In this study, an approach to numerical modeling of short fiber-reinforced composites is presented. Based on the multi-scale approach, elastic material parameters are transferred from the mesoscale to the macroscale, while maintaining their stochastic properties. For this, a complete description of the stochastic properties, consisting of mean, standard deviation and correlation structure in the mesoscale is determined. This information provides a basis for the creation of second order Gaussian random fields, which are used for a FEM simulation. An analysis of the numerical simulation, including a comparison with experimental values, shows the influence of the correlation length on the global material behavior. What is more, it can be demonstrated that the standard deviation is a function of the correlation length, while the mean value remains independent of it.

In this study a methodology for the experimental determination of the probabilistic characteristics induced by injection molding of short‐fibre reinforced composites is presented. This approach is based on a pattern recognition algorithm for the fibre detection on the two‐dimensional micrograph. Since the fibres are of elliptic shape when dealing with a two‐dimensional arbitrary oriented cross‐section, the method of ellipses is used. The stochastic characteristics of the microstructure are described by probability density functions of the fibre length, diameter, and orientation, which are extracted from the two‐dimensional micrograph. For accurate results, sample preparation is of major significance, otherwise the microstructural characteristics are not captured correctly. The result of the fibre orientation over the cross‐section shows a typical layered structure of the microstructure. Furthermore, the extracted probability density function of the fibre diameter present as expected a normal distribution. In general, the results are in good agreement with literature and other scientific works.