A plausible strategy for modeling fatigue life variation

In this paper stochastic modeling is used to predict fatigue life uncertainty by simulating small variations in strain-life material constants. The Monte Carlo method,^[1] using either known cumulative distribution functions (CDFs) for the material constants and/or postulated CDFs, provides the mechanism to generate a set of failure data. These data are analyzed, using ordinary statistical techniques, to develop the Weibull CDF for cycles to failure. It is seen that small variations (∼±15%) in values of the strain-life constants result in large variations (∼600%) in predicted fatigue life at moderate strain. The often-observed phenomenon that spread in fatigue data is greater at lower strain is also described by the stochastic model.