期刊文献+

精细油藏模拟的一种线性求解算法

A NEW LINEAR SOLVER FOR FINE-SCALE RESERVOIR SIMULATION
分享 导出
摘要 本文针对油藏数值模拟中黑油模型方程的各个物理量的性质,利用ABF解耦方法和子空间校正算法提出一种分裂型预条件子,并与Krylov子空间方法结合,设计了一种线性求解算法.我们基于某实际油田区块构建了粗、细两个油藏模型,并将它们模拟计算得到的油产量与油田实际产量进行对比,结果表明精细油藏数值模拟对油田生产实践具有重要指导意义,开展面向精细油藏模拟的大规模数值算法研究是十分必要的.我们在台式工作站上使用所设计的线性求解算法测试了SPE10标准算例及由其拼接而成的千万网格规模算例,计算结果表明该算法能有效求解大规模油藏模拟问题. According to physical variables of black-oil model in reservoir simulation own different characterize, we combine the ABF decoupling method and subspace correction method to design a splitting preconditioner to accelerate the Krylov method. We firstly build two coarse and fine models based on some real reservoir block, and compare predicated daily oil productions of these two models with the observed data, the comparison demonstrates the significance of fine-scale reservoir simulation, which indicates the need to develop efficient linear solver for fine-large reservoir simulation. We then employ the proposed linear solver to solve the SPE10 benchmark and a model with ten millions cells spliced by the SPE10 benchmark on a desktop computer, and numerical results indicate that the proposed linear solver is very efficient.
作者 李政 吴淑红 李巧云 张晨松 王宝华 许进超 赵颖 Li Zheng,Wu Shuhong,Li Qiaoyun,Zhang Chensong,Wang Baohua,Xu Jinchao,Zhao Ying(Kunming University of Science and Technology, Yunnan 650504, China; ( PetroChina Research Institute of Petroleum Exploration and Development, Beijing 100083, China;LSEC & NCMIS, Academy of Mathematics and Systems Science, Beijing 100085, China;Department of Mathematics, Penn State University, University Park, USA;Dagang Oilfield, PetroChina, Tianjin 300000, China)
出处 《数值计算与计算机应用》 2018年第1期1-9,共9页 Journal of Numerical Methods and Computer Applications
基金 中国科学院前沿科学研究重点计划 中国石油天然气股份有限公司“新一代油藏数值模拟软件3.0版研制”课题(2014A-1008) “大规模、高效代数方程解法模块研究”专题(2011A-1010-01)
关键词 精细油藏数值模拟 多层网格法 KRYLOV子空间方法 多阶段预条件技术 Fine-scale reservoir simulation multigrid Krylov subspace method multi-stage preconditioning
  • 相关文献

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部 意见反馈