Restart DEM after gas momentum was disabled

MFiX
#1

Hi All

I am trying to run 3D DEM of a spouted bed. To speed up the settling process I disabled the gas U,V,W momentum equations and set ro_g0 = 0 and allowed the bed to settle. When I try to restart, using “Restart_2” and the .RES files I get singularity matrix problems when switching the gas momentum back on.

The simulation does not give this error when starting a new sim with momentum on.

Any ideas how to fix it? I am unable to share the .mfx file due to ITAR restrictions.

Regards

#2

I’m not sure this will work but here is what I would try:

  • Create a copy of your case file (mfix.dat or mfx).
    • Reduce the number of particles to one or two.
    • Enable the fluid solver (i.e., set ro_g0 to an appropriate value or leave it undefined).
    • If you have a gas mass inflow, set the value to zero.
  • Run this case for a short while (stop time 0.05 sec for example).
  • Copy the .RES file from the new case and the DES.RES from your previous run, and try a RESTART_2.

If this doesn’t work, you could create a UDF that writes particle data in the particle_input.dat format. Restart your original case with your custom executable (with UDF) and let it create a particle input file. Then start a new run with that particle_input.dat file making sure to disable auto particle generation.

You have to be a little careful if you do this. Make sure you write particle positions with a lot of sig-figs otherwise you might get excessive overlaps when you read them into you new run.

#3

Morning Jordan.

I shall try the approach you mentioned.

I am still having issues with particles in ghost cells, even though i increased the wall spring constant to over 100000 and the particle one to 5000. the particles settling velocities do get unrealistic if the spring constant is to high. The geometry is imported via an STL and according to the logs all triangles are included. Increasing the base mesh resolution does not work either.

Regards

#4

Your approach of using the .RES file from the settling and one from the flow development worked. This is valid for a simple geometry with no internal components. Still having issues with ghost particles when the internal geometry is included in the STL.

Regards