Possible bug in nutRoughWall
I think there may be a bug in the rough wall function. Its been discussed a little in the thread:
But basically nutw is set to 0 every time step. This means that
scalar limitingNutw = max(nutw[faceI], nuw[faceI]);
always returns limitingNutw = nuw[faceI].
This in turn means that
*(yPlus*kappa_/log(max(Edash*yPlus, 1+1e-4)) - 1),
always returns nutw[faceI] = 2*limitingNutw
As a result the simulation is subjected to a Ks much less than specified.
In the test case attached the velocity should end up close to:
u = u_star/kappa*log((z+z0)/z0)
where in this case u_star = 0.625137, kappa = 0.4 and z0 = 0.01.
I've fixed the problem buy adding the following lines which evaluate the limitingNutw using the values nut in the cell next to the wall.
const tmp<volScalarField> toldnut = rasModel.nut();
const volScalarField& oldnut = toldnut();
scalar limitingNutw = max(oldnut[faceCellI], nuw[faceI]);
Yes, we fixed this a couple of days ago and are testing the fix at the moment and will push it into 1.6.x shortly.
|All times are GMT -4. The time now is 14:55.|