Nuclear Science and Engineering / Volume 190 / Number 1 / April 2018 / Pages 1-30
Technical Paper / dx.doi.org/10.1080/00295639.2017.1417214
Nodal diffusion is currently the preferred neutronics model for industrial reactor core calculations, which use few-group cross-section libraries generated via standard assembly homogenization. The infinite-medium flux-weighted cross sections fail to capture the spectral effects triggered in the core environment by nonreflective boundary conditions at the fuel-assembly edges. This poses a serious limitation to the numerical simulation of current- and next-generation reactor cores, characterized by strong interassembly heterogeneity.
Recently, a spectral rehomogenization method has been developed at AREVA NP. This approach consists of an on-the-fly modal synthesis of the spectrum variation between the environmental and infinite-medium conditions. It uses information coming from both the nodal simulation and the lattice transport calculation performed to compute the standard cross sections. The accuracy of the spectral corrections depends on the choice of the basis and weighting functions for the expansion and on the definition of a realistic energy distribution of the neutron leakage. In this paper, we focus on the first aspect. Two tracks are researched: a combination of analytical functions (with a physically justified mode) and a mathematical approach building upon the Proper Orthogonal Decomposition. The method is applied to relevant pressurized-water-reactor benchmark problems. We show that the accuracy of the cross sections is significantly improved at reasonably low computational cost and memory requirement. Several aspects of the methodology are discussed, such as the interplay with space-dependent corrections. We demonstrate that this approach can model not only the spectral interactions between dissimilar neighbor assemblies but also the spectral effects due to different physical conditions (namely, multiplicative properties) in the environment and in the infinite medium.