Search NASA⌕ Search

SEARCH · Search NASA

Results for “Reservoir Model”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 739 records · Page 41

Informing Geologic CO 2 Storage Site Management Decisions under Uncertainty: Demonstration of NRAP's Integrated Assessment Model (NRAP-IAM-CS) Application

Managing risks at geologic CO 2 storage sites requires making decisions based on the quantitative predictions of long-term performance that not only take into account various subsurface hydrologic, geochemical processes and interactions but also the related uncertainties. The US DOE (United States Department of Energy) funded National Risk Assessment Partnership (NRAP) has developed a quantitative approach and a model to predict the long-term behavior of GCS sites. The approach uses an integrated assessment model (NRAP-IAM-CS) that couples reduced order models (ROMs) for various GCS components (such as storage reservoir, wells, seals, and groundwater aquifer) in a stochastic modeling framework that allows for rapid simulations of entire system behavior over thousands of years while probing uncertainties. We demonstrate application of NRAP-IAM-CS using two examples that focus on CO 2 leakage and resulting impacts. The examples are used to demonstrate how different uncertain parameters affect overall performance objectives.

03 NATURAL GAS↗

Volumetric Properties and Fluid Phase Equilibria of CO2 + H2O

The need for accurate modeling of fluid-mineral processes over wide ranges of temperature, pressure and composition highlighted considerable uncertainties of available property data and equations of state, even for the CO2 + H2O binary system. In particular, the solubility, activity, and ionic dissociation equilibrium data for the CO2-rich phase, which are essential for understanding dissolution/precipitation, fluid-matrix reactions, and solute transport, are uncertain or missing. In this paper we report the results of a new experimental study of volumetric and phase equilibrium properties of CO2 + H2O, to be followed by measurements for bulk and confined multicomponent fluid mixtures. Mixture densities were measured by vibrating tube densimetry (VTD) over the entire composition range at T = 200 and 250 C and P = 20, 40, 60, and 80 MPa. Initial analysis of the mutual solubilities, determined from volumetric data, shows good agreement with earlier results for the aqueous phase, but finds that the data of Takenouchi and Kennedy (1964) significantly overestimated the solubility of water in supercritical CO2 (by a factor of more than two at 200 C). Resolving this well-known discrepancy will have a direct impact on the accuracy of predictive modeling of CO2 injection in geothermal reservoirs and geological carbon sequestration through improved equations of state, needed for calibration of predictive molecular-scale models and large-scale reactive transport simulations.

Capobianco, Ryan↗

Exploring Effective Best Management Practices in the Miyun Reservoir Watershed, China

Miyun reservoir watershed is a major source of drinking water for China’s capital, Beijing, which has a population of 21.75 million. Recently, the capacity of the Miyun reservoir to supply clean drinking water is threatened by increasing eutrophication (or algae bloom), mainly due to the discharge of wastewater and excessive fertilization application in the upstream watershed. Therefore, there is an urgent need to design effective best management practices (BMPs) to reduce upstream nutrient load and improve water quality in the Miyun reservoir. In this study, we built a watershed model (i.e. the Soil and Water Assessment Tool) in the Miyun Reservoir Watershed (MRW), and calibrated and validated it using long term sediment, nitrogen (N) and phosphorus (P) data. Further, we developed a Markov Chain based multi-objective optimization program to explore optimal BMPs with tradeoffs between economic costs and water quality responses. With the watershed model and the multi-objective optimization algorithms, we explored potential effectiveness of BMPs under two scenarios that are currently being considered: scenario 1 assumes funding of BMP implementation is from national grants and targets high water quality standard, while scenario 2 assumes funding is from farmers and targets water quality that barely meet the drinking water standard. We found substantial discrepancy between these two scenarios with respect to type and spatial configuration of BMPs and associated economic costs, highlighting the need of reconciling concerns from different stakeholders in order to arrive at a BMP plan that all parties will agree upon. In addition, we found that cross-subwatershed coordination and targeting flood season instead of year-round water quality standards could pronouncedly reduce economic cost of BMP implementation while not substantially degrade water quality. The watershed scale optimization method developed here hold the promise serve as an effective tool to explore tradeoffs between economic costs, water quality improvements, and decisions makers and stakeholders’ concerns in BMPs design, thereby informing sustainable watershed scale water resources management.

