy (MMD), or Kullback-Leibler divergence.

[0065] In an embodiment, the trained surrogate models may be deployed on edge-computing devices co-located with the equipment and are asynchronously synchronized with a cloud-based training environment to minimize inference latency without interrupting operations in the facility.

[0066] In an embodiment, the surrogate models may be organized in a directed graph mirroring a process topology of the facility 200, such that an output of an upstream one of the surrogateAttorney Docket No.: IS23.0617-WO-PCT models serves as an input to a downstream one of the surrogate models, thereby enabling end-to-end surrogate simulation of the facility 200.

[0067] In an embodiment, the method 500 may also include automatically switching from the surrogate models to the simulation model in response to a real-time estimate of an error of the surrogate models exceeds a predetermined threshold, thereby ensuring operational reliability under anomalous conditions.Agent Example

[0068] Figures 6A-6F illustrate an end to end reasoning and act-style agent workflow for surrogate model creation, according to an embodiment. More particularly, Figures 6A-6F show (i) problem characterization and structure discovery; (ii) physics-aware data imputation; (iii) space-filling sampling and simulator runs; (iv) baseline model training and overfit diagnosis; (v) physics-informed modeling; (vi) uncertainty-driven evaluation; (vii) adaptive sampling and retraining; (viii) ensemble and uncertainty quantification; (ix) sensitivity analysis; and (x) packaging / deployment. This may include explicit think-act-ob serve traces for auditability, physics-informed regularization to keep outputs physically plausible, automated adaptive sampling focused on high-uncertainty regions, calibrated uncertainty driving recommendations, and one-click deployment artifacts.Exemplary Computing System

[0069] In some embodiments, the methods of the present disclosure may be executed by a computing system. Figure 7 illustrates an example of such a computing system 700, in accordance with some embodiments. The computing system 700 may include a computer or computer system 701A, which may be an individual computer system 701A or an arrangement of distributed computer systems. The computer system 701A includes one or more analysis modules 702 that are configured to perform various tasks according to some embodiments, such as one or more methods disclosed herein. To perform these various tasks, the analysis module 702 executes independently, or in coordination with, one or more processors 704, which is (or are) connected to one or more storage media 706. The processor(s) 704 is (or are) also connected to a network interface 707 to allow the computer system 701A to communicate over a data network 709 with one or more additional computer systems and / or computing systems, such as 70 IB, 701C, and / orAttorney Docket No.: IS23.0617-WO-PCT701D (note that computer systems 701B, 701 C and / or 701D may or may not share the same architecture as computer system 701A, and may be located in different physical locations, e.g., computer systems 701A and 701B may be located in a processing facility, while in communication with one or more computer systems such as 701C and / or 701D that are located in one or more data centers, and / or located in varying countries on different continents).

[0070] A processor may include a microprocessor, microcontroller, processor module or subsystem, programmable integrated circuit, programmable gate array, or another control or computing device.

[0071] The storage media 706 may be implemented as one or more computer-readable or machine-readable storage media. Note that while in the example embodiment of Figure 7 storage media 706 is depicted as within computer system 701A, in some embodiments, storage media 706 may be distributed within and / or across multiple internal and / or external enclosures of computing system 701A and / or additional computing systems. Storage media 706 may include one or more different forms of memory including semiconductor memory devices such as dynamic or static random access memories (DRAMs or SRAMs), erasable and programmable read-only memories (EPROMs), electrically erasable and programmable read-only memories (EEPROMs) and flash memories, magnetic disks such as fixed, floppy and removable disks, other magnetic media including tape, optical media such as compact disks (CDs) or digital video disks (DVDs), BLURAY® disks, or other types of optical storage, or other types of storage devices. Note that the instructions discussed above may be provided on one computer-readable or machine-readable storage medium, or may be provided on multiple computer-readable or machine-readable storage media distributed in a large system having possibly plural nodes. Such computer-readable or machine-readable storage medium or media is (are) considered to be part of an article (or article of manufacture). An article or article of manufacture may refer to any manufactured single component or multiple components. The storage medium or media may be located either in the machine running the machine-readable instructions, or located at a remote site from which machine-readable instructions may be downloaded over a network for execution.

