Dear CFX Users!

I would like to model the dynamic of the opening of a butterfly-valve with a deformed mesh. A similar case can be found by the tutorials in help (the valve vibration) but in my case the mesh will deform more so the mesh quality is going to be very wrong. The further problem is that the mesh adaptation can't be activated in case of a transient process. Is there a possibility the suitable mesh quality to keep with hard mesh deformation?

Thanks in advance Roland

Hi,

Two options:

1) put the valve in a separate domain which only rotates and link it to the rest of the domain with a GGI.

2) When the distortion is enough such that the mesh quality is inadequate you generate a new mesh and interpolate onto this new mesh. You then continue running on this new mesh until that becomes too distorted and your simulation is a sequence of these runs.

CFX V12 has lots of new features for doing this type of simulation. It will be much easier.

Glenn Horrocks

Dear Glenn,

Thanks the quick help to You. Both method are very resourceful I have tried already these possibilities...but both solutions contain errors. In the case of the first one (with the rotating domain) the interface can make inexactness. Moreover there is a centrifugal force in the rotating domain, what is a right solution in the case of a fast rotating machine (for example a compressor) but I think that it is a incorrect approaching in the case of a rotating valve what revolves slowly... In the case of the second solution (remesh and interpolation etc...) my problem is that I always must know the location of the valve (the mesh deformation) in every timesteps. In the case of a dynamic simulation can't be known these locations because the flow is going to shape the defined locations of the valve. Is there so solution which doesn't contain these inprecisions inprecisions?

Thanks Roland

Hi,

GGI causes inexactness - CFD is just a numerical approximation of the equations and a GGI is just another aspect to that. Properly applied the GGI is fine, and in your case it is starting to look like the approximations inherent in the GGI are much more accurate than the alternative methods of doing this simulation.

Centrifugal/coriolis terms - These terms are a function of the rotational speed so if the rotation is slow the terms will be small. It is a valid approach for slow and fast rotations.

Remeshing - This can be done where the motion is not known beforehand but it is significantly more difficult to do.

Glenn Horrocks