论文部分内容阅读
矩阵特征线方法(MOC)通过构造并求解线性方程组,代替传统MOC方法中的反复特征线扫描。幂迭代法求解keff的收敛速度严重依赖于占优比,实际的较大规模的堆芯占优比接近于1,收敛很慢。本研究结合多群耦合GMRES算法直接求解多群问题,采用Wielandt迭代加速矩阵MOC临界问题的求解。对多个基准题的数值结果表明,与幂迭代法相比,结合多群耦合GMRES的Wielandt迭代具有良好的计算精度和更高的计算效率。
The matrix characteristic line method (MOC) replaces the characteristic line scanning in the traditional MOC method by constructing and solving linear equations. The power iteration method to solve keff convergence rate depends heavily on the dominant ratio, the actual larger core occupancy ratio close to 1, convergence is slow. In this study, the multi-group coupled GMRES algorithm is used to solve the multi-group problem directly, and the MOC critical problem of the Wielandt iterative acceleration matrix is solved. The numerical results of multiple benchmark questions show that the Wielandt iteration combined with multi-group coupled GMRES has better computational accuracy and higher computational efficiency than the power iterative method.