基于统计抽样的敏感性系数计算方法

分析核系统的不确定性和敏感性,对于减小核设计的设计余量、提高核系统的经济性具有重要意义。基于统计抽样的不确定度分析方法,由于算法简单、可考虑高阶效应且对响应量没有特殊要求等,越来越受到重视。但之前认为基于统计抽样法很难进行敏感性系数分析,其原因主要是响应量的变化是由多变量同时变化引起,很难把单独一个变量的变化导致的响应量的变化确定出来。本文首先推导了利用统计抽样法进行敏感性系数分析的理论公式,然后利用裸堆双群近似的临界公式和复杂的压水堆单栅元问题进行了验证,验证了统计抽样法的可行性。针对实际问题协方差矩阵求逆困难的问题,本文提出了两种替代解决方法,即采用简化协方差矩阵或统一微扰量的方法,利用²³⁵U裂变截面对上述方法进行了验证分析,证明了方法的可行性和正确性;同时分析了不同敏感性系数对不确定度计算的影响。     

快堆用蒙特卡罗多群数据库制作方法研究

Based on the discrete angle method, a Monte Carlo multi-group cross section generation program MGXSMC was developed. This program can read the cross section data from an input file or read the cross section from a library in a specified format to generate the multi-group cross section for MCNP or RMC. The corresponding index file list can be automatically generated. The two-dimensional two-group IAEA pressurized water reactor (PWR) benchmark and lead-based fast reactor (RBEC-M) benchmark were used to verify the cross section library generated by the MGXSMC program. The calculation results show that the difference between the calculated result of the P5 order approximate multigroup section and the continuous point cross section is 24 pcm (1pcm = 10-5), and the difference of the keff result calculated by the P0 order approximate multigroup section and the continuous point section is large. This shows that the method and the program developed for the Monte Carlo Group Section Library are correct. At the same time, the neutron anisotropic scattering has a large impact on the calculation results of the lead-based fast reactor. Therefore, when the Monte Carlo Group Section library is produced, the neutron scattering angle data should be added.     

快堆用蒙特卡罗多群数据库制作方法研究

基于离散角方法,开发了蒙特卡罗多群数据库生成程序MGXSMC,该程序可以实现从输入文件读取截面数据或者从指定格式的截面库中读取截面,产生可供蒙特卡罗程序MCNP或RMC计算的数据库,并且可自动生成相应的索引文件列表。采用二维两群不带反射层的国际原子能机构(IAEA)压水堆(PWR)基准题和铅基快堆(RBEC-M)基准题对MGXSMC程序加工产生的核数据进行验证,计算结果表明,采用P₅阶近似多群截面与连续点截面计算的有效增殖系数(k_eff)结果相差24 pcm(1pcm=10^-5),而采用P₀阶近似多群截面与连续点截面计算的k_eff结果相差较大。由此说明蒙特卡罗多群数据库的制作方法和所开发的程序是正确的,同时,中子各向异性散射对铅基快堆计算结果影响较大,故制作蒙特卡罗多群数据库时应加入中子散射角数据。