Data transfers for nuclear reactor multiphysics studies using the MOOSE framework

Guillaume Giudicelli^1*Fande Kong¹Roy Stogner¹Logan Harbour¹Derek Gaston¹Zachary Prince¹Lise Charlot¹Stefano Terlizzi²Mahmoud Eltawila³April Novak³Alexander Lindsay¹

• ¹Idaho National Laboratory (DOE), Idaho Falls, United States
• ²The Pennsylvania State University (PSU), University Park, Pennsylvania, United States
• ³University of Illinois at Urbana-Champaign, Champaign, Illinois, United States

High fidelity simulations of nuclear systems generally require a multi-dimensional representation of the system. Advanced nuclear reactor cores are governed by multiple physical phenomena which should be all be resolved, and the coupling of these physics would also need to be resolved spatially in a high-fidelity approach, while lower fidelity may leverage integrated quantities for the coupling instead. Performing a spatially resolved multiphysics simulation can be done on a single mesh with a single coupled numerical system, but this requires catering to each equations' time and spatial discretization needs. Instead, each physics, usually neutronics, thermal hydraulics and fuel performance, are solved individually with the discretization they require, and the equations are coupled by transferring fields between each solver. In our experience coupling applications within the MOOSE framework, mostly for advanced nuclear reactor analysis, there are several challenges to this approach, from non-conservation problems with dissimilar meshes, to losses in order of spatial accuracy. This paper presents the field transfer capabilities implemented in MOOSE, and numerous technical details such as mapping heuristics, conservation techniques and parallel algorithms. Examples are drawn from nuclear systems analysis cases to illustrate the techniques.