Efficient and generalizable nested Fourier-DeepONet for three-dimensional geological carbon sequestration
Geological carbon sequestration (GCS) involves injecting CO[Formula: see text] into subsurface geological formations for permanent storage. Numerical simulations could guide decisions in GCS projects by predicting CO[Formula: see text] migration pathways and the pressure distribution in storage form...
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Taylor & Francis Group
2024-12-01
|
| Series: | Engineering Applications of Computational Fluid Mechanics |
| Subjects: | |
| Online Access: | https://www.tandfonline.com/doi/10.1080/19942060.2024.2435457 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1846136016205774848 |
|---|---|
| author | Jonathan E. Lee Min Zhu Ziqiao Xi Kun Wang Yanhua O. Yuan Lu Lu |
| author_facet | Jonathan E. Lee Min Zhu Ziqiao Xi Kun Wang Yanhua O. Yuan Lu Lu |
| author_sort | Jonathan E. Lee |
| collection | DOAJ |
| description | Geological carbon sequestration (GCS) involves injecting CO[Formula: see text] into subsurface geological formations for permanent storage. Numerical simulations could guide decisions in GCS projects by predicting CO[Formula: see text] migration pathways and the pressure distribution in storage formation. However, these simulations are often computationally expensive due to highly coupled physics and large spatial-temporal simulation domains. Surrogate modelling with data-driven machine learning has become a promising alternative to accelerate physics-based simulations. Among these, the Fourier neural operator (FNO) has been applied to three-dimensional synthetic subsurface models. Despite its good accuracy in simulating CO[Formula: see text] plume migration, it requires large computational resources in training and also lacks generalizability. Here, to further improve performance, we have developed a nested Fourier-DeepONet by combining the expressiveness of the FNO with the modularity of a deep operator network (DeepONet). This new framework is twice as efficient as a nested FNO for training and has at least 80% lower GPU memory requirement due to its flexibility to treat temporal coordinates separately. These performance improvements are achieved without compromising prediction accuracy. In addition, the generalization and extrapolation ability of nested Fourier-DeepONet beyond the training range has been thoroughly evaluated. Nested Fourier-DeepONet outperformed the nested FNO for extrapolation in time with more than 50% reduced error. It also exhibited good extrapolation accuracy beyond the training range in terms of reservoir properties, number of wells, and injection rate. |
| format | Article |
| id | doaj-art-4b62a2d56e4548ed92d38fbf5f2a3da2 |
| institution | Kabale University |
| issn | 1994-2060 1997-003X |
| language | English |
| publishDate | 2024-12-01 |
| publisher | Taylor & Francis Group |
| record_format | Article |
| series | Engineering Applications of Computational Fluid Mechanics |
| spelling | doaj-art-4b62a2d56e4548ed92d38fbf5f2a3da22024-12-09T09:43:45ZengTaylor & Francis GroupEngineering Applications of Computational Fluid Mechanics1994-20601997-003X2024-12-0118110.1080/19942060.2024.2435457Efficient and generalizable nested Fourier-DeepONet for three-dimensional geological carbon sequestrationJonathan E. Lee0Min Zhu1Ziqiao Xi2Kun Wang3Yanhua O. Yuan4Lu Lu5Department of Chemical and Environmental Engineering, Yale University, New Haven, CT, USADepartment of Statistics and Data Science, Yale University, New Haven, CT, USADepartment of Computer Science and Engineering, University of California, San Diego, CA, USAExxonMobil Technology and Engineering Company, Annandale, NJ, USAExxonMobil Technology and Engineering Company, Annandale, NJ, USADepartment of Statistics and Data Science, Yale University, New Haven, CT, USAGeological carbon sequestration (GCS) involves injecting CO[Formula: see text] into subsurface geological formations for permanent storage. Numerical simulations could guide decisions in GCS projects by predicting CO[Formula: see text] migration pathways and the pressure distribution in storage formation. However, these simulations are often computationally expensive due to highly coupled physics and large spatial-temporal simulation domains. Surrogate modelling with data-driven machine learning has become a promising alternative to accelerate physics-based simulations. Among these, the Fourier neural operator (FNO) has been applied to three-dimensional synthetic subsurface models. Despite its good accuracy in simulating CO[Formula: see text] plume migration, it requires large computational resources in training and also lacks generalizability. Here, to further improve performance, we have developed a nested Fourier-DeepONet by combining the expressiveness of the FNO with the modularity of a deep operator network (DeepONet). This new framework is twice as efficient as a nested FNO for training and has at least 80% lower GPU memory requirement due to its flexibility to treat temporal coordinates separately. These performance improvements are achieved without compromising prediction accuracy. In addition, the generalization and extrapolation ability of nested Fourier-DeepONet beyond the training range has been thoroughly evaluated. Nested Fourier-DeepONet outperformed the nested FNO for extrapolation in time with more than 50% reduced error. It also exhibited good extrapolation accuracy beyond the training range in terms of reservoir properties, number of wells, and injection rate.https://www.tandfonline.com/doi/10.1080/19942060.2024.2435457Geological carbon sequestration3D multiphase flow in porous mediadeep neural operatorFourier-DeepONetcomputational costextrapolation |
| spellingShingle | Jonathan E. Lee Min Zhu Ziqiao Xi Kun Wang Yanhua O. Yuan Lu Lu Efficient and generalizable nested Fourier-DeepONet for three-dimensional geological carbon sequestration Engineering Applications of Computational Fluid Mechanics Geological carbon sequestration 3D multiphase flow in porous media deep neural operator Fourier-DeepONet computational cost extrapolation |
| title | Efficient and generalizable nested Fourier-DeepONet for three-dimensional geological carbon sequestration |
| title_full | Efficient and generalizable nested Fourier-DeepONet for three-dimensional geological carbon sequestration |
| title_fullStr | Efficient and generalizable nested Fourier-DeepONet for three-dimensional geological carbon sequestration |
| title_full_unstemmed | Efficient and generalizable nested Fourier-DeepONet for three-dimensional geological carbon sequestration |
| title_short | Efficient and generalizable nested Fourier-DeepONet for three-dimensional geological carbon sequestration |
| title_sort | efficient and generalizable nested fourier deeponet for three dimensional geological carbon sequestration |
| topic | Geological carbon sequestration 3D multiphase flow in porous media deep neural operator Fourier-DeepONet computational cost extrapolation |
| url | https://www.tandfonline.com/doi/10.1080/19942060.2024.2435457 |
| work_keys_str_mv | AT jonathanelee efficientandgeneralizablenestedfourierdeeponetforthreedimensionalgeologicalcarbonsequestration AT minzhu efficientandgeneralizablenestedfourierdeeponetforthreedimensionalgeologicalcarbonsequestration AT ziqiaoxi efficientandgeneralizablenestedfourierdeeponetforthreedimensionalgeologicalcarbonsequestration AT kunwang efficientandgeneralizablenestedfourierdeeponetforthreedimensionalgeologicalcarbonsequestration AT yanhuaoyuan efficientandgeneralizablenestedfourierdeeponetforthreedimensionalgeologicalcarbonsequestration AT lulu efficientandgeneralizablenestedfourierdeeponetforthreedimensionalgeologicalcarbonsequestration |