CFD Online Discussion Forums

CFD Online Discussion Forums (
-   OpenFOAM Running, Solving & CFD (
-   -   Motion diffusivity solver has problems with patches moving toward each other (

bfa July 1, 2008 03:48

Hello community, I encountere
Hello community,
I encountered a strange phenomena using mesh motion and solving motion diffusivity. Here is my dynamicMeshDict:

dynamicFvMesh deformingBoundaryFvMesh;

twoDMotion no;
solver displacementComponentLaplacian y;

diffusivity quadratic inverseDistance (dynBoundaryBottom dynBoundaryTop);

frozenDiffusion off;

Here are some pictures of my case where you can see what's supposed to happen:

The next timestep produces something like this:

As fara as I could find out the cells are moved through each other, which shouldn't happen using the laplace motion solver! But I think it's due to the two opposing boundaries moving towards each other and the diffusivity solver works first on the bottom boundary and afterwards on the top one (or vice versa, no matter).

I tried different diffusivities, but to no account. My question is: How can I make the diffusivity solver solve for both patches at the same time or what other method could I use? I thought about reading diffusivity from file, making the middle axis stiff, but what are the values I need to provide to the diffusivity file?

Your help is greatly appreciated.
Have nice day

elorriaux May 14, 2009 10:49

Hello Bjoern,

I'm getting the same kind of issue. My problem is slghtly different since i have a fixed boundary (straight wall) and a moving boundary (FSI). Even if i use an inverseDistance diffusivity with the FSI patch, the problem occurs. So i don't think the problem comes from the diffusivity solver working on one patch after the other.

I've tried many motion solver and diffusivity models (almost all) without success. I would be interested if you have found one solution or if someone has a hint to deal with this kind of mesh motion.

Regards, Etienne.

richpaj July 8, 2009 21:35

I had a similar problem but managed to apparently overcome (or perhaps merely delay the onset of) the 'overshoot' by grading the mesh so that it was coarser near the fixed patch and finer near the moving patch. For reference I was using OF15dev, the solver laplaceFaceDecomposition, diffusivity = quadratic and frozenDiffusion=off. Perhaps worth a try.


All times are GMT -4. The time now is 00:58.