LAUSR

Initial defects such as microcracks and pores inevitably exist during the fabrication of carbon fiber composites, which can affect design properties of the composites, resulting in loss of structural mechanical performance. In this paper, we develop an approach with which structure failure can be effectively predicated through exact modeling of composite materials. Specially, an L-shaped composite structure with the initial defects is considered. Its macroscopic properties are obtained through an exact 3D finite element model based on Computed Tomography (CT) images of the composite material, in which the fiber orientations are determined through the machine learning. Numerical prediction of the structure failure is then conducted based on 3D-Hashin failure criterion and bilinear cohesive model, which is in good agreement with experiments. Both results show that crack initiation occurs in the defect dense area, and the structure failure is mainly caused by fiber fracture. Based on the validated approach, we further discuss the mechanical properties of the structure without defects, and found that 2% initial defects in the model can lead to a 12% reduction of the bearing capacity. We expect that our failure analysis approach can be critically useful in designing composite materials and structures.