[songyi719]

1. When y+ is between 5 and 30, it locates cell in buffer region, which leads to incorrect representation of boundary layer. When y+ is >30, wall function simulates boundary layer, and when y+ < 5, boundary layer properties are directly calculated through N-S eq.


This is what I know, but i am confused about condition when y+ < 5


Lets say y+ is exactly 5, and have equal size grid all over the world, then third to sixth cell will locate in somewhere btw 15 and 30, and same buffer region problem would occur in those cell.


Is this because wall function is also derived from N-S equation but too fine mesh is required, and can be replaced by directly resolving near-wall calculation?



2. Also, does wall function automatically calculate y+ and determine whether to apply wall function to certain grid? It seems like it would have to, but we use postprocess command to achieve y+ in openfoam after solving.



Does it just approximate y+ and use it?



3. I am considering on switching my current RANS code to DES, and make it possible to switch between two modes with need. I am using k-w SST in openfoam10, and k-w SST DDES/IDDES doesn't exist in .org version, and I am considering to compromise with k-w SST DES.


However, If I add surface layer, it will work great for DES case but reduces accuracy in RANS due to aspect ratio.


But I can't make extreme fine mesh bc it will be too expensive for both memory and computation speed of almost-full LES, so it would lead to creating coarse mesh: which can't properly demonstrate near-wall flow and also create huge amount of grey-area.


What would be the best mesh option?

[LuckyTran]

The problem at hand is how to provide boundary conditions for the turbulence transport equations (i.e. kEqn and omegaEqn in Foam). What is commonly referred to in y+ in CFD refers to the y+ value of the cell adjacent to the wall. It is better to call it Wall y+ or y+ of first cell but too many people have already made the mistake to correct it. The 2nd, 3rd, 4th cells, etc. would obviously at some point cross the buffer region. Those are interior cells and do not need boundary conditions and the wall y+ criteria don't strictly apply. For interior cells you solve whatever is the transport equation for k and omega.


Wall y+ is indeed calculated at runtime on demand and updated every iteration.


It is good to do this feasibility study early so you don't waste time. If you can't afford to run DES, then just don't. Stick to things you can afford. Have you considered not doing DES?