Coherence of inertial particle clusters in the core region of a turbulent channel flow

The morphology and dynamics of clusters of small inertial particles dispersed in a fully developed turbulent flow are investigated for different Reynolds numbers and particle volume fractions. Experiments are performed in a new vertical channel flow facility for Reynolds numbers based on the channel width between 10⁴ and 2.5*10⁴. Glass-microspheres of 54 microns average diameter yield Stokes numbers based on the Kolmogorov timescale of 13 and 48 for the lowest and highest Reynolds numbers, respectively. Planar illumination high-speed imaging is performed on the central plane, with a streamwise field-of-view length of about 5 times the channel width. Clusters are identified in each snapshot using an in-house density-based clustering algorithm specifically designed to reduce the computational cost of cluster identification while facilitating their individual tracking. Shape tracking through structural skeletonization allows studying the evolution of each cluster along its lifetime. The data-analysis tools were first developed and tested on 3D data from point-particle DNS simulations. The same database is used to evaluate the relationship between 2D and 3D statistics. Different cluster formation and disintegration processes and their lifetimes are studied statistically and correlated to their morphological features.