Multiscale quantification of damage in composite structures

Abstract

Fatigue damage in wind turbine blades, made of unidirectional non-crimp glass fibre reinforced polymers (UD-NCF GFRP), has been researched for decades, but the understanding of the damage initiation and progression is not entirely clear. Demand for higher energy output and efficiency drives the need for larger and heavier blades, for which the complex damage interaction needs to be better understood. This study focuses on the tensile fatigue damage mechanisms in a UD-NCF GFRP, their relation to the material architecture, the evolution of strain, and stiffness reduction. A correlative time-lapse workflow was developed using 3D imaging modalities of x-ray computed tomography (XCT) & scanning electron microscopy (SEM), strain characterization from digital image and volume correlation (DIC-DVC), with tension-tension fatigue testing. The stiffness reduction was attributed to damage found from DIC-DVC strain maps, and damage observed in XCT. These damaged regions of interest were excavated in SEM to enable high-resolution studies. The results show that damage initiates independently on the surface and the bulk. Surface imperfections including voids and micro-scale notches lead to damage. The voids give rise to matrix cracking, progressing into off-axis cracks in the supporting backing bundles (BB). Off-axis cracks then propagate into the neighbouring load-carrying UD bundles, bridged by matrix cracking, at which point the deformation is severe and the strain localisation is observed in DIC-DVC strain maps. Due to stronger UD fibres failing, the local region becomes more compliant, leading to UD and BB fibre failures in nearby regions and a significant loss in stiffness. In the bulk, UD fibre breaks originate close to BB and proceed more in width than thickness due to the reduction of BB-induced waviness in width. These clustered UD fibre breaks lead to matrix cracks in resin-rich regions, leading to near-off-axis cracks and UD fibre breaks. UD fibres away from the backing bundle exhibit late-stage failure due to absence of waviness and misalignment. Through DVC, bands of strain concentrations are observed, in higher compliance regions that are resin-rich and have backing bundles. This is also corroborated by the damaged regions observed predominantly in these bands, and with a tensile model that shows higher stresses in these bands. At some point, this surface and bulk damage can possibly join up with larger splits to progress further and can lead to complete failure. This PhD thesis is a two-pronged approach. In addition to the experimental workflow, image analysis methods were developed to quantify fibres and damage. A novel workflow to trace individual fibres and segment phases to study their morphology was developed and compared against existing workflows. Another novel method to automatically detect damage and its evolution in a series of 3D images using fibre tracing and machine learning was developed. A combination of these parts is crucial in understanding complex fatigue behaviour. A better understanding of the fatigue damage mechanisms allows us to design better fibre architecture by optimising the weak links including backing bundles, resin-rich regions, surface defects, and densely packed fibres. The improved composite can withstand higher stresses and longer service life. This also enables accurate modelling of fatigue behaviour, ultimately pushing the design limits and reducing the cost of energy. The correlative workflow and the image analysis methods can be applied to other types of FRPs to study a variety of time-dependent phenomena.

Date of Award	31 Dec 2023
Original language	English
Awarding Institution	The University of Manchester
Supervisor	Philip Withers (Supervisor) & Timothy Burnett (Supervisor)