CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM Native Meshers: snappyHexMesh and Others (http://www.cfd-online.com/Forums/openfoam-meshing-snappyhexmesh/)
-   -   Cyclic BC with snappyHexMesh crashes in multiple processors (http://www.cfd-online.com/Forums/openfoam-meshing-snappyhexmesh/122288-cyclic-bc-snappyhexmesh-crashes-multiple-processors.html)

jgil9 July 25, 2013 09:05

Cyclic BC with snappyHexMesh crashes in multiple processors
 
Hi all,
I am new to Open Foam and Hope you can give me a little hand, my research requires me to model a steady state turbulent case using cyclic boundaries in parallel. Seemed to me that the case motorBike was a good start, and though it handles the cyclic BC in a single processor snappy crashes in multiple processors. I have tried preservePatchs but no luck. I have also read that cyclicAMI could be a solution... I tried and though I get past snappy it crashes in potentialFoam. I have been stuck here for some time and would really appreciate some help.
Thanks In Advanced

jgil9 August 7, 2013 07:11

Help with CyclicAMI
 
Hi Bruno,
I have been trying to Apply Cyclic or CyclicAMI boundaries to the motorBike case with no success. I a farily new to OpenFoam and for my research, I need to apply Cyclic boundaries in a decomposed domain for a k-omeaga model, therefore motorbike fit the bill except for the cyclic boundaries. I appreciate the guidance.
Sincerely
JC
:confused:

wyldckat August 16, 2013 07:02

Hi JC,

I've moved these posts of yours to a single thread, to make it easier to answer and to keep the discussion in a single thread ;)

Have a look into this thread: http://www.cfd-online.com/Forums/ope...tml#post392721 - namely post #9. There you'll see an example case on how you can add cyclic patches, after meshing with snappyHexMesh.
In other words, you first should mesh without cyclic patches and only after the mesh is complete is when you can add the cyclic patches to it.

Best regards,
Bruno

jgil9 August 31, 2013 04:34

Thanks!
 
Thanks Bruno, A couple of weeks or so after I posted the initial question I stumbled on an alternative solution. I was able to run snappy with cyclic boundaries but on a single processor and only decomposing before running the solvers. I will take a look at your suggestion though!
Do you then suggest to blockMesh, topoSet, decompose, run snappy in parallel, reconstructParMesh and the createPatch?

Thanks for the help, sorry for delayed response.
JC

wyldckat August 31, 2013 08:32

Hi JC,

Quote:

Originally Posted by jgil9 (Post 449002)
Do you then suggest to blockMesh, topoSet, decompose, run snappy in parallel, reconstructParMesh and the createPatch?

If you run this command:
Code:

createPatch -help
you'll see that you can also run createPatch in parallel, which means that you can run it before reconstructParMesh ;)

Best regards,
Bruno

jgil9 September 1, 2013 01:25

Case
 
1 Attachment(s)
Bruno,
Thanks for the tip :p, I have set up the case and ran it, but I have been getting an error in createPatch, Face 3616 specified in set is not an external face of the mesh. :confused: I believe I have followed the example case you recommended. I don't mean to impose and I and I a very great-full for the time you have already devoted to this thread but I attaching the case file and I would greatly appreciate your input.
Thanks
JC

wyldckat September 1, 2013 06:17

1 Attachment(s)
Hi JC,

The problem with creating the sets before using snappyHexMesh is that the locations of the face indexes will be completely lost, as shown in the attached image.
I used:
Code:

reconstructParMesh -constant
after snappyHexMesh ran in parallel.
Then ran paraFoam and chose to see only the sets.

Given how you are defining the base mesh, you have two possibilities:
  1. Either you change "blockMeshDict" to create the 6 walls as patches. Then after sHM, use createPatch to change these patches to walls.
  2. Or you can use topoSet and createPatch before sHM for creating those patches. Then after sHM, use createPatch again to change the patches to cyclics.
Best regards,
Bruno

jgil9 September 1, 2013 08:28

1 Attachment(s)
Hi Bruno,
Thanks! Still have not been able to crack this problem. Tried both suggestions but with no luck, I am attaching the case with suggestion no. 1. As per your recommendation I changed blockMeshDict and then tired to create new cyclic patches after running sHM, but have been running into a point matching problem (though I turned off pointsync, and/or relaxed the matchTolerance) I also played around with the separation vector.
Thanks JC

wyldckat September 1, 2013 13:26

Hi JC,

OK, not much time for seeing your case, so here are two examples:
Best regards,
Bruno

jgil9 September 1, 2013 17:50

1 Attachment(s)
Hi Bruno,
I can't thank you enough for your help... really! And I know you must be an extremely busy person (I respect that and try not to waste your time... I had read and tried to use cyclicAMI before, and though that using your suggestion of createPatch might do the trick). Using ciclicAMI the run gets passed sHM and patchSummary which is great :rolleyes: but crashed in potentialFoam with a continuity error :mad:

So only when you have time... here is the case file for cyclicAMI

A very great-full forum user!;)
JC

wyldckat September 7, 2013 14:53

1 Attachment(s)
Hi JC,

Unfortunately something strange is going on with this mesh.
Attached is what should work, but doesn't.

I suggest that you step back a few steps and start by trying to mesh without the motorbike, because I think the bike is introducing several issues into the mesh, leading createPatch to not be able to create the cyclic patches.

Best regards,
Bruno


All times are GMT -4. The time now is 17:11.