Prediction Method of CO2 Plume Distribution Based on Physics-Informed Neural Networks
-
摘要:
为了提高CO2地质封存的有效性和安全性，需要准确预测地层中CO2羽流的分布和迁移规律。为此，利用自动微分技术，将多相渗流偏微分方程约束嵌入模型的损失函数中，建立了多相渗流力学约束的CO2羽流分布深度神经网络预测模型，以确保模型预测结果既符合训练数据样本的分布规律，又严格遵守偏微分方程描述的流体渗流物理规律。为了验证模型的有效性，以枯竭油藏封存CO2的实际案例为研究对象，分别应用多层感知器和长短期记忆深度神经网络构建了2个物理信息深度神经网络（PINNs）模型。研究表明，与纯数据驱动模型的预测结果相比，基于PINNs的模型具有更高的预测精度。研究结果不仅为CO2地质封存项目的设计与实施提供了技术支撑，也为该技术的实际应用提供了理论依据。
Abstract:To enhance the effectivity and safety of CO2 geological storage, accurate prediction of CO2 plume distribution and migration in formations has become essential. Therefore, the partial differential equation (PDE) constraints of multiphase flow were embedded into the loss function of the model by using automatic differential technique, and deep neural network models were developed to predict CO2 plume distributions, with constraints imposed by multiphase flow mechanics, ensuring that the model’s prediction results not only conform to the distribution law of training data samples but also strictly abide by the physical law of fluid seepage described by the PDE. To validate the model’s effectiveness, two PINN models were constructed using a multi-layer perceptron (MLP) and a long short-term memory (LSTM) network. These were applied in a practical case study on CO2 storage within a depleted oil reservoir. The results show that compared with pure data-driven models, the PINNs-based models demonstrate superior prediction accuracy. The findings of this research provide technical support for the design and implementation of CO2 geological storage projects while offering a theoretical foundation for the practical application of this technology.
-
-
![]()
图 2 不同地质模型孔隙度空间分布的对比
Figure 2. Comparison of spatial distribution of porosity in different geological models
![]()
图 3 不同地质模型渗透率和孔隙度的分布频率
Figure 3. Permeability and porosity distribution frequency in different geological models
![]()
图 5 纯数据驱动模型与PINNs模型的CO2羽流分布预测结果
Figure 5. CO2 plume distributions predicted by pure data-driven model and PINNs model
![]()
图 6 不同时刻PINN-LSTM模型的CO2羽流分布预测结果
Figure 6. CO2 plume distributions predicted by PINN-LSTM model at various times
![]()
图 7 纯数据驱动模型与PINNs模型的归一化压力场预测结果
Figure 7. Normalized pressure fields predicted by pure data-driven model and PINNs model
-
[1] 周守为,朱军龙. 助力“碳达峰、碳中和”战略的路径探索[J]. 天然气工业,2021,41(12):1–8. doi: 10.3787/j.issn.1000-0976.2021.12.001 ZHOU Shouwei, ZHU Junlong. Exploration of ways to helping “Carbon Peak and Neutrality” strategy[J]. Natural Gas Industry, 2021, 41(12): 1–8. doi: 10.3787/j.issn.1000-0976.2021.12.001
[2] 霍宏博,刘东东,陶林,等. 基于CO2提高采收率的海上CCUS完整性挑战与对策[J]. 石油钻探技术,2023,51(2):74–80. HUO Hongbo, LIU Dongdong, TAO Lin, et al. Integrity challenges and countermeasures of the offshore CCUS based on CO2-EOR[J]. Petroleum Drilling Techniques, 2023, 51(2): 74–80.
[3] 杨术刚,李兴春,蔡明玉,等. 国外CO2地质封存管理制度、标准体系分析及其启示[J]. 天然气工业,2023,43(12):130–137. YANG Shugang, LI Xingchun, CAI Mingyu, et al. Overseas management systems and standards for CO2 geological storage and their implications for China[J]. Natural Gas Industry, 2023, 43(12): 130–137.
[4] 柏明星,张志超,白华明,等. 二氧化碳地质封存系统泄漏风险研究进展[J]. 特种油气藏,2022,29(4):1–11. BAI Mingxing, ZHANG Zhichao, BAI Huaming, et al. Progress in leakage risk study of CO2 geosequestration system[J]. Special Oil & Gas Reservoirs, 2022, 29(4): 1–11.
