RANS-Modeling. Concepts - Perspectives - Approaches

CFD-based aerodynamic design crucially depends on the capability of predicting turbulent flows, including separation. Limited resources and constraints on the response-times make methods based on the Reynolds-averaged Navier-Stokes (RANS) equations still indispensable for this purpose in industry. The increasing application of CFD towards conditions at the borders of flight is leading to a demand for improved accuracy for the prediction at least of flows with incipient up to mild separation. Inspired by recent discussions, the presentation will address the RANS-concept in terms of ensemble averaging. Eddy-viscosity as well as Reynolds-stress transport models are briefly described as different, but related flavors of RANS-modeling. Observed limitations are motivating DLR's strategy for model improvement. Assuming that a strictly universal model might be impossible, the idea of physics-based targeted modeling and self-adaptation is developed. First results for the plane mixing layer and its impact on reattachment are presented as well as remaining open questions