Multi-Pattern Generalization in CO2‑EOR: Physically Consistent Surrogate for Saturation-Field Evolution
Junwei Zhao, Yukun Dong, Jiyuan Zhang, Kuankuan Wu · 2026
Read Multi-Pattern Generalization in CO2‑EOR: Physically Consistent Surrogate for Saturation-Field Evolution by Junwei Zhao, Yukun Dong, Jiyuan Zhang, Kuankuan Wu (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-13261581.txt.
Description
Junwei Zhao and co-authors built in 2026 a learning-based surrogate that forecasts how injected carbon dioxide spreads through an oil reservoir. Full physics simulators are too slow for testing many injection plans, and simpler data models fail when well layouts change. Their MPG-STNet blends a Fourier neural operator, a graph attention network for well patterns, and a temporal transformer.
A generative network then paints future saturation fields, checked against a hard rule that mass must balance. Training used 4,500 simulated cases spanning nine well patterns over a 150-month horizon. The model scored very low error and near-unity structural similarity on held-out cases. It also held up on well layouts it had never seen. The authors offer the surrogate for fast screening of carbon storage and enhanced recovery plans.
Catalog record
| CREATOR | Junwei Zhao, Yukun Dong, Jiyuan Zhang, Kuankuan Wu |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 55 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |