Understanding granular flow dynamics has been and continues to remain an interesting problem for engineers and physicists. Granular materials attract special attention due to their ability to behave either as solids, as liquids, or neither - depending on the surrounding conditions. Segregation of granular particles has been of particular interest mostly due to its undesirable role in mixing processes although segregation can be desirable in certain applications. The quality of end products in many industrial applications such as pharmaceutical drug manufacturing depend on efficient mixing of ingredients of various shapes, sizes, densities, charge distributions etc. However, mixtures of materials like pharmaceutical excipients and active pharmaceutical ingredients (API) segregate in a variety of ways such as based on size, density, charge, shape etc. The most common modes of segregation encountered are size and charge based segregation. A number of studies have been performed with an attempt to enhance the understanding of particle movement in size-differentiated mixtures. It has been observed that segregation depends on the geometry and environment. Hence it has been studied in relation to specific geometries, for example, in a mixture flowing from a chute, or when being agitated by a moving bar. Study of mixing in long cylindrical geometries is of great interest. In some industrial applications, it is desirable to mix the ingredients slowly without much agitation from baffles etc. Size segregation is seen in the form of banding in such cases. There is both radial as well as axial segregation.

Computational methods using discrete element modeling and simulation (DEM) have been shown to be very useful in complementing experimental analyses in this field. However due to limitations in computational power and effective parallel algorithms, either the particle to vessel diameter ratios had to be kept high or extremely low fill levels had to be used, or the size differential among the particles had to be kept unrealistically high. These constraints, including the limitation on the number of particles that can be simulated (usually kept around 10,000) have limited the scope of computational study in this field. However with increasing computational power and more efficient algorithms becoming accessible some of these limitations have largely been reduced. In this study we apply DEM techniques using a fast, parallel algorithm to simulate the mixing and segregation in a 100,000 particles in a long cylinder with length to diameter ratio of 2.5 and cylinder to particle diameter ratio of 220. The size differential among the particles was kept according to a more realistic normal size distribution with a 20% standard deviation instead of a binary size distribution which was found to segregate very easily. Further studies with mixtures of particles of non-spherical shapes were also performed. The effect of cohesion, rotational speed, wall effects and their absence due to periodic boundary conditions, and fill level were also studied. A summary of the results will be presented with some more insights into the modes of segregation and the effect of the above mentioned variables on it along with a discussion on ways to prevent undesirable particulate segregation.