期刊文献+

千万自由度量级有限元模态分析并行计算研究 被引量:7

Parallel computing for finite element modal analysis with over ten-million-DOF
下载PDF
导出
摘要 大规模有限元模态分析在一些重大装备研制过程中有迫切需求,对于实现装置系统级分析具有重要意义。基于隐式重启动Arnoldi、Krylov-Schur和Jacobi-Davidson三种主流算法和PANDA并行计算框架,构建了大规模模态分析并行计算体系;将并行求解体系应用于某光机主体结构,实现了其上千万自由度、数千核的模态分析并行计算;结合算例对三种主流算法的适应性和并行可扩展性进行了评估。研究结果表明,基于三种算法构建的并行求解体系均可在1小时内求解千万自由度量级的大规模模态分析问题,并行可扩展性非常优异。 In the development of large equipments,the demand of large-scale finite element modal analysis is very strong due to its significance in realizing the systemic analysis of the entire large structure.Based on three predominant algorithms,i.e.,implicitly restarted Arnoldi method,Krylov-Schur method and Jacobi-Davidson method and PANDA framework,a parallel computing system for a large-scale modal analysis was established.As a typical application,it was used to get solutions to the main structure of a certain optical machinery,a parallel modal analysis with over ten-million-DOF was performed and thousands of CPU processors were applied.The adaptability and parallel scalability of the three algorithms with numerical examples.Results showed that the parallel solving system can perform the modal analysis with over ten million-DOF within one hour and its parallel scalability is excellent.
出处 《振动与冲击》 EI CSCD 北大核心 2015年第17期77-82,共6页 Journal of Vibration and Shock
基金 国家自然科学基金面上项目(11472256) 中国工程物理研究院发展基金项目(2012A0202008) 中国工程物理研究院重点学科项目"计算固体力学" 中国工程物理研究院院长基金项目
关键词 有限元法 大规模模态分析 并行计算 finite element method large-scale modal analysis parallel computation
  • 相关文献

参考文献16

  • 1Saad Y. Numerical methods for large eigenvalue problems [ M ]. Halsted Press, Div. of John Wiley &Sons, Inc. , New York : 1992.
  • 2Bai Z, Demmel J, Dongarra J, et al. Templates for the solu- tion of algebraic eigenvalue problems: A practical guide [ M ]. SIAM, Philadelphia, 2000.
  • 3Watkins D S. The matrix eigenvalue problem: GR and krylov subspace methods[ M ]. SIAM, Philadelphia, 2007.
  • 4Arnoldi W E. The principle of minimized iterations in the so- lution of the matrix eigenvalue problem [ J ]. Qaart. Appl. Math. , 1951(9) :17 -29.
  • 5Sorensen D C. Implicit application of polynomial filters in a k- step Arnoldi method[J]. SIAM J. Matrix Anal. Appl., 1992 (13) :357 -385.
  • 6Lehoucq R B, Sorensen D C, Yang C. ARPACK Users' guide : solution of large-scale eigenvalue problems by implicit- ly restarted arnoldi methods [ M ]. SIAM, Philadelphia, PA, 1998.
  • 7Stewart G W. A Krylov-Schur algorithm for large eigenprob- lems[J]. SIAM J. Matrix Anal. Appl., 2001,23(3): 601 -614.
  • 8Meerbergen K, Spence A, Roose D. Shift-invert and cayley transforms for detection of rightmost eigenvalues of nonsym- metric matrices [ J J. BIT Numerical Mathematics, 1994, 3 (34) : 409 - 423.
  • 9Davidson E R. The iterative calculation of a few of the lowest eigenvalues and corresponding eigenvectors of large real sym- metric matrices[J]. J. Comp. Phy. , 1975(17): 87 -94.
  • 10Crouzeix M, Philippe B, Sadkane M. The davidson method [J]. SIAMJ. Sci. Comput., 1994(15):62-67.

同被引文献39

引证文献7

二级引证文献16

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