[dennis722]

hello,
i have some weird results when simulating a multipart and multistory airfoil in 2D. In 2D i always get a stall in the wing in the second floor. But in 3D everything is right.

I dont know what could be wrong here, i also have read the userbook.

I used the polyhedralmesher and the realizable k-epsilon model.

The firstpicture shows the 3D simulation with no stall.
The second picture shows the 2D simulation where the wing in the second floor has differente angle of attacks and positions.

Does somebody know what could be wrong? Does the turbulence modell assumptions things because its a simulation in 2D?

[dennis722]

2D simulation picture

[me3840]

This isn't completely unexpected. Motion in the 3rd dimension (vortices) can help delay separation a lot. In 2d you are relying merely on momentum in 2 directions to keep things attached.

ke is not the best for this type of simulation, kw-SST is generally better at predicting separation here.

Stall prediction is very hard to get right with RANS. Does it separate in the tunnel?

[dennis722]

I have tried everything. Other turbulencemodels, refining the mesh, tried other meshes, change the boundary coniditions of the symmetrie wall to a normal wall with slip. But nothing works, the support from star ccm says it isnt a numerical error. But i dont understand it because in 3D it is a symmetrical wing, so the shouldnt be forces in the z-direction..

[Blanco]

I would add that turbulence is definitely a 3D phenomena and you can't expect to correctly simulate vortexes in 2D, this comes from theory rather than software.
If you're using RANS then try k-w SST as suggested, keep y+ lower than 1 everywhere on the surfaces, try steady state and switch to transient if needed (check residuals + forces on the airfoils).

Hope this could help,
Regards

[me3840]

Originally Posted by dennis722

But i dont understand it because in 3D it is a symmetrical wing, so the shouldnt be forces in the z-direction..

Just because the total integrated force should be zero doesn't mean the local effects of momentum in the 3rd dimension are irrelevant.

[dennis722]

Originally Posted by Blanco

I would add that turbulence is definitely a 3D phenomena and you can't expect to correctly simulate vortexes in 2D, this comes from theory rather than software.
If you're using RANS then try k-w SST as suggested, keep y+ lower than 1 everywhere on the surfaces, try steady state and switch to transient if needed (check residuals + forces on the airfoils).

Hope this could help,
Regards

i tried it already but the 2D stall is still there. i already asked the support of star ccm and they say that it isnt a numerical error.
So i dont know what it could be..

[dennis722]

Originally Posted by me3840

Just because the total integrated force should be zero doesn't mean the local effects of momentum in the 3rd dimension are irrelevant.

can i find this written somewhere in a book? because thats for my bachelor thesis. I didnt find something in the star ccm+ handbook..

[me3840]

Originally Posted by dennis722

can i find this written somewhere in a book? because thats for my bachelor thesis. I didnt find something in the star ccm+ handbook..

This is pretty simple mathematics..

A function f(x) has integral F=0. Because F=0, does this demand that f=0 everywhere? No, obviously, half of f could be -1 and the other half 1, or an infinite number of other possibilities.

It would be a good idea to discuss this subject with your professor.