High-fidelity computational micromechanics of composite materials using image-based periodic representative volume element

This study presents numerical investigations into the progressive damages in unidirectional carbon fiber reinforced plastics using computational micromechanics with an image-based periodic representative volume element (RVE). This RVE modeling approach combines the methodologies of image-based RVE and artificial periodic RVE to reflect the actual fiber packing while retaining geometrical periodicity. For high-fidelity modeling, the RVE model is prepared with more than 200 carbon fibers extracted from X-ray computed tomog-raphy images. Both homogenization simulation for elastic properties and unidirectional loading simulations for plastic and fracture properties are carried out. The constitutive model parameters are identified and verified by comparing the experimental and predicted effective (spatially averaged) elastic and plastic properties, and ultimate strength. Then, with the identified parameters, the local damage initiation and propagation mechanism, and their influence on the macroscopic nonlinear stress-strain behaviors are discussed. Finally, the influences of fiber packing on the prediction accuracy of computational micromechanics are discussed by comparing the results between the proposed image-based RVE and simple RVEs with the square and hexagonal arrays.