Challenge Problem 1: Preliminary Model Development and Assessment of Flexible Heat Transfer Modeling Approaches
- Massachusetts Inst. of Technology (MIT), Cambridge, MA (United States)
- North Carolina State Univ., Raleigh, NC (United States)
- Pennsylvania State Univ., University Park, PA (United States)
- Pennsylvania State Univ., University Park, PA (United States); Argonne National Lab. (ANL), Argonne, IL (United States)
- Argonne National Lab. (ANL), Argonne, IL (United States)
This report presents the modeling progress of a university consortium studying flexible modeling for turbulent heat transfer. In the context of advanced nuclear reactor design, where challenges include non-unity Prandtl fluids, free and mixed convection, and transitional flow, computational fluid dynamics methods are underutilized due to low confidence in modeling approaches and high computational cost. This report evaluates the performance of engineering CFD models in diabatic, buoyant turbulent flow. It finds that all turbulence models including several variants of the k-ε and k-ω models struggle to predict accurate turbulent momentum and heat transfer in such flows. The Nusselt numbers have been compared between the models and the DNS data, where calculations have been performed for each case and trends have shown a good agreement between DNS estimated Nusselt numbers and available correlations as well as experimental data. A novel DNS correlation for high Pr cases as they are transitioning from mixed convection to natural convection has been proposed. While Nusselt number errors relative to DNS range from 20% to 50%, the models capture similar trends to DNS with respect to Nusselt suppression and amplification under varying levels of buoyancy effect. This report also highlights model form error as a significant contributor to CFD predictions and proposes a framework for quantifying model error and improving confidence in CFD calculations. To improve the predictive capability of engineering CFD models, data-driven approaches for turbulence models are investigated. Theoretical frameworks based on the invariant tensor / vector basis neural networks for prediction of Reynolds stresses and turbulent heat fluxes are employed. The models are developed using direct numerical simulations data for forced convection flows of different fluids in vertical planar channel domain. The framework is implemented in spectral element solvers Nek5000 / nekRS and has shown a potential for future development and consideration of mixed convection flows.
- Research Organization:
- Argonne National Laboratory (ANL), Argonne, IL (United States)
- Sponsoring Organization:
- USDOE Office of Nuclear Energy (NE), Nuclear Energy University Program (NEUP); USDOE Office of Nuclear Energy (NE), Nuclear Energy Advanced Modeling and Simulation (NEAMS)
- DOE Contract Number:
- AC02-06CH11357
- OSTI ID:
- 1881860
- Report Number(s):
- ANL/NSE-22/40; 176974; TRN: US2308791
- Country of Publication:
- United States
- Language:
- English
Similar Records
Initial use of Nek5000/Cardinal to improve closure models in Pronghorn
High Fidelity CFD Simulations Supporting the KP-FHR