Non-overlapping near-wall domain decomposition for turbulence modeling

The resolution of near-wall turbulent boundary layer is still a hot topic in fluid dynamics. It is well-known that the resolution of a tiny near-wall region requires the most computational time. Conversely, engineering approaches based on wall functions avoid detailed resolution of the near-wall region. However, they can be too inaccurate and empirical.
The non-overlapping domain decomposition method (NDD) proved to be quite efficient in tackling the problem in question when the model of the RANS equations is exploited. The proposed method is much more universal than conventional approaches based on the wall functions because it does not require any free parameters to be tuned. In this approach the computational domain is split into two non-overlapping domains: inner and outer regions. A key problem is related to the interface boundary conditions (IBCs). IBCs are obtained using the transfer of the boundary conditions from the wall to the interface boundary. It is important that they can be obtained without the solution of the problem in the outer region. This results in IBC that are nonlocal and of Robin type. This approach has been efficiently applied to the RANS-LES decomposition. It is demonstrated that the well-knwon log-layer mismatch is practically eliminated. The accuracy of prediction is much better than that of the coarse LES and comparable with the resolved LES. Finally, the extension of NDD to essentially unsteady problems are also discussed.