Fatigue life prediction of composite structures based on online stiffness monitoring

Abstract

This article presents an online statistical life prediction method based on both historical and local residual-stiffness information to reduce the effect of dispersive mechanical properties of fiber-reinforced plastic structures. Normalized residual-stiffness curves obtained from the laboratory are used as a benchmark, and the model parameters are assumed to be independent random variables. According to Bayesian statistical theory, the prior distribution of fatigue life is updated by the residual-stiffness sample information of the in-service structure through online monitoring; then, a more accurate life estimation is obtained. This method precisely captures the essence of stiffness degradation of a composite structure. The individual dispersion effect is largely eliminated through using the local stiffness data. In general, a more accurate life estimation can be obtained if more sample information is available. The test results of glass-fiber-reinforced plastic and carbon-fiber-reinforced plastic laminates show that the fatigue life can be accurately predicted with only the first 20% residual-stiffness data known.