Towards Reduced Order Modeling of Multiphysics Analysis of Molten Salt Reactors

Open Access
- Author:
- Emler, Casey
- Graduate Program:
- Nuclear Engineering
- Degree:
- Master of Science
- Document Type:
- Master Thesis
- Date of Defense:
- June 25, 2024
- Committee Members:
- Elia Merzari, Thesis Advisor/Co-Advisor
Saya Lee, Committee Member
Dipanjan Pan, Professor in Charge/Director of Graduate Studies - Keywords:
- Thermal-Hydraulics
Neutronics
Reduced Order Models
Multiphysics
CFD
Monte Carlo
POD
Advanced Nuclear Reactors
Molten Salt Reactors - Abstract:
- Molten Salt Reactors (MSRs) present significant modeling challenges due to the inherent coupling between neutron transport and thermal-hydraulics, stemming from the use of a circulating liquid fuel salt. The large-scale multiphysics simulations required demand extensive computational resources, making routine analysis of these systems impractical. This thesis explores the development of reduced-order modeling (ROM) techniques for the multiphysics analysis of MSRs, specifically focusing on the Molten Salt Fast Reactor (MSFR). The study aims to balance computational efficiency with model accuracy through several key tasks. To begin, a high-fidelity neutronics model is developed using the Monte Carlo code OpenMC and verified against results from Serpent. Subsequently, a thermal-hydraulics model is created using the spectral element computational fluid dynamics (CFD) code nekRS and verified against OpenFOAM. This model is then enhanced by incorporating delayed neutron precursor transport. Additionally, the thesis introduces a reduced-order model for thermal-hydraulic behavior using a standard proper orthogonal decomposition (POD) and Galerkin projection procedure. ROM stabilization methods are shown to be needed to generate sufficient small-scale dissipation, with constrained optimization (C-ROM) emerging as the preferred method due to its efficiency and accuracy. The results demonstrate that the C-ROM achieves high accuracy in predicting reactor behavior, comparable to the full-order model, while significantly reducing computational resources. This approach lays the groundwork for future simulations in MSR analysis, enabling more efficient and accurate assessments of reactor performance and safety.