Prediction of the fatigue limit : accuracy of post-processing methods

Abstract: Several fatigue criteria were evaluated with experimental fatigue test data. The fatigue tests contained many different geometries with different degrees of stress concentration. It was concluded that the differences between the criteria were neglectable when compared to the differences between the specimens. The stress concentrations with the stress gradients and sizes of the highly stresses volumes that resulted were very important. In order to understand this, gradient and volume methods were investigated. The gradient methods showed some improvement over the strictly local evaluation first attempted. The volume method, a weakest link application with a Weibull distribution of the fatigue strength, yielded results useable for actual design situations. The volume method is strongly recommended and the choice of criterion does not seem to be very important.