Data-Informed Modelling of Aerosol Resuspension under Aerodynamic Loads

The resuspension of small particles is a phenomenon which influences many processes across a broad range of fields and industries. It is a highly complex mechanism because of the variety of combinations and interactions between particles, surfaces and flow parameters. Due to their ubiquitous nature, resuspended particles can be hazardous and even pose a direct threat to people's health and safety. While numerous models have been developed to predict other particle transport behaviours such as dispersion and dilution, this is not the case for resuspension, which can currently be investigated only through a reduced selection of mechanistic or empirical models with insufficient experimental and numerical sensitivity to fully support them.
By coupling experimental and numerical approaches, this project will investigate the influence of individual parameters involved in the resuspension process on adhesive interactions and intends to develop simplified data-informed models with higher accuracy and applicability. Multiple techniques will be used, including numerical modelling using Computational Fluid Dynamics (CFD) to evaluate a broad range of complex scenarios, and wind tunnel testing to perform measurements on real-life cases and cross-validate simulations. The data obtained from both will be used to develop, train and test models powered by machine-learning-based Physics-Informed Neural Networks (PINNs). Novel resuspension prediction tools will be developed through the combination of these methods, thus broadening the knowledge of the phenomenon and increasing the versatility of scenarios that can be elucidated.