MODELING THE FATIGUE DAMAGE EVOLUTION IN WELDED JOINTS

The present paper presents a two-phase model for the fatigue damage evolution in welded steel joints. The argument for choosing a two-phase model is that crack initiation and subsequent crack propagation involve different damage mechanisms and should be treated separately. The crack initiation phase is defined as the number of cycles to reach a crack depth of 0.1 mm. This phase is modelled based on the Dang Van multiaxial stress approach. Both a multiaxial stress situation introduced by the acting loads and the presence of the multiaxial welding residual stresses are accounted for. The local notch effect at the weld toe becomes very important and the irregular weld toe geometry is characterized by extreme value statistics for the weld toe angle and radius. The subsequent crack growth is based in classical fracture based on the Paris law including the effect of the Stress Intensity Factor Range (SIFR) threshold value. The unique fatigue crack growth rate curve suggested by Huang, Moan and Cui is adopted. This approach keeps the growth rate parameters C and m constant whereas an effective SIFR is calculated for the actual stress range and loading ratio. The model is developed and verified based on fatigue crack growth data from fillet welded joints where cracks are emanating from the weld toe. For this test series measured crack depths below 0.1 mm are available. The two-phase model was in addition calibrated to fit the life prediction in the rule based S-N curve designated category 71 (or class F). A supplementary S-N curve is obtained by the Random Fatigue Limit Method (RFLM). The test results and the fitted model demonstrated that the crack initiation phase in welded joins is significant and cannot be ignored. The results obtained by the Dang Van approach for the initiation phase are promising but the modelling is not yet completed. The fracture mechanics model for the propagation phase gives good agreement with measured crack growth. However, it seems that the prediction of crack retardation based on a threshold value for the SIFR gives a fatigue limit that is overly optimistic for small cracks at the weld toe. The threshold value has been determined based on tests with rather large central cracks in plates. The validity for applying this threshold value for small cracks at the weld toe is questioned. As the present two-phase model is based on applied mechanics for both phases the parameters that have an influence on the fatigue damage evolution are directly entering into the model. Any change in these parameters can then be explicitly taken into account in logical and rational manner for fatigue life predictions. This not the case with the rule based S-N curves that are based on pure statistical treatment of the bulk fatigue life.