Search

Models to predict the split growth in notched AS4/3501-6 graphite/epoxy quasi-isotropic laminates under tension-dominated fatigue are presented. First, a power law model and an artificial neural network (ANN) model are developed to describe the split growth under constant-amplitude fatigue. They are then applied in conjunction with a linear damage growth model to predict the split growth under spectrum fatigue. The ANN model is found to work better than the power law model as a predictive tool for split growth.

The subcritical damage growth in notched AS4/3501-6 graphite/epoxy quasi-isotropic laminates under compression–compression and compression-dominated spectrum fatigue was investigated using the longitudinal split length as a damage measure. A linear split growth model was successfully used to predict the split growth in spectrum fatigue using the corresponding data under constant-amplitude fatigue. However, a linear degradation assumption was found to be inappropriate as it overpredicted lifetime under spectrum fatigue. The residual compressive strength did not show any sign of gradual reduction. The damage mode near the hole changed from longitudinal splitting to transverse extension of fiber failure and delamination as the fatigue stress increased. Thus, while the longitudinal split may serve as a measure of damage in high-cycle fatigue, the extent of transverse fiber failure is a better indication of residual compressive strength. It was shown that a compression-dominated fatigue loading spectrum could be shortened substantially by omitting the two lowest load levels at which no significant damage growth was seen under constant-amplitude fatigue.