Non-Destructive Identification and Numerical Investigations of Manufacturing Defects in CFRP Laminates

Carbon fibre reinforced plastic (cfrp) materials are highly attractive for the development of new lightweight structures in aerospace and automotive industry. During manufacturing, a fully defect free laminate cannot be guaranteed. Instead, numerous manufacturing defects like voids and/or fibre waviness both in-plane and out-of-plane can be induced. To exploit the lightweight potential of modern cfrp material systems, advanced numerical models in conjunction with experimental investigations must be employed.In this work, the influence of artificially induced voids into cfrp prepreg laminates onto the constitutive properties is investigated numerically. Therefore, into a cfrp prepreg laminate consisting of 80 unidirectional plies ammonium hydrogen carbonate was sprinkled between the middle layers to enforce the artificial void nucleation. From this laminate, cuboid shaped specimens were cut out for non-destructive microcomputertomographic (μCT) inspection procedure. A large variance of geometrical dimensions and pore morphologies was found. For a better understanding of the morphological correlations of related pores, a purely qualitatively driven classification into four different groups was carried out.All extracted morphologies from μCT investigations were investigated by representative volume element (RVE) technique in conjunction with structured meshing and application of the Gauss-point method. For this reason, a mapping algorithm for an automated geometry import of real defect morphologies as well as the creation of the corresponding RVEs into the finite element system Abaqus was developed. Compared to reference results obtained from RVEs with defect conform meshing technique, the Gauss-point results lie within a range of 2 % but are commonly slightly stiffer. This is caused by the rough geometric approximation. It is shown that by the application of the mapping algorithm in conjunction with the Gauss-point method, a significant reduction of the number of degrees of freedom within the system of partial differential equations is possible at satisfactory quality of the numerical results