ABSTRACT This study develops a novel mesh‐independent computational framework for prediction of the tensile behavior of woven composites. The core of the methodology is a spatial parameterization approach, where the minimum distances from finite element integration points to the centerlines of fiber yarns are calculated and employed as weighting coefficients. This enables the direct and precise assignment of heterogeneous constituent material properties at the integration point level, bypassing the need for complex, mesh‐conforming geometry discretization. The proposed mesh‐independent approach achieves an optimal balance between predictive accuracy for elastic responses and computational efficiency. The framework's robustness is first established through comprehensive simulations at the Representative Volume Element (RVE) scale, analyzing diverse woven architectures and material combinations. Building on this validated foundation, its predictive capability is successfully extended to the macroscopic component scale. Tensile tests on woven C/SiC composite specimens provide conclusive experimental validation, confirming the framework's accuracy in capturing the full‐field tensile response of an actual structure, beyond idealized RVE models. Consequently, this spatial integration point weighting methodology presents an effective virtual characterization tool. It significantly reduces the dependency on iterative physical testing during the design and development phases of woven composite materials and structures.
Zhao et al. (Wed,) studied this question.