Turbulent natural convection along a vertical plate for different Prandtl number fluids

The design of Liquid Metal cooled reactors relies on Computational Fluid Dynamics (CFD) to evaluate the operational conditions and safety of the reactor. Liquid metals presents a very low Prandtl number value Pr≈O(0.001-0.01). Modeling the turbulent heat transfer for such fluids presents difficulties, and advanced models are required when simulating such kind of fluids. Their development, validation and assessment in different flow configurations will ensure accurate predictions of the reactor. This thesis investigates the turbulent heat transfer along a natural convection boundary layer (NCBL) by means of Large Eddy Simulations (LES) and Reynolds Averaged Navier Stokes simulations (RANS). Three different Prandtl numbers under unity (0.71, 0.2 and 0.025). The LES simulations employ a transition control method to trigger turbulent conditions. This technique is based on the excitation of the most energetic frequency during the transition to turbulence. The present method was developed for forced convection boundary layers in the past, and the present work extends this method to the NCBL flow by means of a best practice guideline. The LES investigation allows for a comprehensive study on the Prandtl number effect on the thermal-hydraulics on the NCBL. The wall heat transfer to the fluid is is dominated by conduction effects as the Prandtl number decreases. Buoyancy effects are more relevant at low Pr values, and large velocity values are reached. The turbulent kinetic energy and turbulent stresses are benefited of this phenomenon, while the thermal fluctuations and turbulent heat fluxes magnitude decrease. The LES results are employed to analyze the behavior of RANS turbulence thermal models for low Prandtl numbers in the NCBL configuration. They have been assessed and calibrated for such flow conditions. Employing a two-equations approach enhances the thermal prediction at different Prandtl number values under unity.