Artificial intelligence-driven rheological identification via the Rayleigh–Plateau instability

A comprehensive understanding of fluid dynamics and the identification of intricate rheological properties are essential for both academic research and industrial applications where fluid behavior is critical. Traditional methodologies for determining these properties have primarily relied on instruments known as rheometers. However, an innovative approach introduced in Maîtrejean et al. in 2022 shifts from direct measurement to the automated identification of rheological characteristics. This method utilizes the morphology of jets formed by the Rayleigh–Plateau instability as distinct signatures, enabling the identification of fluids based on their rheology through artificial intelligence (AI). While the original study, Maîtrejean et al. (2022), was based on a numerically generated dataset focusing solely on variations in viscosity for Newtonian fluids, the current research expands upon this by employing an experimentally generated dataset. This dataset encompasses Newtonian fluids with varying viscosities, densities, and surface tensions. Herein, we demonstrate the capability of our AI-driven model to accurately identify these three properties for Newtonian fluids. This study validates the applicability of the approach to this expanded set of properties for Newtonian fluids using experimental data and sets a new benchmark for precision in this context.