The history-partitioning method for multigroup stochastic cross section generation

Abstract A method is detailed for the stochastic generation of multigroup cross sections. This paper describes a solid theoretical framework for development of multigroup (discrete energy) nuclear data from continuous energy evaluated nuclear data files by Monte Carlo means. Included is a novel means of estimating the angular distribution of particles scattered in different energy groups. The method is demonstrated using a highly heterogeneous VHTR pin cell model. Homogenized cross sections are generated for a fuel region consisting of encapsulated fuel particles dispersed in a graphite matrix, placed back into the pin cell problem, and compared to the original continuous energy heterogeneous model. Results show eigenvalue agreement of approximately 100 pcm, which is within the estimated uncertainty range of the multigroup data.