Large Eddy Simulation of Low-Reynolds-Number Turbulent Flow of Low-Prandtl-Number Fluid in a Tight Lattice Bundle for Assessment of Reynolds-Averaged Navier-Stokes Turbulence Model

The Microreactor Applications Research Validation and Evaluation (MARVEL) microreactor utilizes natural circulation as a core cooling mechanism and liquid metal as a primary coolant. Moreover, the reactor core has a pitch-to-diameter ratio of 1.054, which is considered a tight lattice configuration. Numerous studies have widely reported that Reynolds-averaged Navier-Stokes (RANS) turbulence models inaccurately predict heat transfer in liquid metals and fail to capture flow pulsations that can occur within tight lattices, leading to further inaccuracies in simulation results. Therefore, evaluating the accuracy of RANS turbulence models in the thermal-hydraulic analysis of the MARVEL microreactor core is crucial for assessing reactor safety. In this study, a large eddy simulation (LES) of the MARVEL microreactor core subchannel was conducted and compared with a RANS simulation to evaluate the accuracies and conservatism of the RANS model. In comparison with the RANS model, LES captures flow pulsations in a tight lattice that enhance heat transfer, whereas the RANS model underpredicts heat transfer in liquid metal flow. As a result, the RANS model predicts a higher peak cladding temperature than LES. However, owing to the high thermal conductivity of the liquid metal, the discrepancy between the two approaches is limited. These results indicate that the steady-state RANS model is adequate for the thermal analysis of the liquid metal–cooled MARVEL microreactor core and can provide conservative, yet not excessively overpredicted, results for safety assessment.