LIGGGHTS: Dense packing cannot be achieved in some conditions

Problem: dense packing under gravity will not be formed even the particle velocities are nearly zero.
Conditions: insert particles more than a critical number.

tested versions 2.0.4, 2.1.1

I think this should be a bug where the gravity field is lost when the total particle number is larger than a critical value (I am doing test and will tell the critical value in the following given case).

Files attached: geometry (cubic.stl, extension has to be changed), input (in.packing, extension as well), and a result using liggghts 2.0.4.

Attachment	Size
cubic.txt	3.12 KB
loose_packing.jpg	1.75 MB
log.liggghts.txt	16.07 KB
cubic.avi_.jpg	16.99 MB

Forums:

LIGGGHTS® - User Forum

ckloss | Wed, 09/19/2012 - 16:56

Hi qfhou,can you describe

Hi qfhou,

can you describe more closely what exactly you did and how this behavior deviates from what you expect?

Thanks,
Christoph

qfhou | Thu, 09/20/2012 - 02:02

Hi Chris, Thanks for your

Hi Chris,

Thanks for your reply.

I would like to give more information. A log file and an animation will be attached to the first post.

As you can see: 1) The kinetic energy as indicated in the log file is very small (10E-8).
2) However, the packing structure is quite loose and it changes a lot with time (as shown in the attached animation).

I am wondering 1) something wrong with the input file (but it is generally same to the input file in the tutorial);
2) the gravity field is lost due to some unknown reason(s). Hence the particles would not fall down and the energy is dissipated only through inelastic collision.
3) I am running it on Ubuntu 12.04LTS.
Cheers
qfhou

ckloss | Thu, 09/20/2012 - 11:58

Hi qfhou,I can't open the

Hi qfhou,

>>Ave neighs/atom = 3.7762
This indicates the packing _is_ in fact dense.

Christoph

qfhou | Fri, 09/21/2012 - 01:07

Hi Chris, Thanks a lot for

Hi Chris,

Thanks a lot for your reply!

>>Aver neighs/atom depends on the selection of "neighbor 0.005 bin"

It is not actually the coordination number (I think). and the coordination number of a random loose packing should be around 5.

I am trying to generate the coordination number. and let you know the results later.

Cheers,
qfhou

PaulWinkler | Tue, 07/16/2013 - 19:01

Coordination number

Hi,

coordination number is not equal ave neighs/atom! Simple test, put 8 spheres together in a sc lattice, ave neighs/atom is 1.5 for non-periodic boundaries (there are 12 contacts between 8 spheres), but coordination number is imho 3 for every sphere in the package. The theoretical coordination number of a sc packing is 6.
For periodic boundaries you will get an ave neighs/atom of 17.5. There are 140 contacts!? between 8 spheres. Can someone may explain that please?

Regards
Paul

ckloss | Thu, 07/18/2013 - 18:23

>>coordination number is not

>>coordination number is not equal ave neighs/atom
No, but in this example skin is 0.005, and radius is 0.05, so it's almost the same - except for the wall effect you explained before

Christoph

PaulWinkler | Sun, 07/21/2013 - 21:29

skin is 0.005, and radius is 0.05

Hi Christoph,

I don't get it. Why is coordination number almost the same like ave neighs/atom when skin is 0.005, and radius 0.05?

For anyone who is still searching here for a solution, look for compute contact/atom in the Liggghts pdf manual (not the online ones) or on the Lammps Website (http://lammps.sandia.gov/doc/compute_contact_atom.html).

Best wishes,

Paul