Improving the Discrete Generalized Multigroup Method and Recondensation
Cross-section recondensation using the Discrete Generalized Multigroup method (DGM) has shown promise in improving coarse group neutron transport solutions in a computationally cheap manner. Recondensation can be especially helpful for analyzing very heterogeneous reactors with long mean free paths because it can approximately capture the swapping of fine group spectra between different regions in the core using only the coarse group solution. However, full consistency of DGM with the fine group problem is only assured when using a spatially flat angular flux approximation, such as step difference discrete ordinates.
For spatial methods assuming some shape in the angular flux (e.g. the exponential shape in characteristic type methods), spatial inconsistencies between the DGM equations and the fine group equations may cause recondensation to converge to the incorrect solution. In this work we identify the source of the inconsistency and propose the definition of an exact corrective term that allows recondensation to converge to the true fine group solution. Unfortunately this term must be defined specifically for the spatial method being used and can suffer from large storage requirements. Therefore, finding the best approximations for the corrective term continues to be key in making this method feasible for realistic problems.