[5] 李凤霞,王海波,周彤,等. 页岩油储层裂缝对CO2吞吐效果的影响及孔隙动用特征[J]. 石油钻探技术,2022,50(2):38–44. LI Fengxia, WANG Haibo, ZHOU Tong, et al. The influence of fractures in shale oil reservoirs on CO2 huff and puff and its pore production characteristics[J]. Petroleum Drilling Techniques, 2022, 50(2): 38–44.
[6] 李阳,王敏生,薛兆杰,等. 绿色低碳油气开发工程技术的发展思考[J]. 石油钻探技术,2023,51(4):11–19. LI Yang, WANG Minsheng, XUE Zhaojie, et al. Thoughts on green and low-carbon oil and gas development engineering technologies [J]. Petroleum Drilling Techniques, 2023, 51(4): 11–19.
[7] 张涛,杨若凡,常文杰,等. CO2伴生气混合过程的数值模拟研究[J]. 西南石油大学学报(自然科学版),2023,45(3):143–153. ZHANG Tao, YANG Ruofan, CHANG Wenjie, et al. Numerical simulation of CO2 associated gas mixing process[J]. Journal of Southwest Petroleum University(Science & Technology Edition), 2023, 45(3): 143–153.
[8] 赵鹏,朱海燕,张丰收. CO2增强页岩气开采及地质埋存的三维数值模拟[J]. 天然气工业,2024,44(4):104–114. ZHAO Peng, ZHU Haiyan, ZHANG Fengshou. Three-dimensional numerical simulation of CO2 injection to enhance shale gas recovery and geological storage[J]. Natural Gas Industry, 2024, 44(4): 104–114.
[9] LI Dong, PENG Suping, GUO Yinling, et al. CO2 storage monitoring based on time-lapse seismic data via deep learning[J]. International Journal of Greenhouse Gas Control, 2021, 108: 103336. doi: 10.1016/j.ijggc.2021.103336
[10] SINHA S, DE LIMA R P, LIN Youzuo, et al. Normal or abnormal? Machine learning for the leakage detection in carbon sequestration projects using pressure field data[J]. International Journal of Greenhouse Gas Control, 2020, 103: 103189. doi: 10.1016/j.ijggc.2020.103189
[11] ZHONG Zhi, SUN A Y, YANG Qian, et al. A deep learning approach to anomaly detection in geological carbon sequestration sites using pressure measurements[J]. Journal of Hydrology, 2019, 573: 885–894. doi: 10.1016/j.jhydrol.2019.04.015
[12] KARNIADAKIS G E, KEVREKIDIS I G, LU Lu, et al. Physics-informed machine learning[J]. Nature Reviews Physics, 2021, 3(6): 422–440. doi: 10.1038/s42254-021-00314-5
[13] RAISSI M, PERDIKARIS P, KARNIADAKIS G E. Physics-informed neural networks: a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations[J]. Journal of Computational Physics, 2019, 378: 686–707. doi: 10.1016/j.jcp.2018.10.045
[14] 赵暾,周宇,程艳青,等. 基于内嵌物理机理神经网络的热传导方程的正问题及逆问题求解[J]. 空气动力学学报,2021,39(5):19–26. ZHAO Tun, ZHOU Yu, CHENG Yanqing, et al. Solving forward and inverse problems of the heat conduction equation using physics-informed neural networks[J]. Acta Aerodynamica Sinica, 2021, 39(5): 19–26.
[15] 李野,陈松灿. 基于物理信息的神经网络:最新进展与展望[J]. 计算机科学,2022,49(4):254–262. LI Ye, CHEN Songcan. Physics-informed neural networks: recent advances and prospects[J]. Computer Science, 2022, 49(4): 254–262.
