Speed-up of cross-section collapsing with weight-window for subcritical burnup calculation

In the author’s group, a fusion–fission (FF) hybrid energy system has been analyzed using our own burnup calculation system consisting of Monte Carlo transport code MCNP-4C and point burnup code ORIGEN2.1. Since the neutron energy spectrum changes along with progress of burnup in a subcritical system, it is necessary to update one-group cross-section library in each burnup step. The one-group cross-sections are normally updated by collapsing the evaluated nuclear data such as JENDL and ENDF using a neutron flux calculated by an appropriate transport code such as MCNP. The collapsed cross-sections are handed over to ORIGEN, and the reaction rates for burnup of elements are thereafter estimated accurately. l known, MCNP generates track-length (TL) data in the neutron transport calculation, which are base data to estimate the neutron flux. We thus use the track-length data directly instead of the calculated neutron flux, in order to evaluate the reaction rate as accurately as possible. However, the number of TLs becomes extremely large and thus it takes a longer computation time. We therefore reduce the number of TLs used in the cross-section collapsing process as far as the accuracy is conserved. However, in some energy region the number of TLs is inversely too small to conserve the original cross-section accuracy of the evaluated nuclear data files, because the number of TL data per unit energy is smaller than that of the nuclear data. present study, the weight-window (WW) technique of MCNP was applied to our burnup calculation system in order to control the number of TLs in such an energy region artificially and to complete the collapsing process accurately in the whole energy region. As a result, the variance of the calculated neutron flux thus deteriorates slightly, but the number of TLs could be successfully adjusted to conserve the accuracy of the nuclear data file in the whole energy region. And the accurate reaction rate estimation for burnup with MCNP was finally realized and simultaneously the computation time could be saved reasonably.