GENERATION OF SFR PHYSICS MODELS FOR THE NUCLEAR FUEL CYCLE CODE CLASS

摘要

This participation to PHYSOR 2016 aims to present the methods used to add a full description of a Sodium Fast Reactor (SFR) in the nuclear fuel cycle code CLASS. This description requires two models.The first one manage the determination of the plutonium content in the fresh fuel needed to meet a reactivity criterion according the composition of the available plutonium and depleted uranium storages. The second one aims to predict neutron spectrum averaged cross section needed to solves Bateman equations without calling a neutron transport code. The approach used to create this two models and their performances are shown. Furthermore, the impact of using time independent mean cross sections in the depletion calculation is depicted. These two models are based on applications of machine learning algorithms on a set of depletion calculations. The discrepancies on mean cross sections predicted by the model leads to relatively small errors on nuclei inventories at end of cycle. Indeed, the maximum difference in nuclei inventories calculated with the reference depletion code and CLASS using predicted mean cross sections is 2% for the main nuclei. The accuracy, relatively to the neutron transport code, of the k_(eff) predictor is about 130 pcm and allows to predict the plutonium content in the fresh fuel to meet k_(eff)(t= 0) at beginning of cycle.