[kkpal]

dear foamers,
I am now running cases about flow induced vibration of a circular cylinder at Re=5000 with LES model. The solver is pimpleDyMFoam.
It runs well for the cases in which the amplitude of the cylinder is small. However, when it comes to the large amplitude cases, i.e., lock-in region, the following error message pumps out suddenly (by suddenly I mean the previous time steps are all good in terms of Cd, Cl, displacement, maxCo and so on ) and terminates the simulation.

   Courant Number mean: 0.0407862 max: 2.50865
Time = 109.01


Restraint verticalSpring:  attachmentPt - anchor (0 -0.257633 0) spring length 0.257633 force (-0 3.25133 -0)
Centre of mass: (0 -0.257633 2)
Linear velocity: (0 -0.849337 0)
Angular velocity: (2.0932e-06 -1.04789e-07 -2.56901e-08)
[4] processorPolyPatch::calcGeometry : Writing my 25 faces to OBJ file "/work/Re5000VIV/middle/processor4/procBoundary4to3throughfront_faces.obj"
[4] processorPolyPatch::calcGeometry : Dumping cell centre lines between corresponding face centres to OBJ file"/work/Re5000VIV/middle/processor4/procBoundary4to3throughfront_faceCentresConnections.obj"
[3] processorPolyPatch::calcGeometry : Writing my 25 faces to OBJ file "/work/Re5000VIV/middle/processor3/procBoundary3to4throughback_faces.obj"
[3] processorPolyPatch::calcGeometry : Dumping cell centre lines between corresponding face centres to OBJ file"/work/Re5000VIV/middle/processor3/procBoundary3to4throughback_faceCentresConnections.obj"
--------------------------------------------------------------------------
MPI_ABORT was invoked on rank 4 in communicator MPI_COMM_WORLD 
with errorcode 1.

NOTE: invoking MPI_ABORT causes Open MPI to kill all MPI processes.
You may or may not see output from other processes, depending on
exactly when Open MPI kills them.
--------------------------------------------------------------------------

What does this error message mean?

[wyldckat]

Quick answer: Looks like a physically impossible situation has occurred, probably where the mesh broke inside the cylinder.
In addition, this:

Courant Number mean: 0.0407862 max: 2.50865

is a dangerous situation, since this roughly means that the mass content of one cell can travel to 2-3 cells in a single time step, which would possibly explain why the mesh collapsed into itself.

When in doubt, save more time snapshots near the time of the crash, so that you can visually diagnose what the mesh looks like in each time step.

[davibarreira]

Like Bruno said, it's probably that your mesh deformed too much. Look in Paraview in "surface with edges" mode, and see how the mesh cells are actually deforming. To solve this, you might increase the "outerDistance" in your dynamicMeshDict

[kkpal]

Dear Davi,
I think it might be associated with the mpi communications. I reconstructed the case at the last written time and run the case again with single processor, the same simulation run smoothly past the time at which mpirun stopped.
Besides, when I change the span-wise boundary condition of the computation domain from cyclic to slip, and run in parallel, the problem did not show up.

Originally Posted by davibarreira

Like Bruno said, it's probably that your mesh deformed too much. Look in Paraview in "surface with edges" mode, and see how the mesh cells are actually deforming. To solve this, you might increase the "outerDistance" in your dynamicMeshDict

[wyldckat]

Quick answer: From a blog post of mine: Notes about running OpenFOAM in parallel - see this post:

stop when I run in parallel - post #28 provides a list of known reasons why a run might not go well in parallel

[liuliyun]

Originally Posted by kkpal

dear foamers,
I am now running cases about flow induced vibration of a circular cylinder at Re=5000 with LES model. The solver is pimpleDyMFoam.
It runs well for the cases in which the amplitude of the cylinder is small. However, when it comes to the large amplitude cases, i.e., lock-in region, the following error message pumps out suddenly (by suddenly I mean the previous time steps are all good in terms of Cd, Cl, displacement, maxCo and so on ) and terminates the simulation.

   Courant Number mean: 0.0407862 max: 2.50865
Time = 109.01


Restraint verticalSpring:  attachmentPt - anchor (0 -0.257633 0) spring length 0.257633 force (-0 3.25133 -0)
Centre of mass: (0 -0.257633 2)
Linear velocity: (0 -0.849337 0)
Angular velocity: (2.0932e-06 -1.04789e-07 -2.56901e-08)
[4] processorPolyPatch::calcGeometry : Writing my 25 faces to OBJ file "/work/Re5000VIV/middle/processor4/procBoundary4to3throughfront_faces.obj"
[4] processorPolyPatch::calcGeometry : Dumping cell centre lines between corresponding face centres to OBJ file"/work/Re5000VIV/middle/processor4/procBoundary4to3throughfront_faceCentresConnections.obj"
[3] processorPolyPatch::calcGeometry : Writing my 25 faces to OBJ file "/work/Re5000VIV/middle/processor3/procBoundary3to4throughback_faces.obj"
[3] processorPolyPatch::calcGeometry : Dumping cell centre lines between corresponding face centres to OBJ file"/work/Re5000VIV/middle/processor3/procBoundary3to4throughback_faceCentresConnections.obj"
--------------------------------------------------------------------------
MPI_ABORT was invoked on rank 4 in communicator MPI_COMM_WORLD 
with errorcode 1.

NOTE: invoking MPI_ABORT causes Open MPI to kill all MPI processes.
You may or may not see output from other processes, depending on
exactly when Open MPI kills them.
--------------------------------------------------------------------------

What does this error message mean?

Dear kkpal,

I use the Openfoam of version 4. the pimpleDyMFoam uses the RAS model. I have tried to modified the pimpleDymfoam to use the LES model, and failed for many times .
May I have your case repository about the VIV of cylinder, which uses the pimpleDyMFoam based on the LES .
ps: my email : liuliyun@sjtu.edu.cn

Best regards.
Your sincerely.