Optimising the nuclear data energy group structure used for fusion systems

The generation of reaction rates within Monte-Carlo (MC) transport codes can be accomplished via 1) the standard point-wise estimator approach of tallying the pointwise flux and pointwise partial cross sections and/or 2) the multigroup approach which convolves a pointwise neutron flux with pre-defined binned partial cross-sections. Even fine multigroup sampling is more efficient when compared to pointwise Monte-Carlo sampling due to the removal of the cross-section interpolation from the calculation. This paper describes the four stage optimization procedure used to find an energy binning format which enables an accurate enough and computationally efficient calculation of any reaction rate when applied to MC modeling of fusion devices. This optimum format, named FOMG (Fusion Optimised Multi-Group), was evaluated using a SINBAD (Shielding Integral Benchmark Archive Database) benchmark. Despite the number of multi-group bins exceeding the number of data elements for a significant number of the ENDF files the MCNP simulation took more than one hundred times longer to calculate reaction rates using the pointwise approach when compared to the multi-group method with FOMG structure