Analysis of new nuclear data evaluations and covariances for depletion calculations in light water reactors

This thesis develops a methodology that decomposes the uncertainty on the quantities of interest in depletion calculations into the individual contributions of nuclear data, with the focus on fission yields. While previous authors only extended the decomposition of the uncertainty contributions of fission yields to the quantities of interest to the level of individual fissile nuclides, this new methodology identifies which fission yields with the same mass number contribute most to the uncertainty. This allows the evaluators working on JEFF-4.0 to pinpoint which fission yield uncertainties need to be reduced. The methodology applies the stochastic sampling of JEFF-3.3 fission yields by the SANDY code. The uncertainty on the fission yields is propagated through the Serpent Monte Carlo transport code in a model of the GU3 sample. This sample from an axial section of a PWR UOX fuel rod is selected from the SFCOMPO library for its high burnup. A strategically chosen sampling procedure allows the decomposition of the uncertainty on k∞ and the mass densities of fission products important for criticality safety