Carbon Fiber Reinforced Polymers (CFRPs) are increasingly employed in aerospace applications due to their high strength-to-weight ratio and contribution to fuel efficiency and reducing emissions. Yet, their vulnerability to complex damage modes, particularly impact-induced delami
...
Carbon Fiber Reinforced Polymers (CFRPs) are increasingly employed in aerospace applications due to their high strength-to-weight ratio and contribution to fuel efficiency and reducing emissions. Yet, their vulnerability to complex damage modes, particularly impact-induced delamination, presents critical challenges for structural integrity and airworthiness. Delamination often initiates beneath the surface and evolves under cyclic loading in ways that remain difficult to detect and predict. Addressing this challenge, this thesis develops and validates advanced prognostic methodologies for monitoring delamination progression and predicting the remaining useful life (RUL) of CFRP structures under compressive fatigue loading.
A dedicated experimental campaign was designed, beginning with low-velocity impact testing of woven CFRP specimens, followed by compression–compression fatigue. Two active sensing-based structural health monitoring (SHM) techniques—Guided Waves (GW) and Electromechanical Impedance (EMI)—were utilized. Ultrasonic C-scan inspections were used to label delamination growth, enabling a direct correlation between sensor-derived information and physical damage evolution.
Building on this foundation, the thesis first examines the diagnostic capabilities of GW- and EMI-based indicators through advanced signal processing, establishing their sensitivity to progressive delamination and their relevance for life prediction. The research then explores how integrating the two sensing modalities enhances prognostic accuracy, showing that their complementary nature improves robustness across varying impact severities and fatigue regimes. A further line of investigation focuses on GW-based frameworks that explicitly link delamination size to fatigue life. One framework analyzes the sensitivity of individual sensor–actuator paths, providing spatial insight into delamination growth, while the other develops specimen-level correlations that generalize across different configurations. Finally, a multi-level deep learning approach is introduced, where raw GW signals, transformed components, and engineered features are processed in parallel. Embedding domain knowledge into the model architecture improves predictive accuracy, generalization under diverse loading conditions, and explainability of delamination-driven failure mechanisms.
Overall, this work demonstrates that active sensing techniques with advanced AI-driven models substantially enhance the ability to monitor and predict degradation in composite structures. The contributions provide both methodological advances in SHM and practical pathways toward safer, more efficient, and sustainable aerospace operations by mitigating the hidden risks of delamination-induced failure.