Assessment and improvement on the applicability of turbulent-Prandtl-number models in RANS for liquid metals

Abstract

The liquid-metal cooled fast reactor is known as one of the most promising Gen-IV nuclear power system due to its high safety and superior economy. However, the turbulent heat transfer behaviors of low-Prandtl-number liquid metal in fast reactors are quite different from that of ordinary fluids due to its violation in Reynolds analogy, and there is still lack of a unified turbulent-Prandtl-number (Prt) models in eddy viscous RANS models for liquid metals. In the present study, all existing Prt models were firstly collected and divided into two groups according to their dependent variables, then a comprehensively assessment on the applicability of both two types of Prt models in predicting turbulent heat transfer of liquid metal is conducted. The study found that an optimal Prt number exists in eddy viscous models under fixed Pe conditions, and Prt plays a significant role on heat transfer in the buffer zone and the fully-turbulent region, Under certain conditions, if the averaged second-type Prt in these two regions equals to the Prt in the first type of models, a similar heat transfer result with the same accuracy could be obtained. A big deviation (about 13%) is observed in Kays' model, even though Kays’ model in the second type of Prt models performs best at various Pe. To get more accurate predicting results, an improved Prt model has been proposed based on high fidelity DNS/LES profiles and experimental data, the deviation is less than 5%, which agreed well with the experimental data and correlations.