Influence of molten salt-(FLiNaK) thermophysical properties on a heated tube using CFD RANS turbulence modeling of an experimental testbed

Abstract

In a liquid fuel molten salt reactor (MSR) a key factor to consider upon its design is the strong coupling between different physics present such as neutronics, thermo-mechanics and thermal-hydraulics. Focusing in the thermal-hydraulics aspect, it is required that the heat transfer is well characterized. For this purpose, turbulent models used for FLiNaK flow must be valid, and its thermophysical properties must be accurately described. In the literature, there are several expressions for each material property, with differences that can be significant. The goal of this study is to demonstrate and quantify the impact that the uncertainty in thermophysical properties has on key metrics of thermal hydraulic importance for MSRs, in particular on the heat transfer coefficient. In order to achieve this, computational fluid dynamics (CFD) simulations using the RANS k-ω SST model were compared to published experiment data on molten salt. Various correlations for FLiNaK’s material properties were used. It was observed that the uncertainty in FLiNaK’s thermophysical properties lead to a significant variance in the heat coefficient. Motivated by this, additional CFD simulations were done to obtain sensitivity coefficients for each thermophysical property. With this information, the effect of the variation of each one of the material properties on the heat transfer coefficient was quantified performing a one factor at a time approach (OAT). The results of this sensitivity analysis showed that the most critical thermophysical properties of FLiNaK towards the determination of the heat transfer coefficient are the viscosity and the thermal conductivity. More specifically the dimensionless sensitivity coefficient, which is defined as the percent variation of the heat transfer with respect to the percent variation of the respective property, was −0.51 and 0.67 respectively. According to the different correlations, the maximum percent variations for these properties is 18% and 26% respectively, which yields a variation in the predicted heat transfer coefficient as high as 9% and 17% for the viscosity and thermal conductivity, respectively. It was also demonstrated that the Nusselt number trends found from the simulations were captured much better using the Sieder Tate correlation than the Dittus Boelter correlation. Future work accommodating additional turbulence models and higher fidelity physics will help to determine whether the Sieder Tate expression truly captures the physics of interest or whether the agreement seen in the current work is simply reflective of the single turbulence model employed.