DamageDict - Damage prediction from the fiber level to component level


To be able to exploit the lightweight and performance potential of fiber-reinforced plastic (FRP) composites in the best manner, their anisotropic mechanical properties must be precisely determined. In addition to anisotropy, the complex multiscale FRP structure represents a major challenge. FRP components often consist of multiple textile layers, which in turn are composed by a large number of fiber bundles and the latter typically contain thousands of fibers. In this context, the fibers are the crucial component in FRP for achieving the excellent weight-specific mechanical properties. This makes the knowledge of their orientation essential, followed by the resulting acknowledgment of failure behavior and strength limits of the material - all of this making them critical to the design of FRP components. While a multitude of different strength criteria exists, they introduce problems and inaccuracies. Easy-to-use criteria, such as that of maximum stress or strain criteria, are too inaccurate and do not account adequately for the different failure modes of composites. Puck's criterion, on the other hand, predicts the failure of FRP structures very well - but is very complex in application and requires a very high experimental testing effort to determine all the necessary material parameters to apply the criterion.

In order to keep the effort for the determination of the characteristic values and the component design as low as possible, it is expedient to supplement or even partially replace the experimental investigations by simulative methods. The simulative computational effort can be kept within limits by addressing the multiscale FRP structure with multiscale simulation methods. This means, that entire components do not have to be resolved to fiber level, but the micromechanical properties at the fiber level are calculated in much smaller micro models (Figure 1). These results are used in a processed form for the calculation of textile layers and again this output is in turn used for component simulation. Such a multiscale method allows minimizing the complexity as well as the level of detail and thus the computational effort. In the DamageDict project, such a simulation tool is being developed, which - supported by low experimental effort - predicts the failure of composite materials with high precision, using a multiscale method. Starting at the micro level, the material behavior of fibers and polymer is simulated and crack initiation is investigated in the digital material laboratory GeoDict (Figure 2). To predict failure as accurately as possible, the fiber structure is modeled using micrographs and X-ray microscope images. To optimize the micro material models (modeling of material behavior at the micro level), the properties of the fibers, the matrix and the interface between fiber and matrix are determined experimentally. Based on the structural and material models, crack initiation and propagation can be simulated. Based on these data, a finite element model, representing multiple textile layers, is used to predict the failure of the FRP (Figure 3). In the finite element model, a local weak area is specified in form of a stitching, at which damage preferably occurs and can be observed. With micro models and finite element models of the laminate, a precise component design can be carried out - in which the material properties are taken into account down to the fiber level.

The DamageDict project ("Simulative damage prediction of textile laminates based on microscopic material models") is funded by the German Federal Ministry for Economic Affairs and Energy (BMWi) as part of the "Central Innovation Program for SMEs (ZIM)" program (funding code ZF4052328LF9).

Figure 1: Reconstruction of statistically representative microscopic fiber structures (left) based on computed tomographic images and micrographs (right)

Figure 2: Crack initiation (red) in polymer matrix in a micromodel in a 2D image (left) and in a 3D image (right)

Figure 3: Setup of the finite element unit cell: stacking of the textile layers (top left), meshing with specific mesh density (top right) and stress distribution after failure of the first laminate layer (bottom).

↰ News