Stage M2: Dynamique de particules

Particle-laden flows, where solid particles are transported and interact within a fluid, are ubiquitous in both natural and industrial systems. To name a few examples, in atmospheric sciences, pollens are dispersed by strong winds and can affect the air quality. In marine systems, the water quality can severely deteriorate when plastic debris remobilize from riverbeds or when biological deposits resuspend from the interior of pipes. Near surfaces, particles experience intricate dynamics driven by hydrodynamic forces, particle-particle collisions, and adhesion, leading to phenomena such as clustering, resuspension, and collision propagation. Accurately predicting and optimizing these processes remains challenging due to their nonlinear and multi-scale nature, which requires a combination of physics-based modeling, high-fidelity simulations, and advanced data-driven techniques.

The main goal of the internship is to design effective strategy to displace passive particles by controlling the water flow in order to facilitate collision-propagation effects in near-wall particle dynamics. To answer these questions, the student will develop multi-fidelity numerical simulations of passive particle dynamics, combining simplified asymptotic models with high-fidelity simulations solving the Navier-Stokes equations. These models will be used to optimize flow-control strategies for enhancing particle motion through machine learning and reinforcement learning techniques (e.g., Q-learning, PPO). The approach will be validated with experimental data, ensuring consistency between simulations and real-world measurements.