CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM (http://www.cfd-online.com/Forums/openfoam/)
-   -   Create coarser mesh for post-processing (http://www.cfd-online.com/Forums/openfoam/98093-create-coarser-mesh-post-processing.html)

Horus March 2, 2012 18:52

Create coarser mesh for post-processing
 
Hello!

My current mesh has about 23 M cells. It's fine for solving but somewhat too large for post-processing. Is there any possibility to coarsen an existing OpenFOAM mesh including results?

Thanks!

wyldckat March 2, 2012 19:51

Greetings Florian,

mapFields perhaps? http://www.openfoam.org/docs/user/ca...5-280002.1.6.3

Although I'm not sure that it can work from a high-res mesh to a coarse mesh, since it's usually done the other way around.

Best regards,
Bruno

gschaider March 3, 2012 10:36

Quote:

Originally Posted by wyldckat (Post 347380)
Greetings Florian,

mapFields perhaps? http://www.openfoam.org/docs/user/ca...5-280002.1.6.3

Although I'm not sure that it can work from a high-res mesh to a coarse mesh, since it's usually done the other way around.

Best regards,
Bruno

I think that shouldn't be the problem. What he is asking for is an autoUnrefineMesh-utility. To my knowledge there is no such thing. What I'd do is: write the patches as STL (one of the surface-utilities should do that). Then use snappyHexMesh to generate you a coarse mesh (doesn't have to be high quality too)

Horus March 5, 2012 05:19

I think mapFields would be so way to go. I've created a coarser mesh by setting the base size in my mesher (Spider) to a higher value. Curiously the mesher then avoided creating some patches it should have and consistent mapFields failed.

Ok, I'll be investigating that issue whith the mesher, thanks so far!

(if anyone knows about that behavior with Spider I appreciate a reply of course)

LVDH March 5, 2012 09:03

Hi,
I frequently do this.

Here is my approach:

1. I copy the case and call it _CM at the end.
2. I modify the blockMeshDict so it encloses a smaller volume. I do not need the farfield for visualizations.
3. I modify the snappyhexMeshDict. Here I keep a high refinement on the surfaces. I lower all other refinement levels by 2 (eg from 7 to 5). I only keep one cell between levels and a switch off the surface layers. This gets me from about 70e6 to about 8e6 cells.
4. I map the fields from the original case to the _CM case.
Done!


All times are GMT -4. The time now is 02:12.