[Jinjolee]

I tried to copy the directories in motorBike as a reference, and use them for the simulation for my own bike model. After modifying the dictionaries, I seperately executed command in terminal as follow:

>> surfaceFeatures
>> blockMesh
>> decomposePar
>> snappyHexMesh

I entered paraFoam and the mesh was successfully created as in the figure below.


I then tried to make a script similar with Allrun, but just with the meshing processes, namely "Meshrun". The Meshrun script is as follow:

--------------------------------------------------------------------------
#!/bin/sh
cd ${0%/*} || exit 1 # Run from this directory

# Source tutorial run functions
. $WM_PROJECT_DIR/bin/tools/RunFunctions

runApplication surfaceFeatures

runApplication blockMesh

runApplication decomposePar -copyZero
runParallel snappyHexMesh -overwrite
------------------------------------------------------------------------

However, for this script, I could not see the mesh for my model in paraFoam. Only the blockMesh exist, just like the image below.

There is no error appeared in log.snappyHexMesh.

Does anyone know what might be the possible cause? I could provide further information of my setup if required. Thanks a lot!!!

[Yann]

Hello Jinjolee,

If you have run this exact sequence :

>> surfaceFeatures
>> blockMesh
>> decomposePar
>> snappyHexMesh

Then it means you have run snappyHexMesh in serial mode. To run it in parallel you should run something like that, if you run it on 4 cpus :

mpirun -np 4 snappyHexMesh -parallel

If you run snappyHexMesh in serial without the "overwrite" option, the resulting mesh will be written in timeSteps folders (i.e: 1,2,3) for each snappyHexMesh step (castellated, snap, layers)
If you run is in serial with the overwrite option, the mesh will be written in constant/polyMesh. (and it will overwrite the initial blockMesh mesh)

When running snappyHexMesh in parallel, it works exactly the same way, except the directories 0,1,2,3 and constant will be written in every "processor*" subdirectory.

In paraView, you can choose whether you want to load a reconstructed (serial) case or a decomposed (parallel) case by selecting it in the "Case Type" option of the properties tab, before clicking on "Apply".

In your case, switching the "Case Type" from "Reconstructed Case" to "Decomposed Case" in paraView should allow you to visualize the mesh generated by your Allrun script. If you don't have the "Case Type" option, you might need to run "paraFoam -builtin" instead of "paraFoam".


Yann

[Jinjolee]

Originally Posted by Yann

Hello Jinjolee,

If you have run this exact sequence :

>> surfaceFeatures
>> blockMesh
>> decomposePar
>> snappyHexMesh

Then it means you have run snappyHexMesh in serial mode. To run it in parallel you should run something like that, if you run it on 4 cpus :

mpirun -np 4 snappyHexMesh -parallel

If you run snappyHexMesh in serial without the "overwrite" option, the resulting mesh will be written in timeSteps folders (i.e: 1,2,3) for each snappyHexMesh step (castellated, snap, layers)
If you run is in serial with the overwrite option, the mesh will be written in constant/polyMesh. (and it will overwrite the initial blockMesh mesh)

When running snappyHexMesh in parallel, it works exactly the same way, except the directories 0,1,2,3 and constant will be written in every "processor*" subdirectory.

In paraView, you can choose whether you want to load a reconstructed (serial) case or a decomposed (parallel) case by selecting it in the "Case Type" option of the properties tab, before clicking on "Apply".

In your case, switching the "Case Type" from "Reconstructed Case" to "Decomposed Case" in paraView should allow you to visualize the mesh generated by your Allrun script. If you don't have the "Case Type" option, you might need to run "paraFoam -builtin" instead of "paraFoam".


Yann

Thanks a lot. It works.

[Jinjolee]

Originally Posted by Yann

Hello Jinjolee,

If you have run this exact sequence :

>> surfaceFeatures
>> blockMesh
>> decomposePar
>> snappyHexMesh

Then it means you have run snappyHexMesh in serial mode. To run it in parallel you should run something like that, if you run it on 4 cpus :

mpirun -np 4 snappyHexMesh -parallel

If you run snappyHexMesh in serial without the "overwrite" option, the resulting mesh will be written in timeSteps folders (i.e: 1,2,3) for each snappyHexMesh step (castellated, snap, layers)
If you run is in serial with the overwrite option, the mesh will be written in constant/polyMesh. (and it will overwrite the initial blockMesh mesh)

When running snappyHexMesh in parallel, it works exactly the same way, except the directories 0,1,2,3 and constant will be written in every "processor*" subdirectory.

In paraView, you can choose whether you want to load a reconstructed (serial) case or a decomposed (parallel) case by selecting it in the "Case Type" option of the properties tab, before clicking on "Apply".

In your case, switching the "Case Type" from "Reconstructed Case" to "Decomposed Case" in paraView should allow you to visualize the mesh generated by your Allrun script. If you don't have the "Case Type" option, you might need to run "paraFoam -builtin" instead of "paraFoam".


Yann

BTW, after I successfully reconstruct the mesh, I created a slice and got these:


Is this a poor mesh? I do not know how to identify a "good" mesh.
Thanks a lot!!