Image analysis for time-resolved analysis of microparticle resuspension kinetics on a ventilated duct surface

International audience ; This work aims to develop an experimental method to study the temporal evolution of the resuspension kinetics of microparticles. Particles are deposited as a mono-layer of isolated particles on a surface. It also focuses on airflow properties accountable for the resuspension phenomenon during fan accelerations followed by a steady state. Resuspension kinetics acquisitions provide movies of particle behaviour on the deposit. First, it enables to detect if there is resuspension. Then, by processing the frames, one can obtain parameters as particle number, concentration and repartition, size distribution and information on particle clusters – due to deposit and sedimentation process. The input of the algorithm is an untreated image, and the routines process it following the classical steps of a detection algorithm. The algorithm was validated using two processes. One compared the particle number obtained by manual counting and the algorithm, for the first and last frames of several acquisitions. The other technique consists of creating virtual images with random particle size and location on the frame and the resulting parameters were compared with the initial input values. It leads to a mean difference of about 0.1% between the particle size distributions and a mean Bhattacharyya distance of 0.13%. As it has been previously noted, the larger the particle is the easier it detaches from the surface. The algorithm can separate the particle into size classes, and the results agree with the literature. Finally, the algorithm enables detection and isolating particle clusters to process them separately. Moreover, the study of their dynamic is of interest: a particle cluster is more prominent than an isolated particle. Thus, they are more subject to collisions with other particles. The algorithm can follow the cluster evolution as well as the behaviour of the particle inside it.