Failure probability prediction of delamination under cyclic loading in composite laminates using cohesive interface elements

One major characteristic associated with fatigue behavior in fiber reinforced composites is the large quantity of uncertainty, which needs to be considered for more reliable evaluation of composite structures. In this work, a novel method for predicting the probability of delamination is proposed. Fatigue cohesive elements are utilized to model the propagation of delamination under cyclic loadings, where delamination path tracing is realized by a non-local algorithm, which also enables the propagation of uncertainty information between adjacent elements. A local mean and standard deviation (std) correction scheme is proposed that compensates for the mismatch of uncertainty information caused by the discretization of the delamination area as well as eliminates mesh size sensitivity. Numerical verifications are conducted under both pure-mode and mixed-mode loadings. Kullback-Leibler (KL) divergence is used as a metric to evaluate the performance of the proposed method, where good correlations between the numerical and analytical results are obtained.

Automated summary

A novel method for predicting the probability of delamination in fiber reinforced composites is proposed in this study, utilizing fatigue cohesive elements to model delamination propagation under cyclic loadings. The method uses a non-local algorithm for delamination path tracing and a local mean and standard deviation correction scheme to compensate for uncertainty information mismatch caused by discretization.