Granular simulation lammps

LAMMPS LAMMPS Mailing List Mirror
1

Dear Lammps Users

I am interested in simulating a mixture of spherical and non-spherical particles of finite size. Hence I compiled Lammps with granular and rigid packages.

In particular it is a mixture of spheres and rod-like particles. These rod-like particles are made by spheres (non-overlapping, just touching).

I used the two fixes as follows

fix 3 rods rigid/small molecule

fix 4 spherical nve/sphere

The first fix is to integrate the equation of motion for rod-like particles and the second one is for spherical particles.

In the data file, I assigned molecule ID as zero for spherical particles and I used the command to read the data file as given below

read_data data.sph-rod fix prop NULL Molecules

Lammps version I am using : 30 July 2016.

My question is if we use two fixes like above , does it take care of the interactions between spheres and rods ?

Please let me know your suggestions if you have performed such kind of granular simulations

Thank you

best regards

Anki

2

Dear Lammps Users

I am interested in simulating a mixture of spherical and non-spherical
particles of finite size. Hence I compiled Lammps with granular and rigid
packages.

In particular it is a mixture of spheres and rod-like particles. These
rod-like particles are made by spheres (non-overlapping, just touching).

I used the two fixes as follows

fix 3 rods rigid/small molecule

fix 4 spherical nve/sphere

The first fix is to integrate the equation of motion for rod-like particles
and the second one is for spherical particles.

In the data file, I assigned molecule ID as zero for spherical particles and
I used the command to read the data file as given below

read_data data.sph-rod fix prop NULL Molecules

Lammps version I am using : 30 July 2016.

My question is if we use two fixes like above , does it take care of the
interactions between spheres and rods ?

why should it not?
computing interactions and time integration are two separate things.
but why don't you just make a small test and see for yourself?
it is always better to have first hand proof than go on somebody else's say-so.

axel.

3

Dear Axel

Thank you for your reply. It was observed that there are spherical particles as neighbors of “rod-like” particles and vice versa.

The strange thing is that when we computed the distances between spheres (pairs of spherical particles) , we observed some times these distances are like 0.4 d (where “d” is the particle diameter)

pair_style gran/hertz/history 20000000.0 NULL 425.0 NULL 0.5 1
pair_coeff * *

Kn in the above command is 20000000, this value is reasonably good for granular simulations, despite this high value for “Kn”, spheres overlap to such an extent shouldn’t be possible.

These distances are normal between the contact pairs of “rod-like” particles and pairs of rod and spherical particles. (normal, i mean overlap distance is near 0.99 d).

Is something strange happening when a sphere is colliding with “rod-like” particle ?

Version of lammps : 30 July 2016 and it is compiled with “granular” and “rigid” packages.

First part of the input script

atom_style sphere
atom_modify map array
boundary f f f
newton off
comm_modify vel yes cutoff 2.5

fix prop all property/atom mol ghost yes

read_data data.dumbAR4_segini fix prop NULL Molecules (this data file consists of both kinds of particles)

neighbor 0.2 bin
neigh_modify delay 0 check yes exclude molecule all

part of log file :

60000 atoms in group rods
60000 atoms in group spherical
15000 rigid bodies with 60000 atoms
2 = max distance from body owner to body atom
Neighbor list info …
2 neighbor list requests
update every 1 steps, delay 0 steps, check yes
max neighbors/atom: 2000, page size: 100000
master list distance cutoff = 1.2
ghost atom cutoff = 2.5
binsize = 0.6, bins = 84 84 209

Thank you

best regards

Anki

4

Dear Axel

Thank you for your reply. It was observed that there are spherical
particles as neighbors of "rod-like" particles and vice versa.

The strange thing is that when we computed the distances between spheres
(pairs of spherical particles) , we observed some times these distances are
like 0.4 d (where "d" is the particle diameter)

pair_style gran/hertz/history 20000000.0 NULL 425.0 NULL 0.5 1
pair_coeff * *

Kn in the above command is 20000000, this value is reasonably good for
granular simulations, despite this high value for "Kn", spheres overlap to
such an extent shouldn't be possible.

These distances are normal between the contact pairs of "rod-like"
particles and pairs of rod and spherical particles. (normal, i mean overlap
distance is near 0.99 d).

Is something strange happening when a sphere is colliding with "rod-like"
particle ?

​LAMMPS doesn't know or care what kind of shape you are modeling. however,
it will strictly follow the physics that have been programmed into it. it
is difficult to tell from the extremely limited information you provide,
but i suspect you are ignoring the choice of units and the related impact
of simulation parameters like the length of the time step and its
connection to other simulation parameters.

in short, most of the time when people have similar suspicions as you have,
they just have badly chosen simulation parameters. GI-GO!

axel.