As part of a long-lasting dry storage, the oxidation of the spent fuel linked to an accidental supply of oxygen into the spent fuel rod, may have consequences that must be assessed. Indeed, the oxidation of the UO2 fuel up to U3O8 phase brings about a volume swelling of about 31% together with a bulking which may increase the initial defect of the cladding and thus lead to the release of fission products which remains embedded in nanometric gains or fine particles. From the oxidation model first developed (grain model), a new model has been proposed to describe the oxidation of spent fuel. The latter is based on a transposition of the graded grain model for unirradiated powders to irradiated grains. The modification of the grain model concerns the structural evolutions and the stoichiometric ranges during the oxidation. The convolution between the irradiated grain model and the propagation of the oxidation front along the grain boundaries allow to describe the oxidation kinetics of a spent fuel fragment. A criterion of grain breaking, based on experimental observations, has been proposed. The model has been parametrized from the experimental data of the literature and obtained as part of the French PRECCI's program. The harmony between the model and the experimental data either in terms of mass gain curves or of the oxygen diffusion coefficients in the UO2 matrix is satisfactory. From this mechanistic model, an operational model for dry storage is proposed. On one hand, a simplified model (qualitative one) allowing to link the operational parameters (irradiation history) such as the burn up, the linear power or else the released gas fraction to a time for cracks to appear in the grains has been proposed. This model allows to display the time for the fractures in the fuel to appear (or a time before fracturing) according to the review of the treated fuels. On the other hand, a predictive operational model taking only the temperature into account is proposed.