EVALUATION OF THE PERFORMANCE OF 3 TURBULENCE CLOSURE MODELS IN THE PREDICTION OF CONFINED SWIRLING FLOWS

The performance of three different turbulence closure models in the prediction of turbulent swirling flows is presented. The models evaluated are: a nonlinear k-epsilon Model (NKEM), the Reynolds Stress Transport Model (RSTM), and the Algebraic Stress Model (ASTM). A decomposition of the Reynolds stresses into the isotropic (eddy viscosity) part and the deviatoric part is applied to avoid numerical instability. A bounded streamwise differencing scheme is used to discretize the convection terms in the governing equations, which substantially reduces the numerical diffusion and eliminates oscillation errors in the solution. Four different swirling flows of varying complexity are simulated. Comparisons of predictions with the experiments show clearly the superiority of the RSTM and the ASTM over the NKEM. The RSTM and ASTM provide good agreement with measured mean velocity profiles. However, the turbulent stresses are over-or underpredicted. The NKEM yields unsatisfactory prediction of the mean velocities and turbulent stresses.