Defect-Induced mechanism analysis and surrogate modeling of tensile strength in woven SiC/SiC composites

Porosity defects are inherent in SiC/SiC composites and significantly influence their tensile performance. This study examines the multiscale effects of defect content using high-resolution micro-computed tomography (μCT) and uniaxial tensile testing across low-, medium-, and high-density specimens. Medium-density composites exhibit the highest tensile strength due to more uniform defect distribution and enhanced energy dissipation mechanisms such as crack deflection and fiber bridging. SEM analysis confirms the activation of multiple failure modes in these specimens. A surrogate model was constructed by integrating μCT-informed multiscale simulations with experimental data to predict tensile strength. The resulting double-Gaussian model captures the nonlinear relationship between defect content and strength with high accuracy (R²≈0.98). Compared to full-field simulations, the surrogate model offers significantly improved efficiency while maintaining accuracy, enabling rapid evaluation across a broad defect range. These findings underscore the importance of defect distribution and provide a practical framework for strength prediction in defect-sensitive ceramic matrix composites.