基于非均匀变分节块法的pin-by-pin计算加速算法研究

Acceleration Algorithm to Pin-by-pin Calculation with Heterogeneous Variational Nodal Method

计算效率是制约pin-by-pin计算工程应用的主要因素之一。本文利用三维扩散的非均匀变分节块法的非均匀节块的描述能力,在不改变原问题栅元均匀化材料分布的前提下,将传统pin-by-pin计算中使用的均匀材料细网剖分方式替代为非均匀材料粗网剖分方式(粗网加速方法),既能保证pin-by-pin的计算分辨率,又能显著降低红-黑迭代所需的浮点数操作数目,减小内迭代的计算代价。针对外迭代,运用广义矩阵分离加速(GPM)算法和粗网有限差分(CMFD)算法提高源迭代的收敛速度,降低计算时间。数值结果表明,提出的加速算法能在保证计算精度的前提下,有效提高pin-by-pin计算的效率。

The computational efficiency is one of the main factors restricting the application of pin-by-pin calculation. In this paper, the heterogeneous node capability of the three-dimensional diffusion heterogeneous variational nodal method was utilized to replace the homogeneous fine meshing scheme with heterogeneous coarse meshes (coarse node acceleration method) without altering the pin-cell-homogenized material layout. The method could both preserve the pin-by-pin calculation resolution and significantly reduce the number of floating point operations required in Red-Black iterations, therefore easing the computational burden expended in inner iterations. For outer iterations, the generalized partitioned matrix (GPM) algorithm and the coarse mesh finite difference (CMFD) algorithm were performed to accelerate fission source iterations and reduce computational time. Numerical results show that the proposed acceleration algorithms can effectively enhance the computational efficiency of pin-by-pin calculation without causing any sacrifice in accuracy.