CFD Online Discussion Forums

CFD Online Discussion Forums (
-   OpenFOAM Running, Solving & CFD (
-   -   Inviscid 2D Airfoil Case (

doug May 15, 2007 12:37

I am running an inviscid case
I am running an inviscid case for a 2D airfoil and am running into some snags. I'm using the potentialFoam solver and have tried to mimic the potentialFoam/cylinder case in the tutorials directory. However, my boundary conditions must be a little off because the potential flow field I get as the solution is not correct. I'm attaching a picture of the resultant flowfield. If anyone has any suggestions, I would greatly appreciate them.



vinz May 17, 2007 17:35

Hi Doug, Reading your threa
Hi Doug,

Reading your thread I'm not really surprised you get velocity vectors perpendicular to the exit plane. Actualy, it comes straight forward from your boundary condition which is zeroGradient at the exit.
The zeroGradient boundary condition trough a plane will always give you vectors perpendicular to this plane. Indeed if vectors were not perpendicular, you would end up with a difference in velocity between two points side to side the exit boundary plane. And so, you would have a gradient.

If you really want to have vectors with 45°angle at the exit (which I'm not sure it's correct), you have to prescribe them as you do at the entrance.

Hope that helped,


doug May 17, 2007 17:44

Vincent- Doesn't the zeroGr

Doesn't the zeroGradient velocity boundary condition constrain the velocity vector on one side of the boundary to match the velocity vector on the other side of the boundary? This shouldn't constrain the velocity vector to be perpendicular to the boundary. It should just constrain the difference in velocity vectors from one side of the boundary to the next to be zero.

Also, specifying the velocity field at the exit doesn't allow for mass conservation across the domain. Therefore, I cannot specify the exit velocity profile. I realize the velocity profile at the exit won't be exactly uniform, but it should be very close to uniform because the flowfield is based off of potential flow. Right?

Thanks for your help.


7islands May 17, 2007 19:31

Hi, What about the lateral bo
What about the lateral boundaries? If they are of fixed wall types the outflow naturally gets perpendicular...


doug May 18, 2007 11:16

Takuya- Do you mean the upp

Do you mean the upper and lower sides of the parabola? I'm using a C-grid, so there is a parabolic shape around the front of the airfoil. Yes, the entire length of that parabola is set to a constant velocity and zeroGradient pressure. Is that correct for inlet/outlet boundary conditions of this type? I can't imagine that this is causing the exit velocity to align perpendicular to the exit boundary. The parabola acts as a velocity inlet and is not set as a wall. Therefore, it should not cause the flow to be forced out perpendicular to the exit boundary.

If by lateral boundaries you mean the walls in the z direction (this case is 2D), I have those set to empty as I believe they should be.


richard April 15, 2008 03:51

Hi all, I am wondering we d
Hi all,

I am wondering we deal with the kutta-joukowski condition at the trailing edge in OF potential solutions?

Any help would be greatly appreciated,

bb_ September 9, 2010 07:12

Hello all,

I'm having the same issue with outlet velocity vectors that seem to be forced in a perpendicular direction to the outlet patch face.

I'm dealing with a stator blade which is part of a cascade. Thus, I use periodic boundaries with translational periodicity in the y-direction. Due to this periodicity, inlet and outlet boundaries are planar in y-direction with normal facing in x-direction.
The flow enters the domain with a certain angle to the x direction (achieved with pressureDirectedInletOutletVelocity boundary condition for the velocity at the inlet).

Due to the deflection of the flow caused by the blade, the flow should exit the domain with a certain angle, which differs form that at the inlet face (but is unknown a priori, thus it's not possible to give this direction as condition...).

All simulations I've done so far show that the velocity vectors at the outlet are forced to be in positive x-direction, perpendicular to the face. This is not what one would physically expect and I'm very confused about that behaviour. Does anybody else have some experience with these kinds of problems? Do I have to use different outlet boundary conditions?

Many thanks in advance!

darioth September 9, 2010 08:49

Hi Doug!
Did you fix your problem with the boundaries?
Im having similar problems with a 0015 in the trailing edge with high velocities (60% and 70% over Uinf). The rest of the flowfield appears to be ok except the t.e.
Thanks, regards!!

bb_ October 13, 2010 07:06

Hi all,

just to let you know: I fixed the problem of forced perpendicular velocity vectors at the outlet. The problem was something completely different: I had to recompile my gcc with another set of optimization flags. Apparently they have a significant influence on the outcome of the simulation results. I might be wrong on that, maybe it was not only the optimization but also the compiler version. Anyway, the problem has been resolved for me now. Maybe you want to have a look on your compiler and compiler flags if you face similar problems.

All times are GMT -4. The time now is 20:35.