Dear All,
I would like to create a gaussian or log-normal particle size distribution, I mean for the particles radii. Unfortunately, I cannot find anything in the documentation, even though someone implemented a "radius gaussian number" in the "in.liggghts_init" (input command file for DEM). For example:
https://www.cfdem.com/forums/particletemplate-gaussian-radius-distribution
https://www.cfdem.com/forums/gaussian-distribution-radii-using-particlet...
https://www.cfdem.com/forums/gaussian-particle-distribution
To me is not clear how that gaussian radius distribution works. I would expect a discretised size/radius distribution, but how many sizes/radii are created/selected ?
At the moment I have this:
fix pts1 all particletemplate/sphere 15485863 atom_type 1 density constant 2000 radius constant 0.0005
fix pdd1 all particledistribution/discrete 15485867 1 pts1 1.0
Which is the correct syntax to get a gaussian radius distribution around a certain radius (mean), with a certain variance, and a certain number of radii composing the discretised distribution (for example, 5 or 30 or 100 radii inside the gaussian distribution) ?
Looking forward to hearing from you.
Best regards,
Limone
Best,
Limone
paul | Fri, 10/13/2017 - 12:44
The functionality described
The functionality described in the threads you linked is not present in the current version of LIGGGHTS (or just LIGGGHTS-PUBLIC?), as can be seen in the source:
https://github.com/CFDEMproject/LIGGGHTS-PUBLIC/blob/master/src/fix_temp...
I would work around this by either
a) writing an elaborate LIGGGHTS Input script creating many discrete fix particletemplate/sphere classes according to an Gaussian distribution (Loop within LIGGGHTS using jump and label), or
b) using a simple python script to generate the particletempate/sphere lines according to the distribution.
Note that it is important to truncate the distribution on both ends and adapting the contact detection cutoffs accordingly.
Greetings,
Paul
limone | Fri, 10/13/2017 - 13:59
How many particles per each size class ?
Thank you Paul for your reply.
I could generate the LIGGGHTS Input script with several "fix particletemplate/sphere", like in the following example where there are 10 size classes (i.e. 10 radii):
fix pts1 all particletemplate/sphere 15485863 atom_type 1 density constant 2000 radius constant 0.00046
fix pts2 all particletemplate/sphere 15485863 atom_type 1 density constant 2000 radius constant 0.00047
fix pts3 all particletemplate/sphere 15485863 atom_type 1 density constant 2000 radius constant 0.00048
fix pts4 all particletemplate/sphere 15485863 atom_type 1 density constant 2000 radius constant 0.00049
fix pts5 all particletemplate/sphere 15485863 atom_type 1 density constant 2000 radius constant 0.00050
fix pts6 all particletemplate/sphere 15485863 atom_type 1 density constant 2000 radius constant 0.00051
fix pts7 all particletemplate/sphere 15485863 atom_type 1 density constant 2000 radius constant 0.00052
fix pts8 all particletemplate/sphere 15485863 atom_type 1 density constant 2000 radius constant 0.00053
fix pts9 all particletemplate/sphere 15485863 atom_type 1 density constant 2000 radius constant 0.00054
fix pts10 all particletemplate/sphere 15485863 atom_type 1 density constant 2000 radius constant 0.00055
But where and how can I write how many particles to generate in my simulation for each of the 10 size classes ?
Where can I put the information on the number of particles for each size class ?
Best regards,
Limone
paul | Fri, 10/13/2017 - 14:22
That is specified by fix
That is specified by fix particledistribution/discrete:
https://www.cfdem.com/media/DEM/docu/fix_particledistribution_discrete.html
Greetings,
Paul
limone | Fri, 10/13/2017 - 15:55
Thank you very much! Greeting, Limone
Thank you very much! Greeting, Limone
AndresMM | Tue, 10/17/2017 - 11:25
Why was the gaussian functionality removed?
Using an older version of LIGGGHTS it was possible to obtain a gaussian size distribution. As long as you were careful with the sigma not creating negative radii, it was all ok.
Why was this removed in the current LIGGGHTS version?
limone | Mon, 10/23/2017 - 22:28
No idea mate....I am
No idea mate....I am disappointed too...
In my case I am doing a workaround with fix particletemplate/sphere.
Cheers,
Limo
arnom | Mon, 11/13/2017 - 17:13
The reason for this
The reason for this functionality was a bug in the implementation related to the difference between mass and number distribution. Since this feature does not seem to be in wide use, we decided to remove it. If somebody wants to dig out the old code and fix this bug we would consider adding the functionality again.
Best,
Arno
limone | Fri, 01/12/2018 - 17:51
Many thanks Arno!
Many thanks Arno!