基于蒙特卡罗正算输运的全局减方差方法

Global variance reduction based on forward Monte Carlo calculation

蒙特卡罗方法是当前形势下辐射屏蔽计算的首选分析工具。小概率深穿透问题则是屏蔽计算的关键与亟待解决的核心问题,需要使用有效的减方差技巧。针对全局问题,利用蒙特卡罗正算输运得到的粒子通量或探测响应来构建权重窗参数,将现有的粒子位置偏移拓展到位置和能量偏倚。利用国际屏蔽基准题进行测试验证,通过使用该方法,粒子被引导到模型的所有位置。平均相对误差降低到10%以下,几乎所有网格区域都有粒子统计。结果表明,基于蒙特卡罗正算输运的输运偏倚参数构建方法能够实现全局减方差。

Monte Carlo method is extensively employed in radiation shielding calculation with the advantages of high fidelity geometry modeling,complex radiation source description and continuous-energy cross sections.However,the simulation is impractical for Monte Carlo particles that have little chance of being transported to the far-source area and the statistical error may be unacceptable.Accordingly,effective global variance reduction method（GVR）is significant for Monte Carlo deep-penetration radiation shielding calculation.This work constructs weight window parameters based on forward Monte Carlo calculation to bias the particles position and energy during transport.According to the validation of shielding benchmark,particles have been guided to far-source area by using the weight window.The relative error has been decreased to less than 10%,and almost all the mesh cells have been tallied.The results show that the transport parameters obtained from forward Monte Carlo calculation could realize global variance reduction.