A Long-Life Regime Probability-Based Fatigue Design Method for Weldments
by
Sahng Kyoo Park
F. V. Lawrence, Jr.
Abstract
A probability-based fatigue design methodology for crack initiation and early growth is presented which estimates the average fatigue strength of weldments and its uncertainty using a Monte Carlo simulation. The underlying analytical model considers several different kinds of fatigue variables: applied or induced bending stresses; notch severity; notch-root residual stresses; and notch-root mechanical properties. Each of the fatigue variables is treated as a random variable, and the uncertainty of each is modeled by an appropriate probability distribution function developed from laboratory data.
The suggested design method can determine the fatigue strength for a given design life with a certain reliability, and it can be applied to both constant amplitude and variable load history applications. The method also permits the designer to factorize the contribution of each of the fatigue variables to the overall uncertainty in fatigue strength and thus determine the economic merits of reducing the uncertainty in any one of the fatigue variables.
For this study, fatigue test data of three types of weldments, load-carrying fillet-welded cruciform weldments, non-load carrying fillet-welded cruciform weldments and butt weldments, were used for the simulation of fatigue strength and its related uncertainty.
Download (5 MB)