Structural FE simulations of CT scanned microstructures: A comparison between idealized and real microstructures

Carbon fiber-reinforced polymers (CFRPs) are materials that already meet the demanding lightweight and strength standards imposed by the aeronautic, automotive, and civil industries. For instance, the large scale implementation of composites in the aeronautic industry is basically due to the demand for weight reduction and fuel-efficiency. However, there are still several impediments when considering large scale use of composite materials in structural engineering applications, such as cost of materials, fabrication procedures, and repair difficulties [1-3]. In addition, inspection and testing of CFRP structures are more challenging than the traditional techniques used in non-destructive testing (NDT), mainly due to the complexity of the structures being investigated. Used primarily as an NDT method, for inspecting composites against defects like porosity, delamination, or voids, X-ray computed tomography might be the most convenient way of extracting real defect geometries along with carbon fiber structures for Finite Element (FE) modelling. The models obtained using this approach are strictly valid for the investigated structure, but offer insight on the elastic and damage behavior of other components presenting similar defects. When performing FE mechanical simulations, the accuracy of the results is determined by the finite element size. Smaller element size is generally preferred while paying the cost of increased computational time, in order to ensure accurate results [4, 5]. Thus one aspect of interest for the current work is to determine an optimal element size for the typical micro-volumes being investigated. It is a known fact that beyond a certain amount of mesh refinement, the benefits of reduced computational time trump the increase in accuracy requirements [6]. In order to achieve that goal, a series of static analyses were performed on a given micro-structure model made up of CFRP and including localized defects. When trying to predict the strength of a given material region, the size distribution and the position of existing defects play a critical role [7, 8]. One can differentiate between defects that are critical to the composite strength, and defects that are not determinant for the overall strength of the composite. By focusing only on the critical defects, the computational model size can be drastically reduced. The second part of the current research shows a comparison between FE results obtained when using a real CT scanned micro-structure versus a simplified model that takes into account global porosity and main pore orientations.