Search

• Sensitivity and uncertainty analyses performed with HELIOS-2/XSUSA using covariance data based on ENDF/B-VII.1. • Uncertainties from random sampling of the nuclear reactions. • Main contributors to the total result uncertainties through squared multiple correlation coefficients. • Random sampling with linear regression for evaluating sensitivity vectors. • Sensitivities and uncertainties for various exercises of the UAM-LWR benchmark series Uncertainty "breakdown" over burn-up time in depletion calculations. With the HELIOS–2/XSUSA lattice/depletion code, output uncertainties can be estimated by random sampling of nuclear reactions. The paper presents extensions for performing sensitivity analyses. Main contributors to the total uncertainties are evaluated through squared multiple correlation coefficients. For a direct comparison with sensitivities and partial uncertainties obtained with deterministic linear perturbation theory, a recently introduced extension of XSUSA is used, performing linear regression analysis of the output quantities. A novel feature is to evaluate the uncertainty "breakdown" over burn-up time in depletion calculations. With these extensions, sensitivities and uncertainties are estimated for various exercises of the UAM-LWR benchmark series. The analyses are performed using covariance data mainly based on ENDF/B-VII.1, available in 56-group representation in SCALE 6.2. They were transformed to the 44-group structure used by the HELIOS-2/XSUSA interface. For fresh fuel, the results are in very acceptable agreement with the corresponding results of the SCALE code TSUNAMI. [ABSTRACT FROM AUTHOR]

The paper presents different methods for evaluating sensitivities and uncertainties of the effective delayed neutron fraction (β eff) with respect to microscopic nuclear data in the framework of the SCALE 6.2 code system. The prompt k ratio method is used to calculate β eff approximately. Sensitivities of β eff to microscopic reaction data are calculated with deterministic linear perturbation theory from sensitivities of the multiplication factor, and by performing linear regression analysis on results obtained from random sampling. The latter approach is also applied for estimating sensitivities of kinetic parameters in six delayed neutron groups, as obtained by adjoint flux weighting in lattice calculations, to nuclear reactions. Finally, it is demonstrated that the uncertainties of β eff strongly depend on the choice of the covariance data set; in particular, the use of SCALE 6.2 covariance data for the average number of delayed neutrons per fission in some cases leads to very high β eff uncertainties. [ABSTRACT FROM AUTHOR]