[macfly]

I obtain a wide range of y+ values (5-200) with a coarse mesh, realizable k-e and scalable wall function. And there is lots of y+ values (40 %) in the buffer layer, which is supposed to be pretty bad.

Is it a good idea to use scalable wall function and refine only the first layer of near-wall mesh in order to obtain 100 % of y+<11? Heat transfer will be included in the model eventually. I can't afford a fine mesh with y+<1 and 10-20 layers on the walls. Geometry is too complex and conditions are changing significantly in time, so adjusting the mesh locally is not an option either.

Thanks ahead for your suggestions

[Andrea1984]

Hi,

I know it is a quite an old post but I find myself in exactly the same situation. What approach did you follow?

Cheers
Andrea

[macfly]

In the end I used enhanced wall treatment because it blends linear and logarithmic laws. From theory guide 15.0 section 4.14.5.2: This formula also guarantees the correct asymptotic behavior for large and small values of y+ and reasonable representation of velocity profiles in the cases where y+ falls inside the wall buffer region (3 < y+ < 10).

[Andrea1984]

I was thinking to do the same. What do you reckon is the minimum number of prism layers to be used with this approach? I am aware that if you want to integrate NS equations up to the wall (i.e. EWT with first cell y+ at about 1) you need at least 10 cells within the boundary layer.

However this not the approach I am going to follow, since most of my cells are in the log layer (y+ of about 60) and I only have three prism layers. Is that enough with EWT? and with scalable wall functions?

Andrea

[rolloblues]

Hi all,

since (if I got it right) near wall analysis it's not a goal for you, maybe you could just coarsen the mesh in order to have y+>30 everywhere and then use the standard wf.