Nuclear data uncertainty propagation in continuous-energy Monte Carlo calculations

Abstract

The XSUSA method is a well-established stochastic sampling method for propagating nuclear data uncertainties through multigroup neutron transport calculations. To benefit from the advantages of Monte Carlo transport codes, namely modeling complex geometries and using continuous-energy nuclear data, an extension to XSUSA is proposed which allows perturbing continuous-energy nuclear data using multigroup nuclear data covariances. To verify the extension, sensitivity profiles of nuclear reactions are calculated via direct perturbation for the benchmark problems Jezebel, Godiva, LEU-SOL-THERM-002. The sensitivity profiles agree well with those obtained from TSUNAMI and Serpent. Secondly, the extension to XSUSA is applied to produce randomly sampled continuous-energy data libraries using the covariance libraries of SCALE 6.2. With these data libraries, samples of Serpent calculations are performed for Jezebel, Godiva, LEU-SOL-THERM-002, and the TMI-1 pin cell of the OECD/NEA LWR-UAM benchmark. For each problem, the multiplication factor uncertainty agrees well with the one from TSUNAMI.