[RodriguezFatz]

Dear all,

It is hard to find any help about the scale adaptive simulation here or anywhere else.
I am interested in low-Re boundary conditions at walls.
What I did is using the low-Re boundary conditions of the "non-sas" k-omega-sst model for k and omega, thus:

   // k:
   PIPE_WALL
    {
        type          fixedValue;
        value         uniform 1.0e-12;
    }

   // omega:
    PIPE_WALL
    {
        type             omegaWallFunction;
        value            uniform 10000; 
    }

k should be set to a very low value >0 and omegaWallFunction can be used, since it is both low-Re an high-Re boundary condition.

Now, what about "nut" and "nusgs"? These values don't have any equation to solve for. So first thing that comes in my mind: Why do they need b.c. at all? Is this something that is needed only for high-Re cases, when the calculated values are overwritten by a wall function?
What I found is, people seem to use "zeroGradient" for "nuSgs" and "zero" or "nutLowReWallFunction" for "nut".
But why not using "calculated" for both nut and nuSgs? They can be calculated, once omega and k are known by their equations. Why would I patch them to something different?

Can anyone help me, I am really stuck with that question!

[bmikuz]

Hi Phillip,

I see that it has been almost one year since your post here. Did you find out why there is also a b.c. for turbulent viscosity?
This paper could be useful for you:
http://www.tfd.chalmers.se/~lada/pos..._report_WF.pdf
It investigates low-Re and high-Re implementations of b.c. for turbulence models (mostly k-omega).

[RodriguezFatz]

I think I found some thesis (also from chalmers?) where it was stated, that the nut b.c. is not used at all and can be set to anything. I will try to find that paper...