[ole]

Hi everyone,
I just started to use OpenFOAM and am currently studying the "Flow around a cylinder" tutorial. Suppose I want to see the effect of having the cylinder rotate. This would easily be configured by setting the boundary to moving wall and specifying the angular velocity. Is this possible? I have read about mesh motion mechanisms, but they seem like overkill in this case, since the mesh is not moving, only the cylinder wall is. Any help appreciated.

Thanks,
Ole

[pbohorquez]

That is an interesting problem. Could you use GGI boundary condition on the "inside" cylinder and keep us posted? You could try something similar to mixerGgi that lives into OpenFOAM-1.5-dev/tutorials/icoDyMFoam.
Thanks, Patricio.

[ole]

Using GGI changes (as far as I understand) the problem to an unsteady one. I am interested in the steady solution, just like in the tutorial for the cylinder flow, except the cylinder should rotate with a constant speed. This amounts to setting the tangential velocity of the cylinder boundary to some uniform value, and then solving the steady problem. I can set the velocity on the boundary uniformly to some constant vector in cartesian coordinates, but is it possible to set the tangential velocity to some constant?

[andrea.pasquali]

You can try with this BC in U file

cilindro
{
type rotatingWallVelocity;
origin (0 0 0);
axis (0 0 1);
omega 50;
}

[ole]

Works perfectly! Thank you.

[maddalena]

Hello everybody,
I am wondering if a simpleFoam with a rotatingWallVelocity BC simulation can replace a MRFsimpleFoam simulation. They both are steadyState but:

• simpleFoam + rotatingWallVelocity: update the velocity considering the rotation, thus modify ALL the velocity term in the NS equations;
• MRFsimpleFoam: add a source term to the NS equations (see here).

Anyone has studied the influence on results of these approach?
Thanks to those that will shed some light!

mad

[maddalena]

Ok, I find the answer with the help of a young woman far more expert in cfd modeling than me.

The rotatingWall BC applies the rotation only the flow that is in contact with the wall, since it is a boundary condition. This will propagate to the adjacent cells but the rotating effect will be under-predicted since these cells have their own velocity that has no rotation component. This can be ok if the wall is simple, i.e. there are no blades on the rotating body. This is ok for the rotating cylinder case of this thread.

On the other hand, the MFR approach considers as rotating the whole flow volume around the wall. This case is more adequate if the rotating body has blades. Indeed, the flow between two blades has almost the same velocity of the blades themselves.

The MRF approach is ok if the mean flow far from the rotating body is wanted, or if the blade number is high, since the steady field values can be obtained only. If the blade number is low, the pressure pulse generated by the blade rotation cannot be discarded and a dynamicMesh approach should be used.

Hope to be useful to someone else.

Cheers,

maddalena