Plausible bug in MFIX-25.1 and 25.1.1 excluding particle read in *csv

ebreard6 · May 13, 2025, 1:50pm

Hi,

The particles in x are not excluded as they should based on the dimensions I specified. See case attached; (excluding particles in y and z seems OK)
simple_shearCopy.zip (13.9 MB)

Thanks for the help!

cgw · May 13, 2025, 7:55pm

Running this test with MFiX 25.1.1 (on Windows) I get

Error from des/read_particle_input.f:2405

The expected number of columns does not match the actual number of columns in particle_input.csv,
expected number of columns =   39,
number of columns in file =   14.

I’m not sure where that 39 is coming from, but this is a problem I will follow up on this.

cgw · May 13, 2025, 8:10pm

Ah, it looks like you might have attached the wrong file, the .mfx file has des_usr_var_size=30 while your csv only has 5 user variables. You need to specify all of the user scalars, or none - the current implementation does not allow specifying only a portion of them.

ebreard6 · May 14, 2025, 1:45pm

dem_rheology_dry.zip (13.9 MB)
this was not a mistake, I did not get to the running stage because the filtering did not work. here is a version that was used to settle the very large bed and I was trying to modify the setup to a smaller one and ran into the filtering issue.

You may wonder why setting 1m particle when I need 18k? That is because of another issue at restart that i ran into so i wanted to try again from this particle_input

cgw · May 14, 2025, 1:53pm

But the file you uploaded won’t even load, so the filtering stage is not reached.

Anhyow, we’re looking at this now. There may be some particle overlaps in your data, because some particles are getting ejected immediately. And we think the x-filter issue may be due to cyclic BCs. We will follow up with more details.

ebreard6 · May 14, 2025, 1:56pm

I just ran it with mfix-25.1.1 now, it works so I am confused

thanks for looking into this

jeff.dietiker · May 14, 2025, 2:00pm

One thing we found out with 25.1 is that filtering to x_max=0.01599 instead of 0.016 works. Not sure if this is due to cyclic BC, we are looking into it.

ebreard6 · May 14, 2025, 2:34pm

Ah, I tried, did not work regardless of number of 9’s

cgw · May 14, 2025, 3:04pm

One possible source of confusion here is that the particle preview in the MFIX GUI (Model pane) does not apply filtering to the particles read from particle_input.csv, it shows all of the particles in the file (up to the specified count). Adding filtering to the preview code is on our todo list.

jeff.dietiker · May 14, 2025, 6:01pm

I was talking about the dem particle_input.csv file filtering option part_in_x_max, not the geometry x_max

jeff.dietiker · May 14, 2025, 6:02pm

This setup ( dem_rheology_dry.zip) doesn’t work for me with 25.1.1:

Error 1100: The des search grid is too fine in the X-direction. The
maximum particle diameter is larger than the cell width:
  MAX DIAM:    0.96000E-03
  CELL WIDTH:  0.50000E-03
Decrease the values for DESGRIDSEARCH in the project settings.

ebreard6 · May 15, 2025, 10:34am

Ok let me clarify things:

the restart works with the filtering, in x y z. once I corrected the mesh etc. and also used 0.01599999999 instead of 0.016 :).
I got confused because I expected the GUI to display only particles that will be kept in, sorry about this! If that can happen in the future, great, otherwise it is still working.
The restart with the particles that are really stiff works if we do not reset the particle velocity to 0 at the moment of writing particle_output.csv; this is not something I knew but is somewhat intuitive. Happy days!

cgw · May 21, 2025, 6:35pm

This is fixed in MFiX 25.1.2. Thank you for your help in identifying and fixing this issue.