The research described in this report is intended to provide information on evaluating the fatigue resistance of welded attachments subjected to variable amplitude fatigue loading. The research consisted of laboratory studies of welded attachments under random variable amplitude load spectra defined by a Rayleigh-type distribution with most stress-cycles below the constant amplitude fatigue limit. (Some stress cycles exceeded this limit.) Eight full-size beams with web attachments and cover plates were tested during the program. Fatigue crack growth data were generated using random block variable amplitude stress spectra defined by a Rayleigh-type distribution. Also, nonload-carrying fillet-welded cruciform-type specimens were tested under simple bending using a random variable amplitude block loading to supplement the existing shorter life studies carried out on stiffener details. The results obtained from these variable amplitude tests are consistent with the previously reported constant amplitude test. However, the existence of a fatigue limit below which no fatigue cracks propagate is assured only if none of the stress range cycles exceed this constant amplitude fatigue limit. If any of the stress range cycles (as few as one per thousand cycles) exceed the limit, fatigue crack propagation will likely occur. The random variable test data from the beam specimens generally fell between the upper and lower confidence limits projected from constant cycle data. The smaller simulated details generally resulted in the random variable test data falling near the upper confidence limit of constant amplitude test results.
Abstract