Disabling Wall Functions
 

Hello,
I am trying to simulate a mesh containing walls with y+-Values between 100 and 200, as well as walls with y+-Values well below 1 using kOmegaSST.
Of course this is not really good to begin with but my idea was to use fixedValue for k and zeroGradient for omega on the walls with the low y+-Values.
Unfortunately when I start the simulation the OpenFOAM seems to set the walltypes to wallfunctions on it's own. Is there some way to disable this?
Kind regards
Kaskade

Answer:

k
type fixedValue;
value 0; //some say you should use a value like 1e-11, but it doesn't seem to make any difference

omega
type zeroGradient;

nut
type calculated;
value $internalField;

If you don't set up a file for nut yourself, OpenFOAM will generate it and upgrade your walls with wallfunctions.

Hi,Onno:
Why the nut is set to be calculated instead of zerogradient? I remember that when no wall function is needed, the type should be zerogradient. Please correct me if I'm wrong

Xianbei

That was a long time ago.... right now I what say, you might as well leave the wall functions turned on and use nutUSpaldingWallFunction for nut.


@myself from 2,5 years ago: If your mesh is that bad/uneven -> fix that first.

Hi,Onno:
Thank you for your reply. As omega based turbulence model can solve the finer mesh with y+<3 or less, so no wall function is needed when the mesh is fine enough. What do you mean by saying 'turn on the wall function as well'?

Xianbei

In my original post I was asking for a way to turn of the wall function manually on walls with a small y+. But from I understand now, you don't need to turn off the wall functions, since omegaWallFunction and nutUSpaldingWallFunction can be used for small y+-values.

I see, you mean it's a kind of automatic wall function similar to the commercial codes like CFX, right? Just not long before, I was confused about the wall function that there is no explicit y+ threshold in the source code. So the nuSgsWallFunction is similar to that of nutSpaldingwallfunction, which is also based on Spalding's law, right?

It think it is discussed here: http://www.cfd-online.com/Forums/ope...megasst-2.html

Thank you very much. Now I know that the nut and nuSgs may be a generic wall function. So whatever the resolution of the mesh is , using a wall function will be better.