reconstructPar crash with "denser" mesh

Submitted by evansmuts on Mon, 06/03/2013 - 15:00

Hi

I am not sure if this is the cause of the problem, but a few days ago I reduced my CFD cell size to be about 3 particle diameters. The simulation runs fine in parallel (12 cores), but when I run reconstructPar, I an error when it tries to reconstruct the lagrangian fields (e.g. caseDir/processor0/0.1/lagrangian/). ReconstructPar still runs fine with my original mesh.

I couldn't find any obvious fix after searching online, so I have posted the last bit of the screen output below to see if any one has had this problem as well. I am running OF v2.1.1. and CFDEM v2.5.3.

...
Reconstructing surfaceScalarFields

phi

Reconstructing point fields

No point fields

Reconstructing lagrangian fields for cloud particleCloud

--> FOAM FATAL ERROR:
cell, tetFace and tetPt search failure at position (0.00438061 1.10333 0.000514725)
for requested cell 0

From function void Foam::particle::initCellFacePt()
in file particle/particleI.H at line 758.

FOAM aborting

#0 Foam::error::printStack(Foam::Ostream&) in "/home/cerecam/OpenFOAM/OpenFOAM-2.1.1/platforms/linux64GccDPOpt/lib/libOpenFOAM.so"
#1 Foam::error::abort() in "/home/cerecam/OpenFOAM/OpenFOAM-2.1.1/platforms/linux64GccDPOpt/lib/libOpenFOAM.so"
#2 Foam::particle::initCellFacePt() in "/home/cerecam/OpenFOAM/OpenFOAM-2.1.1/platforms/linux64GccDPOpt/lib/liblagrangian.so"
#3 Foam::Cloud::initCloud(bool) in "/home/cerecam/OpenFOAM/OpenFOAM-2.1.1/platforms/linux64GccDPOpt/lib/liblagrangian.so"
#4 Foam::Cloud::Cloud(Foam::polyMesh const&, Foam::word const&, bool) in "/home/cerecam/OpenFOAM/OpenFOAM-2.1.1/platforms/linux64GccDPOpt/lib/libreconstruct.so"
#5 Foam::reconstructLagrangianPositions(Foam::polyMesh const&, Foam::word const&, Foam::PtrList const&, Foam::PtrList > const&, Foam::PtrList > const&) in "/home/cerecam/OpenFOAM/OpenFOAM-2.1.1/platforms/linux64GccDPOpt/lib/libreconstruct.so"
#6
in "/home/cerecam/OpenFOAM/OpenFOAM-2.1.1/platforms/linux64GccDPOpt/bin/reconstructPar"
#7 __libc_start_main in "/lib/libc.so.6"
#8
in "/home/cerecam/OpenFOAM/OpenFOAM-2.1.1/platforms/linux64GccDPOpt/bin/reconstructPar"
Aborted

Thanks
Evan

a.carmona | Wed, 06/12/2013 - 16:48

I've got the same bug:

Reconstructing lagrangian fields for cloud particleCloud

--> FOAM FATAL ERROR:
cell, tetFace and tetPt search failure at position (0.356894 0.283645 -0.074995)
for requested cell 0

From function void Foam::particle::initCellFacePt()
in file /home/alvaro/OpenFOAM/OpenFOAM-2.2.x/src/lagrangian/basic/lnInclude/particleI.H at line 758.

FOAM aborting

#0 Foam::error::printStack(Foam::Ostream&) at ??:?
#1 Foam::error::abort() at ??:?
#2 Foam::reconstructLagrangianPositions(Foam::polyMesh const&, Foam::word const&, Foam::PtrList const&, Foam::PtrList > const&, Foam::PtrList > const&) at ??:?
#3
at ??:?
#4 __libc_start_main in "/lib/x86_64-linux-gnu/libc.so.6"
#5
at ??:?
Aborted (core dumped)

Regards,
Álvaro.

evansmuts | Tue, 06/18/2013 - 15:57

Hi Alvaro

Can you tell me some more about your model setup? What software versions are you using, what is you mesh cell size, volume fraction, etc.?

Evan

cgoniva's picture

cgoniva | Wed, 06/19/2013 - 16:33

Hi,

this is an annoying bug but not in our influence - OF's reconstruct algorithm for the lagrangian particles fails to locate them.

Please use the parallel (switch to decomposed) visualization in paraview.

Kind Regards,
Chris

evansmuts | Thu, 07/11/2013 - 14:00

Hi Chris

Do you know if this bug has been fixed in OF 2.1.x or 2.2.x? I am just wondering if it is worth upgrading from v.2.1.1.

Cheers
Evan

evansmuts | Mon, 06/24/2013 - 16:13

Hi Chris

Thanks for that idea, I will definitely give it a try.

I also found that increasing "writePrecision" (in controlDict) from 6 to 8 allowed some simulations to get to the second save point before they crashed with reconstructPar. So, maybe increasing this value more can help in some cases... I still need to test with higher values though...

Thanks for the help.

Cheers
Evan

evansmuts | Thu, 07/11/2013 - 17:47

Hi

I just discovered that one can use reconstructPar with the "noLagrangian" flag (i.e. type: reconstructPar -noLagrangian). This will reconstruct all files, EXCEPT those in the "lagrangian" directory in the time folder (i.e. the "positions", "r" and "v" files).

This works fine for me as I don't look as these file (I use the VTK data for the particle information). Or is there some interesting information I can get from these files?

Cheers
Evan