Study on High-performance Whole Core Pin-by-pin Calculation Based on Massage Passing Interface

The high-performance whole core Pin-by-pin calculation is realized by using the MPI-based spatial parallel algorithm, and the corresponding code EFEN has been developed. The code makes full use of the computing and storage capabilities of the parallel CPU through a reasonable design of the domain decomposition scheme to ensure loading balance and minimize the number of communications. The red and black Gauss-Seidel nodal scanning algorithm is applied to avoid the iterative format degradation caused by domain decomposition. With reference to the actual core layout of a commercial reactor, two PWR whole core pin-by-pin calculation problems are designed. The corresponding numerical results show that the accuracy of the calculation results of the code is within the acceptable range. The influence of communication cycle on calculation accuracy and parallel efficiency is very small. The sub-domain with a relatively small surface volume domain decomposition method has higher parallel efficiency. It takes about 900s to perform a PWR whole core Pin-by-pin calculation with 125 CPUs with a spatial meshes number of $289\times 289\times 218$ and an energy group number of 4, and the parallel efficiency is about 90%.