Qiu, Jiali↗

Evaluating proxies for the drivers of natural gas productivity using machine-learning models

We report the extensive development of unconventional reservoirs using horizontal drilling and multistage hydraulic fracturing has generated large volumes of reservoir characterization and production data. The analysis of this abundant data using statistical methods and advanced machine-learning (ML) techniques can provide data-driven insights into well performance. Most predictive modeling studies have focused on the impact that different well completion and stimulation strategies have on well production but have not fully exploited the available in situ rock property data to determine its role in reservoir productivity. We have used machine-learning techniques to rank rock mechanical properties, microseismic attributes, and stimulation parameters in the order of their significance for predicting natural gas production from an unconventional reservoir. The data for this study came from a hydraulically fractured well in the Marcellus Shale in Monongalia County, West Virginia. The data classes included measurements aggregated by well completion stage that included (1) gas production, (2) well-log-derived measurements including bulk density, elastic moduli, shear impedance, compressional impedance, brittleness, and gamma measurements, (3) microseismic attributes, (4) long-period long-duration (LPLD) event counts, (5) fracture counts, and (6) stimulation parameters that included the fluid injection volume and average pumping pressure. To identify observable proxies for the drivers of gas production, we evaluated five commonly used ML approaches including multivariate adaptive regression spline, Gaussian mixture model, random forest, gradient boosting, and neural network. We selected five variables including LPLD event count, seismogenic b-value, hydraulic diffusivity, cumulative moment, and fluid volume as the features most likely to impact gas productivity at the stage level in the study area. The data-driven selection of these parameters for their importance in determining gas production can help reservoir engineers design more effective hydraulic-fracture treatments in the Marcellus Shale and other similar unconventional reservoirs. Plain language summary: We use machine-learning methods and data-driven selection of reservoir parameters to rank and better understand their importance in determining gas production, which can help reservoir engineers design more effective hydraulic-fracture treatments in the Marcellus Shale and other similar unconventional reservoirs.

58 GEOSCIENCES↗

Zircon Lu-Hf systematics: Evidence for the episodic development of Archaean greenstone belts

A combined U-Th-Pb and Lu-Hf isotopic study of zircons was undertaken in order to determine the provenance and age of an Archean granite-greenstone terrain and to test the detailed application of the Lu-Hf system in various Archean zircons. The eastern Wawa subprovince of the Superior province consists of the low grade Michipicoten and Gamitagama greenstone belts and the granitic terrain. The Hf isotopic data indicate that the typical lithological features of a greenstone belt cycle could be accommodated in a crustal growth model that involved decreasing depth of melting in three isotopically distinct reservoirs: mantle, lower crust and upper crust. The model age of the sources given by the intersection of the lower crustal curve with the bulk earth evolution curve is about 2900 My, in good agreement with the zircon U-Pb basement age. This linear array also has a similar intersection age to that of Proterozoic carbonatite complexes. The general convergence of the other reservoir vectors around this age suggests that mantle depletion, crustal extraction and intracrustal differentiation were all part of the same episodic event. It is also apparent that recycling of older basement was important in the formation of many of the later greenstone belt rocks.

Smith, P. E.↗

SNC Meteorites and Martian Reservoirs

Jones first suggested that the inverse covariation of initial epsilon (Nd-143) and Sr-87/Sr-86 of the shergottites could be explained by interaction between mantle-derived magmas with another isotopic reservoir(s) (i.e., assimilation or contamination). In that model, magmas were generated in a source region that was isotopically similar to the Nakhla source and the second reservoir(s) was presumed to be crust. The text also permitted the second reservoir to be another type of mantle, but I can confirm that a second mantle reservoir was never seriously considered by that author. Other features of this model were that (i) it occurred at a particular time, 180 m.y. ago, and (ii) the interacting reservoirs had been separated at approximately 4.5 b.y. In a later paper Jones explored this mixing model more quantitatively and concluded that magmas from a Nakhla-like source region at 180 m.y. would fall on or near an isotopic Nd-Sr-Pb hyperplane defined by the shergottites. This criterion was a necessary prerequisite for the parent magma(s) of the shergottites to have initially been Nakhla-like isotopically. At this juncture, it is perhaps worthwhile to note that this mixing model was not presented to explain geochemical variations but as a justification for a 180 m.y. crystallization age for the shergottites and a 1.3 b.y. crystallization age for the nakhlites. In the mid-1980's crystallization ages estimated for Nakhla ranged from approximately 1.3 b.y to 4.5 b.y. Similarly, preferred crystallization ages for the shergottites ranged from 360 m.y., to 1.3 b.y., to 4.5 b.y. In all these models, the 180 m.y. event seen in the shergottites was deemed to be metamorphic. The fit between the Nakhla-like source region and the shergottite hyperplane was a validation both of the 1.3 b.y. igneous age of Nakhla and the 180 m.y. igneous age of the shergottites.

