Particle 1 is inserted more than particle 2

Submitted by gibel on Wed, 12/29/2021 - 08:38

Hello,
Need small help here.

I am trying to do simulation in Silo.
I am trying to insert 2 types of particles with fraction 0.5 and 0.5. But particle 1 is inserted more than particle 2. It seems particle 1 is inserted almost 80%.

Can I get hint what did I miss here ?

My script file is attach here.

Hoping for some help.

Thank you.

AttachmentSize
Plain text icon light input file7.61 KB

mschramm | Wed, 12/29/2021 - 19:16

Hello,
By default the command

fix pdd1 all particledistribution/discrete 32452867 ${nradii} pts1 ${frac1} pts2 ${frac2}

will perform insertion based on mass and not the raw number of particles. In your log file, you should see something similar to the following (Captured from the packing example).

Fix particledistribution/discrete (id pdd1): distribution based on mass%:
pts1: d=1.000000e-02 (max. bounding sphere) mass%=30.000000%
pts2: d=1.600000e-02 (max. bounding sphere) mass%=70.000000%
Fix particledistribution/discrete (id pdd1): distribution based on number%:
pts1: d=1.000000e-02 (max. bounding sphere) number%=63.708005%
pts2: d=1.600000e-02 (max. bounding sphere) number%=36.291995%

To insert based on number of particles of each type instead of mass of each type use the command,

fix pdd1 all particledistribution/discrete/numberbased 32452867 ${nradii} pts1 ${frac1} pts2 ${frac2}

This information can be found here
https://www.cfdem.com/media/DEM/docu/fix_particledistribution_discrete.html

mschramm | Wed, 12/29/2021 - 19:21

Below is the distribution from your input script (minus the stl files)
Fix particledistribution/discrete (id pdd1): distribution based on mass%:
pts1: d=4.000000e-02 (max. bounding sphere) mass%=50.000000%
pts2: d=8.000000e-02 (max. bounding sphere) mass%=50.000000%
Fix particledistribution/discrete (id pdd1): distribution based on number%:
pts1: d=4.000000e-02 (max. bounding sphere) number%=88.888889%
pts2: d=8.000000e-02 (max. bounding sphere) number%=11.111111%

Here we can see that the mass of each particle will be 50/50 but the number of particles will be 89/11.
Then after using the number based distributions we see the output simple become

Fix particledistribution/discrete (id pdd1): distribution based on number%:
pts1: d=4.000000e-02 (max. bounding sphere) number%=50.000000%
pts2: d=8.000000e-02 (max. bounding sphere) number%=50.000000%

gibel | Thu, 12/30/2021 - 09:10

Let me run my simulation and check the ratio. Thanks a lot. I got it now.

gibel | Mon, 01/03/2022 - 05:34

Hello Mschramm,

I followed your suggestion to insert particles by using number based.
Now it seems particle 2 is inserting 80 % and particle 1 is 20 %?

Did I miss something ?

Thank you.

mschramm | Wed, 01/05/2022 - 22:03

Hello,
Can you post your log file.
Using your script (minus the stl file fix commands), I got 50/50 number based distribution.
Note that I didn't run the code to the point particles were inserted. You may be trying to insert the particles
into a space too small. LIGGGHTS does not insert both particles types at the same time, it will insert the
larger particle first and then insert the smaller one.

