To investigate materials whose properties are determined by effects on multiple scales, homogenization is a promising tool. Possible applications span from apparently homogeneous materials like metals, whose microscopic grain structure is examined, to obviously hierarchically organized fiber-reinforced polymers, on to porous materials, bio materials, etc. In the scientific community the homogenization of elastic, viscous and plastic material is well established and widely used. However, homogenization in case of damage is an ongoing field of research. Different proposals have been made in the literature to handle strain localization in case of strain softening or developing cracks. Mostly these approaches have been applied to examples for numerical verification and to reproduce experimental findings in an empirical approach.
In the present contribution the potential of predictive calculations shall be investigated. Therefor two topics will be addressed. On the one hand, a homogenization scheme suitable for damage has to be applied. Care has to be taken to utilize statistical representative volume elements and to ensure independence of the size of the microstructural volume element when computing effective macroscopic material quantities. Also boundary conditions are necessary, that allow for the localization of strain without leading to spurious localization zones. Herein special attention shall be payed to the latter point. On the other hand, material properties for the constituents on the microscale have to be derived, to render a predictive homogenization possible. The investigation at hand is concerned with glass fiber reinforced epoxy resin. In this example the properties of the fiber and the matrix have to be studied individually by experiments. Furthermore the interface between both materials needs to be examined. To this end experiments on several models of single fiber composites have been developed in the literature. For the present material combination single fiber fragmentation tests and pullout tests have been conducted and evaluated.