Fatigue damage is initiated through some “defects” on the surfaces of and/or inside the component and induced by the fatigue cyclic loadings. These “defects” are randomly scattered in components, and one of these “defects” will be randomly “activated” and finally developed to become the initial crack which causes the final fatigue failure. Therefore, the fatigue strength is inherently a random variable and should be treated by probabilistic models such as typical P-S-N curves. The fatigue cyclic loading could be presented or described in any form. But the fatigue loading spectrum can generally be grouped as and described by these five models: (1) a single constant cyclic stress (loading) with a given cyclic number, (2) a single constant cyclic stress with a distributed cyclic number, (3) a distributed cyclic stress (loading) at a given fatigue life (cyclic number), (4) multiple constant cyclic stress levels with given cyclic numbers, and (5) multiple constant cyclic stress levels with distributed cyclic numbers. The approaches for determining the reliability of components under fatigue loading spectrum of the models 1∼4 are available in literature and books. But few articles and books have addressed an approach for determining the reliability of components under the fatigue loading spectrum of the model 5. This paper will propose two approaches for addressing this unsolved issue. Two examples will be presented to implement the proposed approaches with detailed procedures.

This content is only available via PDF.
You do not currently have access to this content.