Answers are grounded in this paper and cite their sources.
To design offshore wind turbine foundations against fatigue failure, engineers typically rely on the S–N curves provided by international standards and guidelines, such as DNV-RP-C203. These design curves are generally derived using Least Squares Linear Regression applied to large volumes of experimental data from laboratory-scale fatigue tests. However, traditional approaches rarely account for three important factors, which include: (i) the presence of suspended tests (runouts), (ii) unknown misalignment of welded plates, and (iii) heteroskedasticity in fatigue behaviour, whereby the standard deviation of fatigue life increases at lower stress ranges. Despite often being overlooked, these factors can cause significant shifts in the S–N curves within the range of interest for offshore applications (with number of cycles of typically greater than 10 6 ) potentially leading to differences of several years in predicted design life. Bayesian regression offers a powerful alternative. Combined with Hamiltonian Markov Chain Monte Carlo, it enables the calculation of the posterior probability distributions of model parameters, such as the slope and intercept of the S–N curve, through Bayes’ theorem. This approach allows for a more rigorous incorporation of runouts, misalignment, and heteroskedasticity into the statistical analysis. In this study, a new methodology is proposed to analyse S–N fatigue data using Bayesian regression, and results are presented for S355 double-V groove transverse butt welds, a common detail in wind turbine foundation construction. Three models are developed, assuming different distributions for the number of cycles to failure: (i) normal, (ii) Weibull, and (iii) skew-normal. The results are compared and the key differences with conventional standard-based statistical approaches are highlighted. The main findings demonstrate that Bayesian regression effectively captures the heteroskedastic nature of S–N fatigue data and provides a framework to account for misalignment and runouts, ultimately offering a more accurate and reliable tool for fatigue design of offshore wind turbine foundations.
No takes yet. Share an insight, caveat, or question.
Santa et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: