Abstract
Composite airframes are designed to be lightweight and robust. In order to avoid the regions of enhanced stress around mechanical fasteners, composite airframes also make use of adhesives to bond components. In doing so, the stress concentrations arising from joining are distributed over a larger area. When joining composites with adhesives, it is important to consider the surface condition prior to bonding; any contaminant present at the surface may result in a reduction in performance. If contaminants are present, they must be appropriately removed before the adhesive is applied and this is one major area of interest in this work. A second major area relates to understanding how bonded composite structures perform under loading. Advancing up the design tree, mechanical testing becomes more expensive due to the costs of specimens and testing, and thus there is an onus to produce finite element (FE) models that can accurately predict how these structures may perform. FE models require experimental testing to be undertaken for model validation, and the mechanical testing of such a structural element and its behaviour in relation to an FE model of the test specimen is the second major area of this work. With regard to the characterisation of surfaces to be bonded, two aspects were considered. The first aspect considered a limited investigation into two surface inspection techniques for pre-bond qualification that were considered to show potential for automation. Whilst both techniques were able to detect a release agent contaminant on all surfaces tested, both techniques were unable to discern a hydraulic oil or barrier cream contaminant on peel-ply finished surfaces; it was therefore concluded that further development, beyond the scope of this thesis, was required before they could be integrated into a production line. The second aspect of work on surfaces to be bonded considered the atmospheric plasma treatment (APT) of a carbon fibre/epoxy resin composite surface. The removal of a silicone-based release agent using APT, characterised using surface analytical techniques, was coupled with adhesive bond strength measurements. Failure surfaces of adhesively bonded lap-shear joints that had been bonded at various intervals in time after treatment were also investigated. It was concluded that the contaminant was not removed, but rather converted to a more stable silica form. The work on the bonded composite structure investigated the mechanical testing to failure of adhesively bonded composite T-Joint structural elements. The full-field strain contours exhibited by the T-Joints during the load cycle were obtained through digital image correlation and compared to the predictions of an existing finite element model. Good agreement was found between the model predictions and the measured strains. Regions of high strain indicated by both the digital image correlation results and the finite element analysis were indicative of potential failure initiation sites in the fractured specimens and there was some evidence of these initiation sites on the specimen fracture surfaces.