Parameterized Model Order Reduction Strategy for Fast Running CFD-Quality Transient Transport through Nuclear Reactors

Computational fluid dynamics methods are increasingly used for high fidelity modeling of complex systems, including nuclear power systems.For example, thermal stratification in a sodium-cooled fast reactor under natural circulation conditions is resolvable with computational fluid dynamics, but not traditional system-level models.To offset the high cost of computational fluid dynamics, methods providing high quality results at a reduced cost are desired.Software will be developed to enable production of high quality reduced order models using a parameterized proper orthogonal decomposition approach.The technology will enable high delity fast running models for arbitrary transients to be constructed from a set of steady state simulations.The models will then be used to compute boron concentrations in the lower plenum and core region of a reactor or density variations in stratified pools of sodium cooled fast reactors.Phase I Results: A reduced order model production work ow using a parameterized proper orthogonal decomposition approach was proven feasible.Using this new formulation, a series of Large Eddy Simulations with steady state boundary conditions were used to construct a reduced order model with transient boundary conditions running in a fraction of the time.To improve the accuracy and stability of the parameterized approach, various interpolation methods and closure models were investigated.Initial steps were taken to expand this formulation to simulate scalar transport and include the eects of pressure gradients.The feasibility of applying this technology to industrial-scale simulations was also demonstrated.Interfaces to third-party software, including a well-established open source computational fluid dynamics solver and a government-funded platform for uncertainty quantification and stochastic analysis, will facilitate rapid user adoption.The effects of interpolation and closure modeling on runtime, accuracy, and flow dynamics will be studied in greater depth.Scalar transport and pressure modeling capabilities will continue to be developed.The technology will be used to model a challenge problem of interest to stakeholders in the nuclear energy industry.Commercial Applications and Other Benefits: The proper orthogonal decomposition method- ology has been understood for decades, but not widely adopted for several reasons: dicult and lengthy computation, unfavorable stability properties, and requirements of ad hoc software development.By leveraging recent developments and innovative algorithms, these concerns are addressed by this work.A key application in the nuclear engineering community is production of surrogate computational fluid dynamics models to couple with system level codes for design and safety analyses.However, nearly any industry with large modeling and simulation components will benefit.