OF-2.3.x: snappyHexMesh and AMI
 

Having problems with snappyHexMesh and rotating motions using AMIs. I have a case with a propeller, which is rotating inside of that AMI. What I am having trouble with is the interfaces and the minimum weights for the interpolation between them.

I made one mesh, which is running very good and the minimum weights for both interfaces are around 0.5 after a whole rotation. Most of the time the minimum is somewhere at 0.7-0.8. This mesh has 77712 faces on each interface. What I want to do now is to refine just the propeller surface. So I change exactly one parameter in my snappyHexMesh dict, to define a higher maximum level of refinement of my propeller surface. This results in a mesh, which of course has a lot more cells, but leaves the AMI (as before!) with 77712 faces. This makes me think, that my refinement didn't really effect the AMI And if it did change the faces of the AMI, it should be so little, that it doesn't make much of a difference for the rotation. But this is not the case. The refined mesh does not run, because after like 50 time steps there appear lots of "0" minimum weights on the interfaces.

To sum up:
Mesh 1: propeller in AMIs, which each have 77712 faces
Mesh 2: same as mesh 1, but with refined propeller; also 77712 faces on each interface

Does somebody have an idea, what I can do to get my refined mesh running? Do you think the interfaces are the same? Because I can't see a difference. It seems like I can't get my finer mesh running whatever I do. I tried to put single prism layers on the interfaces with an expansion ratio of 1, but that doesn't help either in that specific case.

How did you check it? move Dynamic Mesh? And —check AMI

Hi Tobi,

yes, I used the moveDynamicMesh solver to check it. At some point the weights drop down to 0.
What do you mean with check AMI? The problem is that I can't see any difference at the AMI from one mesh to another. They basically look the same. It seems difficult to find errors in the AMI.

Hi,

there is an option like:
which produces an additional output for you.
Furthermore save each time step and check your AMI rotation! (maybe you do not duplicated the points?)

If you do not find the error you can send me your case and I will check it.

Hi Tobi,

thanks for your quick answer!

I didn't know about that option. I'll try to use that on the not working mesh. I did duplicate the points using "mergeOrSplitBaffles -split". And it does work for multiple other simulations. There are just some specific cases, where the AMI doesn't work... and this is the strangest one, since the AMI of the other mesh has the same number of faces and is running fine.

I found a work around to get a finer mesh now. But I will still check my old mesh and report on what I could achieve.

Thanks for your help!

Hi Pete,

so you are very familiar with AMI mesh generations too. My suggestions is to check out your mesh using moveDynamicMesh with the additional option and save each moving step. Have a look to that time steps with paraview and check out if everything is okay.

Maybe one picture (if its possible for you) could be attached to see the interface or better would be the whole case to check out what is going wrong in your case.

Good luck,
kind regards

Hi Tobi,

I wouldn't say that I'm very familiar, but I have done some mesh generations in the past weeks. ;)

I can spot one difference in the AMIs of the mesh that runs and the mesh that doesn't run. As mentioned, both have the same number of faces, but the AMI that doesn't run seems to have some deformed cells at the top of the curve of the AMI from the back.
The problem is that I don't know how to influence this and why sHM is doing it for this mesh, but not for the other one.

Edit: Sorry, but I think that I can't provide the case, because of the propeller geometrie.

Best regards,
Andreas

Hi Pepe,

is the whole big domain (cylinder) rotating?
If you want you can upload you case to my server, then I can check it.

Kind regards

hi!

I have the same 0 weight AMI problem.
Are you solve this problem yet? Would you mind to tell me that what really the zero weight means? I still try to figure out how to fix this situation.

Moreover, I want to know the why the "mergeOrSplitBaffles -split" needed?
I build the case based on the tutorial case multiphase/interPhaseChangeDyMFoam/propeller, and that doesn't use the command mergeOrSplitBaffles.

Thank you so much.

If you move some special region, lets consider to rotate a cylinder domain in the mesh, then you have to duplicate the points on the interface and spliit them. Otherwise the mesh will be deformed!

Hi,

Thank you for reply.
Now I am using the following way to build my mesh.

I am using "createPatch" to create the AMI patch. Should I add the "mergeOrSplitBaffles -split" after that and before "setFields"?

or any other suggestion of the processes to build my mesh?