论文部分内容阅读
蒙特卡罗粒子输运方法在求解中子深穿透问题时需要多种减方差技巧提高计算效率,减方差参数需要用户凭借经验反复试算,获取可靠有效的减方差参数比较繁琐。离散纵标法相比蒙卡方法具有效率高、可自动获得全局解的优势,但计算精度受多群截面、离散误差等多方面因素影响。蒙卡-确定论耦合计算可结合彼此优点,利用S_N共轭函数估计粒子对探测器的相对贡献,自动生成源偏倚与权重窗参数用于蒙卡减方差,提高蒙卡计算效率。基于S_N输运计算平台ARES编制了三维共轭输运计算模块,根据一致性共轭驱动重要性抽样方法自动生成减方差参数,用于加速MCNP5计算。数值结果表明,自动生成的减方差参数可有效提高蒙卡计算效率,并保证结果无偏。自动减方差技术利用S-N共轭函数可更经济准确的估计粒子重要性,避免手动估算减方差参数的复杂工作,对于复杂屏蔽问题的蒙卡计算具有较好的应用前景。
The Monte Carlo particle transport method needs many kinds of subtraction techniques to improve the computational efficiency in solving neutron deep penetration problems. The subtractive variance parameter requires the user to try it out repeatedly based on experience. It is cumbersome to obtain reliable and effective subtraction variance parameters. Compared with discrete ordinate method, the Monte Carlo method has the advantages of high efficiency and the advantages of global solution can be obtained automatically. However, the accuracy of calculation is affected by many factors such as multistage cross section and discrete error. The Monte Carlo-deterministic coupling calculation can combine the advantages of each other, estimate the relative contribution of particles to the detector by using the S_N conjugate function, automatically generate the source bias and the weight window parameters for reducing the variance of Moncada, and improve the computational efficiency of Moncada. Based on the S_N transport computing platform ARES, a three-dimensional conjugate transport computation module was developed to generate the subtraction variance parameters automatically based on the consensus conjugate-driven importance sampling method to accelerate MCNP5 computation. The numerical results show that the automatically generated subtraction variance parameters can effectively improve the efficiency of Monte Carlo calculation and guarantee the unbiased result. Auto-subtraction technique uses the S-N conjugate function to estimate the particle’s importance more economically and avoids the complicated work of estimating the variance parameter manually. It has a good application prospect for the Monte Carlo calculation of complex shielding problems.