Kink band fracture energy of 2D woven composites with 3D mixed-mode corrections and size effect analysis

This study improves fracture energy estimation for compression-induced kink-band failures in 2D woven composites by integrating size-effect testing with DIC-aided three-dimensional finite element modeling. Geometrically scaled single-edge-notched compression specimens of a carbon/epoxy 2x2 twill woven composite were tested. Their nominal strengths are shown to follow Bažant's Type II size-effect law, confirming quasibrittle behavior, which enables the determination of a size-independent fracture energy. However, the standard 2D dimensionless energy release rate function, which assumes uniform far-field displacement, is found to be invalid for this failure mode. This is due to pronounced three-dimensional effects, including crack-tip tortuosity and 45°out-of-plane kink bands, which were revealed via post-mortem sectioning of failed specimens. Additionally, the composite was also found to lie outside the applicability of Bao's orthotropy rescaling. To overcome these limitations, a full 3D finite element model was developed that incorporates the inclined crack front, mixed-mode loading, and DIC-measured boundary conditions. The 3D analysis shows a substantial mode III contribution and predicts fracture energies nearly 50% lower than a 2D model. The mode III effects can be attributed to tow-level out-of-plane undulations, and can be expected in all 2D woven composites. The size-effect-based R-curve further highlights the discrepancy between 2D and 3D predictions. These findings demonstrate the need to account for quasibrittleness, three-dimensionality, and fully mixed-mode fracture (including mode III effects) when characterizing kink-band failures in woven composites.