A Comparison of Methods for Estimation of Fatigue and Extreme Mooring Response for a Floating Spar Wind Turbine

Abstract

Hywind Demo is one of the worlds first floating offshore wind turbines. It is operated by Statoil and located off the coast of Karmøy in Norway. The substructure is a spar shaped hull, and it is moored with catenary mooring lines. In this study, the motions of the floating turbine and the tension in the mooring lines are investigated in a collaboration with Statoil.

Traditionally the analyses of such structures are preformed using non-linear time domain techniques. To insure validity of the statistical parameters obtained from these analysis it is required to preform a sufficiently high number of time domain simulations. Such processes are time consuming and thus costly. In an early design phase where several designs are to be tested, the time and costs of analysis is critical. A reduction in simulation time could significantly reduce the effort in an early design phase.

In this thesis a suggestion of using a frequency domain analysis to preform the early stage design analysis is investigated. For this purpose the static and dynamic mooring analysis program, MIMOSA, were chosen. A model in frequency domain is established based on a SIMO-RIFLEX time domain model from Statoil. Various tests is preformed on the SIMO-RIFLEX model to extract properties for implementation in the frequency domain model. This results in extensive linearization issues which validity is to be investigated.

Both time domain analysis and frequency domain analysis have been preformed for 9 fatigue and 6 extreme environmental conditions, each lasting 3 hours. To obtain a statistically valid sample a set of 20 time series with varying wave and wind seed for each environmental condition were investigated in time domain. The environmental conditions are chosen based on weather data from the Hywind Demo location. A comparison of pitch and surge motions as well as line tension is investigated to verify the simplified frequency domain method. The static mean, standard deviation and extrema is extracted from each condition and compared.

The comparison of time domain against frequency domain reveals some significant differences. The static mean however approves to behave relatively well in the frequency domain method. Good agreements are also found in dynamic behaviour for weather conditions with moderate waves and a wind velocity below rated. The diversity in the dynamic behavior of the turbine increases as the mean wind speed approaches rated and above. The pitch controller activates at rated wind speed and decreases the thrust as the wind speed increase. This non-linearity is not possible to obtain in frequency domain analysis in MIMOSA and results in large diversities. The frequency domain model established in this thesis is because of this not suited for load estimation on a floating spar wind turbine equipped with an active pitch controller.