The dominant majority of existing offshore wind turbines are supported using monopile foundations which are fabricated by welding thick steel plates. In the current fatigue design recommendations for welded steel structures, which are commonly used in the design of monopile structures, the initial inverse slope of the S-N curve is fixed to −3. The historic rationales for this assumption are the ease of fatigue life calculations, the effect of long-range residual stresses that were not originally captured in the test data on thin welded specimens, and analogy with fatigue crack growth. While this introduces an unquantified level of conservatism to deterministic fatigue calculations, it was an acceptable (and wise) assumption when the S-N design curves were generated a few decades ago; however, current designs and assessments require more accurate calculations and the need to more confidently quantify the likelihoods attached to these evaluations. In the present study, it is argued that unlike large braced structures (e.g. jackets) in which long-range residual stresses may remain in place as fatigue cracks propagate, in circumferential welds of monopile structures the natural relaxation of residual stresses, in the absence of long-range residual stresses, must be accounted for by performing fatigue tests on thick welded samples. This paper presents a thorough analysis of the fatigue test results obtained from 50 mm thick as-welded samples with representative profiles of residual stresses that exist in monopiles. Basic and advanced statistical approaches were employed in the analysis and higher values of inverse slope were found using the large thickness test data. The results from this study are compared with the recommended fatigue design curves available in international standards. The outcomes from this research draw important observations concerning the need to employ an appropriate value of inverse slope in the design of monopile structures by excluding the conservative approach imposed by a fixed value of −3 which was originally implemented for braced structures with the assumption of large and long-range residual stresses.
Read full abstract