A sensitivity and uncertainty analysis of the k_eff values for fast and thermal benchmarks with covariance data

The uncertainties in calculated integral nuclear parameters such as the k_eff values, reaction rates, reactivity worths, spent fuel isotopic concentrations, etc., can be caused by the inevitable approximations in cross-section data, the description of a geometry, material compositions, etc. The uncertainties in the integral parameters due to cross sections can be assessed using cross-section covariance data produced directly from the uncertainties of the measurements. The assessments and analyses of the sensitivity/uncertainty in k_eff for the fast and thermal benchmarks were performed with the available covariance data and analysis tools. The eigenvalues and spectra for this analysis were calculated with the KENO-V.a and ANISN codes, respectively. The sensitivity/uncertainty analyses for the benchmark cores were carried out using the TSUNAMI and SUSD3D codes with the covariance data of JENDL-3.3 and the new ORNL evaluations of the ²³⁵U and ²³⁹Pu covariance data. The uncertainties with the ORNL-generated ²³⁵U covariance data are about two times larger than with JENDL-3.3. The uncertainties due to total inelastic scattering covariance data (MT 4) and individual inelastic data of JENDL-3.3 reveal very large differences for the fast benchmarks. The comparisons and analyses for a detailed sensitivity/uncertainty in k_eff for the benchmarks are presented in this paper.