[0072] In some embodiments, computing system 700 contains one or more method execution module(s) 708. In the example of computing system 700, computer system 701A includes the method execution module 708. In some embodiments, a single method execution module may be used to perform some aspects of one or more embodiments of the methods disclosed herein. InAttorney Docket No.: IS23.0617-WO-PCT other embodiments, a plurality of method execution modules may be used to perform some aspects of methods herein.

[0073] It should be appreciated that computing system 700 is merely one example of a computing system, and that computing system 700 may have more or fewer components than shown, may combine additional components not depicted in the example embodiment of Figure 7, and / or computing system 700 may have a different configuration or arrangement of the components depicted in Figure 7. The various components shown in Figure 7 may be implemented in hardware, software, or a combination of both hardware and software, including one or more signal processing and / or application specific integrated circuits.

[0074] Further, the steps in the processing methods described herein may be implemented by running one or more functional modules in information processing apparatus such as general purpose processors or application specific chips, such as ASICs, FPGAs, PLDs, or other appropriate devices. These modules, combinations of these modules, and / or their combination with general hardware are included within the scope of the present disclosure.

[0075] Computational interpretations, models, and / or other interpretation aids may be refined in an iterative fashion; this concept is applicable to the methods discussed herein. This may include use of feedback loops executed on an algorithmic basis, such as at a computing device (e.g., computing system 700, Figure 7), and / or through manual control by a user who may make determinations regarding whether a given step, action, template, model, or set of curves has become sufficiently accurate for the evaluation of the subsurface three-dimensional geologic formation under consideration.

[0076] The foregoing description, for purposes of explanation, has been described with reference to specific embodiments. However, the illustrative discussions above are not intended to be exhaustive or limiting to the precise forms disclosed. Many modifications and variations are possible in view of the above teachings. Moreover, the order in which the elements of the methods described herein are illustrated and described may be re-arranged, and / or two or more elements may occur simultaneously. The embodiments were chosen and described in order to best explain the principles of the disclosure and its practical applications, to thereby enable others skilled in the art to best utilize the disclosed embodiments and various embodiments with various modifications as are suited to the particular use contemplated.

Claims

Attorney Docket No.: IS23.0617-WO-PCTCLAIMSWhat is claimed is:

1. A method for creating a surrogate model using a generative artificial intelligence (Al) model framework, the method comprising: performing a simulation of a facility using a simulation model to produce a simulation output, wherein the facility is used to process oil and / or gas; receiving first data related to equipment in the facility; generating a plurality of surrogate models based upon the simulation output and the first data; performing simulations using the surrogate models to produce surrogate outputs that predict the simulation output; and generating a recommendation based upon the surrogate outputs.

2. The method of claim 1, wherein the first data is measured by one or more sensors, and wherein the equipment comprises a pump, a heat exchanger, or both.

3. The method of claim 1, wherein each of the surrogate models is generated and trained by iteratively: generating the surrogate model using a first agent; generating training data; training the surrogate model to produce a trained surrogate model based upon the training data; and evaluating the trained surrogate model until it converges.

4. The method of claim 3, wherein the surrogate outputs are predicted by the trained surrogate models faster than the simulation output is predicted by the simulation model, and wherein performing the simulation comprises forward-simulating the surrogate outputs, optimizing the surrogate outputs, forecasting the surrogate outputs, or a combination thereof.

5. The method of claim 3, wherein the first agent implements a generative Al model.Attorney Docket No.: IS23.0617-WO-PCT6. The method of claim 5, wherein the training data is generated based upon (1) simulation queries at points selected by the first agent, (2) synthetic samples from the generative Al model, and / or (3) historical data from one or more sensors.

7. The method of claim 5, wherein the trained surrogate model is evaluated by the first agent and a second agent that critiques the trained surrogate model.

8. The method of claim 1, wherein the recommendation is to repair the equipment, replace the equipment, operate the equipment at a different speed, operate the equipment at a different pressure, operate the equipment at a different temperature, or a combination thereof.

9. The method of claim 1, further comprising automatically switching from the surrogate models to the simulation model in response to a real-time estimate of an error of the surrogate models exceeds a predetermined threshold, thereby ensuring operational reliability under anomalous conditions.

10. The method of claim 1, further comprising generating or transmitting a signal that instructs or causes an automated action to implement the recommendation.

