Random features of the fatigue limit (Q1582508)

The possibility of estimating the fatigue limits of metallic materials for design purposes is investigated. The fatigue limit is defined as the highest stress amplitude for which material in question has an infinite life. The most common method for fatigue limit estimations is to make laboratory tests on a number of specimens and use some statistical method to analyze the results. Another method is to do a microscopic investigation of cross sections of the material and to use an empirical model to determine the fatigue limit from the distribution of the observed defects. The authors first summarize the advantages and disadvantages of these methods. It turns out that one important disadvantage with the established laboratory test method is that the choice of statistical distribution is arbitrary. On the other hand, in the defect analysis method there are physical arguments for the choice of an extreme valued distribution, but the inspection procedure leads to a huge extrapolation in order to obtain the fatigue limit. The authors show how the distributional knowledge from the second method can be used to improve the laboratory test evaluation method.