Multiscale damage modeling of quasi-UD composites using image-based homogenization

Abstract

This study presents a multiscale micro–meso homogenization framework to efficiently capture transverse strength and damage progression in quasi-UD composites. At the micro-scale, representative volume elements with systematically varied fiber volume fraction and inter-fiber spacing are analyzed to quantify their influence on transverse mechanical response. The results demonstrate that reduced inter-fiber spacing accelerates damage initiation under normal transverse loading, particularly at low local fiber volume fractions, while its effect diminishes at higher packing densities. These micro-scale responses are subsequently incorporated into a stochastic homogenization framework and mapped onto a meso-scale geometry reconstructed from experimental microstructural characterization, preserving realistic spatial variability within carbon bundles. Meso-scale simulations reveal that damage initiates preferentially at binder–bundle crossover regions, with the precise onset governed by local fiber density variations rather than average fiber volume fraction. Model predictions show good agreement with experimental transverse tensile results, establishing a computationally efficient and physically grounded approach for architecture-sensitive analysis of quasi-UD composites.