Generating multigroup data stochastically for a highly heterogeneous VHTR problem

While the use of continuous energy cross sections in a whole core Monte Carlo calculation will always yield accurate results if enough particle histories are run, such an approach is so computationally expensive as to be considered not feasible for practical core eigenvalue calculations. It is therefore necessary to produce accurate multigroup cross sections for homogenized regions of the core, and let a faster method determine the core solution. A particular challenge, due to its more complicated spectrum and higher heterogeneity, is the generation of data for the VHTR, a gas-cooled reactor whose fuel is in particle form. In this paper, the stochastic cross section generation method MoCS-Gen is used to generate 26 group data for the VHTR. Problems in a pin cell and a fuel block are solved, and compared to the original fully heterogeneous problem in continuous energy and also to the same problems solved in continuous energy but utilizing a simple volume homogenization strategy. Results are presented which clearly show the benefit of MoCS-Gen as an accurate multigroup cross section generation method.