[16] JAGTAP A D, KHARAZMI E, KARNIADAKIS G E. Conservative physics-informed neural networks on discrete domains for conservation laws: applications to forward and inverse problems[J]. Computer Methods in Applied Mechanics and Engineering, 2020, 365: 113028. doi: 10.1016/j.cma.2020.113028
[17] KHARAZMI E, ZHANG Zhongqiang, KARNIADAKIS G E M. hp-VPINNs: variational physics-informed neural networks with domain decomposition[J]. Computer Methods in Applied Mechanics and Engineering, 2021, 374: 113547. doi: 10.1016/j.cma.2020.113547
[18] YANG Liu, MENG Xuhui, KARNIADAKIS G E. B-PINNs: Bayesian physics-informed neural networks for forward and inverse PDE problems with noisy data[J]. Journal of Computational Physics, 2021, 425: 109913. doi: 10.1016/j.jcp.2020.109913
[19] LU Lu, MENG Xuhui, MAO Zhiping, et al. DeepXDE: a deep learning library for solving differential equations[J]. SIAM Review, 2021, 63(1): 208–228. doi: 10.1137/19M1274067
[20] RACKAUCKAS C, NIE Qing. DifferentialEquations. jl: a performant and feature-rich ecosystem for solving differential equations in Julia[J]. Journal of Open Research Software, 2017, 5: 15. doi: 10.5334/jors.151
[21] 薛亮,戴城,韩江峡,等. 油藏渗流物理和数据联合驱动的深度神经网络模型[J]. 油气地质与采收率,2022,29(1):145–151. XUE Liang, DAI Cheng, HAN Jiangxia, et al. Deep neural network model driven jointly by reservoir seepage physics and data[J]. Petroleum Geology and Recovery Efficiency, 2022, 29(1): 145–151.
[22] SHOKOUHI P, KUMAR V, PRATHIPATI S, et al. Physics-informed deep learning for prediction of CO2 storage site response[J]. Journal of Contaminant Hydrology, 2021, 241: 103835. doi: 10.1016/j.jconhyd.2021.103835
[23] EBIGBO A, CLASS H, HELMIG R. CO2 leakage through an abandoned well: problem-oriented benchmarks[J]. Computational Geosciences, 2007, 11(2): 103–115. doi: 10.1007/s10596-006-9033-7
[24] LALLAHEM S, MANIA J, HANI A, et al. On the use of neural networks to evaluate groundwater levels in fractured media[J]. Journal of Hydrology, 2005, 307: 92–111.
[25] HOCHREITER S, SCHMIDHUBER J. Long short-term memory[J]. Neural Computation, 1997, 9(8): 1735–1780. doi: 10.1162/neco.1997.9.8.1735
-
期刊类型引用(37)
1. 申军武. 井下扭矩保持器设计及实验研究. 中国石油和化工标准与质量. 2025(01): 144-146 . 
百度学术
2. 刘永升，豆子钧，张金成，高德利. 基于地质-工程一体化的纠偏轨道优化设计及产能评价. 中国石油大学学报(自然科学版). 2025(02): 142-150 . 百度学术
3. 许佳鑫，宋明阶，赵红燕，侯亮，李胜楠. 涪陵页岩气田加密井防碰关键技术. 江汉石油职工大学学报. 2024(02): 38-41 . 百度学术
4. 尹虎，范涛，江星宏. 页岩气丛式水平井上部井段防碰关键参数设计. 科学技术与工程. 2024(22): 9342-9349 . 百度学术
