Nuclear data uncertainty and correlation foe spent fuel applications

The computational simulation of the nuclide inventory is paramount for the evaluation of the risk related to the fuel composition, crucial for its role in safety protocols and fuel cycle efficiency. This is a consequence of the complexity and high cost of performing an experimental assessment of each discharged assembly. To validate depletion simulation codes, certain cases from the Spent Fuel Composition (SFCOMPO) assay database are used for comparison. This validation is made comparing the obtained results with the experimentalones. For validation purposes, the results are compared with their respective uncertainties. In this thesis, statistical sampling is used to propagate the uncertainty in the cross-section of different actinides and fission products to the concentration of several isotopes relevant to the different observables of interest in the Spent Nuclear Fuel (SNF). The use of statistical sampling gives an approximation of an output distribution, useful for further analysis. Additionally, a correlation study across different cases complements the uncertainty results and helps identifying common sources of uncertainty within the various systems. This correlation can be performed between different models to evaluate their similarities or within a model to study the influence of an isotope on another. In this thesis, the Gösgen GU1 sample and one of the exercises from the Uncertainty Analysis in Modelling (UAM) framework, the UAM Pin-cell, were selected to propagate the uncertainties in cross-section data from the JEFF-3.3 library through burnup simulations in Serpent. The results obtained are useful as they reflect the impact of nuclear data and other factors to the uncertainty concentration of several nuclides. Also, these results highlight the effects of a certain irradiation environment and initial conditions on uncertainty propagation.