Abstract
This thesis investigates the mechanisms of particle segregation in confined and non-uniform granular shearing flows, combining experimental studies in annular shear cells with discrete element method (DEM) numerical simulations. Key segregation phenomena analyzed include classical vertical segregation, inverse segregation where large particles accumulate at the bottom contrary to expectations, and horizontal segregation characterized by small particle accumulation near sidewalls. The work focuses on bidisperse and tridisperse mixtures, analyzing how segregation patterns are influenced by mixture properties (size ratio, mass fraction) and external conditions (confining pressure, gravity, wall friction). The impact of initial particle arrangement and polydispersity is also evaluated. In addition, the capability and limitations of kinetic theory to predict segregation in these granular flow configurations are tested. Results reveal that segregation rates and segregation degree in steady states depend strongly on mixture properties, with segregation enhanced for larger size ratios and higher mass fractions of large particles, due to asymmetric segregation rates balancing vertical segregation fluxes. Confining pressure can play the opposite role reducing segregation. Further, the steady state is characterized by a balance between segregation and diffusive remixing, and independent of the initial particle arrangement. In the dense granular flows studied, segregation is driven by both pressure and shear rate gradients. The direction of segregation is function of the pressure gradient. Under gravity, large particles segregate to regions of lower pressure, while in the absence of gravity, large particles segregate to regions of high shear rate gradients and high granular temperature. Inverse segregation appears to arise from shear-induced effects dominating over pressure gradients in very confined and dense regimes. Boundary conditions further induce horizontal segregation by promoting small particle layering near walls. Tridisperse simulations highlight an order vertical structure reflecting interplay between species influenced by mixture properties. Finally, kinetic theory effectively predict segregation patterns observed in less confined and shallower non-uniform granular shearing flows. These findings provide new insights into granular segregation mechanisms in realistic confined flow geometries with practical relevance for optimizing industrial processes involving granular materials.