5. 史配铭，贺会锋，朱明明，孟凡金，屈艳平，王玉鹏. 苏里格南部气田Φ152.4 mm小井眼大斜度井快速钻井关键技术. 石油工业技术监督. 2024(09): 51-56 . 百度学术
6. 王胜建，迟焕鹏，庞飞，王都乐，周志，李龙，姜鹍鹏. 黔北正安地区页岩气钻探工程难点与对策研究. 地质与勘探. 2023(01): 162-169 . 百度学术
7. 张淑侠，王振华，郭锦涛，秦芳玲，何焕杰，吕宁超. 除硼树脂在压裂返排液回用中的应用. 水处理技术. 2023(06): 123-127 . 百度学术
8. 邱艳华，吴宇，温庆，林宇，杨建英，何焱，汪刚，文崭，罗彦力. 页岩气平台中压工艺流程的现场应用与评价. 天然气与石油. 2023(03): 27-35 . 百度学术
9. 凡广荣. 页岩气水平井强化钻压提速技术研究. 中国石油和化工标准与质量. 2023(15): 193-195 . 百度学术
10. 黄晶，赵昆，夏绪波，毛庆春. 压裂工程“井工厂”压裂模式提速提效分析. 江汉石油职工大学学报. 2023(04): 15-17 . 百度学术
11. 李阳，赵清民，薛兆杰. 新一代油气开发技术体系构建与创新实践. 中国石油大学学报(自然科学版). 2023(05): 45-54 . 百度学术
12. 姚红生，房启龙，袁明进，张壮. 渝东南常压页岩气工程工艺技术进展及下一步攻关方向. 石油实验地质. 2023(06): 1132-1142 . 百度学术
13. 刘召友，孙永强，郭百利. 苏里格气田?165.1mm小井眼二开一趟钻优快钻井关键技术. 西部探矿工程. 2023(12): 26-30 . 百度学术
14. 张金成. 第一性原理思维法在页岩气革命中的实践与启示. 钻探工程. 2022(02): 1-8 . 百度学术
15. 张东清. 涡轮式水力振荡器在涪陵页岩气水平井中的应用. 科技和产业. 2022(05): 283-287 . 百度学术
16. 石芳，熊青山，李微，王柯，刘恒. 涪陵页岩气田井场规划技术研究. 能源与环保. 2022(06): 104-113 . 百度学术
17. 刘伟，朱礼平，潘登雷，周楚坤，梁霄，刘小斌. WR气田深层页岩气钻井提速提效实践与认识. 天然气技术与经济. 2022(03): 44-50 . 百度学术
18. 郑德帅. 可旋转钻柱定向钻进工具设计及测试. 石油钻探技术. 2021(06): 81-85 . 本站查看
19. 陈亚联. 新型压裂技术应用分析. 化工技术与开发. 2020(01): 38-41 . 百度学术
20. 周亚光. 青海省都兰县八宝山页岩气3井钻探施工技术研究. 能源与环保. 2020(02): 44-46+50 . 百度学术
21. 张辉. 大牛地气田丛式小井眼集约化钻井技术. 天然气技术与经济. 2020(02): 28-33 . 百度学术
22. 周昊. 全自动液体添加剂橇装置的研制. 化学工程与装备. 2020(10): 169+89 . 百度学术
23. 葛明娜，庞飞，包书景. 贵州遵义五峰组—龙马溪组页岩微观孔隙特征及其对含气性控制——以安页1井为例. 石油实验地质. 2019(01): 23-30 . 百度学术
24. 仝少凯，高德利. 水力压裂基础研究进展及发展建议. 石油钻采工艺. 2019(01): 101-115 . 百度学术
25. 樊好福. 页岩气钻完井配套技术集成研究与应用. 探矿工程(岩土钻掘工程). 2019(08): 15-22 . 百度学术
26. 董成林，涂玉林，殷子横，张金成，周号博. 涪陵页岩气田钻井提速集成技术应用研究. 西部探矿工程. 2019(12): 47-50 . 百度学术
27. 孔华，兰凯，刘香峰，刘明国，晁文学，郗刘明. 基于振动实测的非均质地层钻头失效分析与对策. 天然气工业. 2019(12): 110-115 . 百度学术
28. 刘伟，何龙，胡大梁，李文生，焦少卿. 川南海相深层页岩气钻井关键技术. 石油钻探技术. 2019(06): 9-14 . 本站查看
29. 肖佳林，李奎东，高东伟，包汉勇. 涪陵焦石坝区块水平井组拉链压裂实践与认识. 中国石油勘探. 2018(02): 51-58 . 百度学术
30. 王光磊，张金成，赵明琨. 涪陵页岩气田井筒完整性实践与认识. 石油机械. 2018(05): 30-34+59 . 百度学术
31. 臧艳彬. 川东南地区深层页岩气钻井关键技术. 石油钻探技术. 2018(03): 7-12 . 本站查看
32. 路保平，丁士东. 中国石化页岩气工程技术新进展与发展展望. 石油钻探技术. 2018(01): 1-9 . 本站查看
33. 刘小伟. 在涪陵页岩气开发中自动化钻机的现状和发展——“十三五”国家科技重大专项深层页岩气开发关键装备及工具研制. 化工管理. 2017(27): 70 . 百度学术
34. 梅绪东，金吉中，王朝强，何勇，王丹，张春. 涪陵页岩气田绿色开发的实践与探索. 西南石油大学学报(社会科学版). 2017(06): 9-14 . 百度学术
35. 曹明. 页岩气压裂试气工程技术进展. 中国矿业. 2017(S2): 359-362 . 百度学术
36. 李彬，付建红，秦富兵，唐一元. 威远区块页岩气“井工厂”钻井技术. 石油钻探技术. 2017(05): 13-18 . 本站查看
37. 臧艳彬，张金成，赵明琨，宋争，罗锐. 涪陵页岩气田“井工厂”技术经济性评价. 石油钻探技术. 2016(06): 30-35 . 本站查看
其他类型引用(16)
下载:
计量
- 文章访问数: 271
- HTML全文浏览量: 74
- PDF下载量: 84
- 被引次数: 53
