Strange behaviour of rigid mutlisphere/clump using hybird style

Hi LIGGGTHS USER, Hi Chris,

as a commercial DEM code user, I have done realistic particle simulation several years ago. However, it's difficult to modify or add new contact model for commercial software. So I want to implement this job with LIGGGHTS. I have few question concerning the proper "atom_style" for rigid body:

Test example: three non-spherical particles drop under gravity. Their initial positions are (0 0 1), (0 0 1.5), (0 0 2.0) receptively.

1) Method 1: only "atom_style sphere" is used.

The issue is that clumps penetrate each other at certain time step, but the interaction between wall and clumps seems ok.

The fix rigid part is like:
...
read_data clumps.data # sphere input
...
group clump1 id <> 1 61
group clump2 id <> 62 124
group clump3 id <> 125 187

fix 1 all rigid group 3 clump1 clump2 clump3

neigh_modify exclude group clump1 clump1
neigh_modify exclude group clump2 clump2
neigh_modify exclude group clump3 clump3

neigh_modify exclude group clump1 clump3
neigh_modify exclude group clump1 clump2
neigh_modify exclude group clump2 clump3
....

The rest codes is same as tutorials.

So what is wrong with the " fix rigid" and "neigh_modify exclude " command?

2) Method 2: atom_style sphere hybrid sphere molecular.

In this way the clumps do not penetrate each other, but the problem is that the interaction between clump and wall is not correct.
I noticed that once the lowest clump collides with wall, it stops moving like stuck in the wall. But the rest 2 clumps (upper) do behavior OK.

The related codes is like:

......
read_data clumps_Hybrid.data # hybrid style data input, particles within clump have same molecular_ID
...
fix 2 all rigid molecule
neigh_modify exclude molecule all
....

So anyone have ideas to fix the issue?

with best regards,

David
____________________________________________________________________________

Update:

For method 2 adding the following codes will solve this problem, but it looks weird, as "fix 2 all rigid molecule" already defines the rigid clump.
I wonder if we can use pure sphere style to simulate rigid clumps?
...
group clump1 id <> 1 61
group clump2 id <> 62 124
group clump3 id <> 125 187

fix 1 all rigid group 3 clump1 clump2 clump3

neigh_modify exclude group clump1 clump1
neigh_modify exclude group clump2 clump2
neigh_modify exclude group clump3 clump3

....