论文部分内容阅读
基于指数函数展开节块3阶简化球谐函数(EFEN-SP3)方法,通过采用基于标准消息传递界面(MPI)的空间并行算法实现高性能全堆芯Pin-by-pin计算,并开发了相应的程序EFEN。该程序通过合理设计区域划分方案以保证负载平衡并使通信次数最小化,充分发挥并行中央处理器(CPU)的计算和存储能力;通过选择红黑Gauss-Seidel节块扫描算法避免区域分解引起的迭代格式退化。参考实际商用堆的堆芯布置,设计2个压水堆(PWR)全堆芯Pin-by-pin算例,相应的数值结果表明:该程序计算结果的精度在可接受范围内;通信周期对计算精度和并行效率的影响都很小;子区域表面体积比较小的区域划分方式具有较高的并行效率;用125个CPU进行一次空间网格数为289×289×218、能群数为4的PWR全堆芯Pin-by-pin计算所需时间约为900 s,并行效率约为90%。
Based on the exponential function, the 3rd-order simplified spherical harmonic function (EFEN-SP3) method was developed to realize high-performance full-core Pin-by-pin calculation by using spatial parallel algorithm based on standard message passing interface (MPI) Program EFEN. The program can make full use of parallel computing and storage capacity of central processing unit (CPU) by reasonably designing regional partitioning schemes to ensure load balancing and minimize the number of communication; avoiding the regional decomposition by choosing the red Gauss-Seidel block scanning algorithm Iteration format degradation. Based on the core arrangement of practical commercial reactors, two PWR core-in-pin cores are designed. The corresponding numerical results show that the accuracy of the program results is within the acceptable range. The communication cycle pairs The computational precision and the parallel efficiency are very small. The subdivision area with small surface area has higher parallel efficiency. With 125 CPUs, the number of primary spatial grids is 289 × 289 × 218 and the number of energy groups is 4 PWR full core pin-by-pin calculation takes about 900 s, parallel efficiency of about 90%.