赵宁^1,2,,
王绪本³,
余刚²,
何展翔²,
孙卫斌²,
秦策^1,,
1. 河南理工大学物理与电子信息学院, 河南焦作 454000
2. 中国石油集团东方地球物理公司综合物化探处, 河北涿州 072750
3. 成都理工大学"地球勘探与信息技术"教育部重点实验室, 成都 610059

基金项目: 国家自然科学基金项目（41604112，U1704128）资助


详细信息

作者简介: 赵宁, 男, 1982年生, 副教授, 主要从事电磁法数值模拟与反演等方面的研究工作.E-mail:zhaoning@hpu.edu.cn

通讯作者: 秦策, 男, 1988年生, 讲师, 主要从事电磁法正反演研究工作.E-mail:ce.qin@hpu.edu.cn

中图分类号: P319

收稿日期:2018-03-19
修回日期:2018-07-17
上线日期:2019-02-15

3D MCSEM parallel goal-oriented adaptive vector finite element modeling

ZHAO Ning^1,2,,
WANG XuBen³,
YU Gang²,
HE ZhanXiang²,
SUN WeiBin²,
QIN Ce^1,,
1. Department of Physics and Electronic Information, Henan Polytechnic University, Jiaozuo Henan 454000, China
2. Bureau of Geophysical Prospecting, China National Petroleum Corporation, Zhuozhou Hebei 072750, China
3. Key Laboratory of Earth Exploitation and Information Techniques, Ministry of Education, Chengdu University of Technology, Chengdu 610059, China


More Information

Corresponding author: QIN Ce,E-mail:ce.qin@hpu.edu.cn

MSC: P319

--> Received Date: 19 March 2018
Revised Date: 17 July 2018
Available Online: 15 February 2019


摘要
摘要:本文实现了一种面向目标自适应海洋可控源电磁三维矢量有限元方法.为满足三维复杂电性结构模拟的需求，网格剖分采用非结构化六面体.在组装刚度矩阵之后，形成的大型复数线性方程组分解为等价的实数形式，利用带预条件的广义最小残差法进行求解.在获得微分方程的解之后，为提高解的准确性，通过面向目标的自适应误差估计来指示网格细化，重点加密能使观测点数值模拟精度提高的网格.对于大规模三维数据，为了使模型空间的并行计算达到均衡负载的效果，我们使用METIS函数库来进行网格计算任务量的划分.最后，通过对比一维解析解与三维自适应矢量有限元计算结果，验证了程序的正确性；通过自适应过程中误差指示子的分布，验证了面向目标自适应的有效性；通过对三维复杂模型进行均衡负载下的并行计算，测试了程序的可扩展性.
关键词: 总场/
海洋可控源电磁/
面向目标自适应/
矢量有限元/
并行计算

Abstract:We have developed a parallel goal-oriented adaptive finite element method for accurate and efficient electromagnetic modeling. To accommodate arbitrarily complex 3D conductivity variations, we use the total-field solution approaches of unstructured hexahedral grids. After assembling the local element matrices into a global system, one can obtain a sparse linear system of equations. We employed the FGMRES method with a very efficient block-diagonal pre-conditioner for the large conductivity contrasts system. Accuracy of the finite element solution could be achieved through adaptive mesh refinement that is performed iteratively until the solution converges to the desired accuracy tolerance. Refinement is guided by a goal-oriented error estimator that uses a dual-weighted residual method to optimize the mesh for improving observation numerical simulation accuracy. To further improve the computational efficiency, our algorithm is parallelized over mesh partitioning using METIS. In this paper, We validate the newly developed algorithm by comparison with controlled source EM solutions for a 1D layered model. Afterwards, we demonstrate that the estimated error for each element for a variable problem sizes for the effectiveness of the goal-oriented adaptive. In order to verify scalable numerical scheme, we have implemented the dynamic load balancing algorithms used for parallel computations.
Key words:Total field/
MCSEM/
Goal-oriented adaptive method/
Vector finite element/
Parallel computing



PDF全文下载地址:

http://www.geophy.cn/data/article/export-pdf?id=dqwlxb_14881