Unified turbulence models: From FDF to PDF, from LES to RANS

The development of methods which are capable of closing the bridge between established computational methods for turbulent flow calculations (e.g., DNS, RANS and LES) is important with regard to both theoretical and practical questions. Previous developments of such methods were performed on the basis of very different assumptions of heuristic nature, and even with opposite goals. It will be shown here that the consideration of this problem at the level of stochastic turbulence models represents a simple way to construct consistent unified models which can be applied at any level of resolution. Nevertheless, this does not mean that the performance of these models is the same at each level of resolution. This question and some related challenges will be discussed.