Offshore Wind Turbines in the Natural Environment: Statistical Prediction of Power and Fatigue
Abstract
The purpose of this research is to understand the statistical characters of natural wind and further apply this statistical model to improve the power forecast and blade fatigue estimation for FOWTs.
A new methodology is developed for recalibration of cup anemometers and demonstrated by computation of higher order statistics of coastal sea breezes. The development is based on representing the dynamic response of a cup rotor by an equation of motion (EOM) relating rotational motion of the cup rotor to aerodynamic forcing. A numerical method is introduced to determine the dynamic coefficients of the EOM from wind speed measurements. Random process theory is applied to the EOM to develop a recalibration for the mean wind speed without need to recalibrate the entire time history.
A detailed example is presented in which coefficients of the EOM are quantified from laboratory and field data. The overall methodology is then applied to recover the time history and mean of the true wind speed from field-measured cup data. A practical application of the method is demonstrated by recalibrating a larger set of cup data measured during a 2-month field campaign on the Texas coast to assess the higher statistical moments of the wind process. Measured coastal sea breezes in this area are found to be non-Gaussian.
With the deeper understanding of wind process, a new methodology is derived to transform between an ideal zero-turbulence power curve and practical power curves representing wind turbine performance in irregular winds. The derivation is based on substituting a theoretical distribution of the wind process in place of a single mean wind speed in the power computation, and then applying random process theory to derive analytical expressions for the expected power and standard deviation of power. The resulting expressions explicitly include the effects of varying turbulence intensity and higher statistical moments, and enable the performance of an operating wind turbine to be parameterized using a limited number of coefficients. These coefficients can be estimated from limited time-domain simulations or from measured field data, and then applied to predict wind turbine performance in different wind conditions. The accuracy of the new method is demonstrated by benchmarking expected power and standard deviation of power against direct simulation in irregular winds, including winds with various turbulence intensities and non-Gaussian statistics.
The non-Gaussian wind model is used to estimate of the fatigue damage of a composite wind turbine blades and the impact of non-Gaussian winds on fatigue life is also discussed. The Weibull distribution, widely acknowledged to fit the long-term wind speeds, is applied on local buoy data to calculate the probability density function of local winds. Numerical simulations based on OC3-Hywind model with local winds PDF are used to identify the fatigue hot-spot on the blades. Blade fatigue life based on hot-spot are calculated using Gaussian random wind conditions which are simulated through the entire operation wind speed range. Corresponding non-Gaussian random wind conditions using field measured data are simulated as well. The impact of non-Gaussian wind on blade fatigue is analyzed based on the comparison of these results.
Citation
Dai, Shu (2020). Offshore Wind Turbines in the Natural Environment: Statistical Prediction of Power and Fatigue. Doctoral dissertation, Texas A&M University. Available electronically from https : / /hdl .handle .net /1969 .1 /192718.