基于非均匀变分节块法的pin-by-pin计算加速算法研究

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.