Etude d'un schema de calcul pour la determination des sections efficaces d'un coeur de REP.

摘要

The French nuclear reactors dedicated to electricity production are Pressurized Water Reactors (PWR). Their modelization requires using different nuclear simulation softwares, which reflect the fuctionning of different systems (the neutrons in fuel assemblies, the neutrons in the core, the fluids in the core and the fluids in the primary and secondary sides). The lattice code deals with the behaviour of neutrons in an assembly. The full-core code deals with the behaviour of neutrons in the core. Two thermal hydrolics codes calculate the thermal state respectively of the core and of the system as a whole.;The Institut de Radioprotection et de Sûreté Nucléaire (IRSN) is the institution in charge of the technical evaluation of the French power plants safety. As of today, the IRSN uses four codes developed by the French industry. These are APOLLO2 (lattice neutronics code), CRONOS2 (full-core neutronics code), FLICA4 (thermal hydraulics core code) and CATHARE2 (thermal hydraulics system code). The three last codes interact within the calculation chain HEMERA.;In order to provide an input to this chain, this project aimed at studying the possibility of a replacement of APOLLO2 by DRAGON4, which is an academic code developed in Montreal at the Institut de Génie Nucléaire (IGN). Indeed, such a substitution could provide an interesting double-check of the calculations, and a greater independance of the results found at the IRSN. Therefore, when obtaining new alternative results, the goal was not to reproduce exactly the existing calculations, but to remain relevant and to understand where the possible differences come from. The calculations with DRAGON and APOLLO were executed with a different microscopic library and a different calculation scheme (respectively JEFF3.1. and double-level versus JEF2.2. and simple-level).;We were not able to make calculations by coupling DRAGON and HEMERA, as the interface between the two codes proved to be non-functional. The lattice code produces multi-parameter libraries which are read by HEMERA. However, the libraries produced by DRAGON under the SAPHYB format were not read properly. The goal of the project was then to realize the lattice calculations necessary to the implementation of the libraries anyway, without concatenating them. A comparison of DRAGON and APOLLO has then been performed, as well as a comparison between the results given on the one hand by APOLLO and CRONOS and on the other hand by DRAGON and PARCS, an alternative full-core neutronics code.;We concluded that, for each condition of fuel, burnup and control rods position, at least 2340 lattice calculations with DRAGON were mandatory in order to obtain a sufficient precision. We also noticed that the results produced by DRAGON and APOLLO were similar, even if some important differences existed. In particular, given the double-level scheme used with DRAGON, the spatially distributed values tended to show more differences when computed by DRAGON and APPOLO, as opposed to the homogenized values such as the assembly k∞. This suggests that the new implementation could be more precise, for example in flux calculations for heterogeneous assemblies, especially with MOX fuel or with rods poisoned with gadolinium.;In short, DRAGON would be a good candidate for a substitution of APOLLO if needed, under the condition that the generated SAPHYB could be brought up to date. Still, further studies need to be performed in order to understand better the differences between the results of the two modelization strategies.