Quantitative evaluation of process-induced yarn geometric imperfection effect on mechanical response of plain-woven C/SiC composites: X-ray tomography-based finite element analysis

In this paper, the morphology and distribution of yarn imperfections induced by the fabrication process in C/SiC composites were captured and statistically analyzed by X-ray computed tomography using deep learning methods for image processing. Subsequently, these imperfections were decoupled into four typical characteristic imperfections: fiber cross-sectional shape, under-sized (over-sized) yarn cross-section, yarn cross-section variation, and yarn waviness. High-fidelity image-based finite element method (IB-FEM) model was generated from CT images, which considered the mesoscopic geometric morphology. The tensile response and failure mechanism of the as-designed model, IB-FEM model, and statistical models were established and verified by experimental results. The impact of yarn geometric imperfections on tensile properties was systematically discussed and elucidated. The results indicate that the intensification of yarn geometric imperfections (yarn cross-section variation and yarn waviness) significantly weakened the ultimate bearing capacity and failure strain of composites under tension, which have greater impacts than other geometric imperfections.