Resin transfer moulded composite materials : processing, structure, property relationships

Sammanfattning: This thesis contains analysis of the performance of polymer composites manufactured by resin transfer moulding (RTM). The thesis consists of six papers, where the first two deals with the tensile strength of composites reinforced with randomly oriented glass fibres, which still is the most common reinforcement in RTM manufacturing. The following four papers covers aspects of manufacturing induced defects, out-of-plane strength and dimensional tolerances of carbon fibre reinforced high temperature epoxy, which is a relatively new application for RTM. The apparent tensile strength of random fibre composites is usually significantly higher if measured in bending than in tension. In Paper A, it is shown that the deduction of strength from flexural tests is complicated by several factors, leading to large errors if not accounted for. Approximate closed form expressions are derived to account for friction at the support rollers, nonlinear material behaviour and large deflections and suggested for future use. It is then shown that properly deduced flexural strengths can be predicted from tensile test data by Weibull theory, if applied on effective fibre stress. In Paper B, the Weibull theory is extended to biaxial load cases for random fibre composites. Failure strain predictions based on tensile test data are in good agreement with experimental results from three- and four-point bending and biaxial bending. In Paper C, the robustness and geometrical limitations of resin transfer moulding were investigated for single curved laminates. The design of experiments approach was used to determine how the out of plane tensile strength is affected by variations in preforming method, radius, fibre content and vacuum assistance. Carbon fibre /epoxy U-beams with an inner radius of 0.8 mm were manufactured and demonstrated good mechanical performance. The strength of these beams were in fact less sensitive to defects than the strength of beams with a 5 mm radius. Beams with a 5 mm radius were very sensitive to the presence of voids. A good correlation was obtained between local void content and beam strength, indicating that void content is the only important defect. In-situ optical microscopy was used in Paper D to observe the micromechanisms of failure as delaminations develop in the beam radius. Beams with and without voids were studied to determine how the voids affect the failure initiation. The failure mechanism proposed, based on the observations, involves the position of the voids and the microstructure of the reinforcement. The material system investigated was a plain carbon fibre weave impregnated with epoxy by resin transfer moulding. The voids were primarily located in the matrix rich regions created by the crimp in the reinforcement. Voids were found to have no effect on the initial formation of a debond crack at the interface between adjacent fibre bundles. Instead, the voids increase the stress concentration at the crack tip and therefore crack growth starts at a lower load. In Paper E this issue is brought further by numerical modelling of an initial crack adjacent to a void. The calculated increase in energy release rate, due to the presence of a void, suggests a strength reduction in qualitative agreement with the experimental results presented in Paper C. The reasons for why the smaller radius exhibits a lower defect sensitivity is also discussed in Paper E. Finally, Paper F is a study of the springback phenomenon. The mechanisms responsible for springback during a typical RTM cure schedule are discussed. An approximate model for prediction of springback is developed, incorporating the effects of chemical shrinkage and the phase change at the glass transition temperature. The effect from chemical shrinkage is estimated from fibre and matrix properties and it is demonstrated both theoretically and experimentally that the effect from chemical shrinkage is significant. Model predictions of springback angles are in good agreement with experimental data from the beam series presented in Paper C.