Issue with interpolation

[mostanad]

Hi there,
I hope you are doing well and thanks for developing this beautiful PFM version.

I am continuously working with the cfdemSolverMultiphaseScalar solver, and have faced an issue. For interpolating the temperature data from cell center to the particle position within that cell, you have utilized the interpolationCellPoint, which you can see in below:

CFDEMcoupling-PFM/src/lagrangian/cfdemParticle/subModels/energyModel/heatTransferGunn/heatTransferGunn.C

Line 319 in 71acedd

interpolationCellPoint<scalar> TInterpolator_(tempField_);

However, my parallel simulation using this solver and energy model suddenly gets stuck after a number of time steps. This behaviour has also been reported by the OF users getting stuck in parallel without any error:
https://www.cfd-online.com/Forums/openfoam-programming-development/195778-interpolationcellpoint-not-working-parallel.html
and:
https://www.cfd-online.com/Forums/openfoam-programming-development/156754-cannot-use-interpolationcellpoint-parallel.html

just to note that I'm using PFM with OpenFOAM 5.x version. And surprisingly the issue occurs with exactly that line, and not the interpolations above!

What is your approach for solving this? What do you recommend to me as my simulations get stuck?

Regards,
Mohammad

[mostanad]

Any idea on this from developers?!

[danielque]

We are using interpolationCellPoint in quite a few places throughout the code and did not experience that issue so far, though we mostly use OpenFOAM 6 in our department (on the other hand, from a quick glance, I did not notice any relevant changes in that class compared to OpenFOAM 5.x). The second forum link you gave hints at a parallelization issue where the interpolationCellPoint object is not created on all processes, but that should not be the case in our CFDEMcoupling code. Alas, at this point, I cannot provide any solution to your issue.

[mostanad]

Thank you, Daniel, for your reply. I hope you read this message, as I’m back with more details.

I believe nothing demonstrates my point better than an actual case. When I say it "gets stuck," I’m referring to cases that take a considerably long time (or a high number of time iterations). Perhaps these longer cases aren’t common in your department.
To show this, I’ve set up a simulation case with 800 particles using cfdemSolverMultiphaseScalar.

Please run this test.zip with the specified 32 CPUs for about 8 hours, if it's possible. For me, the simulation gets stuck at Time = 139 sec. This issue has been mentioned before in the CFDEM forum by developers, regarding the same style for interpolation variable:
https://www.cfdem.com/comment/11613#comment-11613

In the end, it’s up to you to decide whether to test this. But I’d appreciate it if you could give it a try. Who knows. Maybe I am wrong!
Test case: test.zip