Search Results

The incorporation of reliability theory into a fatigue analysis algorithm is studied. This probabilistic approach gives designers the ability to quantify “real world” variations existing in the material properties, geometry, and loading of engineering components. Such information would serve to enhance the speed and accuracy of current design techniques. An automobile wheel assembly is then introduced as an example of the applications of this durability/reliability design package.

A concise method for modeling nonstationary fatigue loading histories is presented. The mininum number of model parameters is achieved by fitting the variations in mean and variance by a truncated Fourier series. An autoregressive moving average (ARMA) model is used to describe the stationary component. Justification of the method is made by comparing fatigue relevant parameters obtained when subjected to the original and reconstructed histories. In spite of a relatively small number of parameters required, the model is shown to give good results that fall within the bounds predicted by the orginal history.

Fatigue lives are predicted by a variety of methods, including stress-life curves, local stresses and strains, fracture mechanics, and service simulation tests. These approaches are reviewed as to their advantages and disadvantages, and also as to their relationships with one another and their place in the design process. Life prediction for variable amplitude service-type load histories is considered in some detail. It is concluded that it is important to apply appropriate cycle counting and to account for interactions among high and low stresses. It is suggested that life predictions which consider both crack initiation and crack growth are often appropriate.