polydisperse particles command

Submitted by jyothish on Tue, 11/07/2017 - 15:45

Hello everyone

Can anyone give me the syntax for creating spherical particles in range from 50e-6 to 100e-6 in liggghts 3.7.0

Thank you in advance
Jyothish

Forums:

aaigner's picture

aaigner | Tue, 11/07/2017 - 23:43

Hello Jyothish,

start by checking existing examples. For instance, packing-example combines two different particle sizes to one particle distribution.

Best wishes
Andreas

jyothish | Thu, 11/09/2017 - 06:07

Hi Andreas,

I did. However, the above example uses two specific particle sizes for creation. What i would like to create is a uniform distribution of particles of various sizes within a specified range(50e-6 to 100e-6). What command could i use to achieve this in LIGGGHTS 3.7?

Thanks in advance,
Jyothish.

aaigner's picture

aaigner | Fri, 11/10/2017 - 21:54

Hello Jyothish,

in a previous version there was a function to create a uniform distribution (with an internal random generator), but it was very(!) inefficient and much more important there was a logical problem with it. Therefore, this function was removed again.

Thus, the best way is to create it manually with a number of discrete templates (something between 5 and 10 is usually sufficient).

Best wishes
Andreas

paul | Sat, 11/11/2017 - 23:26

Hi,

The ability to select common PSDs seems like a good feature for a DEM code, judging by the threads made in the forum in the last few weeks.
As the common "solution"/workaround seems somewhat algorithmic, it might be desirable to just implement it directly :)
I'd imagine a "nclasses" parameter, and a cutoff for asymptotic distributions like Gaussian ones.

After all, others like the OpenFOAM guys seem to have it figured out:
https://github.com/OpenFOAM/OpenFOAM-5.x/tree/master/src/lagrangian/dist...

Or is a reimplementation of these features in the works?

Greetings,
Paul

AndresMM | Thu, 11/09/2017 - 08:14

Do it manually.

Discretize your gaussian distibution either with matlab/excel/python in, let's say, 10 bins. And then use ten templates.

The more bins you use, the more "continuous" the distribution is.