Investigating the effect of polydispersity on the dynamics of multiphase flows using computational fluid dynamics tools

Granular flows consist of discrete macroscopic particles. If they are non-cohesive, their status is determined by the interaction of particle-particle frictional forces, external boundaries and gravity. In particular, the understanding of the transport mechanisms of granular materials is of paramount importance for the characterization of volcanic granular flows and for hazard assessments associated with these flows. In order to investigate dynamics of these kinds of flows, we replicated large-scale experiments with multiphase computational fluid dynamic (CFD) simulations using the Two-Fluid Model approach, with an emphasis on the polydispersity effect on the flow behaviour. The CFD simulations were run using the software MFIX. The present work consists of: 1) investigations on the drag force relationships implemented in MFIX; 2) applications of MFIX to replicate largescale experiments on volcanic dry granular flows sliding on an inclined channel; 3) comparisons between experimental and simulated data with particular emphasis on the velocity of the granular flow front. Simulations on polydisperse granular flows demonstrated the simulated flows capability to replicate segregation dynamics active in real granular flows, and the polydispersity effects on velocities and shapes of granular flows. The nonuniformity of solid phases highly affects the dynamic of the whole flow and results in a better agreement between simulated and experimental flow velocities than the simplest monodisperse particles systems. In particular, the greater the number of the solid phases, the lower the velocity of granular flows and the mean square error, which decreases by ca. 50%.