CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM Running, Solving & CFD (http://www.cfd-online.com/Forums/openfoam-solving/)
-   -   cyclic boundary with same geometric size but different number of faces? (http://www.cfd-online.com/Forums/openfoam-solving/99988-cyclic-boundary-same-geometric-size-but-different-number-faces.html)

strikeraj April 17, 2012 21:53

cyclic boundary with same geometric size but different number of faces?
 
Hi

I am just wondering if this is possible.
I have a 3D rectangular domain. It is meshed in ICEM with unstructured grid, save to fluent .msh and converted in openfoam using fluent3DMeshToFoam.
But when I try to run it, it says the number of faces does not match on the two patches that I set as cyclic. Is there a way to work this out without remeshing the faces with a structured grid? If not, does my way of generating and exporting mesh works when I put a structured surface mesh on both sides?

Thanks in advance

Cheers,
Tom

strikeraj May 6, 2012 16:16

sorry to bump this up but can anyone give me an idea on how to put on cyclic BC on an unstructured mesh?

Many thanks

danishdude May 7, 2012 10:50

The original cyclic condition requires point matched faces. However, the cyclicAMI condition allows you to use a cyclic condition without point matching. It's available in OF 2.1.

brunotojo May 8, 2012 12:44

Yes, it is possible. In ICEM you need to define the mesh as periodic in Mesh>Global Mesh Setup>Set up periodicity.

And then, after converting to OpenFoam format, you will need to use createPatch (there are a lot of post about this here) to define your periodic boundary.

I am also new to OpenFoam so, I am not sure if this will work for every case, but for me it did.

Best regards,
Bruno

strikeraj May 8, 2012 12:46

Thanks Michael and Bruno for your reply.
I will take a look into both methods after I finish my work on hand and I will let you know how it works out

strikeraj May 15, 2012 18:58

1 Attachment(s)
Quote:

Originally Posted by danishdude (Post 359760)
The original cyclic condition requires point matched faces. However, the cyclicAMI condition allows you to use a cyclic condition without point matching. It's available in OF 2.1.

Hi Michael

I have just tried my case with the cyclicAMI BC and it works if i run it serial.
But when I try to use decomposePar and run in parallel with mpirun, error occurs. Do you have clue on what might be wrong?

Thanks very much for your time

Cheers,
Tom

attached is the error log file

danishdude May 16, 2012 11:10

Any chance you can upload the case somewhere so I can take a closer look? If it's not too big, you can e-mail it to me at mahlmann@ucdavis.edu

alquimista May 21, 2012 07:02

Hello,

I had the same questions months ago and in my opinion the best option was to use the extend version of OpenFOAM. It has General Interface Grid implemented. There's more information available on internet.

https://cmg.soton.ac.uk/community/at...LloydAug11.pdf
http://www.tfd.chalmers.se/~hani/kur...dy_2010_OP.pdf

Good luck.

strikeraj May 21, 2012 10:43

Update:

I have posted this on the bug tracker and got a reply with working solution:

Quote:

The AMI assumes the patches are collocated - if not you have to specify the 'transform' type. (This does not affect serial running since it uses a different code path).

Add transform=translational and a separationVector to your constant/polyMesh/boundary file and it should work:

CYCLIC1
{
type cyclicAMI;
neighbourPatch CYCLIC2;
transform translational;
separationVector (0 0 -0.0914);
nFaces 9824;
startFace 593664;
}

CYCLIC2
{
type cyclicAMI;
neighbourPatch CYCLIC1;
transform translational;
separationVector (0 0 0.0914);
nFaces 11028;
startFace 615000;
}

swapnilsalokhe December 13, 2013 13:35

decompose par for cyclicAMI bc
 
Hi Tom Li,

When you are using any cyclic boundary in decomposePar you have to set both cyclic planes on one processor then only it will work on parallel processors.


Quote:

Originally Posted by strikeraj (Post 361306)
Hi Michael

I have just tried my case with the cyclicAMI BC and it works if i run it serial.
But when I try to use decomposePar and run in parallel with mpirun, error occurs. Do you have clue on what might be wrong?

Thanks very much for your time

Cheers,
Tom

attached is the error log file



All times are GMT -4. The time now is 18:43.