Particles crossing wall

Submitted by rob1 on Thu, 11/12/2015 - 16:51

Hello,
I am a new user of LIGGGHTS and I am simulating a shear cell with:
- upper surface as servo, to apply a force in y direction
- lower surface moving in x direction
- fixed walls in the z direction
- boundary p f f
All the surfaces are imported as STL files.

I have the problem that after the compression of the particles and during the motion, at a certain time some particles disappear because cross the lower wall and are lost from the domain. The same happens even without movement of the lower wall, i.e. simply with a compression.
This is my script, can you help me?!

Thanks a lot
============================================
atom_style sphere
boundary p f f
newton off
units si
communicate single vel yes

region reg block 0 0.01 0 0.01 0.0 0.01
create_box 1 reg
neigh_modify delay 0

fix m1 all property/global youngsModulus peratomtype 1.0e7
fix m2 all property/global poissonsRatio peratomtype 0.25
fix m3 all property/global coefficientRestitution peratomtypepair 1 0.5
fix m4 all property/global coefficientFriction peratomtypepair 1 0.5

pair_style gran model hertz tangential history
pair_coeff * *

fix part all particletemplate/sphere 1 atom_type 1 density constant 1000 radius constant 0.0003
fix distr all particledistribution/discrete 63243 1 part 1.0

fix bottom all mesh/surface/stress file mesh/bottom_triang1mm_sal.stl type 1 scale 0.001 stress on
fix top all mesh/surface/stress/servo file mesh/top_l1.stl type 1 scale 0.001 move 0.0 -0.007 0.0 &
stress on com 0.0 0.0 0.005 axis 0.0 1.0 0.0 ctrlPV force target_val -0.1 vel_max 0.5
fix front all mesh/surface/stress file mesh/lateral_front.stl type 1 scale 0.001 stress on
fix back all mesh/surface/stress file mesh/lateral_back.stl type 1 scale 0.001 stress on
fix geometry all wall/gran model hertz tangential history mesh n_meshes 4 meshes bottom top front back

region insreg block 0 0.01 0.001 0.003 0.0 0.01
group nve_group region insreg
fix ins all insert/pack seed 5330 distributiontemplate distr maxattempt 200 insert_every once &
overlapcheck yes all_in yes vel constant 0.0 0.0 0.0 region insreg particles_in_region 1000

fix integrator all nve/sphere
compute forces all wall/gran/local force_normal force_tangential

timestep 0.0000001

run 5000

dump dmp all custom 5000 post4/dump*.triangular id type type x y z ix iy iz vx vy vz fx fy fz &
omegax omegay omegaz radius
dump dmpstl1 all mesh/stl 5000 post4/bottom*.stl bottom
dump dmpstl2 all mesh/stl 5000 post4/top*.stl top
fix motion all move/mesh mesh bottom linear 0.01 0.0 0.0
fix time all check/timestep/gran 1000 0.2 0.2

run 3000000
===============================================

AttachmentSize
Image icon image.png67.15 KB
Image icon image2.png146.1 KB

rob1 | Wed, 11/18/2015 - 16:56

Probably is it due to a bug in LIGGGHTS?
I've read of problems related to particles crossing walls, but I have the latest release and I think these bugs have been fixed.

I've tried everything: timestep, mesh size, particle size, skin.
Can somebody help me?
Thanks a lot.

rob1 | Thu, 11/19/2015 - 11:24

Yes, I've tried different neighbor sizes, even in the order of magnitude of the particle radius and diameter...

JoshuaP | Thu, 11/19/2015 - 11:32

maybe you put a too large force on the particles? what is the pressure on the top plate? ... I saw u put in a negativ value for target val of the top wall. Try to put a positive value and change in axis the 1 to a -1.

rob1 | Thu, 11/19/2015 - 14:08

No, it is not working... I always lost particles like in the picture.
I believe it is a problem of a moving mesh crossing a periodic boundary... what do you think? And is there any solution?

It's strange that it is not possible to simulate a shear cell with imported stl surfaces... what boundary conditions should I use?
Thanks,
rob1

rob1 | Thu, 11/19/2015 - 15:48

It's incredible, I am experiencing this problem (particles crossing stl walls) even without using periodic boundary conditions!
I have used a longer bottom surface and a longer simulation box, constraining the particles between two primitive planes.

Can somebody tell me if this is due to a bug?
I have used different mesh refinements and particle radii...

Thank you,
rob1

JoshuaP | Fri, 11/20/2015 - 09:20

on your attached figure it looks like that the mesh output and the particle output doesnt coincide on time. do you always clear your post folder before you run a new simulation?
on the left side of your picture is no mesh but the particle stay in the triangle position, so I assume that the mesh is just wrong in the output. Remove all mesh outputs and rerun it.
If nothing works try to build up a new model and post it here again if it is still not working.

rob1 | Tue, 11/24/2015 - 09:05

Hello Andreas,
Yes I am using stl mesh crossing a periodic boundary.
I don't understand why it does not work... so how can I model a shear cell with a complex surface?
Do you have any suggestion?

Thanks a lot!

rob1 | Tue, 11/24/2015 - 09:23

I am trying to use different boundary conditions to simulate this shear cell: no periodic boundaries and particles contained between top and bottom meshes, and primitive planes on x axis. See the attached picture! I am losing one entire row of particles!

So this time it is not due to periodic BCs, maybe something regarding elasticity?
I specifiy that I have tried meshes with different refinements (triangle size)...

Thanks again,
rob1

Riccardo Maione's picture

Riccardo Maione | Wed, 11/25/2015 - 12:09

Hello rob1,

make sure your stl is a good quality mesh, you may look to non manifold points for a start

Regards
Riccardo

rob1 | Wed, 11/25/2015 - 15:12

Hello Riccardo thank you for your help.
Anyway, I have already checked the mesh and there are no problems... think that it is not working even with simple meshes (e.g. 2 triangles per face)...

ckloss's picture

ckloss | Thu, 12/10/2015 - 22:53

Hi,

not a bug - 99.9% (my guess)
I guess this solves the mystery: Y= 1.0e7 Pa is pretty low...

Best wishes
Christoph