PWR MOX FUEL PHYSICS MODELS FOR THE DYNAMIC FUEL CYCLE SIMULATION TOOL CLASS

摘要

Dynamic fuel cycle simulation tools model the entire fuel cycle, from fuel fabrication to spent fuel storage, and calculate nuclei inventories and material flows evolution in each unit. In the fuel cycle simulation tool CLASS, a fuel loading model builds a fresh fuel meeting the reactor requirements at each reactor loading and a cross-section predictor provides the mean cross-sections required to calculate its depletion during irradiation. This paper focuses on the elaboration of these models in the case of a PWR MOx-extended fuel, which includes MOx and multi-recycled MOx on uranium enriched support (MOxEUS) fuels. These models are built using neural networks applied on a databank gathering a thousand depletion simulations results. A fuel loading model based on a infinite multiplication factor (fcoo) calculation predicts the plutonium content, and in specific cases the uranium enrichment, required in the fuel to reach the target burn-up. This method allows an accuracy of 270 pcm on the reconstructed k_∞. The neural network mean cross-sections reconstruction coupled to the Bateman equation solving allows a depletion calculation with a deviation at the end of cycle (EOC) lower than 4 % on main nuclei compared to the fuel depletion reference calculation. PWR MOx-extended models are also compared to a previous equilibrium MOxEUS strategy calculation, showing a good general agreement on calculated equilibrium isotopic vectors.