How does the selection of wave hindcast datasets and statistical models influence the probabilistic design of offshore scour protections?

Abstract

Offshore wind structures often use bottom-fixed foundations that are susceptible to scour phenomena. The design of scour protections for such structures requires defining the physical actions that they must be able to withstand during their lifetime. In this context, hindcasts providing estimates of necessary metocean variables play a key role. However, it is unclear how the selection of hindcast datasets can influence the design of scour protections at a given location. Moreover, defining design actions requires estimating extreme sea states with return periods exceeding the length of hindcast time series, which introduces additional uncertainty in the modelling process, as different statistical models may perform differently when applied to different hindcast datasets. A better un-derstanding of the effects of these modelling choices is therefore necessary for the design and optimization of scour protections. The present study compares different wave hindcasts for the North Sea, and analyses how their selection and the subsequent application of statistical models affect the estimated probabilities of failure of the armour layer for bottom-fixed monopile foundations. Results show that the selection of different hindcast datasets and statistical models can significantly influence the probabilistic-based design of this type of scour protection, based on which practical recommendations are provided.