The settling of small particles in airflows has been a subject of wide interest for some years now. Refined models for the particle trajectories exist (e.g. Guha, Annu. Rev. Fluid Mech. 2008, and references therein). Even though many terms of the models have been corroborated using numerical and experimental studies, some uncertainties remain. One of them is a precise and quantitative description of the effect of turbulent diffusion. We tackle this challenging issue in this paper thanks to an experimental study that involves multiple interactions between particles, turbulent flow and passive obstacles and surfaces. 

Namely, we try to better understand the transport, the settling and the trapping of micrometric particles in an airflow that becomes suddenly turbulent due to a geometrical trigger (referred to as the obstacle). Particles are made of starch powder with a size range of 1-30 µm. The considered obstacle is a circular cylinder, with a corresponding Reynolds number of Re=800. A collecting surface (referred to as the splitter plate) is located downstream the obstacle in order to trap the particles advected by the flow. Particle deposition is measured using two kinds of weighing scale: one precision scale we use as a measure of the deposited particles after a long-time experiment; and another one adapted from an aerodynamic scale that gives time-resolved but less precise measurements. Two types of splitter plates (solid plate and grids) are used to study the effect of the collecting surface porosity on the flow and the possible fluxes going through the plates, yielding different particle trajectories and consequently different particle deposition. The splitter plates are also either attached to the cylinder or located in the wake of the cylinder, one at the end of the vortex formation region and the other further downstream. This enables to create different large-scale flow dynamics, from a back-step-like wake to a cylinder wake impacting a plane plate. 

Flow statistics (mean and fluctuating velocities) are obtained using a standard 2D2C PIV setup with oil droplets seeding, giving an insight into the flow topology and the large-scale structures therein. The results are then compared to statistics obtained for starch particles, and correlated with statistics of deposition from the two balances. Hence, this study allows us to study a set of configurations with different turbulence ratios and different large-scale structures. From the configurations that favor the maximum particle deposition on the collecting surface, we finally infer how the turbulence level and the large-scale flow organization influence the mean deposition of particles.