Preferential concentration of heavy particles in turbulence

Particle-laden flows are of relevant interest in many industrial and natural systems. When the carrier flow is turbulent, a striking feature is the phenomenon called preferential concentration: particles denser than the fluid have the tendency to inhomogeneously distribute in space, forming clusters and depleted regions. We present an investigation of clustering of small water droplets in homogeneous and isotropic active-grid-generated turbulence. We investigate the effect of Reynolds number (R-lambda) and Stokes number (St) on particles clustering in the range R-lambda similar to 200-400 and St similar to 2-10. Using Voronoi diagrams, we characterise clustering level and cluster properties (geometry, typical dimension and fractality). The exact same Voronoi analysis is then applied to investigate clustering properties of specific topological points of the velocity field of homogeneous isotropic turbulence obtained from direct numerical simulations at R-lambda similar to 220 and 300. The goal is to compare clustering properties of actual particles with those of such points in order to explore the relevance of possible clustering mechanisms, including centrifugal effects (heavy particles sampling preferentially low-vorticity regions) and sweep-stick mechanisms (heavy particles preferentially sticking to low-acceleration points). Our study points towards a leading role of zero-acceleration points and sweep-stick effects, at least for the experimental conditions considered in this study.