[lammps-users] granular particle inside fix wall/region

LAMMPS LAMMPS Mailing List Mirror
1

Dear lammps-users,

Originally I tried to simulate a funnel using a cone region in combination with fix wall/region and particles with gravity but as soon as the falling particles (inside the cone) touch the wall the error "Particle on or inside fix wall/region surface" occurs.

In a simulation with just one particle it depends on the position of the particle. If it is located above the flat part of the "funnel" (where later should be the hole) the simulation works fine (shift the "#" in the example code one line up):

# === Code 1 ===
atom_style granular
dimension 3
boundary f f f
newton off
communicate single vel yes
region R_gesamt block 0 10. 0 10. 0 10. units box
create_box 1 R_gesamt
neighbor 0.2 bin
neigh_modify delay 0
pair_style gran/hooke/history 2. NULL 50.0 NULL 0.5 0
pair_coeff * *
timestep 0.001
fix F_1 all nve/sphere
fix F_2 all gravity 1.00 spherical 0.0 -180
lattice bcc 10.
region R_trichter cone z 5.0 5.0 0.2 15.0 4.1 9.9 side in units box
region R_insert cone z 5.0 5.0 0.2 2.0 6.0 8.0 units box
# ==========
create_atoms 1 single 6.0 6.0 8.0 units box
#create_atoms 1 single 5.0 5.0 8.0 units box
# ==========
group G_korn type 1
fix F_trichter G_korn wall/region R_trichter lj126 0.00001 0.04 2.1
set group G_korn diameter 0.02
fix F_visco G_korn viscous 0.002 scale 1 0.003
compute C_1 G_korn erotate/sphere
thermo_style custom step atoms ke c_C_1 vol
thermo 100
compute_modify thermo_temp dynamic yes
dump D_2 all custom 100.0 dump.test04 id type x y z &
      ix iy iz vx vy vz fx fy fz omegax omegay omegaz radius
run 20000.0
# === Code 1 ===

Trying to find the reason I used a box instead of the cone. There it depends on the position of the bottom (z_lo) whether the simulation works or not (using a smaller diameter would also help in this simulation).

# === Code 2 ===
atom_style granular
dimension 3
boundary f f f
newton off
communicate single vel yes
region R_gesamt block 0 10. 0 10. 0 10. units box
create_box 1 R_gesamt
neighbor 0.2 bin
neigh_modify delay 0
pair_style gran/hooke/history 2 NULL 50.5 NULL 0.5 0
pair_coeff * *
timestep 0.001
fix F_1 all nve/sphere
fix F_2 all gravity 1.00 spherical 0.0 -180.0
lattice bcc 10.
# =========
#region R_box block 3 7 3 7 2.96 8 side in units box
region R_box block 3 7 3 7 2.97 8 side in units box
# =========
lattice bcc 100.
region R_insert cone z 5.0 5.0 0.2 2.0 6.0 8.0 units box
create_atoms 1 region R_insert units box
group G_korn type 1
fix F_form1 G_korn wall/region R_box lj126 0.00001 0.04 2.1
# =========
set group G_korn diameter 0.02
#set group G_korn diameter 0.01
# =========
fix F_visco G_korn viscous 0.002 scale 1 0.003
compute C_1 G_korn erotate/sphere
thermo_style custom step atoms ke c_C_1 vol
thermo 100
compute_modify thermo_temp dynamic yes
dump D_2 all custom 100.0 dump.test04 id type x y z &
      ix iy iz vx vy vz fx fy fz omegax omegay omegaz radius
run 20000.0
# === Code 2 ===

Maybe there are any suggestions for creating boundary regions in combination with the granular style. The idea is to use "region union" for developing complex geometries in future.

Regards,
Christian

2

I think you are confusing 2 different kinds of walls. For granular systems, you
normally want walls that interact via a granular potential with the particles,
i.e. the particles come in contact with the wall and there are normal and
tangential forces. This is only done by fix wall/gran and for a few
simple geometries
(plane, cylinder). All the other fix wall/xxx commands do not involve
granular interactions - they only generate normal forces, no
tangential friction.
This includes fix wall/region. They are all designed for use with
non-granular systems.

What I think you want is something like fix wall/region/gran, which does not
exist (yet) in LAMMPS.

Steve

3

Thank you Steve for your hint. You are right that I confused the two different kinds of walls. I must have missed that in the documentation.

Will there be any further development of different wall geometries for granular systems in the near future?
Would it be easier to use the geometries in the granular package and try to modify them or to use the fix wall/xxx commands and try to add friction, if I tried it on my own?

Another question: Is it possible to cluster granular atoms somehow? The error message says: No bonds allowed with this atom style -- what else would you suggest?

Christian

4

You can use a hybrid atom style, e.g. granular bond. They you could
add bonds. You can also create rigid bodies that are collections of
granular particles.
We do have plans to enhance the walls for granular. Check out this WWW page:
http://web678.public1.linz.at/Drupal/?q=node/10
for some add-ons to LAMMPS for granular from Christoph Kloss - he has added
triangulated walls.

Steve