用于迷宫屏蔽计算的蒙特卡罗减方差方法研究

Study on Monte Carlo Variance-reduction Method Used for Maze Shielding Calculation

摘要: 本文应用多种减方差技巧提出了强迫指向自动重要抽样（FPAIS）方法，并在MCNP5程序平台实现了该方法。采用该方法对1个多折迷宫算例进行了模拟计算，计算结果与MCNP5程序的直接模拟、DXTRAN球、点通量3种方法的结果进行了比对。基于此算例对FPAIS方法进行了引导面设置和粒子数敏感性分析。结果表明，FPAIS方法在保证一定计算精度的前提下，比其他3种方法的FOM提高2~3个量级，且该方法对引导面设置不敏感、可用性强，对于迷宫屏蔽计算是一种准确、高效的解决方案。

Abstract: The forced pointing auto-importance sampling (FPAIS) method was presented using multiple variance-reduction methods in this paper, and was realized in MCNP5 code. This method was applied to the calculation of a multiple-bend maze example, and the calculation result was compared with natural Monte Carlo calculation, DXTRAN sphere and point flux methods using MCNP5 code. Besides, the sensitivity for setting the guiding surface and particle number of FPAIS method was analyzed with this example. The results show that the FOM of FPAIS significantly increases with 2.3 orders of magnitude compared with the other three methods in the premise of ensuring high calculation accuracy. FPAIS method is insensitive to the setting of the guiding surface and easy to use, and it is an accurate and efficient solution for the calculation of maze shielding.