Analysis and modelling of induced-turbulence in two-phase flows

The present thesis deals with the study and modeling of turbulent two-phase flows and in particular with the mechanisms of turbulence production by a dispersed phase. The objective is to propose reduced models for practical applications and to understand the underlying physics thanks to direct numerical simulations (DNS). The first part covers gas-solid particles flows with a Eulerian (for the fluid) – Lagrangian (for particles) approach. We have proposed a stochastic model for tracking particles in a turbulent Large-Eddy velocity field (LES) in dilute flows, with results very close to DNS. When the mass fraction of the particle phase is significant (dense flows) the effect of the particles on the fluid is accounted for with a two-way coupled stochastic model, which considers also particle collisions, in a Reynolds-Average equations context (RANS). This model has been tested and validated in different homogeneous and non-homogeneous cases, and in a Cluster Induced Turbulence configuration. The second part focuses on the study of the upward motion of bubbles in a liquid. The numerical code (Basilisk) that we have used has been validated against different literature studies. We have then investigated several technical problems, like the criteria for the local refinement of the adaptive mesh. Finally, we have studied a bubble column configuration in two and three dimensions, analyzing the spectra and PDFs to describe the fluctuations induced by the bubbles in the liquid.