[SuperSonicCat]

Hello forum users,


there is a question for which i cant find an answer in literature:


I have a High- Re- turbulence-model and a wall function which switches between vscous sublayer (U+=y+) and log-layer at y+ =11,225 (Launder&Spalding or Enhanced). From literature i know, that i should place the first node in the log-law-region 30<y+<300. But since i have this wall-functions which switch between linear and log- region, i am asking myself whether its possible to use a fine mesh such that y+<=3.
In my case i have very low y+- values (very thick boundary layer) and i have to use a wall function (LS-Dyna ICFD)
Has someone experience in such a case?
Sorry for my very bad english.


Best regards.

[sbaffini]

It really depends from what you mean by High-Re turbulence model. Unfortunately, this confusing, wrong terminology spreaded a lot in the past.

You either have a turbulence model that is able to solve the viscous sub-layer (i.e., work at low y+) or you don't. Also, some models can in theory but require some specific implementations to then also work in practice.

Spalart-Allmaras will work in either case without modifications. k-omega models will as well, but with modifications for the boundary conditions at walls.

Most k-epsilon models (standard, realizable, RNG) WON'T work with low y+ values without modifications.

Now, if I can correctly guess what software you are using (Fluent), and the turbulence model (k-epsilon), you shouldn't really bother about any of this.

Still, I would probably switch to the k-omega family if there isn't any specific rationale for the k-epsilon in your case.