Development of transport calculation for unstructured mesh geometry in cosRMC

Abstract

The traditional constructive solid geometry (CSG) modeling approach fails to meet the requirements of complex geometric transport, fusion neutronics analysis, multi-physics coupling calculations, and large-scale nuclear radiation effect simulations. To tackle these challenges, this study developed unstructured mesh (UM) transport and statistical functionalities in the Monte Carlo particle transport code, cosRMC, based on the MOAB and libMesh libraries. These functionalities allow for accurate particle positioning and tracking within UM elements, and the transport process is accelerated through a neighbor element acceleration technique. The track-length estimator is then used for tally results. To verify the computational accuracy and efficiency of the UM feature, three test cases were constructed and analyzed: a box case, a fuel rod case, and a tokamak case. The computational results demonstrate that the k_eff values calculated by UM-cosRMC geometry transport function align well with those from traditional CSG and DAGMC geometries, and the UM tally results are consistent with those of UM-MCNP calculations. In terms of computational efficiency, after employing the neighbor mesh acceleration technique, the speed of UM geometry calculations improved significantly. The UM geometry calculation efficiency based on the libMesh library is higher than that based on the MOAB library, with the former taking 1.18 to 10.42 times longer than CSG geometry calculations and 0.37 to 0.96 times that of DAGMC geometry. Compared to UM-MCNP calculation efficiency, UM-cosRMC shows a distinct advantage, reducing computation time by >70 % with the same mesh. Additionally, this study analyzed the computational efficiency of UM transport based on both libraries, and the results indicate that the libMesh-based UM calculations outperform in terms of UM reading speed, memory usage, and transport computation speed.