Local buckling, post-buckling and failure of pultruded GFRP I-section columns with narrow and wide flanges

This study examines the local buckling (LB) behaviour of pultruded glass fibre-reinforced polymer (pGFRP) I-section columns through experimental, computational and design-based approaches. The experimental program targeted compact to slender wide- and narrow-flange I-section columns to clearly capture LB modes. Digital Image Correlation (DIC) was used to monitor strains and out-of-plane displacements during the post-critical phase up to failure. Prior to testing, high-precision geometric imperfection (GI) measurements were obtained using a 3D Coordinate Measuring Machine (CMM). Material properties were characterized through coupon and full-section compression tests. Experimental critical buckling loads were estimated using the P – w 2 method and compared with analytical and computational predictions obtained from linear bifurcation analyses performed with both finite element (FEM) and finite strip (FSM) models. A shell-based FEM model incorporating Hashin’s failure criteria—including damage propagation via fracture energies—and measured initial GIs was developed to assess post-buckling behaviour. Results demonstrate a significant post-buckling strength reserve in slender pGFRP columns, up to 85%. Current design standards are found to be overly conservative in estimating their ultimate capacity under LB-driven failure (up to 370%). The study identifies possible sources of uncertainty that affect the computational and analytical predictions, including: (i) subjectivity in experimental determination of critical load, (ii) material property variability, and (iii) limitations of homogenized models—both, computational and analytical—in capturing complex material characteristics.