提高多群截面蒙特卡罗计算效率的方法研究

Study on Method of Improving Efficiency of Monte Carlo Calculation of Multi-group Cross Section

  • 摘要: 确定论方法求解中子输运方程时, 需要用精确的中子通量谱估算多群截面, 该通量谱可通过蒙特卡罗中子输运过程获知。已有的使用蒙特卡罗方法计算多群截面与散射矩阵的程序MCNPNG(蒙特卡罗多群截面统计分析软件), 在计算材料种类多、几何结构复杂的模型时时间开销非常大, 不利于工程应用。本文分析了其原因, 并针对性地开展了优化, 有效提升了计算效率; 开展了多群截面与散射矩阵计算的并行方法研究, 研制了相应的程序模块, 显著降低了运行时间, 且在调用数百CPU的情况下仍能保持85%以上的并行效率。通过对OKTAVIAN基准题的计算分析, 验证了多群截面与散射矩阵并行计算方法的正确性和高效性。

     

    Abstract: When solving neutron transport equation by deterministic method, it is necessary to estimate multi-group cross sections with accurate neutron flux spectrum. This flux spectrum can be obtained by MC neutron transport process. The existing program MCNPNG (Monte Carlo Multi-group Cross-section Statistical Analysis Software) for calculating multi-group cross-sections and scattering matrices by using the Monte Carlo method is very time-consuming when calculating models with various materials and complex geometric structures, which is not conducive to engineering application. In this paper, the reasons are analyzed firstly, and the optimization is carried out, which effectively improves the calculation efficiency; Then, the parallel method of multi-group cross section and scattering matrix calculation is studied, and the corresponding program module is developed, which significantly reduces the running time and can still maintain more than 85% parallel efficiency when hundreds of CPU are called. Finally, through the calculation and analysis of OKTAVIAN benchmark problem, the correctness and efficiency of the parallel calculation method of multi-group cross section and scattering matrix are verified.

     

/

返回文章
返回