Investigation of a multiple energy grid deterministic method – Application to core-reflector problems. (1st June 2022)

• Record Type: Journal Article
• Authors: Desplats, Fiona; Archier, Pascal; Vidal, Jean-François; Palau, Jean-Marc; Lenain, Roland; Masiello, Emiliano
• Is Part Of: Annals of nuclear energy. Volume 170(2022)
• Publication Date: 2022-06-01
• Subjects: Multi-grid -- Domain decomposition method -- Transport calculation -- Core/reflector interfaces; Nuclear energy -- Periodicals; Nuclear engineering -- Periodicals; 621.4805
• DOI: 10.1016/j.anucene.2022.108970
• Languages: English
• ISSNs: 0306-4549

Abstract

Highlights: Whole core energetic refinement can be necessary to treat a few resonances. Domain decomposition methods allow the user to refine only a part of the problem. The user can set multiple energy grids to treat the problem geometry. Local energetic refinement of the core leads to gains in precision and time. Abstract: Incorrect representations of flux variations in a reactor core can introduce significant errors in the resulting calculation. In cases with strong energetic flux variations, it becomes important to refine the deterministic calculation to have a better description of resonances. Refining the energy mesh for the whole core can become quite computationally costly. Domain decomposition methods and parallel architecture seem to offer a solution, by allowing the user to refine only part of the problem. The present work aims to provide a domain decomposition method in a deterministic solver with the capability of better describing local heterogeneities and strong transients by adapting the energy refinement. The goal is to accomplish this without sacrificing calculation time. This paper investigates the use of domain decomposition methods with cross sections of different energy meshes within the IDT deterministic solver. In this novel method, subdomains have different energy discretizations and no overlap. Boundary fluxes are condensed or reconstructed to account for spectral differences between subdomains to make the boundary condition fully consistent across the whole core. Multi-grid cases, where subdomains have different energy discretizations are compared to classic cases with the same energy discretizations over the entire configuration. This method has been tested on two cases with different reflectors, to investigate the potential for this method to deal with heavy-material reflector design such as iron reflectors in modern PWR. Two boundary flux expressions have been examined. The gains in precision and computation time using this methodology have been estimated on a core configuration inspired by the well-known C5G7-benchmark. Preliminary results are promising both in terms of time and precision.