Geometric aspects of particle segregation

Richard Caulkin, Xiaodong Jia, Michael Fairweather, Richard A Williams

Research output: Contribution to journalArticlepeer-review

24 Citations (Scopus)


Size segregation is a natural occurrence both in everyday life and in industrial processes. Understanding and research of the phenomenon has overwhelmingly been from a mechanistic point of view. This paper demonstrates through simulations that segregation can also be explained and trends predicted geometrically. The algorithm used in this study contains three simple elements: random walks combined with a rebounding probability to encourage particles to settle, plus the nonoverlap constraint. It is implemented digitally in a regular lattice grid, to make it easy to deal with arbitrary shapes. It does not explicitly consider any particle interaction forces, and it does not include any rules specifically designed to promote or suppress segregation. Yet particle movement, which occurs within a digitized cubic grid, leads to shaking-induced segregation comparable to that observed in physical tests. The paper details the comparison of shaking-induced particle segregation between a series of computer based simulations and those of physical experiments undertaken in the laboratory. A range of mixtures, comprising nonspherical, arbitrary shaped/sized particles are investigated, having been packed into pseudo-two-dimensional containers. The simulation results suggest that segregation can be adequately explained, from a geometrical point of view, as a result of the relative motion between particles of different sizes and shapes. The geometrical algorithm thus provides a fast and qualitative prediction as to how likely segregation is to occur for any given mixture of arbitrary shapes.
Original languageEnglish
Article number051302
Number of pages9
JournalPhysical Review E
Issue number5 Pt 1
Publication statusPublished - 6 May 2010


Dive into the research topics of 'Geometric aspects of particle segregation'. Together they form a unique fingerprint.

Cite this