论文成果

首页 - 论文成果

Physics-Informed Neural Network for CH4/CO2 Adsorption Characterization

摘要:This study addresses the critical need for accurate characterization of methane (CH4) and carbon dioxide (CO2) adsorption behavior in shale formations, pivotal for optimizing hydrocarbon extraction and advancing carbon neutrality goals. The research introduces a novel approach utilizing Physics-Informed Neural Networks (PINNs) to predict adsorption isotherms across diverse shale cores, integrating Langmuir adsorption theory into a data-driven model. By collecting a limited core dataset and leveraging automatic differentiation techniques, the PINN systematically incorporates physics knowledge into neural networks, compensating for data scarcity and enhancing predictive robustness. The method is validated through statistical analysis, feature selection, and cross-validation, demonstrating its superior performance compared to conventional Machine Learning (ML) models like Random Forest, with a 4.75% improvement in R2 for model performance. Overall, this approach represents a valuable tool for optimizing hydrocarbon recovery, offering insights into competitive adsorption phenomena and paving the way for more efficient and environmentally friendly extraction techniques in complex subsurface environments. Copyright 2024, Society of Petroleum Engineers.

ISSN号:9781959025214

卷、期、页:Society of Petroleum Engineers - SPE Canadian Energy Technology Conference and Exhibition, CET 2024, 2024

发表日期:2024-03-13

期刊分区(SCI为中科院分区):无

收录情况:EI(工程索引)

发表期刊名称:Society of Petroleum Engineers - SPE Canadian Energy Technology Conference and Exhibition, CET 2024

参与作者:Wang, Hai,Chen, Shengnan,Wang, Muming,Wu, Zhengbin

第一作者:惠钢

论文类型:会议论文

论文概要:Wang, Hai,Chen, Shengnan,Wang, Muming,Wu, Zhengbin,惠钢,Physics-Informed Neural Network for CH4/CO2 Adsorption Characterization,Society of Petroleum Engineers - SPE Canadian Energy Technology Conference and Exhibition, CET 2024,2024,Society of Petroleum Engineers - SPE Canadian Energy Technology Conference and Exhibition, CET 2024, 2024

论文题目:Physics-Informed Neural Network for CH4/CO2 Adsorption Characterization

分享到:
分享到
×
分享到微信朋友圈
×
Scan me!