Sensitivity and uncertainty analysis in nuclear data with SCALE 6.2 code. Application to LWRs assemblies

This work aims to compare the resulting sensitivity and uncertainty information of k_eff and the most sensitive nuclide-reaction cross-section data between two modules of SCALE. In addition, a bias calculation was estimated in order to provide the margins needed to address uncertainty in the computational method, primarily due to neutron energy parameters. With Sensitivity Analysis (SA) and Uncertainty Quantification (UQ) it is possible to analyze which input parameters are more influential in the predicted response, identifying differences due to changes in the energy-dependent, nuclide-reaction cross-section data. This approach gives significant information in reactor modelling with Best-Estimate codes, especially in criticality calculations for Light Water Reactors (LWR) fuel applications. In this study neutron transport simulations are performed with two modules of SCALE6.2.1 in order to compare the deterministic and stochastic methodology and to assess the similarity in output parameters. The sensitivity and uncertainty (S/U) analysis sequences employed were TSUNAMI-2D and SAMPLER. TSUNAMI-2D estimates the uncertainty in the basic data using the sensitivity of neutron cross-section laid in the covariance library of SCALE. In contrast, SAMPLER module employs the same information to produce a stochastic sampling technique of the input parameters and gives the correlation coefficients and covariance matrix, as well as more statistical information. Two models of Light Water Reactors were studied: a Boiling Water Reactor (BWR Peach Bottom-2) and a Pressurized Water Reactor (PWR Three Miles Island-1). Both of them at Hot Full Power (HFP) and Hot Zero Power (HZP) conditions, with and without control rod. Moreover, USLSTATS code was used to quantify bias, bias uncertainties, and trending analysis, providing detailed assessments of sources of biases and their uncertainties. (author)