AuthorsA. G. Buchan, P. E. Farrell, G. J. Gorman, A. J. H. Goddard, M. D. Eaton, E. T. Nygaard, P. L. Angelo, R. P. Smedley-Stevenson, S. R. Merton and P. N. Smith
TitleThe Immersed Body Supermeshing Method for Modelling Reactor Physics Problems With Complex Internal Structures
Afilliation, Scientific Computing
Project(s)Center for Biomedical Computing (SFF)
StatusPublished
Publication TypeJournal Article
Year of Publication2014
JournalAnnals of Nuclear Energy
Volume63
Pagination399-408
Date PublishedJanuary
Abstract

This article describes a new immersed body method for the efficient modelling of complex reactor physics problems. The approach is based on a projection method that maps geometric diagnostics of internal bodies onto practical computational meshes. It applies a recently developed supermeshing algorithm originally developed for data transfer problems to parameterise the effects of internal bodies on the reactor dynamics. This projects meshes of internal bodies onto a mesh that encompasses the entire problem domain. With this mapping, all necessary information about the intersection of an element with the internal body is known. This includes information about the volume, surface area and distances along the internal bodies; importantly, these quantities are always conserved. The appropriate material cross-sections for each element are then calculated from the volume information to account for all the internal bodies they contain. This in turn enables the problem to be solved efficiently on meshes that are practical to generate. The method is demonstrated on two eigenvalue problems where the domain contains fuel pins, cooling pipes, control rods and guide tubes. The first problem is used to demonstrate convergence when the mesh fully resolves the internal bodies and the geometric details of the problem are completely recovered. The second problem models the SUPO (Super POwer) solution reactor which contains many complex and detailed internal components. It is shown that the internal structures of the problem can be parameterised efficiently without the use of computationally expensive geometry-conforming meshes.

DOI10.1016/j.anucene.2013.07.044
Citation KeySimula.simula.2410