High cycle spectrum fatigue of welded components: Dissertation

High cycle spectrum fatigue of welded components has been investigated. The experimental program included fatigue tests of longitudinal non-load carrying fillet weld specimens manufactured from carbon and stainless steels, fatigue of large welded box beam components, and static hot spot strain measurements. The loading spectra, weld quality, and geometry of the box beam components were typical of that found in the rail-car industry. Based on the experimental results and a review of published literature, the questions of small cycle fatigue damage and small cycle cut-off limit below which cycles need not to be considered were addressed. Three load spectra were defined from a single peak-valley sequence by imposing different omission levels for small cycles and the damage contribution of these cycles could be estimated. For long- life tests, a straight line extension of the S-N curve below the fatigue limit using a slope of 3.0 best fits the data for tests as long as 2 x 107 cycles to failure. For the longest tests, the majority of fatigue damage was contributed by cycles with stress ranges less than the constant amplitude fatigue limit. In this study, cycles in the range 50% to 100% of *So caused significant damage in welds from both carbon and stainless steel and cycles as small as 33% to 50% of *So caused significant damage in the carbon steel welds. Hot spot measurements were made for the simple weld specimen geometry and for seven locations on the large box beam component. A single hot spot design curve for butt welds was found to be a conservative estimate for the longitudinal fillet weld fatigue data when **hs rmc was plotted against Nf. This curve also provided a conservative estimate for most of the failures observed in the box beam components when the failure criterion was defined as a 10 mm long crack. The fillet weld design curve was a conservative estimate for all the data.