I am new to LIGGGHTS software. Does surface mesh size effect the particle flow simulation?? I am doing the simulation for particle flow inside the screw feeder where only surface mesh i generated. what about the volume mesh and how the calculations are occurring??? also please explain the role of neighbor listing.
Thanks in advance
This is my LIGGGHTS file.
units si
atom_style sphere
boundary ff ff ff
newton off
processors 1 1 3
communicate single vel yes
# Declare domain
region reg block -0.025 0.025 -0.025 0.035 -0.01 0.20 units box
create_box 2 reg
# Neighbor listing
neighbor 0.001 bin
neigh_modify delay 0
### Setup
# Material and interaction properties required
fix m1 all property/global youngsModulus peratomtype 5e6 5e9
fix m2 all property/global poissonsRatio peratomtype 0.4 0.3
fix m3 all property/global coefficientRestitution peratomtypepair 2 0.5 0.7 0.7 0.3
fix m4 all property/global coefficientFriction peratomtypepair 2 0.5 0.25 0.25 0.3
#fix m5 all property/global coefficientRollingFriction peratomtypepair 1 0.25
# Gravity
fix 2 all gravity 9.81 vector 0.0 -1.0 0.0
# Particle insertion
region factory cylinder y 0.0 0.0175 0.014 0.0196 0.0294 units box
fix pts all particletemplate/sphere 1 atom_type 1 density constant 1670 &
radius constant 0.001
fix pdd all particledistribution/discrete/numberbased 14127 1 pts 1.0
fix ins all insert/rate/region seed 132412 distributiontemplate pdd &
nparticles 1000000 particlerate 1000 insert_every 1000 &
overlapcheck yes vel constant 0. 0. 0. region factory ntry_mc 10000
# Import mesh from cad:
fix cad1 all mesh/surface file mesh/trough_fine.stl type 1 scale 0.001 curvature 1e-5
#heal & auto_remove_duplicates
fix cad2 all mesh/surface file mesh/Screw_fine.stl type 1 scale 0.001 curvature 1e-5
# Use the imported mesh as granular wall
fix mixer all wall/gran model hertz tangential history mesh n_meshes 2 meshes cad1 cad2
# Define the physics
pair_style gran model hertz tangential history
pair_coeff * *
### Detailed settings
# Integrator
fix integrate all nve/sphere
# Timestep (keep < 20% T_Rayleigh)
timestep 0.0000125
mschramm | Wed, 10/03/2018 - 18:45
Some answers
Hello,
The mesh size can effect the flow but as you increase the number of surface elements, your simulation will slow down (I think I saw someone say each triangle is equivalent to 10 extra particles...).
All meshes are used to determine a surface. This surface is then used in contact force calculations with the particles.
The neighbor list keeps track of what particles "neighbor" a particular particle during the simulation.
This is a technique so while you are looping over all your particles to find contact forces, you only need to loop over particles that neighbor your current particle.
The first neighbor command controls how "fine" this list is and the second command tells LIGGGHTS to update this list every time step.
dheeraj | Thu, 10/04/2018 - 07:31
SOME QUESTIONS
Thanks a lot for replying.
Slowing down of simulation with finer mesh that is not the concern.
More finer the mesh more is the accuracy. Like in CFD the mesh size does effect the timestep but in LIGGGHTS is this the case??
The contact forces calculation between the particle and wall that is fine but what about the particle-particle interaction?? how that calculation occur without the volume mesh??
mschramm | Thu, 10/04/2018 - 16:55
meshless simulation
DEM does not use a mesh like with cfd.
Theoretically, there should be no dependence on the time step for the mesh size.
As for particle-particle forces, these are based on how the particle overlaps (This is very simplified).
I would recommend reading the documentation.
https://www.cfdem.com/media/DEM/docu/gran_model_hertz.html
dheeraj | Fri, 10/05/2018 - 08:52
Neighbor listing question
Thanks a lot.
In my simulation a have spherical particles of multiple radii (1mm, 2mm upto 10mm). How should i define neighbor listing for this case?
mschramm | Fri, 10/05/2018 - 15:28
rule of thumb
LIGGGHTS automatically sets a value to the skin of 0.001 meters when using si units.
The larger the skin value, the less time is needed to build the neighbor lists but more time is needed to check potential interactions.
I started out using EDEM and they recommend setting the bin size to be 2 - 3 times the smallest radius, so for you
I would say a bin neighbor skin of 0.001 would be fine since, I believe, setting a skin == 0.0 would result in bins the same size
as your smallest particle diameter.
So the command
neighbor 0.001 bin
is saying to set the skin to 0.001 and to use the bin algorithm (which is recommended).
For more info
https://www.cfdem.com/media/DEM/docu/neighbor.html