An Experimental Investigation of Fatigue Reliability Laws

Abstract

This paper begins with a review of the more familiar fatigue reliability laws, such as the normal, log normal, Weibull, etc., and the not so familiar laws of Birnbaum and Saunders [1] and Provan [2], which have been proposed as suitable descriptions of the scatter in fatigue data and the reliability of components being subjected to fatigue loading situations. Except for the latter two, all of the above laws are, to some extent, based on empiricism with little or no effort being made to relate the form of these laws to the microstructural behavior of the material being investigated. The Birnbaum-Saunders model, on the other hand, is based on a probabilistic description of the microstructural fatigue crack growth process but is, unfortunately, based on the known to be false premise that the fatigue crack grows per fatigue cycle an amount which is independent of the size of the crack. The reliability law of Provan, however, does depend on the crack size in the sense that it is based on the stochastic fatigue crack growth being described by a Markov linear birth process.The main purpose of the present paper is, then, to present the results of an experimental program developed to ascertain the validity of these laws, especially in relation to the “Provan” law. The program involved fatigue failing a statistical number of specimens whose manufacture and preparation were carefully controlled thereby being able to infer that the resulting scatter in the fatigue data is due, in part, to the microstructure itself rather than to variations in the surface and testing conditions. The program further involved the viewing of a selected number of specimen fracture surfaces using a scanning electron microscope to determine the basic material crack growth transition intensity required for the implementation of the “Provan” law. In this way, the applicability of both the empirical reliability laws and the microstructural laws is assessed.KeywordsFatigue CrackFatigue Crack GrowthFatigue Crack PropagationFatigue DataFatigue ReliabilityThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.