Computational analysis for evaluating the dynamic behavior of birefringent samples in digital photoelasticity

Abstract

Analyzing mechanical stresses in loaded bodies is relevant for diverse engineering areas. Among other possibilities, this kind of analysis is useful to identify failure conditions. One experimental technique for analyzing mechanical stresses is digital photoelasticity, which allows visualizing the stresses on the surface of a loaded body as fringe patterns. Digital photoelasticity stands out from other techniques for evaluating mechanical stresses, due to their non-contact, full-field and visual capabilities. Nonetheless, using photoelasticity to characterize zones with different stress concentration levels is not an easy task. This is due to the complexity of conventional photoelasticity algorithms for decoding stress information modulated by the fringe patterns. This manuscript presents a computational strategy to characterize zones with different stress concentration levels. The strategy relies on analyzing pixel trajectories extracted from sequences of photoelasticity images. The trajectories are transformed into histograms to quantify their changes of orientation. Finally, the characterized trajectories are grouped by using clustering.