TWI Industrial Member Report Summary 196/1982
By S J Maddox
Background
Recently introduced fatigue design rules for steel welded structures relate specifically to aswelded joints, containing high tensile residual stresses. The same rules are known to be applicable to stress-relieved joints which are subjected to fully tensile fatigue loading but they are likely to be over-conservative for stress-relieved joints under partly compressive loading. In practice, full or even partial relief of residual stresses may not be achieved, depending on the size and geometry of the welded structure, but for circumstances in which stress relief can be assumed to have been effective there may be scope for relaxing the design rules. The aim of the present investigation was to study the fatigue behaviour of stress-relieved fillet welded joints with a view to formulating design rules. Fatigue tests were carried out on structural steel specimens consisting of 12.5mm thick plates with fillet welded longitudinal attachments on each surface under R = 0, -1, -2 and -4. The specimens were thermally stress-relieved but subsequent measurements showed that tensile residual stresses up to around 60N/mm2 could still be present. Even so, the fatigue strength of the joint increased as R became more negative, such that the S-N curve was shallower than that for R = 0, intercepting it in the high stress region.
Unsuccessful attempts were made to correlate the results in terms of effective stresses, including one derived from fatigue crack propagation results. However, the increased fatigue limits werc reasonably compatible with the values deduced by considering the effect of R, taking into account the presence of residual stresses, on the threshold stress intensity factor of a crack at the weld toe. By considering the effect of R on the fatigue limit and assuming that all the S-N curves for stress-relieved joints should converge in the low-cycle regime, design S-N curves for stress-relieved Class F fillet welds were proposed.