Hi,
I am running a relativley large case with 2.56 Million Particles and around 800’000 cells. I have stability issues when I use a constant density with that grid size. If I choose a very coarse grid or ideal gas law it works. Could you tell me why this error appears and are there any solutions to use a constant density with this amount of cells ?
I attached the mfix bug report:
largescalecase_2021-07-13T165444.672839.zip (39.8 KB)
Best regards
Janna
Hi Janna. We are looking at the non-convergence, but we noticed in the attached bug report that you are running MFiX version 21.1.4 in a Conda environment called mfix-20.4.3. This looks like a possible installation error. Can you tell us a little bit more about how you set this up - when you downloaded it, which instructions you followed, whether this was an upgrade from an older version, etc - so we can understand whether there’s a problem with the installation command or instructions. We will follow up separately about the convergence question.
– Charles
Hi Charles,
thanks for the reply.
I upgraded it from the mfix-20.4.3 to the newer version and I had some problems removing the old version. I followed instructions removing the old mfix but I couldn’t open the mfix command to open mfix as it was still connected to the mfix-20.4.3 version. I fixed it like this, as it was easiest way for me, and didn’t have problems with it yet, probably not the cleanest way I guess? Could it affect the convergence ?
Best regards, Janna
Thanks. This is probably not the reason it’s failing to converge - we’re looking into it.
Please try the attached. I set an initial constant pressure, used a line preconditionner, turned off the stall detection and set the residual normalization to zero and increased the max number of iterations (see numerics section). It seems to be running better.
LargeScaleCase.mfx (16.6 KB)
Regarding your installation issue, I am not sure what is going on, maybe it would help to do a
conda deactivate
at the prompt to get out of the mfix environment (once the GUI is closed) before installing a new version of MFiX.
Hi Jeff,
Thank you, the case works now! I figured that if I turn on/off the pre-conditioner for the gas pressure it has the most effect on the stability and the other factors more on the computational performance.
With the installation, I tried conda deactivate but it didn’t help also deinstalling anaconda and installing it again didn’t help. I assumed the connection of the “mfix” command to the old version doesn’t get deleted on my computer, but I also checked the bash.rc file and it looked okay. I will try once again following the deinstallation and installation process, maybe I forgot a line and let you know.
Best regards, Janna