Abstract
Three-dimensional woven fabric composites (3DWFCs) provide superior through-thickness reinforcement and delamination resistance under ballistic impact, yet no multiscale modeling framework has achieved broad, independently validated predictive capability across architectures and impact regimes. This systematic review, conducted following PRISMA-guided screening of the literature from 1956 to December 2025, synthesizes multiscale modeling strategies for 3DWFCs across four hierarchical scales—fiber, yarn, fabric, and structure—with explicit classification of cited evidence as direct 3D woven data or background analogy from 2D, braided, or laminate studies. A central contribution is the systematic distinction between model calibration and independent predictive validation, revealing that reported accuracy (e.g. 8–15% V
50
error for yarn-scale finite element models) is frequently calibration-dependent rather than genuinely predictive. The review further establishes an architecture-specific framework for matrix-system effects, showing that dry, thermoset, thermoplastic, and shear-thickening-fluid impregnation interact differently with binder-yarn kinematics unique to 3D architectures. Persistent gaps are identified, including violation of scale-separation assumptions under ballistic-frequency loading, absent dynamic interfacial fracture toughness data, and the near-absence of validated multi-hit models. The review concludes with a prioritized roadmap emphasizing standardized validation protocols, physics-informed machine learning, and digital-twin frameworks, while cautioning against overstating the current maturity of these emerging tools.