Probabilistic models of fatigue crack propagation and their experimental verification

Abstract

To study the fatigue crack growth problems and to emphasize the variability of the growth curves in addition to their average growth trend, three stochastic fatigue crack growth models are presented. These models include the Markov chain model, Yang's power law model and a polynomial model. Experimental work is carried out to produce the required fatigue crack growth data, which are then used for verification of the models. Two sets of statistically meaningful data, one under constant-amplitude loading and the other under random loading, are produced. Numerical study shows that all three models can be used to depict the experimentally obtained fatigue crack growth data with certain degrees of accuracy. However, out of the three models, one is superior to the others concerning a certain data set while the other two models are better for other data sets. Comparison and comments on employing individual model are made at the end of the paper based on experience gained in the present study.