Jones, J. H.↗

Numerical modeling to assess the sensitivity and resolution of long-electrode electrical resistance tomography (LEERT) surveys to monitor CO 2 migration, Phase 1B area

This document describes the results of a numerical modeling study that evaluated whether LEERT could be used successfully to monitor CO 2 distribution in the Weyburn- Midale reservoir, Phase 1B area. The magnitude of electrical resistivity changes and the technique’s resolution depend on many site-specific factors including well separation distances, casing lengths, reservoir depth, thickness, and composition, and the effect of CO 2 on the electrical properties of the reservoir. Phase 1B-specific numerical modeling of the electrical response to CO 2 injection has been performed to assess sensitivity and resolution of the electrical surveys.

54 ENVIRONMENTAL SCIENCES↗

EGS Collab Experiment 1: 3D Seismic Velocity Model and Updated Microseismic Catalog Using Transfer-Learning Aided Double-Difference Tomography

This package contains a 3D Seismic velocity model and an updated microseismic catalog associated with a proceedings paper (Chai et al., 2020) published in the 45th Workshop on Geothermal Reservoir Engineering. The 3D_seismic_velocity_model text file contains x (m), y(m), z(m), P-wave velocity (km/s), P-wave velocity quality indicator (1 for well-constrained; 0 for poorly constrained), S-wave velocity (km/s), and S-wave velocity quality indicator (1 for well-constrained; 0 for poorly constrained). The Updated_MEQ_catalog text file contains event origin time, x(m), y(m), z(m), error in x (m), error in y (m), error in z (m), and RMS misfit (millisecond). The 3D_seismic_P-wave_velocity_model animation file shows slices of the 3D P-wave velocity model. The 3D_seismic_S-wave_velocity_model animation file shows slices of the 3D S-wave velocity model. The Interactive_MEQ_locations API file is an interactive visualization of the updated microseismic event locations. The visualization allows users to view the event locations by dragging, rotating, and zooming in. References: Chai, C., Maceira, M., Santos-Villalobos, H. J., Venkatakrishnan, S. V., Schoenball, M., and EGS Collab Team, 2020, Automatic Seismic Phase Picking Using Deep Learning for the EGS Collab Project, in PROCEEDINGS, 45th Workshop on Geothermal Reservoir Engineering, edited, Stanford University, Stanford, California, 45, 1266-1276.

15 GEOTHERMAL ENERGY↗

Phase Behavior of Methane–Ethane Mixtures in Nanopores

The behavior of fluids in confined spaces is important in several research fields, including the oil and gas industry, because of the importance of confined hydrocarbons in unconventional shale reservoirs. A large percentage of the hydrocarbons are contained in nanoscale shale pores, and as a result, interactions between fluid molecules and the pore walls become significant to the fluid behavior and cause deviations of the fluid properties from bulk properties. In this work, classical molecular dynamics (MD) simulations and grand canonical Monte Carlo (GCMC) simulations are performed to model kerogen pores in shale reservoirs and their effect on adsorption selectivity and phase equilibria of hydrocarbon mixtures in a wide range of temperatures and pressures. A slit pore made of graphite walls is used to represent the basic unit where a fluid may be stored in shale reservoirs. The separation between graphite layers is varied to observe the effect of confinement on the adsorption selectivity and phase equilibrium of methane-ethane mixtures. Results from GCMC and MD simulations show that ethane is preferred over methane in the confined system. Moreover, most fluid molecules in the confined system tend to be adsorbed on the pore walls instead of remaining in the gas phase. Preferential interactions between the fluid molecules and the pore walls induce changes in the phase diagram in the confined system. Furthermore, the results from GCMC simulations show shifts with respect to the bulk pressure-composition phase diagram that becomes more pronounced as the pore size decreases.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

