École Nationale Supérieure de
Techniques Avancées

Abstract

Increasingly heterogeneous core configurations require a more rigorous treatment of the scattering term, i.e., increasing the order of the approximation, which in turn increases memory space and CPU time. The transport correction simplifies the treatment of the scattering term, traditionally based on the so-called outflow approximation. In today's reactor configurations, this correction does not produce satisfactory results, especially for the reflector zones. The more accurate inflow approximation allows for a better representation of these zones. However, it increases the memory space compared to the outflow approximation. This has slowed down its implementation in neutronic codes. In the context of a two-step reactor calculation, particular importance is attributed to the quality of the diffusion coefficient D for the 3D reactor calculation in diffusion theory, and especially for the reflector region. We have shown that the use of the formula inflow allows to improve the estimation of the coefficient D for one-dimensional geometries. This work focuses on the implementation of the inflow correction and its use in the estimation of the diffusion coefficient D for 3D reactor calculations. Two methods were investigated for the estimation of the diffusion coefficients of the macrogroups: the weighting of the transport cross sections or the weighting of the diffusion coefficients. Both the inflow correction and the outflow correction were used for these two methods. The weighting of the diffusion coefficients allows a significant gain in accuracy compared to the weighting of the effective transport sections, independently of the correction used. The choice of the transport correction has a lesser effect. The inflow correction allows a slight gain in accuracy compared to the outflow correction. However, some stability problems appear.