Influence of frequency-directional wave spectra on the wave induced fatigue loads on offshore wind turbines

Abstract

The current trend in the offshore renewable energy industry is to install ever larger offshore wind tur- bines (OWT) in ever deeper waters. This results in an increased capacity requirement of the substruc- ture which results in increased dimensions and thus higher hydrodynamic loads.
For the design of OWT the fatigue limit state is next to the ultimate limit state a crucial design loading condition to consider. In the current engineering practice, wave spectra are mostly reconstructed with the JONSWAP spectral parametrization which gives a simplified representation of the actual sea surface based on bulk wave parameters for wave components, such as wind-sea and swell. By using realistic frequency-directional (2D)-wave spectra instead of the JONSWAP spectral model, the wave spectral density distribution can be showed in its most authentic form.
This study investigates the influence of 2D-wave spectra on the wave-induced fatigue loads on OWT’s, as a comparison to the commonly used JONSWAP spectral parametrization. Therefore the relation of different wind and wave parameter as well as of the wave spectral characteristics on the wave-induced loads and the load differences between both spectra are investigated. Wave spectral data as well as bulk wave parameters for a wind-sea and swell wave component were obtained from the DHI MetOcean database for the Dutch North Sea. Based on those the fatigue loads were calculated in the frequency domain using Dirlik’s method.
Results show that the JONSWAP spectral parametrization leads to an overall overestimation in the range of ∼ 12 - 15 % of the total wave-induced fatigue loads compared to those obtained from 2D-wave spectra. The highest fatigue load differences and DEL overestimation between the JONSWAP and 2D- wave spectra were observed for significant wave heights between 2 and 3.8 m (14.70 % overestimation), peak periods between 8 and 15 s (13.63 % overestimation) and wind speeds between 10 and 20 m/s (15.04 % overestimation).
Furthermore, it was observed that narrow frequency and directional width, between 0.1 - 0.3 Hz and 20 and 30 deg, lead to the highest discrepancies for the calculated fatigue loads between the JONSWAP and 2D-wave spectra (∼ 13 % overestimation). For extreme sea states, that correspond to the upper 5 % of the observed data it appears that the JONSWAP spectral parametrization leads mostly to an underestimation of the calculated fatigue loads.
With the distance correlation calculation the dependencies between the wave characteristics, such as the frequency width, directional width and the peakedness parameter, and the fatigue load differences was quantified. But this analysis did not result in additional research insights since the obtained corre- lation coefficients were almost identical and only exhibiting weak dependencies.
Still, it can be concluded that using 2D-wave spectra in the engineering practice and for the fatigue load calculation on OWT’s with monopiles as substructures, physically more representative loads can be obtained compared to those computed with the simplified JONSWAP spectral model. This more realistic load estimation could finally lead to an improved design of OWT’s.