Investigation of Heat Convection using Liquid Metals

Abstract

Liquid metals are commonly used as nuclear reactor coolants to maintain the core temperature within a safe range. Their low Prandtl numbers indicate high molecular thermal conductivity, which Investigation of Heat Convection using Liquid Metals enhances heat transfer capabilities. However, most turbulent heat flux models, based on eddy-diffusivity approaches with the constant turbulent Prandtl number, have been developed for fluids with Pr approaching 1 and may not perform well for simulating liquid metals. The present thesis addresses the issue by thoroughly assessing a proposed Prt model through computations of forced convection flows in five different geometries with various low Prandtl number fluids. Low-Re turbulence models were tested in all cases. The proposed Prt model was derived from a DNS database of thermal channel flow at moderate and low molecular Prandtl numbers (Alcantara-Avila et al., 2018). Initially, a study was conducted on flow and heat transfer in a 2-D fully developed turbulent straight channel. Reynolds numbers of Re = 2846, 10200, 23000, and 48500, along with Prandtl numbers ranging from Pr = 0.007 to 0.71, were simulated. The constant turbulent Prandtl number and Kays correlation (Kays, 1994) were compared with the proposed Prt model. Results from dynamic and thermal fields indicated good agreement with DNS data, particularly at higher Reynolds numbers. Available DNS results for a wavy wall channel at a low Reynolds number of Re = 4720 were used to show limited improvements from the proposed Prt model in this case. In the case of a 2D plane free jet, with heated co-flow, the thermal field was dominated by turbulent mixing rather than molecular effects, and the proposed Prt model behaved satisfactorily. Testing was also conducted on a backward-facing step flow in a turbulent forced convection regime at Re = 4805, with a range of low Prandtl numbers. The proposed Prt model, which includes a modified Yap term weakening the effect of the standard Yap term, presented improved prediction accuracy compared to DNS results. Lastly, for an impinging jet at Re = 5700, the modified Yap term and proposed Prt model demonstrated superior thermal performance to other models tested at Pr = 0.01 and 0.1. Based on the findings from simulations across various geometries and flow conditions, the proposed Prt model with the L-S k-epsilon turbulence model, validated against DNS data, demonstrates robust performance in predicting thermal characteristics in turbulent flows with low Prandtl numbers.

Date of Award	10 Feb 2025
Original language	English
Awarding Institution	The University of Manchester
Supervisor	Hector Iacovides (Main Supervisor), Timothy Craft (Co Supervisor) & Andrea Cioncolini (Co Supervisor)