CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM Running, Solving & CFD (http://www.cfd-online.com/Forums/openfoam-solving/)
-   -   High U and P forming on borders (http://www.cfd-online.com/Forums/openfoam-solving/67581-high-u-p-forming-borders.html)

hansel August 19, 2009 13:57

High U and P forming on borders
 
I'm trying to simulate VAWTs in 2d. My simulations will run for 1000's of iterations (100s of ms of simulated time) and eventually somewhere on the edge (not in a corner) the U and P starts to go crazy. It's like a black hole forms in my simulated universe, and there is no pressure or velocity track that would indicate the cause is in anyway the turbine being tested.

My boundary conditions are:

Inlet fixed uniform U
outlet fixed uniform pressure of 0
The turbine is moving wall velocity, fixed U of 0.
Top and bottom are empty
I used zeroGradient for every other factor not mentioned, for U, P, epsilon, and k.

Do my boundary conditions provide too much freedom? I'm using turbDyMFoam but the problem happens in icoDyMFoam also and I believe simple icofoam.

Can anyone suggest better boundary conditions for a simulated 2d wind tunnel with an object in it? Should I have used slip or uniform U on the sides?

I realize I haven't posted all the properties, cooefficients, etc. If they are needed, please ask and I will post them.

hansel August 30, 2009 18:02

I didn't get much help here, but through trial and error found that non-slip walls in a sufficiently large wind tunnel gave the best results. You can expect a high pressure cell in the corner, where the fixed velocity side meets the no velocity side. I was a little surprised that didn't blow up, but it seems to be tolerated by the solvers.

chiven August 30, 2009 22:28

Hi, dear Hansel, for outflows with zeroGradient for U don't use zeroGradient for transported quantities like k, epsilon and T. Please use inletOutlet. Otherwise things might explode when you get a backflow there.

best regards,
Chiven

hansel August 31, 2009 00:46

Quote:

Originally Posted by chiven (Post 227964)
Hi, dear Hansel, for outflows with zeroGradient for U don't use zeroGradient for transported quantities like k, epsilon and T. Please use inletOutlet. Otherwise things might explode when you get a backflow there.

best regards,
Chiven

Thanks for the pointer, that was probably my original problem. My sides were zeroGradient for U and everything else.


All times are GMT -4. The time now is 12:42.