Clustered Particles

Hi Toketo,

This is a very good question because it is hard to give a definitive answer. You will need to experiment a bit!

Usually I start by trying to determine a suitable maximum sphere radius. If you wish for your clusters to be of dimensions between 1-2mm and to contain 3-5 particles, then I would select a maximum particle radius so that there is, on average, about 3 particles across the diameter of each cluster. In this case, I would select a maximum particle radius in the range 0.15-0.3mm or perhaps a little larger.

Once you have settled on a maximum sphere radius, the choice of minimum sphere radius largely determines the total number of spheres in the final model. I usually start by trying Rmin = 0.3 x Rmax and see how many spheres the final model contains. If that number of spheres is not too large, given your computational resources, I would stick with it. In my experience this radius ratio is the bare minimum to achieve a reasonable sphere packing arrangement in clustered models. My preference is to set Rmin = 0.2 x Rmax but this typically results in models with a considerably larger number of spheres.

When all is said and done, the choice of Rmin and Rmax largely depends on what phenomena are the most important to simulate for your particular application. If you can obtain useful results for rather bumpy clusters (with a large Rmin/Rmax ratio) then that will be computationally more efficient than a model with much smoother clusters comprising many more spheres.

Part of the fun with DEM, in my experience, is in determining the optimal resolution (sphere dimensions vs. model dimensions) to capture the phenomena you are most interested to study. Sometimes this can be a rather subjective choice to make. Experimentation is often the only option in such cases.

Cheers,

Dion