gibel | Fri, 01/07/2022 - 04:18

Hello Mschramm sorry for late reply:
I re run the simulation and this time only type 2 is inserted.
Furthermore particles didn't completely drop on the floor from silo drum.
```
This is my log file:
#################################RUN the simulation Outflow###############################################################

run ${dischargesteps}
run 250000
Setting up run at Fri Jan 7 09:34:49 2022

Memory usage per processor = 160.77 Mbytes
Step Atoms KinEng CPU
70000 63541 281.93443 0
320000 63541 167.9888 8919.2075
Loop time of 8919.21 on 1 procs for 250000 steps with 63541 atoms, finish time Fri Jan 7 12:03:29 2022

Pair time (%) = 4701.9 (52.7166)
Neigh time (%) = 1934.52 (21.6894)
Comm time (%) = 24.1962 (0.271282)
Outpt time (%) = 175.524 (1.96793)
Other time (%) = 2083.07 (23.3548)

Nlocal: 63541 ave 63541 max 63541 min
Histogram: 1 0 0 0 0 0 0 0 0 0
Nghost: 0 ave 0 max 0 min
Histogram: 1 0 0 0 0 0 0 0 0 0
Neighs: 207609 ave 207609 max 207609 min
Histogram: 1 0 0 0 0 0 0 0 0 0

Total # of neighbors = 207609
Ave neighs/atom = 3.26732
Neighbor list builds = 46966
Dangerous builds = 0
```

My run script is :
```
fix pts1 all particletemplate/sphere 10487 atom_type 1 density constant ${density} radius constant ${radius1}

fix pts2 all particletemplate/sphere 11887 atom_type 2 density constant ${density} radius constant ${radius2}

fix pdd1 all particledistribution/discrete/numberbased 32452867 ${nradii} pts1 ${frac1} pts2 ${frac2}

fix ins_mesh all mesh/surface/planar file Insertionsface.stl type 1 scale 0.001

fix ins all insert/stream seed 86028157 distributiontemplate pdd1 &
mass ${fillmass} massrate ${fillmassrate} overlapcheck yes all_in yes vel constant 0 0 -0.75 &
insertion_face ins_mesh extrude_length 0.25
```

Did I miss anything?

Hoping for some help.

I just need to run simulation with 2 types with insertaion rate 50 50.

I did try another approch as below:

```
#fix pts1 all particletemplate/sphere 15485863 atom_type 1 density constant ${density} radius constant ${radius1}
#fix pts2 all particletemplate/sphere 15485867 atom_type 2 density constant ${density} radius constant ${radius2}
#fix pdd1 all particledistribution/discrete/numberbased 32452843 ${nradii} pts1 ${frac1} pts2 ${frac2}
#fix ins_mesh all mesh/surface/planar file Insertionsface.stl type 1 scale 0.001
#fix ins all insert/stream seed 32452867 distributiontemplate pdd1 &
# nparticles ${fillmass} massrate ${fillmassrate} insert_every 2000 overlapcheck yes all_in no vel constant 0 0 #-0.75 &
# insertion_face ins_mesh
```

this gave me around particles type in 50 50 rate. When i increased the n particles with some number like I need 10000 , I got same problem like above
where particles doesn't drop completely on the floor.

Thank you. Hoping for you suggestion.

gibel | Fri, 01/07/2022 - 04:20

My two types of particles are differ in radius.
Radius of particles A is 0.01 and radius of particles B is 0.02.

mschramm | Fri, 01/14/2022 - 05:00

Hello,
It appears that your log file is missing the part that shows the particle insertion distributions.
Given your particle sizes, it makes sense to me that you are trying to insert too many particles into your insertion space making liggghts give up on trying to do the insertion.
This coupled with the fact that liggghts will insert the larger particles first (your type 2 particle), I can see why type 1 is not being inserted.

To get around this, you need to either lower your massflow value or increase your insertion area.

gibel | Fri, 01/14/2022 - 07:11

Let me re do my simulation.

One more questions.

I feel like particles id are repeated in LIGGGHTS simulation. One particle which is dropped is again repeated. Am I right?

If so,
Can I just do simulation with all different id? LIke fill Silo with 2000 particles and droped them without repeating them.

Thank you.

gibel | Mon, 01/17/2022 - 03:33

I will check LIGGGHTS examples after I solve my problems.
I am still not able to get atom type 1 - 50% and atom type 2 - 50% in insertion.

I will do more reasearch about liggghts and I will come back.

I really appreciate your time and answers. Thanks a lot Mschramm.