March 2, 2011, 03:57
|
|
#11
|
Senior Member
Ivan Flaminio Cozza
Join Date: Mar 2009
Location: Torino, Piemonte, Italia
Posts: 210
Rep Power: 18
|
Hi Bernhard, and thank you for your answer.
Quote:
Originally Posted by gschaider
I trust you had a look at the Wiki-page T.D. gave.
|
Yes, I take a look there sometimes ago.
Quote:
Originally Posted by gschaider
So at least you stumbled upon normal() for the normal vecotor. U.snGrad() should give you the surface normal of U, but that is not what you want.
|
Correct, I don't need U.snGrad(), or at least not only it.
Quote:
Originally Posted by gschaider
Another problem is your tau. For a patch as it is 2D you'll always have 2 orthogonal tangential directions so you'll probably want to reformulate (x dot tau) as mag(x-normal()*(normal() dot x)).
|
Again correct, in my case tau is the tangential vector aligned with the flow velocity U, so (U dot tau) = mag(U-normal()*(normal() dot U)).
Quote:
Originally Posted by gschaider
Did I understand your formulation correctly that if you can calculate U_slip correctly from the current flow conditions, then you can set it on the boundary as a Dirichlet? Otherwise you'll have to reformulate it. But as I said: you'll have to play around (not sur how to implement the d/dTau in you equation) and I can't guarantee success
|
What I want to do is set the slip velocity at wall, so at the end a Dirichlet b.c. for U is possible. On the other end, for stability reasons I need to evaluate U_slip in an implicit way, so I can't just discretize the derivatives and solve.
For d(U dot n)/dtau, maybe it could be rewritten as grad(U dot n) dot tau
Any other ideas? I definitely hope that groovyBc would do the trick...
Thanks, Ivan
|
|
|