Mechanical and Civil Engineering Seminar

Particle-laden flows with inertial particles display a rich array of fluid dynam- ics depending on the particle Stokes number and mass loading. In this talk, I will present recent work in collaboration with researchers at Cornell Univer- sity on gravity-driven, gas-particle flow with sufficient mass loading to generate cluster-induced turbulence in an otherwise statistically homogeneous system. Such flows are relevant to many technological and environmental applications, and provide a non-trivial canonical flow for validation of multiphase turbulence models. In these flows, due to the mean velocity difference between the gas and par- ticle phases, particles cluster spontaneously on characteristic length scales much larger than one particle diameter (dp ). In turn, the presence of clusters leads to spatially nonuniform momentum coupling with the gas phase and, hence, to gas-phase turbulence that maintains the cluster size distribution. Analysis of the Reynolds-averaged governing equations points to the importance of particle- volume-fraction fluctuations (i.e. clusters) and their correlation with gas-velocity fluctuations in determining the statistical properties of the fully developed flow. In our work, mesoscale direct-numerical simulations (DNS) are used to inves- tigate the flow physics and to extract turbulence statistics needed for model vali- dation. Following the pioneering work of Simonin and coworkers, the Lagrangian particle data is filtered to the Eulerian grid to decompose the particle fluctuating velocity into its correlated and uncorrelated components. For cluster-induced turbulence, large periodic domains (1400dp )3 with over 100 million particles are required to obtain fully developed energy spectra and turbulence statistics. I will describe how we use the mesoscale DNS data to investigate the flow physics and to validate turbulence models for gravity-driven, gas-particle flows.