Design for fatigue resistance has been judged as critical to the integrity of welded connections of offshore structures. While considerable understanding of this failure process and of how it can best be mitigated has been realized in recent years, much of this information has been based upon data and analyses from constant amplitude or deflection experiments. In actual service, however, the stressing is a random non-stationary process; and questions have been raised regarding distinctions, if any, between constant and variable amplitude (or deflection) results and applicability of linear damage summation techniques. The present paper describes an ongoing research project investigating variable deflection fatigue of welded structural steel connections in sea water. Two spectra, one of narrow band and the other wide band, and three levels of cathodic protection are employed. Results to date are presented along with a perspective of the rationale being employed to evaluate the results, first, with regard to our present understanding of variable amplitude (deflection) fatigue and, second, as these pertain to design of welded connections for marine service.
Fatigue has been judged to be of major importance with regard to integrity of offshore structures (1-3). This failure mode may be particularly critical at welded tubular connections, where both macro- and micro-geometrical irregularities contribute to a concentration of stress. Figure 1 illustrates this schematically for the case of a "Y" connection. Present design procedure for fixed offshore structures of the jacket type typically employs the following sequential steps (4) :
Definition of the long-term wave climate based upon the best available information.
Global analysis of the structure to determine its response to (1). In the case of deep water structures (period greater than three seconds) dynamic amplification effects are included.
Evaluation of local stresses at tubular connections in terms of sea state.
Assemblage of stresses to develop a long-term distribution.
Calculation of the cumulative damage, D, based upon (4) and the appropriate design S-N curve, according to the expression(Formula is available in full paper) where ni is the number of stress cycles corresponding to a discrete sea state and Ni is the number of cycles for the corresponding stress range of the design curve .Typically, D should not exceed unity; and the design fatigue life of each joint is often required to be twice the intended service life of the structure.
An important assumption upon which accuracy of this design technique depends is that the S-N curve adequately reflect the fatigue response of a particular connection in question. Past research activities sponsored by the American Petroleum Institute have produced fatigue data (5-7) which have enhanced confidence in the high cycle regime of the S-N curve, and these are referenced in the design commentary (4). Historically, these curves have been based upon data from constant amplitude or deflection fatigue experiments; however, it is not known how accurately these represent behavior of actual welded connections which experience random stressing in sea water from wind, wave, tidal, duty loadings and, perhaps, dynamic amplification. Similarly, this design procedure can be no more correct than the linear damage summation technique which is generally employed in the analysis