[foamiste]

hello foamers,

I need to simulate a mixing tank, I have 4 processor, so I used the option decomposePar. But I am not sure if it's good decomposition or not.

Please find attached my decomposeParDict and screenshot of my geometry.

Any remarks or reply will be appreciated.

[foamiste]

No one has an idea about what I am saying?

[open]

Why not try one of the libraries that doesn't rely on the user e.g. scotch or metis

http://www.cfd-online.com/Forums/ope...mulations.html

[foamiste]

Thank you for your reply, when I use metis method I get this error :

--> FOAM FATAL ERROR:
You are trying to use metis but do not have the metisDecomp library loaded.
This message is from the dummy metisDecomp stub library instead.

Please install metis and make sure that libmetis.so is in your LD_LIBRARY_PATH.
The metisDecomp library can then be built from $FOAM_SRC/parallel/decompose/metisDecomp and dynamically loading or linking this library will add metis as a decomposition method.
Please be aware that there are license restrictions on using Metis.

From function labelList metisDecomp::decompose(const pointField&, const scalarField&)
in file dummyMetisDecomp.C at line 108.

FOAM exiting


And when I use scotch method, I think it's the worst decompostion that I can have, You can have a look in the attached screenshot.

[open]

I have to admit I'm a beginner so not sure how to interpret that. It doesn't look great though!

Try running checkMesh and compare results (also compare checkMesh/U/p post simulation).

Metis is 3rd party so has to be manually installed.