TFM, same setup, different results (slight timestep change, one diverging)

MFiX Bug report
#1

Hi,
I have been puzzled by a 3D TFM bubbling fluidized bed for a while, never really able to run it in a stable manner, and not being able to pinpoint exactly what is the problem. Every now and then simulation plunge into divergence without recovery. I have been experimenting with keeping timestep fixed, holding on to upwind numerics for a while, improving mesh and so on. But the errors keeps coming at approximately the same time every time, regardless of adjustments.

Recently I found this to happen with only one of two identical setups, the only difference being that they where executed on different systems (one GUI and the other compiled from source). Could this be of any help? I have marked the time where they first differ in timesteps.

I have also attached the setup file and .stl.

The working run:

bilde
bilde515×563 50.3 KB

The diverging run:

bilde
bilde569×632 39.6 KB

3D_bfb_KBM.mfx (18.0 KB)
tank_and_cone.stl (22.0 KB)

#2

This does seem a bit odd…

Are you saying that one solver was compiled from source, while the other was the “default” binary we distribute with the GUI? Which one is failing?

– Charles

#3

I tried to run this but I think you did not include all of the needed STLs.

mfix - WARNING - Error loading geometry: 1400mm bunn 20gr - rett1_utentopp.stl