NUSS: A tool for propagating multigroup nuclear data covariances in pointwise ACE-formatted nuclear data using stochastic sampling method

Abstract

Stochastic sampling (SS) method for quantifying nuclear data uncertainties is accomplished by using perturbed nuclear data in routine neutronics calculations and determining the variance of output parameters due to the input nuclear data uncertainties. Existing SS-based methods have demonstrated the feasibility and efficiency of propagating uncertainties in multigroup nuclear data. However, in fields such as criticality safety assessment, pointwise representation of nuclear data is more appropriate in order to corroborate the increasing safety demand and best-estimate modeling capabilities. In this work, an SS-based tool, called NUSS is implemented which perturbs pointwise ACE-formatted nuclear data using multigroup nuclear data covariance. The use of pointwise ACE-formatted nuclear data in NUSS can accommodate flexible multigroup covariance structures and allows for nuclear data uncertainty propagation through the continuous/pointwise-energy transport code MCNPX. As a first step of the NUSS development and verification, uncertainty contributions from 239Pu and 235U nuclear data were assessed for Jezebel (Pu-fueled) and Godiva (U-fueled) fast-spectrum criticality benchmarks. NUSS results are compared to those by other uncertainty quantification methods such as TSUNAMI and MCNPX PERT CARD. Next, Light Water Reactor (LWR) pin cell models from the OECD/NEA UAM Phase-1 benchmark were analyzed. Results of cross section and kinf uncertainties in consideration of different nuclear data covariance libraries are presented.