Lack of Mesh-Particle Interaction for Hopper Discharge

Submitted by apvyas on Thu, 07/14/2022 - 22:13

I am working on a standard hopper discharge simulation to create a basis for some future work in my program. In trying to learn LIGGGHTS by itself, I am able to create particles and load the "STL" file for the hopper mesh, However, when I run the particles, there is no contact detection it seems between the particle and the mesh wall supplied.

In reading other posts on this forum & DEM in general, I know that the timestep plays a critical role. I have played around with the timestep and found that ~1e-7 seems to work but am not totally convinced. Given the particle properties, any other timestep was causing the particles to explode. I have not tried anything lower than that. As for the particle and wall properties, they are from papers my professor sent to me.

My main question is regarding the lack of particle-wall contact detection. Why are the particles and mesh not interacting? How can I troubleshoot this to get the proper interaction? Any suggestions or ideas are greatly appreciated! I have been struggling with this for some time.

(The mesh file and input script I submitted are attached in a zipped 'tar.gz').

AttachmentSize
Binary Data Input script and "STL" file56.18 KB

mschramm | Fri, 07/15/2022 - 15:41

Hello,
When you added your stl file you forgot to add a contact model (see the chute_wear example).
The missing line is:
fix granwalls all wall/gran model hertz tangential history mesh n_meshes 1 meshes hopper

Also, you can check your time step by using the command:
fix ts_check all check/timestep/gran 100 0.1 0.1

The above command will check your time step and warn you if you exceed 10% ray time or 10% hertz time

apvyas | Wed, 07/20/2022 - 01:14

Thanks! I submitted the simulation to my cluster and hopefully that works regarding the contact model.

As for the timestep, why is the threshold variable set at 10%? is that more of a benchmark or better practice to ensure the simulation does not blow up?

efrizz | Wed, 08/03/2022 - 21:40

The timestep should be much less than the collision time so I think 10% is an order of magnitude check