Hi,
Ran into the “pop-corn effect” upon reading the particle_input.csv (created from settling a bed and wrote particle_output.csv).
- when settling with cyclic_x and cyclic_z ON the restart does pop-corning on sides. It works better by increasing the domain size very slightly.
- this issue arise because our particles are extremely stiff.
- never reset velocities to zero is a good call when particles are stiff… but even when not resetting them, we do not have enough digits to capture their position and velocities accurately enough (they are tiny in our case).
Overall, as I wrote, the problem is that the number of digits written for the particle position and velocities is too low, written with format F18.6. This leads in our case to having columns with values as shown below, despite having none 0 velocities in “reality” as shown by loading them in Paraview: X,Y,Z,phase_id,diameter,density,U,V,W,usr_var1,usr_var2,usr_var3,usr_var4,usr_var5
0.000170,0.000320,0.000959,1,0.000640,2500.000000,0.000018,-0.000000,0.000031,0.000000,0.000000,0.000000,0.000000,0.000000
0.001803,0.000379,0.000459,1,0.000757,2500.000000,-0.000022,0.000000,0.000010,0.000000,0.000000,0.000000,0.000000,0.000000
0.002583,0.000375,0.000727,1,0.000750,2500.000000,-0.000006,0.000000,-0.000006,0.000000,0.000000,0.000000,0.000000,0.000000
0.003769,0.000339,0.000348,1,0.000679,2500.000000,0.000000,0.000000,-0.000000,0.000000,0.000000,0.000000,0.000000,0.000000
0.004622,0.000457,0.000468,1,0.000914,2500.000000,0.000000,0.000000,-0.000000,0.000000,0.000000,0.000000,0.000000,0.000000
0.005417,0.000392,0.000177,1,0.000783,2500.000000,0.000000,0.000000,-0.000000,0.000000,0.000000,0.000000,0.000000,0.000000 .
I have solved it by modifying the format from F18.6 to E24.15 in the subroutine attached (search for “ERIC BREARD” to see my tiny modifs. See particle_output.csv attached ran on a tutorial case for a quick example;
The default is that of current MFIX-25.1.1 and the other one is that written with my modified *.f attached. Note that perhaps other properties should have more digits?
particle_output_DEFAULT.csv (143.1 KB)
read_particle_input.f (239.0 KB)
particle_output_AFTER_MODIF_digits.csv (246.3 KB)
Yours volcanologically 