Nondestructive examination of polymer composites by analysis of polymer-water interactions and damage-dependent hysteresis

Polymer composites are currently replacing metals in applications requiring design flexibility, high strength-to weight ratio, and corrosion resistance. However, the damage modes in these materials are very different from metals and require specialized techniques to detect internal flaws which may exist even in the absence of visible surface damage. This study proposes a technique for damage detection in polymer composites which uses naturally absorbed moisture as an 'imaging' agent. The locally higher concentration of water in the 'free' state at damaged regions and the tendency of such water to quickly migrate to and from damage sites-exhibiting damage-dependent hysteresis-is leveraged for damage detection. To identify damaged regions, a machine learning approach is adopted using logistic regression to classify local regions as 'undamaged' or 'damaged'. New possibilities resulting from higher sensitivity levels achievable by damage-dependent hysteresis are highlighted, providing a pathway to field deployment of the novel damage detection technique.