Prediction of Airfoil Characteristics With Higher Order Turbulence Models

The assessment of turbulence model performance in predicting ow elds that are directly relevant to industrial needs has become increasingly important. In aerodynamics, many problems exist for which the performance of some turbulence models is ambiguous at best and incorrect at worst. The application of many di erent turbulence models to a particular problem is a lengthy process that can lead to confusing results simply as a result of the volume of data produced. Given the motivation to develop turbulence models that do have some applicability to solving industry's problems, this study focuses on the prediction of airfoil characteristics, including lift and drag, over a range of Reynolds numbers. Two di erent turbulence models, which represent two di erent types of models, are tested. The rst is a standard isotropic eddy-viscosity two-equation K " model, and the second is an explicit algebraic stress model (EASM). The EASM model is an extension of the K " model because it introduces nonlinear dependency of the Reynolds stresses on the mean strain and rotation-rate tensors and, therefore, better accounts for the turbulent stress anisotropy more e ectively. The turbulent ow eld over a general-aviation airfoil (GA(W)-2) at three Reynolds numbers ranging from 2:1 10 to 6:3 10 is studied. Experimental pressure coe cients are compared with model predictions for Re = 4:3 10. At each Reynolds number, predicted lift and drag values at di erent angles of attack are compared with experimental results, and predicted variations of stall locations with Reynolds number are compared with experimental data. Finally, the size of the separation zone predicted by each model is analyzed, and correlated with the behavior of the lift coe cient near stall. In summary, the EASM model is able to predict the lift and drag coe cients over a wider range of angles of attack than the K " model for the three Reynolds numbers studied. However, both models are unable to predict the correct lift and drag behavior near the stall angle, and for the Re = 2:1 10 case, the K " model did not predict separation on the airfoil near stall.