Search citation statements
Paper Sections
Citation Types
Year Published
Publication Types
Relationship
Authors
Journals
Avoiding over-pressurization in subsurface reservoirs is critical for applications like CO2 sequestration and wastewater injection. Managing the pressures by controlling injection/extraction are challenging because of complex heterogeneity in the subsurface. The heterogeneity typically requires high-fidelity physics-based models to make predictions on CO2 fate. Furthermore, characterizing the heterogeneity accurately is fraught with parametric uncertainty. Accounting for both, heterogeneity and uncertainty, makes this a computationally-intensive problem challenging for current reservoir simulators. To tackle this, we use differentiable programming with a full-physics model and machine learning to determine the fluid extraction rates that prevent over-pressurization at critical reservoir locations. We use DPFEHM framework, which has trustworthy physics based on the standard two-point flux finite volume discretization and is also automatically differentiable like machine learning models. Our physics-informed machine learning framework uses convolutional neural networks to learn an appropriate extraction rate based on the permeability field. We also perform a hyperparameter search to improve the model’s accuracy. Training and testing scenarios are executed to evaluate the feasibility of using physics-informed machine learning to manage reservoir pressures. We constructed and tested a sufficiently accurate simulator that is 400 000 times faster than the underlying physics-based simulator, allowing for near real-time analysis and robust uncertainty quantification.
Avoiding over-pressurization in subsurface reservoirs is critical for applications like CO2 sequestration and wastewater injection. Managing the pressures by controlling injection/extraction are challenging because of complex heterogeneity in the subsurface. The heterogeneity typically requires high-fidelity physics-based models to make predictions on CO2 fate. Furthermore, characterizing the heterogeneity accurately is fraught with parametric uncertainty. Accounting for both, heterogeneity and uncertainty, makes this a computationally-intensive problem challenging for current reservoir simulators. To tackle this, we use differentiable programming with a full-physics model and machine learning to determine the fluid extraction rates that prevent over-pressurization at critical reservoir locations. We use DPFEHM framework, which has trustworthy physics based on the standard two-point flux finite volume discretization and is also automatically differentiable like machine learning models. Our physics-informed machine learning framework uses convolutional neural networks to learn an appropriate extraction rate based on the permeability field. We also perform a hyperparameter search to improve the model’s accuracy. Training and testing scenarios are executed to evaluate the feasibility of using physics-informed machine learning to manage reservoir pressures. We constructed and tested a sufficiently accurate simulator that is 400 000 times faster than the underlying physics-based simulator, allowing for near real-time analysis and robust uncertainty quantification.
Avoiding over-pressurization in subsurface reservoirs is critical for applications like CO$$_2$$ 2 sequestration and wastewater injection. Managing the pressures by controlling injection/extraction are challenging because of complex heterogeneity in the subsurface. The heterogeneity typically requires high-fidelity physics-based models to make predictions on CO$$_2$$ 2 fate. Furthermore, characterizing the heterogeneity accurately is fraught with parametric uncertainty. Accounting for both, heterogeneity and uncertainty, makes this a computationally-intensive problem challenging for current reservoir simulators. To tackle this, we use differentiable programming with a full-physics model and machine learning to determine the fluid extraction rates that prevent over-pressurization at critical reservoir locations. We use DPFEHM framework, which has trustworthy physics based on the standard two-point flux finite volume discretization and is also automatically differentiable like machine learning models. Our physics-informed machine learning framework uses convolutional neural networks to learn an appropriate extraction rate based on the permeability field. We also perform a hyperparameter search to improve the model’s accuracy. Training and testing scenarios are executed to evaluate the feasibility of using physics-informed machine learning to manage reservoir pressures. We constructed and tested a sufficiently accurate simulator that is 400 000 times faster than the underlying physics-based simulator, allowing for near real-time analysis and robust uncertainty quantification.
Interest in CCS project development is accelerating in SE Asia, driven by the need to monetize emission-intensive assets in the region while complying with increasingly ambitious GHG emissions targets. Depleted hydrocarbon fields represent an attractive storage option for early CCS project due the enhanced understanding of the reservoir, its dynamic behavior, and proven storage capability. Re-use of existing infrastructure also presents the potential to reduce both project costs and time to first injection, however, these brownfield sites also carry significant risk to the long-term, safe containment of injected CO2 through risk of leakage via legacy wells. A methodology is presented in this paper to investigate the risk-reward balance of developing a depleted gas field as a storage site in the Gulf of Thailand. A screening process to assess all abandoned, suspended, and active wells is used to identify wells with re-use potential as CO2 injectors or CO2 plume monitoring wells, and those which represent a leakage risk to the project. A set of legacy well risk identifiers is generated for the field based on well construction records, descriptions of current well barriers, well utilization history, and current best practice guidelines. Southeast Asia has significant remaining reserves of oil and gas, and coal, and an active liquefied natural gas (LNG) export industry. The region's energy demand is increasing rapidly and is forecast to continue to grow over the next decades (World Economic Forum, 2019). To date, fossil fuels have supplied nearly 90% of this growth in the demand for energy in the region (IEA, 2021). To meet this growing energy demand, several new gas projects are under development across Southeast Asia, but many of these are associated with high CO2 gas fields where the produced gas contains significant (up to 70% by volume) CO2 (GCCSI, 2020). In Thailand, where nearly 94% of the primary energy is met by fossil fuels (BP Statistical Review, 2022), the energy sector represents the biggest contributor (74% in 2013) to the country's greenhouse gas emissions (GHG; UNFCCC, 2020). However, as per the nationally determined contribution to the United Nations Framework Convention on Climate Change (UNFCCC), Thailand intends to reduce its GHG emissions by at least 20% from projected business as usual levels by the year 2030 (UNFCCC, 2020). Carbon capture and storage (CCS) represents one option to help meet this increased demand in fossil energy while also reducing GHG emissions. An approach which is gaining traction across the region is to utilize the high concentrations of CO2 stripped out of the raw gas streams at gas processing plants and, instead of venting to atmosphere, the CO2 can be compressed, dehydrated, and transported to suitable long-term storage locations. Depleted oil and gas fields form an attractive opportunity for long-term storage of CO2 due to the wealth of both static and dynamic knowledge available from appraisal through production activities. Depleted fields also have the advantage that they have a working primary seal for hydrocarbons, which has been proven over geological time and so can be considered, in general, to carry low risk of leakage through geological means. Brownfield sites can, however, also represent a challenge to project success through an increased risk to the containment of the injected CO2 due to the presence of legacy wells. These existing wells represent a variable risk to containment depending on well age and type, well history, well design, and plug and abandonment methodology applied. This paper presents the outcomes of a CO2 storage feasibility study for a depleted gas-condensate field in the Gulf of Thailand. The main aims of the study were to:1) identify the project risk associated with the integrity of the field legacy wells, and 2) to evaluate the potential for well re-use for the CO2 injection project. Reusing an existing field offers new life to an otherwise end-of-life asset, inching towards decommissioning and site closure. As commercial scale CO2 storage in depleted hydrocarbon fields represents a ‘First of a Kind’ project, the feasibility study is designed to evaluate the current status of the field and surface facilities with respect to CO2 injection and long-term storage. As a feasibility study, the focus of the technical work was to identify any ‘showstoppers’ which might indicate that the selected site was not suitable for long-term CO2 storage and, if sufficient positive storage indicators were identified, to select the most appropriate options for progression into a Concept Selection study in which more detailed engineering studies will be completed.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.