Using MFiX-25.3, I am getting a segmentation fault when I use a moving internal surface defined by an STL (generated using the MFiX GUI). MFiX run files and outputs are attached in a .zip file below.
If the internal geometry is not used, the simulation seems generally stable — the particle properties/DEM time step are standard. If the internal surface is prescribed a velocity of 0 (i.e not moving), the simulation will still fail, but can be restarted and it will then run past the most recent restart time. Even very low velocities of the internal geometry (vertical at 1 mm/s) will cause the segmentation fault to occur. The image below is from a case where the internal geometry was moving at 1 mm/s and output at a high dt to show that the simulation is failing just as particles contact the internal surface.
Moving the stl geometry should only be done with pure granular flow because the fluid phase doesn’t see the moving internal stl surface.
When you are troubleshooting, I would suggest simplifying the model as much as possible until you can reproduce the issue quickly. Here, disable the fluid solver (Model setup pane), set the IC such that it is just above the cylinder and use fewer particles. You also don’t need the other stl files to define a rectangular box since the MFiX domain is rectangular by default. Then you can play with the particle’s properties (Young’s modulus, restitution coef. etc.), use monodisperse particles and see if this helps.
Thanks for this. That is interesting about the fluid not seeing the internal surface - I didn’t realise that. Jeff, in your above image is that with the fluid solver disabled?
Charles - I was using a decomposition of 10x1x12, but it was not evenly decomposed. The gridmap.dat file I was using is attached, sorry I thought I had included it in my .zip folder earlier but I did not.
The image I sent did not have the fluid solver disabled, although I also ran it with the fluid solver disables and it also ran fine. Does it still crashes for you without gridmap.dat ?
