The impact of boundary treatment and turbulence model on CFD simulations of the Ranque-Hilsch vortex tube

In this work the relative impacts that different boundary conditions have on the temperature separation pre-dictions of 3D Computational Fluid Dynamics (CFD) models of the Ranque-Hilsch Vortex Tube (RHVT) are explored. Concurrently, in the interest of comparing to experimental results, the impact of including the exit plenums and inlet shroud in the computational models of temperature separation are examined. Simulations of the 3D model using the k -epsilon, k -omega, k -omega SST and Scale-Adaptive-Simulation SST turbulence models are presented and their relative impacts on flow structure and energy separation are discussed. The results of this study show that imposition of the inlet mass flow combined with mass flow rate at one outlet and pressure at the other leads to reasonable predictions, whereas multiple pressure conditions at RHVT boundaries leads to errors in inlet flow or mass flow split. It is also shown that each turbulence model predicts a unique eddy viscosity dis-tribution inside the vortex tube and that the length of the stagnation stream surface is negatively correlated to the (average) eddy viscosity. By this observation, it is noted that the integral results for energy separation are not sufficient to determine which turbulence model is best.

Automated summary

This work investigates the impacts of different boundary conditions on the temperature separation predictions of 3D Computational Fluid Dynamics (CFD) models of the RHVT. The inclusion of exit plenums and inlet shroud in the computational models is also examined. The study shows that reasonable predictions can be obtained by imposing the inlet mass flow combined with mass flow rate at one outlet and pressure at the other, while errors in inlet flow or mass flow split can occur with multiple pressure conditions at RHVT boundaries. Additionally, each turbulence model predicts a unique eddy viscosity distribution and the length of the stagnation stream surface is negatively correlated to the eddy viscosity. The integral results alone are not sufficient to determine the best turbulence model.