LES and Unsteady RANS Computations of Natural Convection Cooling Loops

Natural Circulation Loops (NCLs), where fluid is driven through a closed circuit solely by thermal imbalance, offer potential for use in passive cooling systems within nuclear power plants. However, the combination of uniform heat flux at the hot end and uniform temperature on the cold side promotes thermal imbalances which can cause large-scale flow instabilities, including flow reversal, that can severely challenge numerical methods. This study employs both unsteady RANS models and Large Eddy Simulation for the conjugate heat transfer analysis of an experimental NCL across a range of different heating powers, with a focus on comparing several widely used URANS models (Launder-Sharma k−ε, k−ω SST and Elliptic-Blending RSM) against the higher-fidelity LES. The results indicate that the more advanced EB-RSM provides the best agreement with the LES, with the computationally less demanding Launder-Sharma (LS) k−ε model also performing well. The k−ω SST model performed less well, tending to overpredict turbulence levels within the loop, leading to significantly cooler temperatures, higher heat transfer coefficients and flatter velocity profiles. The subsequent application of the LS k−ε model to a wide range of heating powers demonstrated good agreement with the experimental data, correctly capturing the trends presented by both the experimental data and existing correlations. Overall, the results demonstrate the suitability of the unsteady RANS approach in modelling flow behaviour typical of NCLs and contrast the performance of several widely used modelling approaches, providing insight for those designing, optimizing or experimenting with NCLs.