Automated crack propagation detection in adhesively bonded composite joints with Digital Image Correlation techniques

This study presents and validates a set of automated techniques for tracking crack propagation in adhesively bonded composite joints using Digital Image Correlation (DIC). Two approaches were developed: one based on vertical displacement thresholds and another on localized strain fields. The methods are fully automated from the initial crack length identification stage, using a region-growing segmentation algorithm, and are validated under both quasi-static and fatigue loading for joints with different adhesives and thicknesses. The displacement-based method proved highly effective in capturing crack growth during stable propagation, showing excellent agreement with visual measurements. For fatigue tests, Paris-law parameters extracted from the automated routine differed by less than 6% compared to manual reference values. The strain-based method was more robust under complex conditions, enabling the detection of crack trajectories in thicker joints and in cases where the crack propagated within the GFRP adherend. Validation through energy release rate (G I ) and J-integral calculations confirmed the applicability of the methods within linear elastic fracture mechanics. The developed framework offers a reliable, scalable alternative to manual tracking and is suited for routine testing, parameter identification, and future integration with data-driven fracture mechanics tools.