Influence of weld geometry and residual stresses on the fatigue strength of longitudinal stiffeners

Abstract

Typically, for a fatigue strength assessment of a welded structure, the influence of welding residual stresses has to be considered. An easy applicable approach is given within the framework of the IIW recommendations; the design S-N curves can be upgraded in case of medium or low residual stresses. A relaxation or redistribution of tensile residual stresses, which is often accompanied by an increase in fatigue life, cannot be considered. To obtain a better understanding of the residual stresses and their influence on the fatigue strength, fatigue tests have been performed on longitudinal stiffeners in as-welded and stress relieved state with both stress ratios R = −1 and R = 0. Local strains and stresses were measured using strain gauges and X-ray diffraction technique to get insight into the local material’s response due to global loading close to the weld toe. The crack initiation and propagation phase was detected by a camera. FE models were set up to reproduce the observed material behavior in the notch for the crack initiation phase. By numerical analysis it could be shown that, due to sharp notches at the weld toe with an average toe radius of r = 0.05 mm, a stress redistribution occurs during the first load cycle. This leads to a fatigue strength, which is, particularly at higher load amplitudes, independent from mean as well as from initial residual stresses. A comparison of fatigue data derived on longitudinal stiffeners from literature confirms this effect. For lower load amplitudes instead, mean and residual stresses gain more influence. In this investigation a reliable assessment of the crack initiation period could be performed with strain-based concepts using damage parameter PSWT for a mean stress assessment. All fatigue tests could be transformed into a single damage parameter PSWT-N curve with a low scatter if residual stresses as well as the distortion of the specimens are regarded.