CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   OpenFOAM Meshing & Mesh Conversion (https://www.cfd-online.com/Forums/openfoam-meshing/)
-   -   [snappyHexMesh] snappyHexMesh not snapping perfectly to surface and failing checkMesh (https://www.cfd-online.com/Forums/openfoam-meshing/80444-snappyhexmesh-not-snapping-perfectly-surface-failing-checkmesh.html)

Greg Givogue September 26, 2010 15:58

snappyHexMesh not snapping perfectly to surface and failing checkMesh
 
4 Attachment(s)
Hi,

I've been trying to create a mesh for a pod geometry (similar to a fighter jet centerline fuel tank) using sHM. After several iterations I've got it to the point were the surface is fairly well represented. The geometry is represented by a pod.stl file (high degree of tessellation for the streamlined portion) and 66 face stl files that make up the support structure. I've also defined several edge features to optimize the refinement around the edges.

It takes about 4 GB of RAM to add and snap the mesh to the surface and then another 2GB when it moves onto the addLayers portion of sHM. In the end it produces about 440k cells. I'm using OF 1.7.1 in Ubuntu 10.04 (guest) in VMWare Fusion 3 (max 7.8 GB RAM).

I'm having 2 problems now with sHM.

The 1st is that the mesh fails checkMesh with 43 skewed faces. I've tried correcting this problem by changing the skew settings in meshQualityControls of sHM.

The 2nd problem is that there are a few spots where the mesh did not snap well to the surface. I've tried increasing both the nsmoothPatches and the nRelaxIter but that did not seem to help.

I've attached the sHMDict, checkMesh.log and mpirun.log files. If you require more information just let me know and I will post more stuff.

Thanks in advance for the help,

Greg

Greg Givogue September 27, 2010 12:27

I tried snapEdge on this and it only caused the mesh to look worse and created more errors in checkMesh.

ziad September 27, 2010 18:58

jmf fixed something similar with maxConcave = 90
 
Hey Greg, check this out http://www.cfd-online.com/Forums/ope...pyhexmesh.html

jmf seems to have solved a similar problem.

Ziad

Greg Givogue September 28, 2010 18:43

It turns out that the problems with the surface mesh and the skewed faces that are causing the meshCheck to fail are a result of addLayers. When I turn it off I get 25 skewed faces instead of 43 and the surface of my geometry looks really good. If I can't figure out why addLayers is doing this than I'll just turn it off and use other refinement tools. Any suggestions?

owayz February 28, 2011 10:45

Hi
 
I am also working with sHM. And I am also having the RAM problem. But I think thats because of the surface refinement and layer addition processes. And it depends on number of cells produced or the number of cells that the processor has to deal with during the process.
I have another problem. I have sharp features in my geometry and snapEdge seems to make things worse for me. But i think it worked before the layer addition process. As I added the layers it stopped working. Can you check if it works for you without layer addition.
Plus I just read a post which says that you can reduce the maxcanveness in mesh quality controls to make the mesh follow the sharp edges stringently. You can also try that.
Regards,
Awais

Greg Givogue February 28, 2011 11:43

Hi,

I understand your frustration, but I am no longer using sHM.

Instead of using snapEdge - try using edge refinement in sHM-features - edges... but first create a .eMesh file in the Trisurface folder to define the edge. A good explanation can be found if you search my other posts.

Good Luck!


All times are GMT -4. The time now is 08:56.