First-order perturbation-based stochastic homogenization method applied to microscopic damage prediction for composite materials

In this paper, a first-order perturbation-based stochastic homogenization method was developed to predict the probabilities of not only macroscopic properties but also microscopic strain damage in multiphase composite materials, considering many random physical parameters. From the stochastic solution of microscopic strains, damage propagation was analyzed to predict where progressive damage would occur in the microstructures of composites subject to a given macroscopic strain. As an example, a short fiber-reinforced plastic, consisting of short fibers, matrix, and interphase, was used to show the influence of random physical parameters for each constituent material on the variability of the homogenized properties and microscopic strain. In another example, a coated particle-embedded composite material was stochastically analyzed to consider even slight influences of uncertainty in the mechanical properties of the coating material and show damage propagation in this coating layer. Characteristic displacements representing material heterogeneity were thoroughly investigated and extensively used with the aim of reducing the computational cost of finding them in a nonlinear analysis of the microscopic damage propagation.