Approach to better assess fission gas behaviors, applicable to fuels with complex microstructures

International audienceThe industrial fission gas behavior model CARACAS developed and validated for LWR-UO2 fuels in different fuel performance codes at CEA, EDF and AREVA is now being adapted to simulate the behavior of fuels with complex microstructures. In the current industrial licensed code, the description of such fuel is based on homogenization of the pellet, whereas for CEA's fuel performance code ALCYONE, this microstructure can be modelled by inclusions surrounded by a continuous matrix. The main objectives of the CARACAS fission gas model are to propose a continuous approach in both steady-state and off-normal conditions, but also to make it depend on the initial microstructure of the material. To model the fission gas behavior in such a complex microstructure, we assume that the behavior of each phase can be modeled as a homogeneous phase. First, neutron model provides CARACAS model the average local power rate, fission density and xenon/krypton production in each phase. Then, in CARACAS model, the first way of coupling between phases is done by the fission gas production mechanism, in which the recoil of fission gas atoms and their implantation from one phase to the other is taken into account. This heterogeneity of fission distribution and fission gas atoms production leads also to a heterogeneous swelling of the material, and so to a heterogeneous distribution of stress in the microstructure.It is now possible to take into account the heterogeneity of stress induced by this heterogeneous swelling and its feedback on the fission gas behavior model, thanks to the micro-mechanical behavior law. As the fission gas behavior model (particularly the gaseous swelling and the intragranular/intergranular gas distribution) depends at first order on the hydrostatic stress, the distribution of stress in the microstructure is very important. As this approach is very time-consuming, it is more dedicated to research applications or to very specific design calculations such as refined accidental initializations. For standard industrial applications, some simplifications are done. Depending on the representation of the microstructure in the fuel performance code where the model is implemented, different hypothesis are presented homogeneous or heterogeneous. Comparison of calculated results in both configurations with experimental data is also presented. An attempt of uncertainty quantification in relation to experimental conditions of the validation database is also performed