Modeling the variable amplitude fatigue of composite materials: A review and evaluation of the state of the art for spectrum loading

In this article, we review the experimental research and modeling of fiber reinforced polymer composite materials subjected to variable amplitude fatigue. In general, these models are empirical or phenomenological and contain parameters that must be determined using constant, and in some cases variable, amplitude fatigue data for the material system in question. In many cases, the authors who proposed these models simply fit them to their experimental variable amplitude results and thus their predictive ability has remained uncertain. The predictive accuracy of the fatigue models examined in this article was compared by applying them to four material systems for which extensive fatigue data was available. Several approaches are rejected because the data required to fit them are not available or their formulation prevents straightforward application to a spectrum loading with blocks of one cycle in length. Four damage accumulation rules and eight residual strength approaches are fit to the static, constant amplitude fatigue, residual strength, and two block repeated fatigue loading for each data set. Then the models are used to predict the fatigue life under variations of WISPER, WISPERX, and Rayleigh distributed spectrum fatigue loads that were applied experimentally. Results show that the Palmgren–Miner rule is non-conservative in every case and gains in accuracy are possible by using a simple residual strength law such as the one proposed by Broutman and Sahu. More complex residual strength approaches gave more accurate results in some cases, but overall did not provide a significant improvement relative to Broutman and Sahu’s model.