Maximum entropy method for probabilistic bearing strength prediction of pin joints in composite laminates

The mechanical properties of composite materials exhibit a much wide range of scatter than that of metallic materials, even under the same loading condition. A general computational framework for probabilistic strength prediction of pin joints in composite laminates using maximum entropy method is proposed and developed in this study. The deterministic progressive failure analysis of composite pinned joint is based on a two-dimensional ABAQUS model, which takes into account large deformation theory and the non-linear shear behavior of composite material. Uncertainties arising in the mechanical properties and ultimate strengths of one lamina are modeled as random variables and considered in the process of probabilistic strength prediction. Point estimate method is used to estimate the first four statistical moments of bearing strength of composite pinned joint. Incorporating all those statistical moments, a maximum entropy method is used to approximate its probability distribution to avoid the pitfall encountered in Edgeworth expansions, which is a popular tool to approximate a probability distribution. An example of T300/QY8911 composite plate is given to demonstrate the proposed computational framework. Comparisons between the proposed framework and Edgeworth expansions based method are also performed to illustrate the advantage of the proposed framework.