3D quantification of spatial variability of microscale features in yarns of plain woven composites

To investigate the spatial variability of microscale features within the yarns and their impact on the mechanical properties of yarns from a three-dimensional perspective, micro-computed tomography (CT) with a resolution of 1 µm was conducted to capture the microscale structures of 2D plain woven composites. A fiber path reconstruction procedure was developed to extract the fiber path. The variability in key geometrical parameters—fiber volume fraction (V_f), and fiber crimp angles (Angle-x, Angle-z)—was characterized using both the Voronoi cell method and moving window method. The random field characterization method was employed to characterize the spatial distribution of these key geometrical parameters. Finite element models incorporating these microscale geometrical features were analyzed to predict elastic and strength parameters of yarns under displacement boundary conditions. Results indicate that the geometrical parameters exhibit significant spatial variability in both transverse and longitudinal directions, which is intrinsically linked to the manufacturing process, particularly Angle-x. Random field modeling of microstructural parameters reveals that Angle-x exhibits larger correlation lengths and more regular spatial distributions than V_f and Angle-z. Besides, Angle-x and Angle-z correlation lengths vary significantly across different yarn cross-sections. While elastic properties are primarily governed by V_f and are predictable using the Chamis model, strength values are significantly affected by both fiber distribution and misalignment. Longitudinal strength decreases with misalignment but remains correlated with V_f, whereas transverse and shear strengths are strongly influenced by local fiber distribution patterns. This work underscores the importance of incorporating spatial variability of microscale features for accurate multi-scale damage analyses in woven composites.