The Contribution of Reservoirs to Global Land Surface Water Storage Variations

Man-made reservoirs play a key role in the terrestrial water system. They alter water fluxes at the land surface and impact surface water storage through water management regulations for diverse purposes such as irrigation, municipal water supply, hydropower generation, and flood control. Although most developed countries have established sophisticated observing systems for many variables in the land surface water cycle, long-term and consistent records of reservoir storage are much more limited and not always shared. Furthermore, most land surface hydrological models do not represent the effects of water management activities. Here, the contribution of reservoirs to seasonal water storage variations is investigated using a large-scale water management model to simulate the effects of reservoir management at basin and continental scales. The model was run from 1948 to 2010 at a spatial resolution of 0.258 latitude–longitude. A total of 166 of the largest reservoirs in the world with a total capacity of about 3900 km3 (nearly 60%of the globally integrated reservoir capacity) were simulated. The global reservoir storage time series reflects the massive expansion of global reservoir capacity; over 30 000 reservoirs have been constructed during the past half century, with a mean absolute interannual storage variation of 89 km3. The results indicate that the average reservoir-induced seasonal storage variation is nearly 700 km3 or about 10%of the global reservoir storage. For some river basins, such as the Yellow River, seasonal reservoir storage variations can be as large as 72%of combined snow water equivalent and soil moisture storage.

58 GEOSCIENCES↗

Using well casing as an electrical source to monitor hydraulic fracture fluid injection

The depth to surface resistivity (DSR) method transmits current from a source located in a cased or openhole well to a distant surface return electrode while electric field measurements are made at the surface over the target of interest. This paper presents both numerical modelling results and measured data from a hydraulic fracturing field test where conductive water was injected into a resistive shale reservoir during a hydraulic fracturing operation. Modelling experiments show that anomalies due to hydraulic fracturing are small but measureable with highly sensitive sensor technology. The field measurements confirm the model results,showing that measured differences in the surface fields due to hydraulic fracturing have been detected above the noise floor. Our results show that the DSR method is sensitive to the injection of frac fluids; they are detectable above the noise floor in a commercially active hydraulic fracturing operation, and therefore this method can be used for monitoring fracture fluid movement.

02 PETROLEUM↗

Strain Measurement of Geological Samples Subjected to Triaxial Stresses Experienced During Hydraulic Loading

Understanding stress and strain relationships and having the ability to predict these quantities for known load conditions is crucial to all geomechanical and, in some instances, reservoir flow applications. The constitutive equations governing the deformation of geological materials are typically adequate for bulk or large scale deformation and stress analyses. However, these rules are generally less precise in their ability to make accurate predictions in physical processes where highly localized material heterogeneity exists or where the presence of geometric irregularities such as micro-cracks may be present. This is especially relevant to EGS where hydraulic fracture propagation models are needed to develop optimal reservoir creation strategies and where fracture permeability is significantly influenced by regional stress states and may affect reservoir operation strategies. The deficiencies of the models used to describe these physical processes are a practical reality necessitated by the manner in which rock properties must be obtained. Conventional rock mechanics tests subject samples to controlled load conditions and measure bulk deformations of the sample or more localized deformations only on exposed surfaces of the sample. They are currently unable to comprehensively map the deformation state within the sample. For processes such as fracture, however, the state of a particular region within the rock drives the overall failure behavior of the sample. The authors believe that possessing a means to measure strains within samples subjected to hydraulic fracture loading conditions will provide a useful tool for understanding the localized effects not captured by conventional techniques and may serve as a method for improving hydraulic fracture models. An ongoing effort at Oak Ridge National Laboratory endeavors to develop a neutron diffraction based strain measurement capability to interrogate the strain state of a geological sample, at arbitrary internal locations, subjected to a triaxial stress state. The basis of the method and initial results for simple load conditions were reported at last year s Stanford Geothermal Workshop. This work will report results from recent neutron diffraction strain measurement experiments in which marble samples were subjected to load conditions more representative of hydraulic fracturing operations within a pressure cell specially designed for the reported strain measurement technique.

