On the effect of random defects on the fatigue notch factor at different stress ratios

The effect of randomly distributed defects of different sizes on the fatigue performance of a notched member has been studied. Both the effect of the volume density of defects and the size distribution of the defects have been studied. The fatigue life and the fatigue limit distributions of a smooth or notched specimen have been estimated. Using this probabilistic approach, the fatigue notch factor as function of failure probability can be estimated. It is shown that the approach with randomly distributed defects has better predictive capabilities than classical methods. The weakest link approach can be seen as an alternative to simulation of the randomly distributed defects. The two methods have similar predictive capabilities. The stress based methods generally do not take the load ratio into account when estimating the fatigue notch factor. It is also shown that the fatigue limit estimated with defect analysis is lower than the experimental one.