Mercredi 5 novembre 2025 à 14h00, salle des séminaires IRPHE
Abstract
Solid aggregates in dilute turbulent colloidal suspensions can break due to strong local shear fluctuations. This process is relevant for a broad variety of problems, from environmental sciences, to engineering, pharmaceutics and food industry.
In this talk, I will discuss a data-driven approach that we recently developed. This is based on a neural network - which includes geometrical features of the aggregates-, that can be trained to predict if an aggregate breaks or not, once the turbulent velocity gradient statistics is known.
Alessandra Lanotte, Istituto di Nanotecnologia, Consiglio Nazionale delle Ricerche