Particles of differing size or density often segregate in industrial flows such as chutes, silos, conveyor belts and rotating drums. This is the single biggest cause of material non-uniformity, which poses significant problems in handling and processing the grains, leading to plant downtime and product wastage. The most common form of segregation occurs in surface avalanches, which develop whenever a static granular material is tipped above its angle of repose. For example, pouring one's muesli into a bowl at breakfast! These avalanches are very efficient at sorting particles by size, with the large ones rising to the surface and the small ones percolating down to the base. The density of the grains may enhance or counteract this effect. When these flows come to rest a rich variety of particle size and density distributions develop in the deposit, sometimes with large regions of just one particle type. This naturally presents a major problem in processes that are supposed to be well-mixed. Understanding the segregation process and being able to model it effectively is the first step in being able to develop strategies to mitigate its effects. This proposal aims to use a powerful combination of small scale experiments, theory, continuum simulation and discrete element simulations (where the interactions of every single particle are modeled) to determine the functional dependence of the segregation rates on particle properties, as well as the applied shear-rate and pressure. The resulting mathematical model will then be applied to more complex flows, where there is mass transport between the the surface avalanche and the static, or slowly moving, grains beneath. This presents the project with its biggest challenge, because the rheology of granular materials is still very poorly understood, compared to fluids, which makes simulating the flow in a silo problematical. Over the past decade there has however been significant progress in the development of the so called mu(I)-rheology, which works over a large range of parameter space. Our aim is to regularize the model, by including additional physics, so that it can be applied in all regions of the flow and hence solve for the bulk velocity field. This will then allow the evolving particle-size and density distribution to be computed, so that we can understand in detail how pockets of just one particle type form. With our industrial partners we develop mitigation strategies, that use our knowledge of segregation to design clever chutes and silos that greatly reduce its effects.