Multiscale XFEM–CZM simulation of crack propagation in silica/epoxy nanocomposites with interfacial contact effects

Understanding the fracture behavior of nanoparticle–reinforced polymer composites requires an accurate description of damage mechanisms operating across multiple length scales. In this study, a unified multiscale numerical framework is established to investigate crack initiation and propagation in silica/epoxy nanocomposites from the nano– to macroscale. The approach integrates representative volume element (RVE) modeling with the extended finite element method (XFEM) coupled to a cohesive zone model (CZM), enabling simultaneous simulation of matrix cracking, particle–matrix debonding, and interfacial damage. At the nano and microscale, unit–cell and clustered RVEs are employed to quantify the effective elastic modulus, tensile strength, and fracture energy under three interfacial bonding conditions: perfect bonding, frictional sliding, and cohesive interaction. Microscale responses are subsequently transferred to meso– and macroscale models to predict heterogeneous and homogenized fracture behavior. The modified Halpin–Tsai–Lewis–Nielsen model accurately captures interphase effects and demonstrates good agreement with RVE–based finite element predictions. Across all scales, interfacial bonding quality emerges as the dominant factor governing stiffness, strength, and fracture resistance. The CZM provides the most realistic representation of particle–matrix interactions and is therefore recommended for predictive fracture analysis and design optimization of particulate polymer nanocomposites.