Evaluation of RANS k – ε calculations for turbulent stably stratified layers based on GEMIX experiments using the CUPID code

In the present study, the prediction capability of RANS k − ε turbulence models implemented in the CUPID code which has been developed by KAERI for nuclear safety assessment was evaluated based on GEMIX (GEneric MIxing eXperiment) experiments. The present study is a part of the work participating in the International Benchmark Exercise based on the GEMIX experiment done by OECD/NEA and PSI. The GEMIX experiment performed by PSI is focused on the basic turbulent mixing mechanisms in unstratified and stably stratified conditions. For both the unstratified and stably stratified conditions considered, the RANS k − ε turbulence model calculations show reasonably good agreement with experimental data for time-averaged axial velocity. However, the turbulent kinetic energy measured in the GEMIX experiment was not well predicted by RANS k − ε model, which becomes worse in the stably stratified conditions. The influence of grid resolution and the inclusion of buoyancy terms in k − ε model were not significant. Also, the different RANS models such as SST and RST showed similar results to k − ε model, which indicates the fact that the drawbacks of k − ε model such as the isotropic feature do not play significant role of underestimation of turbulent kinetic energy in the present GEMIX benchmark problems. On the other hand, with the tuning of the model constant C_μ based on the experimental data of k and uv¯, the prediction capability of RANS turbulence models is shown to improve the estimation of the turbulent kinetic energy in stably stratified conditions, which is consistent with the turbulence modeling theory reported in the literature. The present sensitivity approach for C_μ could be elaborated by a calibration approach in order to improve the prediction capability of the current RANS k − ε calculations. As has been seen by the RANS society, such a pragmatic modeling could be a successful means.