The Effects of Time Step on Reynolds Averaged Navier-Stokes Simulations of High-Lift Flows

The linear finite element, time-implicit PHASTA flow solver was employed to perform Reynolds Averaged Navier-Stokes (RANS) simulations of the DLR F11 model studied for the 2nd AIAA CFD High Lift Prediction Workshop. The geometry is representative of a commercial aircraft in landing configuration, with full-span slats and Fowler flaps deployed. The effects of time and spatial resolution of the numerical computation on the predicting capability of the solver were investigated for the geometry at two angles of attack; namely 7° and 21°. At 7°, the geometry is in the linear section of the lift curve, however 21° is a post-stall stall condition for the F11 wing. A time step study was conducted with RANS and unsteady RANS simulations, as well as a grid refinement study with both uniform refinement and local, error-based adaptation.;Analysis of the results obtained from the time step study showed that at 7° angle of attack, time step does not have a significant effect on the predicting capability of the solver. However, for the F11 geometry at 21° angle of attack, time accuracy for both the transient phase and the limit cycle is important in order to avoid significant modeling errors in the solution due to the unsteady flow features present. The grid refinement study indicated that mesh refinement can be a valuable tool to improve the flow prediction, however overprediction of separation regions cannot be corrected. Furthermore, error-based, local adaptation was proved to be a more efficient approach compared uniform refinement of the entire grid.