A possible bug of Write_Start when coupling CFD-DEM

Submitted by NTT1508 on Wed, 09/14/2016 - 08:02

Hi all,

I have probably got a bug when running LIGGGHTS coupled with CFD. I found that the write_start command in LIGGGHTS does not work properly when they are working in parallel with CFD. Write_start works fine in single running of DEM, but it does not save any updated state when running through CFD. I suppose specialists of CFDEM project can shed a light on this issue.

Thank you,

Regards,

Nathan,

Forums:

alice's picture

alice | Tue, 09/20/2016 - 15:12

Hello Nathan,
I'm not entirely sure what you mean with write_start, but I assume you mean the write_restart command within LIGGGHTS?
If you want restart files to be written in a coupled calculation (coupled calculations can ONLY be run in parallel and never serially), you need to do this via a liggghts command model. Have a look at the turorial cases, e.g. cfdemSolverPiso/settlingTestMPI. You'll find the command writeLiggghts there.
Best regards,
Alice

NTT1508 | Mon, 09/26/2016 - 05:22

Hi Alice,

Nice to have your involvement. Absolutely I work well with Write_Restart on LIGGGHTS for single DEM running in many cases of my practice. Recently I tried with Write_Restart for coupling where Write_Restart was used in LIGGGHTS side but it did not work well. I further tested for 1 simple case where there is only a particle dropping in a fluid domain and found that this problem truly exists. For a better understanding, I would like to describe briefly below:

STAGE 01: A simple case is used for investigation: A particle drops from the surface of a fluid domain. The particle travels to the bottom of the fluid domain and STAGE 01 finishes here. Note that in this step, I use Write_Restart in LIGGGHTS while coupling. So I get a saved file liggghts.restart01 for next running.

STAGE 02: I use such liggghts.restart01 obtained in STAGE 01 by using Read_Restart liggghts.restart01 in this running. So this means I continue previous stage and therefore the particle should be in the bottom of fluid domain at the beginning of this step. However, the particle comes from the surface of fluid domain, exactly the same as STAGE 01. This means Write_Restart does not save information at the last step of STAGE 01 properly. It might just store information at the initial steps.

I think there should be something wrong here when coupling. I agree that this does not happen when running only LIGGGHTS (without fluid).

I hope you can clarify this issue.

Thank you,

Regards,

alice's picture

alice | Mon, 09/26/2016 - 13:45

Hi Nathan,
as already stated in my previous comment, write_restart does not (and is not meant to) work for coupled cases. ErgunTestMPI_restart is a tutorial for cfdemSolverPiso that shows how to restart a coupled calculation, it would be best to have a look at it. The generation of a restart file is not achieved with a write_restart command on the LIGGGHTS side but with a writeLiggghts command in the liggghtsCommands dictionary on the CFD side.
I hope this solves your issue.
Best regards,
Alice