Machine Learning and Dynamics: Soft Materials Under Stress

A yield stress fluid (YSF) is a core component of many industrial formulations. A YSF is a material that reversibly transitions from solid to liquid state under application of a load. YSFs are susceptible to delayed failure, which occurs unpredictably as the internal structure changes over time. Failure could be caused by, for example, gravitational stresses due to density differences in the material itself. Prior to material failure, microscopic precursor events can be observed which, over time, accumulate into an "avalanche", leading to macroscopic failure.1

Utilising laser sheet microscopy, the displacement of fluorescent tracer beads embedded in the matrix of a model YSF will be imaged, allowing precursor events to be identified and characterised. This method will provide a 3-dimensional image of a relatively large sample volume. Due to the large number of tracer beads present in the image, a vast amount of data will be generated; this data would be difficult and time-consuming to analyse using traditional methods. An aim is to develop and train machine learning algorithms to extract the most relevant information from, and enable analysis of, this large data set. In the future, these machine learning algorithms could be used in the prediction of delayed failure in a YSF.

[1] L. Cipelletti, K. Martens and L. Ramos, Soft Matter, 2020, 16, 82-93.