CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   OpenFOAM Meshing & Mesh Conversion (https://www.cfd-online.com/Forums/openfoam-meshing/)
-   -   [snappyHexMesh] Able to run snappyHexMesh in parallel on local machine but unable to run on linux clu (https://www.cfd-online.com/Forums/openfoam-meshing/139635-able-run-snappyhexmesh-parallel-local-machine-but-unable-run-linux-clu.html)

abhinav2601 July 28, 2014 09:48

Able to run snappyHexMesh in parallel on local machine but unable to run on linux clu
 
4 Attachment(s)
Hello everyone,
I am trying to solve a ship resistance calculation problem using interDyMFoam on OpenFoam version 2.3. The OS is Red Hat. The desired mesh size is between 3.5 million to 6 million for half-body( generated using blockMesh followed by snappyHexMesh).
The problem I am encountering is that: The same sequence of commands submitted through command line will generate the mesh( in parallel ) , initialize the fields and submit the solution (in parallel) and reconstruct the time directories. But when the same sequence of commands is submitted with a slight change to the same version of OpenFoam (2.3) installed on a RedHat based cluster I am getting some errors.

The error is the following: There is a
Code:

Time = 0
No mesh.

Message in both cases at the end of reconstructParMesh –constant
But in local machine the checkMesh right after this displays increased number of cells and a ‘hull’ patch. In case of the cluster the checkMesh at this stage returns the original number of cells(as after blockMesh) and no ‘hull’ patch.
Thus the solution can proceed correctly on the local machine but not on the cluster.

I am submitting the following sequence of commands through command line:
Code:

surfaceFeatureExtract > log.surfaceFeatureExtract
blockMesh > log.blockMesh
checkMesh > log.checkMesh_blockMesh
decomposePar –force –latestTime > log.decomposePar_blockMesh
foamJob –screen –parallel –snappyHexMesh –overwrite
mv log log.snappyHexMesh
reconstructParMesh –constant > log.reconstructParMesh
checkMesh > log.checkMesh_snappyHexMesh
setFields > log.setFields
decomposePar –force –latestTime > log.decomposePar_snappyHexMesh
foamJob –screen –parallel –interDyMFoam
mv log log.interDyMFoam
reconstructPar > log.reconstructPar

The above log files are attached as localmachine.zip


I am submitting the following sequence of commands through a queue script
Code:

surfaceFeatureExtract
blockMesh
checkMesh
decomposePar –force –latestTime
time mpirun –hostfile \$PBS_NODEFILE –np \$np snappyHexMesh –overwrite
reconstructParMesh –constant
checkMesh
setFields
decomposePar –force –latestTime
time mpirun –hostfile \$PBS_NODEFILE –np \$np interDyMFoam –parallel
reconstructPar

The foam.log file generated by the above commands is attached as cluster.zip

For better illustration the cluster script is also attached as queue.zip. The job is submitted using
Code:

./foam_combined.run -Np 8 -Both
I have also attached my 0.org, constant and system directories for a cleaned case as case.zip for further clarity.
Any help will be much appreciated.

Termo January 26, 2015 05:42

solution?
 
Did you find why you got the "No mesh"? I seem to get the same, and I can not see why reconstructPar doesn't work as it should.


All times are GMT -4. The time now is 03:03.