节点文献
汽车碰撞有限元仿真的并行计算及其性能研究
Finite Element Simulation and Performance Study of Vehicle Crash Based on Parallel Computing
【摘要】 并行计算是提高汽车碰撞有限元仿真的计算速度和求解能力的有效途径。文章简要介绍了汽车碰撞有限元仿真并行计算的区域分解方法,给出了处理接触时的并行计算流程,分析了并行求解过程以及主要时间开销。在此基础上,研究了两种硬件体系结构和不同处理器数目对计算效率的影响。最后研究了不同的区域分解策略对整车碰撞计算效率的影响。研究结果对最终用户选择合适的硬件平台和计算软件具有指导意义。
【Abstract】 Parallel computing is an efficient approach to speed up and enhance the solving ability of finite element simulation of automobile crash. In this paper, domain decomposition for parallel computing is introduced simply, and the algorithm flowchart including contact is drawn, and the parallel computing process and primary overhead are analyzed. Furthermore, effect of two hardware platforms and the number of processors on computing efficiency are studied. At last, effect of different domain decomposition strategy on automobile crash computing efficiency is presented. To end users, research result is a guidance to choose appropriate hardware platform and computing software.
【Key words】 parallel computing; crash; finite element simulation; domain decomposition;
- 【文献出处】 系统仿真学报 ,Acta Simulata Systematica Sinica , 编辑部邮箱 ,2004年11期
- 【分类号】U467.13
- 【被引频次】39
- 【下载频次】687