Abstract
In-situ X-ray computed tomography (CT) techniques enable to examine microstructural evolution of composite materials continuously and nondestructively under loading for better understanding the initiation and propagation of microscopic damages. This study explores the tensile damage behavior of notched woven carbon/epoxy composites under successive loading process. Since the image contrast of different composite material phases is normally low and conventional image segmentation techniques are hard to identify microstructures and defects in different directions distinctively, a U-Net image segmentation model based on deep machine learning algorithms is employed to characterize microstructures and material damages precisely and consistently. Digital volume correlation (DVC) approach is proposed here tocharacterize the three-dimensional (3D) deformation fields of carbon/epoxy composites at different loading steps, and the DVC results are used to predict location of damage initiation and propagation. It is found that the damage modes of notched carbon/epoxy composites mainly include fiber breakage, longitudinal cracks in warp fiber tows, and transverse cracks in weft fiber tows. The individual fiber breakage occurs in warp fiber tows at the intermediate loading level, and longitudinal and transverse cracks appear in further loading levels. Longitudinal cracks and fiber breakage events take place in a region tangent to the notch tip and exhibit a strong interaction. The development of longitudinal cracks promotes the initiation and propagation of transverse cracks. In addition, the strain concentration regions in 3D deformation fields match well with the microscopic cracks segmented by the U-Net deep learning model. This study demonstrated that the integration of X-ray computed tomography, digital volume correlation and image semantic segmentation enables to identify damage initiation, evolution, extent and mechanism.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have