dynamicRefineFvMesh on polyhedral meshes
 

Hi everyone!

It's been a while since my last forum post, I was extremely busy wrapping up my PhD thesis. If time allows me, I will increase my message board involvement! Now the problem.

I am interested in using automatic refinement on arbitrary polyhedral meshes, which can be generated using ansys, snappy and some other tools. However, when running interDymFoam on the damBreak utility the following error occurs:

============
Selected 1178 cells for refinement out of 7854.
hexRef8 : Dumping cell as obj to "/media/data/OpenFOAM/frank-1.6/run/tutorials/multiphase/interDyMFoam/ras/damBreakWithObstacle/cell_9.obj"

cell 9 of level 0 does not seem to have 8 points of equal or lower level
cellPoints:6(28 29 30 31 32 33)
pointLevels:6{0}
============

Has anyone used dynamicRefineFvMesh before? Did anyone encounter and solve this problem?

Best regards,
Frank Bos

I believe dynamic refinement is currently supported only for 3D meshes with hexahedral cells. That would explain the error - it expects cells containing exactly 8-points for octree sub-division.

Polyhedral mesh refinement is not a trivial problem. I think it might help to decompose polyhedral cells into tetrahedra, refine/de-refine/optimize, and then agglomerate appropriately. But to do this automatically and intelligently would be quite an accomplishment indeed.

Hope this helps.

Also face this problem: I intended to refine initially coarse grid according to my solution, but it fails on tet, prisms and pyramids, all that don't have 8 vertices should not pass)))

Is dynamicRefineFvMesh improved in OF1.7?

Dear Foamers,

I also face the same problem using adaptive/ dynamic mesh refinement in Foam (here OF version 3.0.1). The next figures show the refinement of a hexahedral mesh (generated with Hexpress5.1) behind a foil.
This is the original mesh:
https://s31.postimg.org/cgxpifntz/original.png
And this is the refined mesh (refined cells chosen according to a vortex identification criterion; some dark grey cells fulfill this criterion too):
https://s31.postimg.org/n12jg79gn/refined.png

The dark grey colored cells mark 'protedtedCells' which are detected by 'dynamicFvMesh'. These cells contain hanging nodes and hence can be describes as 'polyhedral' cells. Besides, Foam seems not to be able to refine these cells. But: I really want to use some kind of adaptvie/ dynamic mesh refinement with this kind of grid.

Can someone help me out?

Regards,
Peter

I found the solution here in conjunction with snappyHexMesh :)
http://www.openfoam.com/version-v1606+/meshing.php
under "snappyHexMesh refinement and unrefinement".

Thanks a lot to OpenCFD Ltd. for their valuable extensions!

Best Regards,
Peter

Hello Everyone,
I have experienced similar problem with AMR on snappyhexmesh case :
The problem was that cells with refinement level 2 were immediately next to level 0 cell.

Doing some distance mode region refinement solved the problem :

The surface refinement is of level 3, buffer layers are of level 2.

Region refinement :
stlsurface
{
mode distance;
levels ((0.5 1));
}