Influence of various aspects of low-Reynolds number turbulence models on predicting flow characteristics and transport variables in a horizontal direct-chill casting

Abstract

Computational Fluid Dynamics (CFD) was used to simulate the solidification in a horizontal direct-chill (HDC) casting using five low-Re k-ɛ turbulence model: including the v2¯−f model of Durbin (V2F), Yang-Shih (YS), Chang-Hsieh-Chen (CHC), Abid (AB), and Abe-Kondoh-Nagano (AKN). The model incorporated turbulence energy production due to buoyancy in the k and ɛ equations to predict flow pattern, eddy viscosity ratio, turbulence intensity, turbulence velocity, turbulence energy production due to buoyancy, solid fraction, and temperature. Three sets of grids were systematically chosen in such a way that the grid refinement factor was above 1.3. The results obtained indicate that the solution was independent of grid size. The flow pattern predicted using the turbulence models are generally similar, but the V2F indicated additional minor counter-clockwise recirculating cells toward the bottom mold. Further, the solutions indicate that turbulence energy production due to buoyancy contributes to the strong turbulence levels predicted towards the top mold region. The predicted velocity of the mean flow for the V2F model in the flow domain is higher, followed by the YS model and least for the CHC, AB, and AKN models. The eddy viscosity ratio is highest with a value of 144 at y+<100 for the CHC model, followed by AKN and AB with eddy viscosity ratios of 144 and 137 respectively. Then, the YS has a maximum eddy viscosity ratio of 114 at y+<50, V2F has the least maximum eddy viscosity ratio of 78 at y+<150. The turbulence intensity, the turbulence velocity, and the turbulence energy production due to buoyancy were highest towards the top mold for the CHC, AB, and AKN than the YS and V2F models. The V2F and YS predicted lower temperature curves and higher solid fraction curves towards the mid-section of the profiles than the CHC, AB, and AKN models.