Understanding Material Response Across Length Scales Through the Use of Displacement Tracking and the Analysis of Large, Multi-Modal Data: Recent Advances and Challenges

Abstract: As in problems relating to the flow of fluids, experiments in solid mechanics often rely on minimally-invasive, optical measurement techniques to characterize system response. One approach that is under rapidly increasing utilization in the experimental mechanics community is that of digital image correlation (DIC), which is an optical metrology technique wherein random patterns on a specimen are tracked during thermo-mechanical testing. The subsequent correlation of these patterns results in high-resolution displacement fields in a manner that is length- and time- insensitive, assuming that clear, undistorted images of the tracking patterns can be captured. To achieve high-resolution, accurate displacement fields from DIC, several factors during imaging, such as pattern quality and dispersion, must also be tuned. The resultant data sets can be on the order of millions of data points in a single test, where each of these data points can themselves have an associated 100+ variables regarding its relationship to the surrounding material landscape. The capture and alignment of these large multi-modal data sets, at different resolutions and length scales, will be discussed, using optical and scanning electron microscopy at the micron length scale to illustrate the approaches. The application of large data analytics and machine learning on this multi-dimensional data structure in order to pull out fundamental, complex interactions will be addressed. Finally, there will be a brief discussion of exciting new directions and inherent challenges in these experimental and analytical approaches, particularly with respect to the large data analysis of multi-modal data.