Physical and numerical issues in the prediction of free wake hover performance

This paper concerns the persistent difficulty in predicting hover performance to a level of accuracy required for engineering analysis and design. The method used is a hybrid employing a standard RANS solver for the blade and an Eulerian potential flow solver with an embedded Lagrangian wake. The hybrid nature of the solver makes it possible to separately investigate wake and near-blade viscous effects. It was found that grid density changes in the wake solver lead to small changes in wake solutions, which in turn lead to more significant effects on computed rotor performance. This grid dependence of the wake solution was significantly reduced by slightly increasing the numerical core size of the embedded vortical wake. This increased core size was purely for numerical reasons, and had no physical effect on the self-induced convection velocities of the wake vortices. With the hybrid solution being grid independent, comparison with the model UH-60 performance data of Lorber et al is promising. However, computed wake trajectories are closer to the blade than in the data. Because the solutions are now grid independent, this inconsistency must result either from some physical modeling errors, or, equally likely, some physical features of the test that are not currently well understood.