Fatigue assessment of welded joints taking into account effects of residual stress
MetadataShow full item record
- Institutt for marin teknikk 
The aim of the study is to contribute towards a better understanding of the effect of residual stress and shakedown on fatigue strength of welded joints. Residual stress may have a significant effect on the fatigue strength of welded joints. As a nonfluctuating stress, it has an effect similar to that of the mean stress. Recently the International Association of Ship Classification Societies (IACS) has issued Common Structural Rules (CSR) for respectively tankers and bulk carriers. The effect of mean stress in fatigue design is taken into account in both sets of rules. However, the treatment is quite different, in particular with regard to residual stress and shakedown effects. A comparative study of fatigue design procedures of several rules/standards is reported, with emphasis on residual stress effects. This work involves of several tasks, firstly the experimental study to investigate behaviour of welding residual stress. Measurements of welding residual stress on totally 26 of welded joint specimens have been performed using the sectioning method. Various load cases comprising of static preloads and cyclic loads have been applied to simulate loading history in marine structures. The initial magnitude of residual stress was determined in as welded condition. The static preloads and cyclic loads had effects in reducing or relaxing residual stress. The magnitude of residual stress relaxation was determined by the highest tensile stress in the load history. The compressive preloads had only a minor effect on residual stress relaxation. The cyclic loads had effect on residual stress relaxation only in the first cycle. The empirical formulae showing relationship between residual stress magnitudes after relaxation and applied loads had been established that has aim to be incorporated in fatigue assessment procedure. Hence, a fatigue assessment procedure which considers residual stress in more realistic manner can be established. Secondly, stress categories have to be determined in fatigue analysis. Structural discontinuities lead in general to stress concentrations that significantly affect fatigue strength. The effect of stress concentrations may be accounted by the hot-spot stress approach using linear extrapolation of surface stresses. A longitudinal attachment weld is studied experimentally and by finite element analyses (FEA). The detailed stress measurements are used to evaluate various modelling techniques in FEA. The solid element and the shell element models are studied. The different models for inclusion of weld stiffness are studied, with solid elements and shell elements. Based on the results, recommendations are given for choice of element model and mesh size. Lastly, fatigue tests have been carried out using totally 59 of the small scale specimens of welded joints applying stress ratio, R=0 and R=-1, as well as various pre-load conditions. S-N curves have been plotted with different fatigue assessment procedure, namely IACS CSR bulk-carrier (2006a), IACS CSR tanker (2006b), DNV CN30.7 (2008), and IIW. Zhang-Moan proposed procedure based on Yamamoto & Matsuoka’s work (2001) was also included. It has been found that by applying the different procedures, scatter on S-N data were somewhat higher than the standard design curves indicating that residual stress is a hidden parameter causing increased scatter. Hence, the new improvement procedures based on modification of the IACS CSR bulk-carrier procedure have been proposed. The empirical formulae of residual stress relaxation established from previous study had been explicitly incorporated in the mean stress correction factor led to reducing scatter of S-N data.