航空发动机包容性数值仿真并行计算效率研究  

Research on Parallel Computing Efficiency of Aero Engine Containment Analysis

作　　者：史同承 柴象海 张晓云[2] 

机构地区：[1]中航商用航空发动机有限责任公司,上海200241 [2]上海交通大学,上海200240

出　　处：《计算机仿真》2013年第1期86-90,共5页Computer Simulation

摘　　要：研究如何提高航空发动机包容性数值仿真并行计算效率的问题。由于仿真需要庞大的网格数量、高度的非线性和复杂的接触算法,并行计算效率一直比较低,已经成为制约工程应用的重要因素。为了提高航空发动机包容性数值仿真并行计算效率,提出了在共享内存并行模式(Share-Memory Parallel,SMP)下,采用自接触算法进行显式动力学分析,提高并行计算效率的方法。实际算例的比较表明,相比传统的面-面接触算法,采用自接触算法可以有效提高航空发动机包容性数值仿真并行计算效率。The problems of improving the parallel computing efficiency of the aero engine containment analysis were studied. Due to the tremendous number of grids, highly non-linear equations and complex contact algorithm, the low parallel computing efficiency has been one of the major restrictions of the industrial application of the aero engine containment simulation. In order to improve the parallel computing efficiency, the dynamic explicit analysis with single-surface contact algorithm was carried out, under the shared-memory parallel （SMP） computing mode. The practices show that the computing efficiency of the analysis with single-surface contact algorithm is much higher than the analysis with surface-to-surface contact algorithm.

关 键 词：航空发动机 包容性 数值仿真 并行计算效率 

分 类 号：V235.1[航空宇航科学与技术—航空宇航推进理论与工程]