Fatigue Damage Evaluation of 3D Woven Composites

Abstract

Fibre-reinforced composite materials are used for the load-bearing and structural materials for different applications, including wind turbine blades and aerospace structures. Fatigue failure is one of the most common failure types when the materials are in service. It is important for us to understand the fatigue damage mechanism to prevent and predict catastrophic fatigue failure. This work uses advanced X-ray CT characterisation technique to investigate the fatigue damage development of 3D woven composites under tension-tension and short beam shear fatigue loading. The initial work began on the tension-tension fatigue testing of hybrid glass/carbon fibre 3D woven composites. It is the continuous work from a previous PhD research. There is no significant difference in the fatigue damage mechanism between hybrid composites and pure glass or pure carbon fibre composites. Then, the research moved to the second part. I first investigate the inter-laminar fatigue performance and the fatigue damage development of different Z-binder architecture 3D woven composites and the hybrid one using the in-situ time-lapse synchrotron X-ray CT experimental environment. The damage initiated from the bottom of the specimen as the matrix cracking or weft tow debonding. Then delamination happened at the mid-plane of the specimen. The top surface of the specimen under the loading point experienced the compressional stress concentration and has the Z-binder tow fracture and pull out. Different damage modes were distributed among the specimens with the increase of fatigue cycle but not connected. The warp tow shear fracture due to the large deflection is the main reason for the fatigue failure. Finally, I studied and compared different X-ray CT imaging segmentation techniques for carbon fibre-reinforced composites. The suitable segmentation method can be selected based on the imaging quality and desired segmentation accuracy. Overall, I applied the deep learning for carbon fibre bundle segmentation for the first time. The segmentation results are highly accurate, and the processing can easily be repeated for other data.

Date of Award	15 Mar 2024
Original language	English
Awarding Institution	The University of Manchester
Supervisor	Prasad Potluri (Supervisor) & Philip Withers (Supervisor)