Pultruded composite materials are typically characterised by a high strength-to-weight ratio, excellent corrosion resistance, high durability and high fatigue resistance, making these materials competitive with metallic alloys in many applications. Pultruded fibre-reinforced polymer matrix composites are typically characterised by a high percentage of unidirectional (UD) fibres running parallel to the pultrusion direction. The uniaxial fibres result in anisotropic properties, with higher stiffness and strength in the longitudinal direction. However, off-axis fibres are often included in the pultruded profile to provide additional load-bearing capacity in response to non-axial loads. The present work examines the influence of fibre architecture on the mechanical performance of GFRP pultrudates and the accuracy of analytical methods for predicting the mechanical response of pultrudates. Specifically, the validity of modelling pultrudates as a laminated composite material is investigated by comparing experimental results with those from classical lamination theory (CLT) and finite element (FE) modelling.

In this work, four different types of planar fibre architecture typical in pultrusion manufacturing are characterised in terms of their mechanical properties and microstructures. The fibre architecture are unidirectional rovings, chopped strand mad, biaxial mat, and quad mat. The fibre architectures were subject to tensile, flexural, and shear testing with coupons obtained from pultruded and hot-pressed samples. The experimentally determined properties were compared to one another and with those reported in the literature.

Additionally, three laminate pultrudates consisting of two types of fibre architecture consolidated through pultrusion were studied. The laminates were subject to tensile and flexural testing. CLT was then used to predict laminate elastic properties. The experimentally determined properties were compared to the CLT calculations. The CLT results were accurate for the chopped strand laminates, however, not for the quad mat laminate. The discrepancy between CLT and testing most likely stems from the CLT assumptions of thin lamina, presence of undulation, and hot-pressed specimens not accurately representing the microstructure of the pultruded composites.

Finally, full-scale flexural testing of a rectangular hollow section (RHS) pultrudate was undertaken to determine the load-deflection and stress-strain relationships. The FE model was used to replicate flexural loading of the laminate pultrudate at both the coupon and full-scale levels. The results from FE analysis and CLT were compared with experimental test results. In general, close agreement of the flexural response was observed between FE analysis and CLT and experimental results at the coupon level; however, the results from FE analysis overpredicted the flexural properties of the RHS at the full-scale level.