How to change particle shape in DEM

Is there any option in mfix to change the particle shape in mfix with or without coding?

No. Particles are always assumed to be spheres.

What will we do in case of non spherical particles?

you would have to scrap the collision detection and spherical collision model and code up your own non-spherical ones. this is non-trivial.
if you are simulating particles which are modestly non-spherical, the typical trick is adjust the drag law, with a single sphericity if \Psi > ~0.9, or multiple shape characterizing parameters, see, eg Loth Powder Tech 2008

For non-spherical particle simulation in MFiX, the simplest way is that you assume the collisions between particles ( and particle-wall) are spherical, while use a non-spherical drag law. Recently, I am working on a bioenergy project, we found the Gansor (1993) non-spherical drag coefficient ( couple with Gidaspow drag) works pretty well for non-spherical biomass particle fluidization. If you really want to include the non-spherical particle-particle collision, you have to modify the exciting MFiX code by implementing either multi-spherical particle model (or glued spherical model, bond spherical model) , superquadric particle approach, polygon approach or real particle geometry (like cylinder).

Well you explained very detailed. I got your point. But is there any code already made for non spherical particles. I need this in order to work on it according to my designed shape.

There is a commercial code you could look into called EDEM: https://www.edemsimulation.com/

Not sure what your application is but I played with Blender (https://www.blender.org/) to determine packing arrangments of cylinders. Blender is for graphics, not physics though.

Thank You dear. I am trying to explore it.

I wanted to explore a different aspect of this question. How might one imitate the tendency to stick-slip of a non-spherical particle on a wall, preferably within the existing DEM solver framework? As an extreme and “academic exercise”, let’s assume the particles are cubes instead of spheres. Technically they can roll a little but sliding will be the dominant motion.

Thanks for reading -any ideas even if unproved are appreciated.

There is work underway adding superquadrics to the DEM model in MFiX. This will not be fully integrated with the solver for a while, but the initial release of the code, for testing purposes, will be toward the end of March. We will announce this when it becomes available for download. You may be interested in being one of the beta testers.

– Charles

2 Likes

I also found this paper, which addresses modeling of angle of repose.

I may look into using the friction coefficients to mimic the behavior of my “sharp” material while maintaining the spherical model. Charles, I would also be interested in trying the superquadratic parcels. How might I apply to the beta-release? I forget if it’s open for download or released only to the beta tester group.

1 Like

We’ll send an announcement when it’s available for download. Thanks for your interest.

1 Like