Particle size distribution

Hi Christian,

Thanks for your interest in ESyS-Particle and your question.

It is indeed sometimes inconvenient to construct particle samples with a specific size-distribution. The particle packing algorithm is not easily adapted to this purpose, however that algorithm can be applied to construct a broad range of complex particle samples. A new python module called LSMGenGeo will be released in the new few months that will attest to that. This new module will not address your specific concern regarding size-distributions.

To the best of my knowledge there are no generic geometrical algorithms for generating particle assemblies with a specified size distribution. I have experimented recently with a few ideas but have only a prototype I am not ready to release until the work is published. Sorry about that.

In addition to geometrical packing algorithms, there are a number of algorithmic techniques utilised by various DEM practitioners over the years. Algorithmic techniques involve executing a simulation to achieve an initial particle sample e.g. adding particles to a box and letting them settle under gravity.

For your purposes, perhaps one method would be insert particles at random locations one-by-one into a simulation with fixed walls on all sides of a cube. You can use the sim.createParticle(..) method described in the Tutorial called bingle.py to insert particles of specified radii. Of course many of the particles will overlap one-another so specify NRotElastic interactions between unbonded particles with a relatively high stiffness (normalK). You will also need to add viscosity (as described in gravity.py) to damp the motion of particles so they do not escape from the simulation. Then run a simulation and record the total kinetic energy of the particle assembly (e.g. via a FieldSaver). Eventually the particles will reach an equilibrium packing and the kinetic energy will drop to a near constant low value. Finally write out the locations of all particles to a file and use this file to construct your particle sample for your angle-of-repose simulations.

Whilst not ideal, algorithmic techniques can help when you require a particle sample with specific properties such as size-distribution or range of densities. It can be difficult at times to perfect a simulation to achieve a desirable packing though.

Good luck and have fun with ESyS-Particle!

Cheers,

Dion.

Hello!!

Regarding with the particle size distribution, suppose I have particle radius ranging from 0.1mm-011mm. What is smallest next particle radius to 0.1mm or what is the smallest particle size present between the given range?

Jessica

Hi Jessica,

I'm not sure I understand your question. If you are asking about the distribution of particle sizes one obtains from LSMGenGeo, then here's a rough answer:

Since LSMGenGeo uses a space-filling packing algorithm, the distribution of particle sizes tends towards a power-law (i.e. fractal) within the specified particle size range. For a sufficiently broad range of sizes and large enough volume, the fractal dimension approaches D=3 i.e. the fractal fills 3-dimensional space. If you use a narrow range of sizes, LSMGenGeo is not so efficient and you will tend to get a fairly flat, uniform distribution of particle sizes with somewhat of a bias towards smaller particles.

In case you were really asking about the specific radii of particles inserted by LSMGenGeo. LSMGenGeo does not select particles for insertion from discrete bins. Particles can have any radius at all, as long as it is within the specified maximum range of sizes. LSMGenGeo seeks holes to fill with particles then expands a particle to fill that hole. If the particle radius lies in the specified range, the particle remains. If not, LSMGenGeo will try again to find a hole to fill.

I hope this helps. If not, please feel free to ask more questions!

Cheers,

Dion.

Hello Sir Dion!!!

Thank you so much for the answers!! That solved my problem. =)