CFD analysis and experimental validation of steady state mixed convection sodium flow

Formation and destruction of thermal stratification can occur under certain flow conditions in the upper plenum of sodium cooled fast breeder reactors (SFR). The flow patterns in the hot sodium pool of SFR upper plenum are very complex and include zones of free and wall-bounded jets, recirculation and stagnation areas. The interaction of sodium flow and thermal stratification has been analyzed experimentally at CEA in the SUPERCAVNA facility. The facility consists of a rectangular cavity with heated side walls. The flow is driven by a wall-bounded cold jet at the bottom of the cavity. Experimental data of the temperature distribution inside the cavity are available for steady-state and transient flow conditions. In the present study, the steady state experiments are analyzed with the CEA CFD reference code TrioCFD and the commercial code FLUENT employing Reynolds Averaged Navier-Stokes (RANS) equations. The SUPERCAVNA modeling is based on preliminary separate effects’ studies of wall-bounded jets and free convection benchmark simulations for low Prandtl number fluids. It was found that a two-dimensional treatment is sufficient to reproduce correctly the measured thermal stratification for steady-state SUPERCAVNA experiments. However, it is necessary to take into account conjugated heat transfer between cavity sodium flow and side-wall heating channel. Using temperature-dependent physical properties was also found to be an important factor in simulating the experiments correctly. Applying the Boussinesq approximation to study the impact of buoyancy on the vertical flow momentum was found to be justified. Turbulence modelling is necessary for a successful simulation of the experiments. For practical reasons, the analysis was restricted to turbulence models of the k-ε family. High Re k-ε models, in either standard, realizable or RNG formulations, do not lead to significant differences in the calculated temperature fields. Non-linear eddy viscosity modelling might improve the quality of the simulation.