|
[Sponsors] |
[snappyHexMesh] Inconsistent and Unstable SnappyHexMesh results |
|
LinkBack | Thread Tools | Search this Thread | Display Modes |
October 7, 2015, 19:47 |
Inconsistent and Unstable SnappyHexMesh results
|
#1 | ||||
New Member
Join Date: Jul 2015
Posts: 23
Rep Power: 10 |
All,
I am trying to complete the meshing of a cylinder tightly packed with particles. On few occasions, I get a stable mesh, which has been attached. However, if I were to use the same snappyHexMesh dict, while doubling the number of cells in the blockMeshDict (a=b=32, c=48), the simulation crashes before a single iteration, giving the following error (when not run in parallel, to keep it smaller): Quote:
Quote:
Quote:
I have tried running unset FOAM_SIGFPE, although the simulation still crashes, citing a nan input. I can typically generate meshes for fewer particles, but once I increase beyond 20, the majority of my meshes crash on first iteration. I have tried toying with snappyHexMeshDict and blockMeshDict, including fine grids to start, with few refinement iterations, or coarse grids to start, with many refinement iterations. Any ideas how to solve this issue? .stl files are available at the following link: http://s000.tinyupload.com/index.php...75582890393339 controlDict is below: Quote:
Any help/suggestions are appreciated! |
|||||
October 8, 2015, 13:12 |
|
#2 |
Member
DanielP
Join Date: Jan 2015
Posts: 33
Rep Power: 11 |
Hello Nicolas,
The crash seems to be coming when you try to solve the matrix to solve the pressure equation. The error seems to indicate that the solver is excepting an integer but is getting something else. The first step would be to check the mesh quality. Regarding the mesh, you can check the quality of the mesh with the following command: checkMesh -allGeometry -allTopology And compare between the two meshes (20 particles and more that 20 particles). Compare the logs. After that check your boundary conditions. Are the boundary conditions dependent on the number of particles ? Good luck, Daniel |
|
October 8, 2015, 17:57 |
|
#3 | ||||
New Member
Join Date: Jul 2015
Posts: 23
Rep Power: 10 |
Thanks for the tip!
I have been doing a standard checkMesh, the added parameters should be a useful tool. I will be comparing the stable and unstable results tomorrow. If you are interested/willing, I have attached the boundary conditions and checkMesh results. In both tests, boundary conditions are unchanged, and are listed below. I have tried to keep things simple. Quote:
Quote:
Quote:
Quote:
|
|||||
October 9, 2015, 09:22 |
|
#4 |
Member
DanielP
Join Date: Jan 2015
Posts: 33
Rep Power: 11 |
Good morning Nicolas,
What version of openFOAM are you using ? If you are using version 2.3.0, your dictionary file seems to have missing parameters.....Check the tutorials in your folder openFOAM/openFOAM-2.3.0/tutorials/mesh/snappyHexMesh/ I have notice that the number of regions is different between the two mesh.You have one region for the first mesh and 5 regions in the second mesh ? Also, there seems to have cells that are not connected. See the message from the second mesh. "The mesh has multiple regions which are not connected by any face." Daniel |
|
October 9, 2015, 11:43 |
|
#5 |
New Member
Join Date: Jul 2015
Posts: 23
Rep Power: 10 |
Hey Daniel,
Thanks for the analysis! The missing entries have been added to the snappyHexMeshDict. Do you have any idea why there are multiple regions being created? How can I control this beyond trial and error? Edit: solved by using splitMeshRegions -largestOnly. However, any insight you can provide as to why this is happening is appreciated! |
|
October 10, 2015, 11:52 |
|
#6 |
Member
DanielP
Join Date: Jan 2015
Posts: 33
Rep Power: 11 |
Hello,
One cannot be sure, but I would speculate that when snappyHexMesh tried to determine which part of the initial mesh (blockMes) was to be kept, some sections of the mesh where not connect, so it created different regions instead of one region (created by default if not specified). Does your simulation work now ? Thanks, Daniel |
|
October 12, 2015, 13:31 |
|
#7 |
New Member
Join Date: Jul 2015
Posts: 23
Rep Power: 10 |
Hey Daniel,
The splitMeshRegions -largestOnly worked, the resulting polymesh is used with no further modifications to any dict/boundary files. Thanks for the help! |
|
Thread Tools | Search this Thread |
Display Modes | |
|
|