EI、Scopus 收录
中文核心期刊
王福军 王利萍 程建钢 姚振汉. 并行有限元计算中的接触算法[J]. 力学学报, 2007, 23(3): 422-427. DOI: 10.6052/0459-1879-2007-3-2006-407
引用本文: 王福军 王利萍 程建钢 姚振汉. 并行有限元计算中的接触算法[J]. 力学学报, 2007, 23(3): 422-427. DOI: 10.6052/0459-1879-2007-3-2006-407
Fujun Wang, Liping Wang, Jiangang Cheng, Zhenhan Yao. A contact algorithm for parallel computation of FEM[J]. Chinese Journal of Theoretical and Applied Mechanics, 2007, 23(3): 422-427. DOI: 10.6052/0459-1879-2007-3-2006-407
Citation: Fujun Wang, Liping Wang, Jiangang Cheng, Zhenhan Yao. A contact algorithm for parallel computation of FEM[J]. Chinese Journal of Theoretical and Applied Mechanics, 2007, 23(3): 422-427. DOI: 10.6052/0459-1879-2007-3-2006-407

并行有限元计算中的接触算法

A contact algorithm for parallel computation of FEM

  • 摘要: 针对动态接触问题的有限元并行计算,提出了一种新的接触算法. 新算法引入局部拉氏乘子技术来计算接触力. 由于同时考虑了无穿透的接触约束条件和相邻接触对的相互影响,较之广泛使用的罚参数法,新算法使接触约束条件和系统平衡方程得到更充分的满足. 虽然为提高接触计算精度而在局部采用了迭代技术,但算法仍然具有较高的效率,且与显式时间积分方案完全相容. 此外,通过构造专门的区域分解方案,实现了将现有为串行程序开发的搜索算法平滑移植到并行环境的目标. 数值算例表明,所提出的接触算法具有很好的并行性,在保证了接触问题并行计算精度的同时,取得了满意的并行效率.

     

    Abstract: A general contact algorithm for parallel finite element simulation ofdynamic contact problem is presented in this paper. The local Lagrangemultiplier approach is introduced into the new contact algorithm. Since theimpenetrability condition and the interaction of adjacent contact pair areall considered in this algorithm, the contact constraint and equilibriumequations of the system are satisfied sufficiently as compared withtraditional penalty method. Although some local iterations have to be done,the algorithm has high computational efficiency, and is totally consistentwith explicit time integration. In addition, the existing contact searchalgorithms developed for serial finite element program could be transferedinto the parallel finite element program without any modification bywonderful design of a special domain decomposition scheme. Numericalexamples show that the new contact algorithm is suitable for parallel finiteelement computation. The new algorithm ensures not only the good simulationaccuracy, but also the high parallel-execution efficiency.

     

/

返回文章
返回