|
[Sponsors] |
Looking for an example/help with manual decomposition |
|
LinkBack | Thread Tools | Search this Thread | Display Modes |
March 20, 2017, 14:10 |
Looking for an example/help with manual decomposition
|
#1 |
Senior Member
Abe
Join Date: May 2016
Posts: 119
Rep Power: 10 |
Hello Foamers,
I am trying to construct an internal mesh for a geometry that occupies only a small fraction of its bounding box. Please see the attached figure for an image of the .stl surface and blockMesh. It is one interior region, and I am using snappyHexMesh. As you can imagine, when using simple decomposition, many processors end up with a section of the base mesh that immediately gets thrown away during snappy. I tried to get around this by just decomposing in only one direction, but it seemed to cause an MPI nightmare because I got back an error about openfabrics from the cluster (it was a lame idea, but worth a try). I have been scouring the internet to try to figure out how to use manual coefficients, but I can't find and example of the format the file needs to be, and don't know the best way to break up the interior cells into chunks with minimal boundaries. If anyone has an example case file, or some advice, I am not even sure where to start. Thanks! |
|
March 21, 2017, 17:47 |
|
#2 |
Senior Member
Abe
Join Date: May 2016
Posts: 119
Rep Power: 10 |
I found that using decomposePar -cellDist writes a file that can be used, but I can't open / manipulate this file... It seems to be a labelList, but I am not entirely sure what that is. Does anyone know how to write a "labelList" file given a base mesh and some .stl files?
|
|
March 28, 2017, 00:06 |
|
#3 |
Senior Member
Abe
Join Date: May 2016
Posts: 119
Rep Power: 10 |
Well, if you are a lost soul who happens across this thread, here is a dirty work around for geometries with this sort of problem:
1) Set you snappyHexMeshDict to run only castellated mesh controls, and set all refinement levels to 1. 2) decompose with the simple method, and run snappyHexMesh 3) copy the resulting mesh to the constant folder (replace the base mesh), throw away processor* folders 4) set decomposeParDict to scotch. 5) set snappyHexMesh to the settings you want for the final mesh. All refinement levels must me greater than 1, or snappy will crash 6) decomposePar and run snappy again. Or alternatively, give up on your career and become an alcoholic - I am about one core dump from drowning in my own vomit. I would still like to know a cleaner way to do this - if anyone has suggestions, please feel free to share. |
|
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
How to build current blower fan model in Flotherm | eric0722 | FloEFD, FloWorks & FloTHERM | 3 | January 2, 2021 03:36 |
[snappyHexMesh] How to define to right point for locationInMesh | Mirage12 | OpenFOAM Meshing & Mesh Conversion | 7 | March 13, 2016 15:07 |
Manual Decomposition Method | smraniaki | OpenFOAM Pre-Processing | 3 | November 24, 2013 17:53 |
Manual decomposition of domain | pss | OpenFOAM Pre-Processing | 0 | April 26, 2012 02:33 |
Description file for manual decomposition | A.Devesa | OpenFOAM Running, Solving & CFD | 2 | July 4, 2011 09:09 |