An, Ke↗

Spontaneous imbibition of water and determination of effective contact angles in the Eagle Ford Shale Formation using neutron imaging

Understanding of fundamental processes and prediction of optimal parameters during the horizontal drilling and hydraulic fracturing process results in economically effective improvement of oil and natural gas extraction. Although, the modern analytical and computational models can capture fracture growth, there is a lack of experimental data on spontaneous imbibition and wettability in oil and gas reservoirs for the validation of further model development. In this work, we used neutron imaging to measure the spontaneous imbibition of water into fractures of Eagle Ford Shale with known geometries and fracture orientations. An analytical solution for a set of nonlinear second-order differential equations was applied to the measured imbibition data to determine effective contact angles. The analytical solution fit the measured imbibition data reasonably well and determined effective contact angles were slightly higher than static contact angles due to effects of in-situ changes in velocity, surface roughness, and heterogeneity of mineral surfaces on the fracture surface. Additionally, small fracture widths may have retarded imbibition and affected model fits, which suggests that average fracture widths are not satisfactory for modeling imbibition in natural systems.

DiStefano, Victoria H. (ORCID:0000000160239716)↗

CD8 + lymphocyte control of SIV infection during antiretroviral therapy

CD8 + lymphocytes play an important role in suppressing in vivo viral replication in HIV infection. However, both the extent to which and the mechanisms by which CD8 + lymphocytes contribute to viral control are not completely understood. A recent experiment depleted CD8 + lymphocytes in simian immunodeficiency virus (SIV)-infected rhesus macaques (RMs) on antiretroviral treatment (ART) to study the role of CD8 + lymphocytes. CD8 + lymphocytes depletion resulted in temporary plasma viremia in all studied RMs. Viral control was restored when CD8 + lymphocytes repopulated. We developed a viral dynamic model to fit the viral load (VL) data from the CD8 depletion experiment. We explicitly modeled the dynamics of the latent reservoir and the SIV-specific effector cell population including their exhaustion and their potential cytolytic and noncytolytic functions. We found that the latent reservoir significantly contributes to the size of the peak VL after CD8 depletion, while drug efficacy plays a lesser role. Our model suggests that the overall CD8 + lymphocyte cytolytic killing rate is dynamically changing depending on the levels of antigen-induced effector cell activation and exhaustion. Based on estimated parameters, our model suggests that before ART or without ART the overall CD8 cytolytic killing rate is small due to exhaustion. However, after the start of ART, the overall CD8 cytolytic killing rate increases due to an expansion of SIV-specific CD8 effector cells. Further, we estimate that the cytolytic killing rate can be significantly larger than the cytopathic death rate in some animals during the second phase of ART-induced viral decay. Lastly, our model provides a new explanation for the puzzling findings by Klatt et al. and Wong et al. that CD8 depletion done immediately before ART has no noticeable effect on the first phase viral decay slope seen after ART initiation Overall, by incorporating effector cells and their exhaustion, our model can explain the effects of CD8 depletion on VL during ART, reveals a detailed dynamic role of CD8 + lymphocytes in controlling viral infection, and provides a unified explanation for CD8 depletion experimental data.

59 BASIC BIOLOGICAL SCIENCES↗

Application of predictive data analytics to model daily hydrocarbon production using petrophysical, geomechanical, fiber-optic, completions, and surface data: A case study from the Marcellus Shale, North America

Predicting gas production from stimulated unconventional reservoirs has always been a challenge to oil and gas companies because the physics of fluid flow through such reservoirs is not well understood. On the other hand, modeling gas production from shale reservoirs is a complex multi-variate problem that requires proper integration of data from multiple disciplines, such as geology, petrophysics, geomechanics, completions, and surface-based measurements, etc. Here this study demonstrates the application of data-driven machine learning algorithms, integrating geoscientific, distributed acoustic sensing (DAS), distributed temperature sensing (DTS) fiber-optic, completions, flow scanner production log, and surface data to model daily gas production from a 28-stage stimulated horizontal well drilled in the Marcellus Shale of the Appalachian basin in North America. In addition, this study aims to utilize the data from a fiber-optic monitoring system, such as DAS and DTS to evaluate the well performance. We build supervised data-driven machine learning models using Random Forest (RF), Artificial Neural Network (ANN), and Support Vector Machine (SVM) algorithms to understand the well performance and forecast the daily gas production. The study compares the forecasted results, tests the level of accuracy, and addresses different issues with these machine learning algorithms. A spatio-temporal database is constructed and used to perform sensitivity analyses to identify the key drivers for gas production. The results show that RF and ANN algorithms can be used to predict daily gas production with significantly high accuracy; however, RF algorithm is the best predictor in terms of highest accuracy (96%), less computational time and cost. Based on sensitivity analysis, it appears that Poisson's ratio, minimum horizontal stress, DAS, casing pressure, gamma log are the most important parameters to predict gas production.

