Random field modeling of local strength of a randomly arranged unidirectional fiber-reinforced composite material under transverse tensile loading

This study examines the random field modeling of local strength, which is the apparent strength of each window extracted from an SEM image of a cross-section of a randomly arranged unidirectional fiber-reinforced composite material specimen. In particular, computation of the spatially varying random apparent fracture strength based on the microstructure simulation is performed. To analyze the apparent fracture strength, single-scale analyses assuming plane-stress and plane-strain conditions, as well as multiscale stress analysis based on the homogeni-zation theory, are attempted. In addition, the moving window method is employed to compute the respective random fields and their autocorrelation functions. First, the influence of the analysis method on the identified random field is investigated. Furthermore, the influence of the window and image size is discussed. The nu-merical results revealed that a smaller window leads to a more accurate evaluation of the apparent strength, and thus, is preferable for the identification of the respective random field.