Is it possible to generate non-sphere particle in LIGGGHTS, by bonding spheres together?

Submitted by keepfit on Fri, 09/14/2012 - 15:38

Hi Christoph,

I have done such similar job by using commercial codes. There are 2 methods to generate a non-sphere particle: 1) cluster spheres together as a single particle, the bonds between spheres can be broken; 2) particle is formed by number of overlapping spheres as a rigid body, bonds can not be broken.

Since the feature 'multisphere' is not yet available in both 1.5.3 and the latest version, according to previous posts. So i was wondering, is it possible to bond spheres ( overlapping or non-overlapping) together using method 1 in Liggghts/Lammps.

Cheers,

David

Forums:

msbentley's picture

msbentley | Sat, 09/15/2012 - 15:58

Yes, you can define non-overlapping clusters or clumps and use them with a hybrid sphere/molecular pair style. I usually define the initial clumps and their positions externally and read them into LIGGGHTS with read_data.

You can read the docs for more info, but I think you just define a data file along the lines of:

atom-ID atom-type x y z diameter density clump-ID

Cheers, Mark

keepfit | Sat, 09/15/2012 - 21:32

thanks for your reply, Mark.

I used to generate non-sphere particle (with realistic shape) by a scanned STL mesh. I noticed that Liggghts has the capability to pack spheres together via tetrahedral mesh by means of "polydisperse particle packing", have you ever tried this feature?

Cheers, David

ckloss's picture

ckloss | Sun, 09/16/2012 - 23:45

Hi David,

I posted an early version of a bonded particle model here: node/525
The code is for the 1.5.X version, but it should be pretty straight-forward to update it to 2.X
Feel free to use it and ask questions in the dev forum that may arrise

Cheers, Christoph