cyclic CFD and DEM working but not cyclic coupling (cfdemSolverIB)

Submitted by cigaletranquille on Fri, 07/14/2023 - 19:04

Hello,

My eventual goal is to run a resolved channel flow case using cfdemSolverIB, including periodic conditions. I've set up a simple falling sphere test to check that I've set up my periodic conditions correctly using cyclicAMI. I don't run into errors, but the results appear unphysical and about what I would expect if the periodic conditions were set up correctly for DEM and CFD sides individually, but not for the coupling. I assume it's something in my set up, which I based on the DKT example. I've seen others mention this issue but it's unclear in those cases how they resolved it, if they did.

I've set up a minimal example here: https://github.com/acatwithacomputer/CFDEMperiodicTest

It takes about 15 minutes to run on my machine. Any help is much appreciated.

-CT

Solution: I have done as Daniel suggested below and switched to using CFDEM-PFM. My problem seems to have resolved! Thank you!

AttachmentSize
Image icon view of velocity, void fraction, and interface fields5.56 MB
Plain text icon oscillations in the velocity as it crosses the boundary6.4 KB
Plain text icon (after fix) velocity no longer has the oscillations5.55 KB
Plain text icon (after fix) particle position showed as going beyond the box, so something to look for if anyone else runs into this issue5.68 KB

Forums:

cigaletranquille | Tue, 08/01/2023 - 21:36

I set up a little warning after line 93 of cfdemCloudIB.C:

Warning << "Approaching checkPeriodicCells..." << endl;
if(checkPeriodicCells_)
{
Warning << "checkPeriodicCells is true!" << endl;
.....
}

and have in my couplingProperties file


voidFractionModel IB;
locateModel engineIB;
....
IBProps
{
maxCellsPerParticle 100000;
alphaMin 0.30;
scaleUpVol 1.0;
checkPeriodicCells;
}
...
engineIBProps
{
treeSearch false;
zSplit 8;
xySplit 16;
checkPeriodicCells;
}

I'm still getting the same behavior. Moreover, while the solver does output the message "Approaching checkPeriodicCells...", I don't get the notice that it's true, which seems to be telling me that I've either set checkPeriodicCells incorrectly or I need to set it somewhere else and am unable to find documentation so far. Any help would very much be appreciated.

Daniel Queteschiner | Thu, 08/03/2023 - 11:20

checkPeriodicCells is neither a property of the IB voidfraction model nor or of the engineIB locate model as falsely claimed by the documentation.
It is rather a global property property such as solveFlow or couplingInterval, i.e. should be simply used like
couplingInterval 10;
checkPeriodicCells true;
// more settings ...

In case you are interested, we have examples of periodic flow in our version of CFDEMcoupling here: https://github.com/ParticulateFlow/CFDEMcoupling-PFM/tree/master/tutoria...

cigaletranquille | Thu, 08/03/2023 - 19:19

Ah, I see. Thank you; that is enlightening. I also had a look at some of the other examples in the github repository that you linked as well.

I put in the line checkPeriodicCells true (current version of the couplingProperties file here) . Two things:

1. The little message I set up in the if statement of the solver is showing up now, so that seems like progress.
2. I now get this new error

[2]
[2] --> FOAM FATAL IO ERROR:
[2] keyword wall_blockPeriodicityCheck is undefined in dictionary "IOstream"
[2]
[2] file: IOstream from line 0 to line 0.
[2]
[2] From function const Foam::dictionary& Foam::dictionary::subDict(const Foam::word&) const
[2] in file db/dictionary/dictionary.C at line 701.
[2]
FOAM parallel run exiting
[2]

I see in the examples that this wall_blockPeriodicityCheck is not set anywhere. May I ask if you may know what to look at next? And/or, since the setup appears a little different and I notice that this is for your modified code, may I ask what the differences are in the solver?

I'm doing this little test case to work out the periodic boundaries, but ultimately I want to simulate several particles in a channel.

-CT

cigaletranquille | Thu, 08/10/2023 - 02:31

Please disregard my previous post. I switched over to the PFM version and resolved the issue with your instructions. Sorry, I didn't realize that the code originally came out of this university group (or... that is my current very rough understanding).

Thank you so much!

amiya1202 | Fri, 08/25/2023 - 11:19

Can someone tell me the process for compiling the PFM version? And can we have both versions on your system?

Daniel Queteschiner | Sun, 08/27/2023 - 10:10

>> Can someone tell me the process for compiling the PFM version?
Installation instructions are given in the manual. A precompiled html version of the manual can always be found on the release site of the repository, i.e.
https://github.com/ParticulateFlow/LIGGGHTS-PFM/releases
https://github.com/ParticulateFlow/CFDEMcoupling-PFM/releases
For the latest release these are
https://github.com/ParticulateFlow/LIGGGHTS-PFM/releases/download/23.02/...
https://github.com/ParticulateFlow/CFDEMcoupling-PFM/releases/download/2...
The installation procedure is not too different from the DCS/PUBLIC version, though we recommend using cmake for compiling LIGGGHTS.

>> And can we have both versions on your system?
That shouldn't be a problem, you just need to source the correct environment variables to switch between versions.