Understanding 3D Foam Plasticity: An Integrated Approach

ongoing

Liquid foams possess distinct structural and mechanical properties, making them essential to various industries ranging from food production to ore flotation. When subjected to shear, foams exhibit complex rheological behaviors: responding elastically to small deformations, plastically to larger ones, and flow with a shear-rate-dependent viscosity once their yield stress is exceeded. These behaviors are governed by local plastic rearrangements of bubbles. Although much is known about local rearrangements in 2D foams, the understanding of 3D foams remains limited. This knowledge gap hinders accurate predictions of dynamical heterogeneities in plastic rearrangements, often leading to trial-and-error approaches when engineering foam properties for applications. This project aims to bridge that gap by combining X-ray tomographic microscopy with numerical simulations and advanced AI techniques. I will begin by analyzing data from fast X-ray tomographic microscopy experiments combined with rheological measurements, which simultaneously capture the macroscopic response of 3D foams and the detection of plastic rearrangements at the bubble scale. However, the limited temporal resolution of X-ray tomography necessitates the development of experimentally informed numerical simulations to further explore foam dynamics. Predicting the occurrence of plastic rearrangements remains a significant challenge. To address this, I will develop graph neural networks (GNNs) capable of predicting the spatiotemporal occurrence of plastic events in foams during a secondment at a company specialized in AI. GNNs have already proven effective in predicting plastic events in glasses, making them an ideal tool for this task. Since foams are an athermal analog of amorphous solids, insights from this research will be applicable to other yield-stress materials, where the interaction of localized plastic rearrangements is crucial for predicting fracture and flow.