02 PETROLEUM↗

Optimal carbon storage reservoir management through deep reinforcement learning

Model-based optimization plays a central role in energy system design and management. The complexity and high-dimensionality of many process-level models, especially those used for geosystem energy exploration and utilization, often lead to formidable computational costs when the dimension of decision space is also large. This work adopts elements of recently advanced deep learning techniques to solve a sequential decision-making problem in applied geosystem management. Specifically, a deep reinforcement learning framework was formed for optimal multiperiod planning, in which a deep Q-learning network (DQN) agent was trained to maximize rewards by learning from high-dimensional inputs and from exploitation of its past experiences. To expedite computation, deep multitask learning was used to approximate high-dimensional, multistate transition functions. Both DQN and deep multitask learning are pattern based. As a demonstration, the framework was applied to optimal carbon sequestration reservoir planning using two different types of management strategies: monitoring only and brine extraction. Both strategies are designed to mitigate potential risks due to pressure buildup. Results show that the DQN agent can identify the optimal policies to maximize the reward for given risk and cost constraints. Finally, experiments also show that knowledge the agent gained from interacting with one environment is largely preserved when deploying the same agent in other similar environments.

15 GEOTHERMAL ENERGY↗

Impacts of irregularly-distributed acidified brine flow on geo-chemo-mechanical alteration in an artificial shale fracture under differential stress

The efficacy of geological carbon sequestration is reliant on the integrity of the caprock and its resistance to physical and chemical alteration. Caprocks with high abundance of reactive carbonates like calcite are susceptible to acid-promoted dissolution and can result in structural weakening. This work investigates the effect of acidified brine flow through an artificially fractured, high-carbonate (30 % by XRD) shale under differential compressive stress. Cylindrical samples were cut in half vertically and milled to create an artificial fracture with interlocking asperities and open channels. Samples were sheared with a single applied stress in a custom flow cell housed within an industrial CT scanner. Further, either acidic (pH 4) or reservoir-simulated (pH 9.5) brine was flowed through the artificial fracture for 7–8 days under reservoir pressure and room temperature. Model simulations indicate flow mainly occurred in open channels, with limited flow between overlapping asperities. Analysis of fracture surfaces by optical and scanning electron microscopy show increased surface alteration and roughness after exposure to pH 4 versus pH 9.5 brine indicating mineral dissolution/loss, and this effect is greater in areas that receive the highest brine flows. Similarly, surface analysis by scratch testing shows fracture toughness decreases more after exposure to acidic versus reservoir-simulated brine, with the greatest alteration in areas of highest acidic brine flows. Despite weakening, no shear slip occurred. Overall, the results indicate that acidified brine can result in significant physical and geomechanical alteration of irregular fracture surfaces in shale caprock, with greatest effects in preferential flow regions.

58 GEOSCIENCES↗

Recurrent neural networks for short-term and long-term prediction of geothermal reservoirs

Accurate prediction of geothermal reservoir responses to alternative energy production scenarios is critical for optimizing the development of the underlying resources. While the conventional physics-based models offer a comprehensive prediction tool, data-driven models provide an efficient alternative to build fit-for-purpose predictive models by extracting and using the statistical patterns in the collected data to make predictions. The recurrent neural network (RNN) is a data-driven model that is commonly applied to predict time series sequences. This paper presents a variant of RNN that also utilizes the efficiency of convolutional neural networks (CNN) for the prediction of energy production from geothermal reservoirs. Specifically, a CNN–RNN architecture is developed that takes historical well controls as input (features) and their corresponding production response data as output (labels) to learn an input-output mapping that can predict the future well production responses/performance for any given future well control inputs. The model is paired with a labeling scheme to handle real field disturbances that create data gaps. In addition to the model structure, we introduce a thorough workflow for applying the model, which includes data pre-processing, feature selection, as well as different training strategies for short-term and long-term prediction. Finally, the performance and accuracy of the model are evaluated by applying it to multiple datasets, including a field reservoir model.

15 GEOTHERMAL ENERGY↗