Pilot study on the constant and variable amplitude behavior of transverse stiffener welds: Fid-Bau Portal

Pilot study on the constant and variable amplitude behavior of transverse stiffener welds

Klippstein, Karl H. / Schilling, Charles G.

AbstractAs part of a pilot study on the long-life variable-amplitude behavior of transverse stiffeners, thirty-nine fatigue specimens were tested as follows: seventeen specimens were tested to establish the constant-amplitude fatigue limit, twelve specimens were used to determine the sloped portion of the constant-amplitude S-N line, and ten tests were performed to study the fatigue behavior under a variable-amplitude random-sequence load-stress spectrum with a truncated Rayleigh distribution. The remaining variable-amplitude random sequence originally planned, which included additional tests with a truncated Rayleigh and the trimodal spectrum based on the combined 1974 truck traffic, were not tested because the project was discontinued for economic reasons. However, because the findings for the completed part of the program might be useful for future investigations, the procedures followed and the results obtained are reported in this paper.Special care was taken during the fabrication of the specimens to reduce geometrical and residual-stress variations. This allows a more meaningful evaluation of the effect of the constant-amplitude fatigue limit on the fatigue life of a specimen exposed to a variable-amplitude random-sequence load-stress spectrum.The constant-amplitude fatigue limit determined by the modified staircase method was 18·2 ksi. This is significantly higher than the lower-bound value of 12 ksi used in the AASHTO fatigue provisions and as sometimes assumed in the literature. The slope of the constant-amplitude tests was close to −4, which is also different from the generally assumed slope of −3.The limited number of variable-amplitude fatigue-life data obtained, which includes tests up to 15 million cycles, are compared to (1) a lower-bound curve obtained by ignoring the effects of the constant-amplitude fatigue limit, and to (2) an upper-bound curve obtained by assuming that the stress cycles below the constant-amplitude fatigue limit cause no fatigue damage.It is anticipated that this information may be useful for future investigations related to the behavior of welded-steel fabrication details exposed to variable-amplitude-random-sequence loads, as found predominantly in steel bridge applications.