Flowing granular mixtures with species differing in some property, such as size, shape or density of the grains, tend to segregate (de-mix) during flow. Since flowing granular materials are common in industry, and since segregation is often unwanted and costly, understanding and mitigating segregation is an important concern in industrial solids processing. In this dissertation, segregation is studied in granular shear flows under confining pressure. A method is also developed to experimentally determine the segregation parameter of industrially-relevant particle mixtures for a continuum transport model of segregation. First, segregation is studied in confined granular shear flows in which the overburden pressure can be significant. Discrete Element Method (DEM) simulations of size and density bidisperse shear flows under confining (overburden) pressure are performed to understand the influence of overburden pressure on segregation and diffusion rates. The rate of segregation varies inversely with overburden pressure, and close to linearly with a non-dimensional quantity frequently used to characterize granular flows called the inertial number, for both size and density bidisperse flows. The rate of diffusion is independent of overburden pressure for the range of simulation conditions tested. A continuum transport model for segregation is adapted for granular shear flow under pressure according to the inertial number dependence of the segregation to predict the steady state segregated concentration profile. The model results match the results of the DEM simulations, demonstrating that the approach works. It also demonstrates the applicability of the continuum transport model in the presence of an overburden pressure. Furthermore, the dependence of segregation and diffusion on pressure suggest that mixing can be induced at high pressures, because segregation diminishes while diffusion is unaffected. Indeed, simulations from initially segregated conditions in a shear flow under relatively high confining pressure demonstrate that good mixing is induced in size or density bidisperse mixtures that would otherwise segregate at free surface conditions. To better predict segregation of industrially relevant granular material, an approach is developed to extract the segregation model parameter for industrially relevant particle mixtures using concentration profiles deposited on a heap in a small scale experiment. The measured concentration profiles are used in a continuum transport model for segregation to determine the segregation model parameter that best matches the experiments. First, the concept for the approach is demonstrated using DEM simulations of quasi-2D bounded heap flow. A sensitivity analysis of the continuum model solution in the quasi-2D heap flow geometry indicates that the approach is best implemented by estimating the segregation parameter, while providing a rough estimate for the diffusion coefficient and an accurate estimate of the flow kinematics, in particular the flowing layer thickness, as inputs to the model. The segregation model parameter estimated using the approach based on the deposited concentration profiles is compared to the segregation model parameter measured directly in the flowing portion of the simulation, demonstrating reasonably good agreement. Optimal conditions for implementing the approach experimentally are explored using DEM simulations, yielding the requirement that the influence of gap width between the sidewalls of the bounded heap be minimized to reduce its effect on flow kinematics and segregation. The method is next implemented experimentally using a modified quasi-2D bounded heap experimental apparatus that allows the deposited heap to be partitioned and sampled. The approach is tested using bidisperse (two component) mixtures of Cellets\textsuperscript{\textregistered} (microcrystalline cellulose) particles, and the resulting segregation parameter estimated from the experiments matches closely with a correlation previously derived for bidisperse mixtures of glass-like particles. To demonstrate the effectiveness of the approach, the estimated segregation parameter for each of the species interactions is used to predict segregation in a tridisperse (three component) mixture using the continuum transport model, which yields good agreement with the experimental data from a tridisperse experiment.

Creator

• Fry, Alexander Michael

DOI

• https://doi.org/10.21985/n2-kdx1-d716

Subject

• Mechanical engineering

Language

• en

Alternate Identifier

• http://dissertations.umi.com/northwestern:14808
• etdadmin_upload_685275

Date created

• 2019-01-01

Resource type

• Dissertation

Rights statement

• In Copyright