11. A computing system, comprising: one or more processors; and a memory system comprising one or more non-transitory computer-readable media storing instructions that, when executed by at least one of the one or more processors, cause the computing system to perform operations, the operations comprising: performing a simulation of a facility using a simulation model to produce a simulation output, wherein the facility is used to process oil and / or gas; receiving first data related to equipment in the facility, wherein the first data is measured by one or more sensors; generating a plurality of surrogate models based upon the simulation output and the first data, wherein each of the surrogate models is generated and trained by iteratively:Attorney Docket No.: IS23.0617-WO-PCT generating the surrogate model using a first agent; generating training data; training the surrogate model to produce a trained surrogate model based upon the training data; and evaluating the trained surrogate model until it converges; performing simulations using the surrogate models to produce surrogate outputs that predict the simulation output; and generating a recommendation based upon the surrogate outputs.

12. The computing system of claim 11, wherein the operations further comprise augmenting the training data with synthetic operating scenarios generated by the first agent, such that rare or safety-critical regimes that are under-represented in the first data are probabilistically up-sampled before training the surrogate models.

13. The computing system of claim 11, wherein each surrogate model is regularized by one or more physics-informed loss terms that enforce conservation of mass, energy, or momentum, thereby constraining the surrogate outputs of the surrogate models to remain physically plausible, and wherein the physics-informed loss terms comprise (1) residual penalties enforcing conservation of mass, energy, or momentum, (2) constraint penalties for operational bounds on pressure, temperature, or flow, and (3) governing-equation residuals for ordinary or partialdifferential equations.

14. The computing system of claim 11, wherein the operations further comprise quantifying predictive uncertainty for each trained surrogate model via an ensemble, Bayesian neural -network, or Monte-Carlo dropout technique, and wherein the recommendation includes a confidence score derived from the quantified predictive uncertainty.

15. The computing system of claim 11, wherein the operations further comprise an active-learning loop that: automatically identifies regions where an uncertainty of the surrogate outputs of the surrogate models is greater than a threshold;Attorney Docket No.: IS23.0617-WO-PCT triggers additional high-fidelity simulations of the simulation model in the identified regions; and retrains the surrogate models with outputs from the additional high-fidelity simulations and / or new data from the one or more sensors until a predefined accuracy threshold is met.

16. A non-transitory computer-readable medium storing instructions that, when executed by one or more processors of a computing system, cause the computing system to perform operations, the operations comprising: performing a simulation of a facility using a simulation model to produce a simulation output, wherein the facility is used to process oil and / or gas; receiving first data related to equipment in the facility, wherein the first data is measured by one or more sensors; generating a plurality of surrogate models based upon the simulation output and the first data, wherein each of the surrogate models is generated and trained by iteratively: generating the surrogate model using a first agent; generating training data; training the surrogate model to produce a trained surrogate model based upon the training data; and evaluating the trained surrogate model until it converges; performing simulations using the surrogate models to produce surrogate outputs that predict the simulation output; and generating a recommendation based upon the surrogate outputs.

17. The non-transitory computer-readable medium of claim 16, wherein convergence of the trained surrogate models is assessed within a sliding time window of the first data to detect non-stationary behavior in the facility, and wherein failure to converge within the sliding time window automatically invokes a model-drift alarm.

18. The non-transitory computer-readable medium of claim 16, wherein the first agent comprises a generative- adversarial -network (GAN) generator and a second agent comprises a GAN discriminator, wherein the GAN discriminator is configured to minimize a divergence metricAttorney Docket No.: IS23.0617-WO-PCT between a distribution of the surrogate outputs from the surrogate models and a distribution of the simulation outputs from the simulation model, and wherein the divergence metric comprises a Jensen-Shannon divergence, a Wasserstein-1 distance, a maximum mean discrepancy (MMD), or Kullback-Leibler divergence.

19. The non-transitory computer-readable medium of claim 16, wherein the trained surrogate models are deployed on edge-computing devices co-located with the equipment and are asynchronously synchronized with a cloud-based training environment to minimize inference latency without interrupting operations in the facility.

20. The non-transitory computer-readable medium of claim 16, wherein the surrogate models are organized in a directed graph mirroring a process topology of the facility, such that an output of an upstream one of the surrogate models serves as an input to a downstream one of the surrogate models, thereby enabling end-to-end surrogate simulation of the facility.

Images