Brief description

The prediction of the fatigue life of a steel welded joint, which is frequently subjected to repeated stresses, is usually carried out using notch case catalogues based on experimentally determined Wöhler curves. The methods currently used to determine them, based on fatigue tests carried out in the time-strength range, do not statistically allow these curves to be extrapolated into the High Cycle Fatigue (HCF) range.

Various methods can be used to computationally estimate the fatigue life of a notch detail.

One approach consists in the statistical evaluation of test data by a given function (Wöhler curve), which is used to describe the number of fatigue cycles as a function of the applied stress ranges.

The description of a Wöhler curve is a complex mathematical problem due to the stochastic nature of the fatigue process. Starting from the Weibull distribution, a dimensionless model is proposed to describe the Wöhler curves. There are several mathematical methods to determine the parameters of the Weibull model, and these methods depend on the experimental data.

One advantage of the Weibull model, in addition to the stochastic consideration of runs

is the ability to include the data from a runaway sample in the analysis when it is reused. This means that a sample that has not failed is further investigated after classification as a runaway with a higher stress range.

The Weibull model provides a general representation of the Wöhler curve and an estimate of the confidence intervals of stress ranges in the HCF range.